ZRIMA
ZRIMA
ZRIMA
B
, and then for another point, and so on.
A fluid in a region which contains only points for which convergence occurs,
i.e., in which the density is defined everywhere, is denoted a continuum with
respect to density. A thermodynamic system defined inside such a region need not
be in equilibrium, yet the density at each point in it is well defined. A fluid can also
satisfy the requirement of being a continuum with respect to other fluid properties,
e.g., a continuum with respect to temperature. A fluid which is a continuum with
respect to all relevant properties, i.e., all properties of interest in the considered
problem, shall be denoted a continuum with no further qualifications.
The requirements to be a continuum may be satisfied in one region but not in
another or at one moment and not at another. Thus, a fluid may be considered a
continuum in parts of the region or part of the time only. The flow of a rarefied gas
is a typical example of such a case. When the continuum model breaks down, some
other models may become appropriate, e.g., those used in statistical mechanics or
statistical thermodynamics. These are outside the scope of the book.
It is noted that the continuum model may have to be abandoned in cases
where very fine details of the flow are required, i.e., in examples where regions
smaller than V
c
are of interest.
Local Properties in a Continuum
The mental experiment, which has led to the concepts of density and
continuum with respect to density, can be adjusted to yield some other properties,
such as temperature, thermodynamic pressure, etc. The decision of what must be
S
Body of fluid
A
Figure 1.1 Point A in a fluid.
1. Introduction 3
measured in these mental experiments is not always simple and is not necessarily
unique. The emphasis, however, is on the realization that such an experiment can
be performed and that the concept of, say, temperature can be defined as a result of
a limiting process and with it the continuum with respect to temperature. The
same, of course, holds for other measurable properties.
Properties, which cannot be measured directly, but rather are calculated using
other properties, are assumed to be here defined only after the measurable proper-
ties have been defined. With this understanding a continuum is now considered to
have all its relevant properties defined everywhere. A somewhat picturesque
description of what has been achieved up to this point is that of the fluid being
divided into many small subregions, each of the order of V
c
in size and each having
a list of its properties. This description is indeed correct. What follows is an
elaboration, which does not increase accuracy but is more amenable to analysis.
Hence, by changing somewhat the considered model now, fewer difficulties arise
later in the analysis.
Denote the considered property, say, the density, by p. This p is known in any
subregion of the continuum. Let a continuous function be defined, such that for
some reasonable subdivision of the region its values coincide with the values of p
in the subregions. The definition of such a function does not add information but
rather smoothes out the description. Let this continuous function also have contin-
uous partial derivatives up to some order to be specified as needed. At a point
where p is specified, this function coincides with its value. Between two adjacent
points of specification in that particular division, the function interpolates between
the two values and, therefore, does not change the approximation. By considering
this function as p, instead of the previous discrete p values, no error is introduced.
The continuous p function has, however, the advantage of having derivatives and
being integrable, which proves invaluable in what follows.
A local property in a continuum is henceforth considered as the local value of
this continuous function. What has been done so far is the establishment of a basis
for translating physical reality into a form, which is amenable to mathematical
treatment. The part lost in this translation process is the molecular structure of
matter. Hence, we consider regions of matter as continua of properties that change
in a fairly smooth way. We may talk about density and temperature fields or
velocity profiles. In describing the density of a continuum we disregard the fact
that matter is composed of very dense protons and neutrons surrounded by regions
of zero density. Similarly, while describing a velocity profile of a gas flowing in a
pipe, we disregard the velocity of any given molecule and treat the velocity profile
of the continuum as a continuous, well-behaved mathematical function.
4 Fluid Mechanics
Body and Surface Forces
Forces acting on a fluid divide naturally into body forces and surface forces.
Body forces act at a distance and need not be associated with any transmission
agent, or point of contact. In many cases, they are the "effects" of external fields.
Some typical examples of body forces are those of gravity, electrostatics and
electromagnetism.
Surface forces are viewed as acting on any surface defined in the fluid,
including its boundaries. Because each point in the fluid may belong to several
surfaces, several such surface forces seem to coexist at the point. This is more
complex than a single body force acting on a mass element, and therefore surface
forces deserve some more elaborate consideration.
Stress at a Point
Let a surface passing through point A in a fluid be defined. Let a plane
tangent to the surface pass also through A and a small disk be drawn on this plane,
as shown in Fig. 1.2. This disk has two sides, one of which is assigned to the outer
normal n. Let the area of the disk be
S and the force acting on its n sur-
face be F. The average stress acting
on the surface is now defined as the
force per unit surface area
T = F / S .
(1.3)
We now shrink S around A and look
for a limit of the average stress. This
limit is called the stress at point A.
The considerations here are some-
what similar to those leading to the
definition of the concept of the local
properties. The limit in Eq. (1.3) is a vector, which depends not only on the
location of point A but also on the direction of the unit normal n.
Henceforth, the term stress is used as a short form for stress at a point. The
stress is a vector, which depends on another vector n. The component of the stress
in the direction of this normal n is called normal stress and the component
perpendicular to n, i.e., which lies in the plane of the disk and is, therefore, tangent
to the considered surface, is called tangential stress or shear stress. Intuitively, the
normal stress may be considered as that tending to pull the disk away from the
Figure 1.2 A small disk tangent to a surface
in the fluid at point A.
1. Introduction 5
surface while the shear stress is the one that tends to shear the disk off the surface
while sliding on it tangentially.
Example 1.1
A body of water inside a container, as shown in Fig. 1.3, is moving to the
right with a constant velocity of
U = 6 m/s . The temperature in the water changes
from
T = 20
o
C at the top to
T = 10
o
C at the bottom of the container.
Figure 1.3 A body of water in a moving container.
Is the water a continuum with respect to density? With respect to tempera-
ture? What are the forces acting on the water? What are the stresses in the water?
Solution
The water is homogeneous and it is a continuum with respect to density. Its
density, , is constant, hence well defined at every point. Although the water
temperature does change from top to bottom, still, we do not expect it to be
discontinuous; hence, the water is a continuum with respect to temperature too.
The pressure in the water changes continuously from
p
o
at the top surface,
increasing as we go down. The water is, therefore, a continuum with respect to
pressure. In Chapter 3, we will show that the hydrostatic pressure at any point in
the body of water is given as
p = p
o
+ g(2 z) .
The forces acting on the water are body forces and surface forces. The body
force, G, is due to gravity, g; it is proportional to the mass of the water, and its
direction is that of the vector g, i.e., in the ( k)-direction,
d G = gdm = kgdV
6 Fluid Mechanics
The surface force S acting on any plane in the water is due to the normal
stress,
S
n
, as explained in the next chapter. There are no shear stresses in this body
of fluid.
Definition of a Fluid
We are now in a position to introduce the definition of a fluid. A fluid is
defined as a continuum, which cannot support a shear stress while at rest. The
terms continuum and shear stress that appear here make it obvious why this
definition had to wait until now.
Figure 1.4 Material subjected to shear stress.
Consider an experiment designed to use this definition to decide whether a
given material is a fluid. Let the material be placed between two parallel plates and
be subjected to an external shearing stress. An example of such an experiment is
shown in Fig. 1.4, where the shear force is applied by means of a weight. A fluid
should not be
able to support any shear stress while at rest. The material will be
considered a fluid if
the top plate moves to the right as long as a shear stress is
applied, no matter how small.
Example 1.2
A cylindrical rod is left on the surface of still water in a tank. It sinks to the
bottom, Fig. 1.5. Neglecting forces at the bottom of the rod, which is thin and long,
it is the shear stresses on its circumference that resist the sinking.
The experiment is repeated with tar instead of water. The same results are
obtained, but the sinking takes many hours. Is water a fluid? Is tar?
1. Introduction 7
Solution
The water is a fluid. It cannot support
shear stresses while at rest. The decision con-
cerning the tar depends on what the observer
considers "at rest."
If the whole situation is of interest only for a
few seconds, the tar is approximately at rest;
it does support shear stresses, and a model of
a solid fits better its behavior. However, for
longer periods, e.g., to build a house on it,
fluid mechanics considerations must be used
and the tar must be considered a fluid. (Can a
house be put safely on tar?)
Units and Dimensions
Fluid mechanics, as many other branches of science and engineering, cannot
be completely formulated by the use of pure numbers. Therefore, dimensions and
basic units must be employed. The dimensions are considered modifiers for the
pure numbers; the pure numbers describe quantities, i.e., "how much," while the
dimensions tell us "of what." With this in mind unit vectors are also identified as
modifiers, which tell us "in what direction." Thus, a vector is given by its three
components, which are pure numbers, modified by the system's three unit vectors,
and with some possible additional modifiers
_
the physical dimensions of the
vector, i.e., what it represents.
Transformations between different systems of units as well as between
different systems of coordinates are possible. The point of view underlying these
transformations is that all observers must see the same physical phenomena. The
observers may choose their own means (units, coordinate systems) of reporting
what they see, but they may not tamper with the phenomenon itself. Thus, if one
"wouldn't touch it with a ten-foot pole" he or she would not touch it with a
3.048-meter pole either. Because no matter what units are used to report the length
of the pole, it is still the same pole. Similarly, a vector viewed in different coordi-
nate systems remains the same, no matter what systems the observer happens to
favor.
We now return to the 10-foot pole and find what was involved in turning it
into a 3.048-meter stick. The original equation for the length l of the pole was
Figure 1.5 Cylindrical rod before
and after sinking.
8 Fluid Mechanics
l = 10 ft. (1.4)
We also know that
1 ft = 0.3048 m. (1.5)
Treating the dimension "ft" as an algebraic quantity, we now rewrite Eq. (1.5) as
l = 0.3048 m / ft. (1.6)
As seen from the left-hand side of Eq. (1.6), the value of 0.3048 m/ft is unity (a
pure number), and multiplication by 1 is always permissible. Performing this
multiplication on Eq. (1.4) results in the length of the pole in meters:
l = 10 ft 0.3048 m/ft = 3.048 m, (1.7)
where again physical dimensions are treated as algebraic quantities.
Conversion factors similar to that of Eq. (1.6) are presented in Appendix A.
The following example illustrates the use of these conversion factors.
Example 1.3
A small economy car runs 12 km per liter of gasoline. What is the car mileage
in MPG (miles per gallon)?
Solution
Using the conversion factors in Appendix A,
12
km
L
3.7854L/ gal
1.609 km/mi
= 28.23
mi
gal
.
The decision whether to multiply or divide by "one," i.e., by the conversion factor,
is made such that the desired units are obtained after cancellation.
In principle the description of a vector in various coordinate systems, i.e., the
conversion from one coordinate system to another, is analogous to the conversion
of units. However, because vector algebra does not have the operation of division,
one is restricted to multiplication only, and more details of this operation are
necessary.
1. Introduction 9
Dimensional Homogeneity
An equation is said to be dimensionally homogeneous if it does not depend on
the system of units used. A dimensionally homogeneous equation has, therefore,
the same dimensions for each term on both sides of the equation. In practice, this
means that the numerical constants appearing in the equation are dimensionless.
Thus, the equation
h =
1
2
gt
2
(1.8)
is dimensionally homogeneous and the constant 1/2 is dimensionless. This homo-
geneous equation holds for any system of units.
A common practice in engineering is to write equations in a nonhomogeneous
form. If, for example, we substitute in Eq. (1.8) the value g = 32.174 ft/s
2
, we
obtain
h = 16.09t
2
.
(1.9)
This equation, although easier to use for repeated calculations, is nonhomogeneous
and holds only for the case where t is given in seconds and h in feet. Here the
constant 16.09 has dimensions of ft/s
2
. In order to convert Eq. (1.9) to a different
set of units, a new constant, c, has to be recalculated using the method outlined
above. Thus, in SI units (i.e., International System of Units) this constant will be
c =16.09 ft/s
2
0.3048 m/ft = 4.90 m/s
2
, (1.10)
resulting in a new dimensionally nonhomogeneous equation
h = 4.90t
2
.
(1.11)
As demonstrated by this example, nonhomogeneous equations arise for prac-
tical reasons, such as to shorten repeated calculations or use more easily available
data. It is rarely used in scientific presentations and, when used, must always be
followed by a clear statement as to what units are employed.
Example 1.4
Atmospheric pressure at sea level is p
a
= 14.7 psia, where the term "psia"
stands for pounds force per square inch absolute. What is the pressure in N/m
2
, i.e.,
in Pa (pascals)?
10 Fluid Mechanics
Solution
Using the conversion factors in Appendix A, the pressure is
2 2 2 2 2 2
2
14.7 lbf / in 14.7 lbf / in 144 in / ft 10.764 ft / m 4.448 N/ lbf
101,300 N/ m 101,300 Pa.
a
p = =
= =
Fluid Properties
In this section, some of the basic fluid properties used in this book are de-
scribed.
Density, Specific Weight and Specific Gravity
The property density was used in the discussion of the concept of a contin-
uum. It has been defined as
= lim
V V
c
m
V
.
(1.12)
The units of density are mass per unit volume. Hence, in SI units density is
measured in kilograms per cubic meter (kg/m
3
). Other units frequently used for
density are g/cm
3
, lbm/ft
3
and slug /ft
3
.
The density of water at room temperature and atmospheric pressure is
= 1,000 kg/m
3
= 1 g/cm
3
= 62.4 lbm/ft
3
= 1.94 slug/ft
3
. (1.13)
The density of a fluid may change from point to point. However, if the fluid
is of a single phase, the change in density will be almost everywhere continuous.
The density may depend on both the pressure and the temperature. A fluid whose
density may be assumed not to vary with pressure is called incompressible. Most
liquids may be considered incompressible.
The reciprocal of the density is the specific volume, v:
v =
1
(1.14)
A quantity related to the density is the specific weight, , defined as the
weight per unit volume. Hence,
1. Introduction 11
= g. (1.15)
The SI units of specific weight are N/m
3
. The specific weight of water at
room temperature and atmospheric pressure is
= 9806 N/m
3
= 62.4 lbf/ft
3
.
(1.16)
Occasionally one may encounter a term "specific gravity" which is defined as
the ratio of the density of a given fluid to that of water. Specific gravity is
therefore dimensionless, and its numerical value is equal to that of the density
expressed in CGS (centimeter-gram-second) units, i.e., in g/cm
3
.
Viscosity
While deciding whether a given material is a fluid, we used an experiment in
which the material was placed between two parallel plates, Fig. 1.4. One of the
plates was held stationary while the other could move pulled by a force acting in a
direction parallel to the plate. We have stated that if the plate moves, the material
is considered a fluid. Now, the speed at which the plate moves depends on a prop-
erty of the fluid called viscosity. The more viscous the fluid is, the slower the plate
moves.
The force per unit plate area, i.e., the shear stress, T, exerted on the plate is
proportional to the plate velocity V and inversely proportional to the gap between
the plates, h:
T =
F
A
V
h
.
(1.17)
The proportionality constant that converts Eq. (1.17) into an equality is called
dynamic viscosity and is denoted by . Thus, for that particular experiment
T =
V
h
.
(1.18)
Equation (1.18) can be generalized for two adjacent fluid layers separated by
a distance dy, both moving in the x-direction. If the difference in velocities be-
tween the layers is du, then Eq. (1.18) becomes
T =
du
dy
|
\
|
.
|
|
.
(1.19)
Equation (1.19) is sometimes referred to as Newton's law of viscosity. The
term du/dy is called the rate of strain, or simply shear rate. The relationships
between shear stress and the rate of strain are considered in detail in Chapter 5.
The dimensions of viscosity must be chosen such that Eqs. (1.19) and (1.18)
are rendered homogeneous. Hence, as the shear stress, T, is given in N/m
2
and the
12 Fluid Mechanics
shear rate in s
-1
, the unit of viscosity becomes
1
| |
= 1
N s
m
2
= 1
kg
ms
. (1.20)
In the British system, the unit of viscosity is lbm/ft
.
s, while in the CGS
system it is g/cm
.
s and is called the poise. Thus,
1poise= 1
g
cms
= 0.1
kg
m s
.
(1.21)
The viscosity of water at room temperature is roughly 0.01 poise, or 1 centipoise,
which is written as 1 cp.
The viscosity of gases increases with temperature. This is due to the more
vigorous molecular movement at higher temperatures. In liquids, the viscosity
depends mainly on the intermolecular cohesive forces. The forces between the
molecules of a liquid decrease with an increase in temperature; so does the
viscosity.
The ratio of the dynamic viscosity, , of a fluid to its density, , is called
kinematic viscosity and is denoted by v,
v =
.
(1.22)
The unit of kinematic viscosity is cm
2
/s and is called the stoke. It is related to
other units of kinematic viscosity as
1 m
2
/s = 10
4
st = 10.764 ft
2
/s. (1.23)
Compressibility and Bulk Modulus of Elasticity
Fluid compressibility, k, is defined as a relative change in fluid volume under
the action of an external pressure, i.e.,
k =
1
V
o
AV
Ap
.
(1.24)
A more careful definition specifies under what conditions the compression occurs.
*
Hence, isothermal compressibility is defined as
k
T
=
1
V
cV
cp
|
\
|
.
|
|
T
.
(1.25)
while adiabatic compressibility performed reversibly (i.e., isentropically) is
*
See, for example, A. Shavit and C. Gutfinger, "Thermodynamics From Concepts to Applications", 2
nd
Edition, CRC Press, 2009, p. 384.
1. Introduction 13
k
s
=
1
V
cV
cp
|
\
|
.
|
|
s
.
(1.26)
Equations (1.24) and (1.26) may be rewritten in terms of densities as
k
T
=
1
c
cp
|
\
|
.
|
T
(1.27)
and
k
s
=
1
c
cp
|
\
|
.
|
|
s
.
(1.28)
The units of compressibility are reciprocal of those of pressure, e.g., atm
-1
,
psi
-1
, m
2
/N, etc. For example, the compressibility of water at room temperature and
atmospheric pressure is 5 10
-5
atm
-1
. Hence, a pressure increase of 1 atm results in
a relative volume reduction of 1/20,000.
The compressibility of a gas may be found from the perfect gas relationship
p = RT, (1.29)
which upon substitution into Eq. (1.27) yields, for isothermal compressibility,
k
T
=
1
p
.
(1.30)
Adiabatic compressibility is found by using the isentropic relationship
p
-k
= const, (1.31)
where k = c
p
/c
v
. Equation (1.31) together with Eq. (1.28) result in the following
expression for adiabatic compressibility of a perfect gas:
k
s
=
1
pk
. (1.32)
The reciprocal of compressibility is known as the volumetric or bulk modulus of
elasticity. The isothermal bulk modulus of elasticity is defined as
E
T
=
1
k
T
=
cp
c
|
\
|
.
|
T
.
(1.33)
Similarly, the isentropic bulk modulus is
E
s
=
1
k
s
=
cp
c
|
\
|
.
|
s
.
(1.34)
The bulk modulus of elasticity is expressed in units of pressure. The concepts
14 Fluid Mechanics
of compressibility and bulk modulus of elasticity are used in the study of com-
pressible flow. In calculations of flow phenomena liquids are usually considered
incompressible. The assumption of incompressibility holds for gases at low speed
only. In many flows, at speeds of up to 30% of the speed of sound, a gas may be
treated as incompressible. The concept of compressibility is considered some more
in connection with the equation of continuity, in Chapter 5, and in Chapter 13 on
compressible flow.
Problems
1.1 In a certain city the amount of money people carry in their pockets is
between $1 and $1,000. The average amount per person depends on the zone
of the city. In trying to represent the city as a continuum with respect to
solvency, i.e., amount of money per person, find the smallest number of
people you have to include in V
c
such that the error caused by one person
going in or out of V
c
is less than 1%.
1.2 A cylindrical container is filled with water
of density = 1,000 kg/m
3
up to the height
h = 5 m, Fig. P1.2. The outside pressures
p
o
= 10
5
Pa.
a. Is the water a continuum with respect to
density? Is it with respect to pressure?
What is the height Az
c
of your V
c
if a
deviation of 0.1% in p is negligible?
b. Is the water a simple thermodynamic
system in equilibrium?
c. How are the answers modified for an
outside pressure of p
o
= 610
7
Pa?
d. The top layer of the water is held for a long time at 310 K, while the
bottom is held at 290 K. Is it a continuum with respect to temperature?
Explain.
1.3 A body of water in the shape of a cube is selected inside the container of
Fig. P1.2 such that the lower side of the cube coincides with z = 2.0 m. The
side of the cube is 0.01 m.
a. What is the body force acting on the fluid inside the cube?
b. What are the surface forces acting on the six sides?
Figure P1.2 Water in a tank
and small cube.
1. Introduction 15
What are the stresses?
c. Is the whole cube in mechanical equilibrium? Is it a stable equilibrium?
d. What are the forces and the stresses if the cylinder is put in space, i.e.,
for g = 0?
e. What are the forces if the cylinder falls freely? How would you keep the
water together?
1.4 A boxlike block of wood has the dimensions of 1 m 2 m 3 m. The density
of the wood is = 800 kg/m
3
. The coefficient of dry friction between wood
and concrete is 0.4, i.e., when the block is drawn on a concrete floor, it is
pulled with a force of 0.4 N per each 1 N force pushing the block normal to
the floor. Calculate and draw the average stress on that side of the block that
touches the floor when this side is:
a. The 1 m 2 m side.
b. The 2 m 3 m side.
Note that stress is a vector.
1.5 A certain oil has the viscosity of 2 poise. Its density is 62 lbm/ft
3
. What is its
kinematic viscosity in m
2
/s?
1.6 A certain slurry is filtered at constant pressure at the rate of
52.5 6.2
,
V
V
t
+
=
where V [liters/s]
is
the volumetric flowrate, t [s] is the time and V [liters] is
the filtered volume. Is the equation dimensionally homogeneous? Rewrite
the equation with V in cubic feet. Can you rewrite the equation in a homo-
geneous form?
1.7 A falling body has its z-coordinate change in time as
z = z
o
4. 9t
2
,
where t [s] is the time and z [m] is the height. Is the equation dimensionally
homogeneous? Rewrite the equation with z [ft].
a. Can you rewrite the equation in a homogeneous form. Why is the answer
here different from that in Problem 1.6?
b. Why do dimensionally homogeneous equations give more information?
What is this information?
1.8 The following dimensionless numbers are defined:
16 Fluid Mechanics
Re = dU /, Reynolds number;
Pr = c
p
/k, Prandtl number;
Pe = Re
.
Pr, Peclet number;
where U = 2 m/s is the flow velocity in the pipe, d = 2 in. is the pipe diame-
ter, = 1,000 kg/m
3
is the fluid density, = 3 cp is its viscosity, c
p
= 0.5
Btu/(lbm
.o
F) is the specific heat of the fluid and k = 0.65 W/m
.o
C is its
thermal conductivity.
What are the numerical values of Re, Pr and Pe? In what system of units?
1.9 Using Appendix A, change into the S.I. system:
density = 120 lbm/ft
3
,
thermal conductivity k = 170 Btu/(hr
.
ft
.o
F),
thermal convection coefficient h = 211 Btu/(hr
.
ft
2.
F),
specific heat c
p
= 175 Btu/(lbm
.o
F),
viscosity = 20 cp,
viscosity = 77 lbf
.
s/ft
2
,
kinematic viscosity v = 3 ft
2
/s,
StefanBoltzmann constant o = 0.171310
-8
Btu/(ft
2.
hr
.o
R
4
),
acceleration a = 12 ft/s
2
.
1.10 The following empirical equation gives the wall shear stress exerted on a
fluid flowing in a concrete pipe:
t
w
= 0.0021V
2
r
1
3
,
where t
w
is the shear stress in lbf/in.
2
, the fluid density in slug/ft
3
, V the
average velocity of the fluid in ft/s and r the hydraulic radius of the pipe in
feet. Rewrite the equation in terms of SI units.
1.11 The distance between the plates in an experimental system, as shown in Fig.
1.4, is h = 1 in. When the upper plate is pulled with the velocity
V = 40 ft/min, the shear stress is T = 12 lbf/ft
2
. Using Eq. (1.18) find the
viscosity of the fluid, , in SI units.
1.12 A metal sphere of 1 ft in diameter is put on a scale. The scale shows a read-
ing of 200 kg. Find the volume, the mass, the density, the specific volume
and the specific weight of the sphere in:
a. SI units.
b. British units.
1. Introduction 17
1.13 A volume of 30 liters of alcohol, subjected to a pressure of 500 atm at 25
o
C
contracts to 28.8 liters.
a. What is the modulus of elasticity of alcohol?
b. What is its compressibility?
18 Fluid Mechanics
19
2. STRESS IN A FLUID
In this chapter, we consider stresses in a fluid. We start by setting the forces
resulting from these stresses in their proper perspective, i.e., in relation to body forces,
together with which they raise accelerations. This results in a set of momentum
equations, which are needed later.
We then consider the relations between the various stress components and
proceed to inspect stress in fluids at rest and in moving fluids.
The Momentum Equations
In this section, we establish relations between body forces, stresses and their
corresponding surface forces and accelerations. Newton's second law of motion is
used, and the results are the general momentum equations for fluid flow.
Consider a system consisting of a small cube of fluid, as shown in Fig. 2.1. A
system is defined in classical thermodynamics as a given amount of matter with well-
defined boundaries. The system always contains the same matter and none may flow
through its boundaries.
As a rule, a fluid system does not retain its shape, unless, of course, the fluid is at
rest. This does not prevent the choice of a system with a certain particular shape, e.g.,
a cube. The choice means that imaginary surfaces are defined inside the fluid such that
at the considered moment they enclose a system of fluid with a given shape. A
moment later, the system may have a different shape, because the shape is not a
property of the fluid or of the location. Systems of different shapes may be chosen
simultaneously at the same point, and the choice is made for convenience. Thus, two
different shapes are chosen in this chapter, a cube and a tetrahedron, each leading
conveniently to some particular conclusions.
Let a set of cartesian coordinates be selected and a system of fluid in the form of
a small cube be defined, with its planes parallel to the coordinates directions as shown
in Fig. 2.1.
20 Fluid Mechanics
z
y
x
Ax
Ay
Az
n = + k
n = + i
n = - j
xx
xz
xy
T
T
T
T
x
( x, y, z )
(x+Ax, y+Ay, z+Az)
Figure 2.1 Small cube.
Now let all forces acting on this cube be considered. Using Newton's second law
we may write for our system
m a = G +S,
(2.1)
where G denotes the total body force acting on the cube and S is the surface force
resulting from the stresses acting on the six sides of the cube.
Equation (2.1) is a vectorial equation and may be rewritten in component form.
The x-component of Eq. (2.1) is
ma
x
= G
x
+ S
x
.
(2.2)
Each surface of the cube is named after the direction of its outer normal n. Thus,
there is a positive x-surface located at x + Ax in Fig. 2.1 and a negative x-surface at x.
These two surfaces are denoted, respectively, as
n = i for positive x-surface,
n = i for negative x-surface.
Similarly, the two y-surfaces located at y + Ay and at y are n = j, respectively; and the
z-surfaces are n = k. We denote the stress vector acting on a surface with the outer
normal n by T
n
. Newton's third law then requires
T
n
= T
n
.
(2.3)
2. Stress in a Fluid 21
We denote the stress vector acting on the positive x-surface of the cube by T
x
,
where the subscript x denotes the face of the cube. The force acting on this side is then
F
x
=T
x
AyAz (2.4)
The term F
x
can be resolved into components,
( ) ( ) ( ) ( ) ( ) ( ) .
x x x x x x x
x y z x y z
F F F T T T y z
(
= + + = + + A A
F i j k i j k (2.5)
It is customary in fluid dynamics to drop the parentheses around the F
i
and T
i
,
with the understanding that the first subscript indicates the surface on which the force
or the stress acts while the second subscript is used to show the direction of the
component. Thus, T
xx
, T
xy
and T
xz
are, respectively, the x-, y- and z-components of T
x
,
which is the stress acting on the positive x-surface. Here T
xx
is identified as the normal
stress on the x-surface of the cube, while T
xy
and T
xz
, lying in the x-plane itself, are
recognized as shear stresses.
The force on the positive x-surface, i.e., the one located at x+Ax, is therefore
F
x x +Ax
= iT
xx
+ jT
xy
+kT
xz
| |
AyAz
(2.6)
while the force on the negative x-surface is
F
x x
= iT
xx
+ jT
xy
+kT
xz
| |
Ay Az.
(2.7)
Similarly, we have on the other four surfaces
F
y y+Ay
= iT
yx
+ jT
yy
+ kT
yz
| |
Ax Az,
F
y y
= iT
yx
+ jT
yy
+ kT
yz
| |
Ax Az,
F
z z +Az
= iT
zx
+ jT
zy
+ kT
zz
| |
Ax Ay,
F
z z
= iT
zx
+ jT
zy
+ kT
zz
| |
Ax Ay.
(2.8)
We may now compute the x-component of the resultant surface force S acting on
all the surfaces of the cube, Eq. (2.2). It is comprised of the normal stresses acting on
the x-surfaces and the shear stresses acting on the y- and z-surfaces,
S
x
= T
xx x+Ax
T
xx x
| |
AyAz + T
yx y+Ay
T
yx y | |
AxAz + T
zx z +Az
T
zx z
| |
Ax Ay.
(2.9)
Equation (2.9) may be rewritten as
S
x
=
T
xx x+Ax
T
xx x
Ax
+
T
yx y+Ay
T
yx y
Ay
+
T
zx z +Az
T
zx z
Az
(
(
(
(
Ax Ay Az.
(2.10)
22 Fluid Mechanics
Similar expressions may be formed for S
y
and S
z
.
Returning to Eq. (2.2), let a general body force per unit mass g be defined. The
x-component of this body force acting on the cube is
G
x
= g
x
AxAyAz.
(2.11)
The left-hand side of Eq. (2.2) is written for the small cube as
ma
x
= a
x
Ax AyAz.
(2.12)
Substitution of Eqs. (2.10), (2.11) and (2.12) into (2.2) and division by Ax Ay Az result
in
a
x
= g
x
+
T
xx x+Ax
T
xx x
Ax
+
T
yx y+Ay
T
yx y
Ay
+
T
zx z+Az
T
zx z
Az
.
Shrinking the cube into a point, i.e., taking the limit as AxC0, AyC0 and AzC0, leads to
a
x
= g
x
+
cT
xx
cx
+
cT
yx
cy
+
cT
zx
cz
.
(2.13)
Similarly, for the y- and z-directions,
a
y
= g
y
+
cT
xy
cx
+
cT
yy
cy
+
cT
zy
cz
,
(2.14)
a
z
= g
z
+
cT
xz
cx
+
cT
yz
cy
+
cT
zz
cz
.
(2.15)
Equations (2.13), (2.14) and (2.15) are, respectively, the x-, y- and z-components
of the balance between forces and rates of change of momentum for a cube whose
sides approach zero. They therefore hold at any point in the continuum and are called
momentum equations.
Index Notation
Equations (2.13), (2.14) and (2.15) may be written in a more compact form by
using index notation.
Let the subscript i have the range i = 1,2,3, x
i
stand for x
1
= x, x
2
= y, x
3
= z and g
i
for g
x
, g
y
, g
z
. Let also a repeated subscript indicate summation over that subscript, e.g.,
2. Stress in a Fluid 23
A
ji
x
j
= A
1i
x
1
+ A
2i
x
2
+A
3i
x
3
, i =1,2,3.
(2.16)
Making use of these range and summation conventions, the momentum equations
(2.13) - (2.15) may be written as
a
i
= g
i
+
cT
ji
cx
j
.
(2.17)
The momentum equation, as given in Eq. (2.17), is entirely equivalent to
Eqs. (2.13) - (2.15). It is yet another form of Newton's second law of motion as applied
to fluids.
Equation (2.17) includes the acceleration a. From mechanics of solid bodies, one
expects this acceleration to be eventually expressed by the derivatives of the velocity
of the system. However, to take these derivatives, one must know where the
thermodynamic system is at the time t +At, which is not known. Moreover, connections
between the stress components, all nine of them, and the velocity must also be
established, or Eq. (2.17) could not be solved. Both these problems, i.e., that of the
acceleration and that of relating stresses to velocities, are treated in Chapter 5.
Because both i and j in T
ij
have each the range 3, T
ij
represents nine numbers, the
so-called stress components, which may be arranged in a matrix form, the stress
matrix:
T
ij
| |
=
T
11
T
12
T
13
T
21
T
22
T
23
T
31
T
32
T
33
(
(
(
(
(
=
T
xx
T
xy
T
xz
T
yx
T
yy
T
yz
T
zx
T
zy
T
zz
(
(
(
(
(
.
We now proceed to establish relations between the nine stress components at a
point in one cartesian coordinate system and those at the same point in other cartesian
systems, rotated with respect to the original one.
Moments on a Cube
In this section we show that the stress matrix is symmetrical, in the sense that
T
yx
= T
xy
. This important feature of the stress matrix applies both for stationary and
moving fluids.
Let the moments acting on the small cube be now computed. To obtain the
x-component of the moments, we imagine a shaft piercing the centers of the n = i and
the n = i sides of the cube, Fig. 2.2, and compute the turning moments around this
shaft.
24 Fluid Mechanics
The only contributions to the
turning moment come from the shear
stresses. For greater clarity, these
shear stresses are shown in detail in
Fig. 2.2. We employ here the con-
vention of taking the stresses repre-
sented by the arrows as positive when
they act on positive surfaces, n = +j
and n = +k. Newton's third law, that
of action and reaction, then implies
that the stresses acting on the negative
surface n = j and n = k are negative.
Note, that the cube is small and
in the limit it shrinks to a point. Com-
puting moments we obtain
( ) ( )
2 2
2 2
1 1
.
2 2
x yz yz
y y y
zy zy
z z z
yz yz zy zy
y y y z z z
y y
M T x z T x z
z z
T x y T x y
x y z T T T T
+A
+A
+A +A
A A
( (
= A A + A A
A A
( (
A A A A
(
= A A A + +
(
(2.18)
Newton's second law of motion states that the angular acceleration e
x
of the cube is
M
x
= I
x
e
x
,
(2.19)
where I
x
is the polar moment of inertia of the cube around the piercing axis, i.e.,
( ) ( ) ( )
/2 /2
2 2
2 2
/2 /2
12
y z
x
y z
I x y z dzdy x y z y z
A A
A A
(
= A + = A A A A + A
} }
(2.20)
Substitution of Eqs. (2.18) and (2.20) into Eq. (2.19) and division by the volume
of the cube, Ax Ay Az, yield
( ) ( )
( ) ( )
2 2 1 1
.
2 2 12
x
yz y y yz y zy z z zy z
T T T T y z
A A
e
A A
+ +
(
+ + = +
(2.21)
Now let the cube shrink to a point, i.e., let Ax0, Ay0 and Az0, resulting in
Figure 2.2 Shear stresses on a cube.
2. Stress in a Fluid 25
T
yz
T
zy
= 0 (2.22)
with T
yz
and T
zy
now acting at the same point. Repeating those considerations for M
y
and M
z
, two additional relations are obtained, i.e.,
T
xy
= T
yx
,
T
yz
= T
zy
,
T
xz
= T
zx
. (2.23)
Equation (2.23) holds at a point. Furthermore, because there is nothing particular
about the coordinate system used to derive it, it holds in any set of orthogonal coordi-
nates.
We note that each of the three Eqs. (2.23) relates a stress on a plane to a stress on
another plane perpendicular to the first plane. Thus the first equation states that the
y-component of the stress acting on the x-plane equals the x-component of the stress
acting on the y-plane. There is, therefore, a coupling between the stresses on the
various planes passing through the same point. This coupling is investigated in more
detail in the next section.
It is noted that normal stresses on the cube sides and body forces may also con-
tribute to the moments. However, it can be shown that their contributions are of a
higher differential order than those of the shear stresses, and when the cube shrinks to
a point, i.e., in going from Eq. (2.21) to Eq. (2.22), their contributions vanish.
Forces at a Point on a Plane
In this section we find that once the stress matrix is known at a point in one carte-
sian system of coordinates, it is known at that point in all cartesian systems rotated
with respect to the original one.
As stated earlier, surface forces acting at a point on a plane depend on the
location of the point and on the plane orientation, i.e., on the direction of its outer
normal. That the forces should depend on the coordinates of the considered point is
quite clear. After all, the fluid properties may vary from point to point, and the velocity
vector does, and it seems reasonable that the stresses and the forces should also
change.
When stress at a point was defined, it seemed plausible that it should depend on
the orientation of the plane's outer normal. The plane, or at least a small disk of it, was
necessary to define the stress, and therefore the stress might depend on the plane. The
previous section, however, indicates that there are fairly simple and quite clear
relations between stresses acting at a point located on a particular plane and those
acting at the same point on other planes, perpendicular to the original one.
We already know that there are nine stress components T
ij
, which can be
expressed in cartesian coordinates in the form of a 33 square matrix:
26 Fluid Mechanics
T
11
T
12
T
13
T
21
T
22
T
23
T
31
T
32
T
33
(
(
(
(
(
.
(2.24)
In this matrix the subscripts 1, 2, 3 stand for x, y, z, respectively. Equation (2.23),
derived in the previous section, indicates that this matrix is symmetrical. It is quite
important to realize that this symmetry establishes a relation between a stress acting on
a particular plane and those acting on other planes perpendicular to it. However, Eq.
(2.23) holds for any coordinate system. Thus one may choose a new pair of y'- and z'-
coordinates while x remains the same, and yet Eq. (2.23) connects T
xy'
with T
y'x
. There-
fore the components of the stress acting on one plane must be related to those acting
on any other plane passing through the same point.
Consider now the stresses at a point on the x-plane, n = i, as shown in Fig. 2.3.
The x-plane coincides with the printed page. The shear stress vector T
s
in Fig. 2.3 may
be expressed in terms of its components as
T
s
= jT
xy
+ kT
xz
.
These components, in turn, uniquely determine T
yx
and T
zx
in the planes n = j and n =
k. Now let the coordinate system rotate around the x-axis. In the new (primed) system
the shear stress vector remains the same and only its resolved components change:
T
s
= jT
xy
+ kT
xz
= ' j ' T
xy
+ ' k ' T
xz
.
(2.25)
These in turn uniquely de-
termine
' T
yx
and
' T
zx
. Thus the
shear stress vector seems to
determine the shear stresses
T
yx
and T
zx
for all orientations
of the coordinate system. In
other words, once the stress
matrix (2.24) is known at a
point for a particular set of
coordinates, it is already
uniquely determined at that
point for all orientations. In
the next section we develop
the rules for this determina-
tion.
y
T
s
T
xy
T
xz
z
T
xz
'
T
xy
'
z'
y'
Figure 2.3 Shear stress at a point.
2. Stress in a Fluid 27
The Elementary Tetrahedron
We now proceed to obtain the rules that express the stress at a point on an arbi-
trary plane in terms of the components of the stress matrix T
ij
in a given coordinate
system.
A small system of fluid in the form of a tetrahedron is drawn in Fig. 2.4. The
tetrahedron is generated by the intersection of four planes: three with their outer
normals i, j, k, i.e., perpendicular to the coordinate axes, and the fourth with the
outer normal n = in
1
+jn
2
+kn
3
, where n
1
, n
2
and n
3
are the direction cosines of this
normal.
Figure 2.4 Elementary tetrahedron and stresses.
The tetrahedron thus obtained is convenient for what we have in mind. It has
three sides on which the stress components T
ij
are defined, as given in the stress matrix
(2.24), and then a fourth side, with its normal n. The n vector may have any
orientation, and therefore the fourth side represents just any plane. Thus, the
expression obtained for the stress on the n plane in terms of the stresses acting on the
28 Fluid Mechanics
i, j, k planes will be general and hold for any plane.
Figure 2.4 shows the stresses which act on the n plane and on one of the
coordinate planes, i.e., the j plane. Denoting the area of the n plane as A
n
and those of
the i, j and k planes as A
1
, A
2
and A
3
, respectively, we note that A
1
, A
2
and A
3
are
the x, y and z projections of A
n
, or
A
1
= n
1
A
n
, A
2
= n
2
A
n
, A
3
= n
3
A
n
.
Let the volume of the tetrahedron be V. Then the net force acting on the system is
T
1
n
1
A
n
( ) T
2
n
2
A
n
( ) T
3
n
3
A
n
( )+ T
n
A
n
+Vg
(2.26)
and Newton's second law of motion requires
V (a g) = T
n
T
1
n
1
T
2
n
2
T
3
n
3
( )A
n
.
(2.27)
Equation (2.27) is now divided by A
n
with V/A
n
= h
n
/3, where h
n
is the "height" of
the tetrahedron and A
n
is its "base." We now let the tetrahedron shrink to a point,
remaining geometrically similar to itself while shrinking. The shrinking is necessary
because relations at a point are sought; its remaining similar to itself is essential if the
n direction is to be preserved. As a result of this shrinking, h
n
C 0 and with it V/A
n
.
Equation (2.27) thus becomes
T
n
= T
1
n
1
+ T
2
n
2
+ T
3
n
3
.
(2.28)
Equation (2.28) may be also written in component form as
T
n1
= T
11
n
1
+ T
21
n
2
+ T
31
n
3
,
T
n 2
= T
12
n
1
+ T
22
n
2
+ T
32
n
3
,
T
n 3
= T
13
n
1
+ T
23
n
2
+ T
33
n
3
.
(2.29)
Using index notation, Eq. (2.29) is recast in a more compact form,
T
nj
=T
ij
n
i
.
(2.30)
The desired rule has thus been obtained. Indeed, for a given T
ij
the stress vector T
n
on
any plane n can be computed, provided the direction cosines n
1
, n
2
and n
3
of the
normal to the plane are known.
Equations (2.28) - (2.30) indeed show coupling between the stresses acting on the
various planes passing through the same point. Once the stress matrix is given in any
one coordinate system, these equations show how the stresses may be obtained on any
plane passing through the same point. All one has to do is specify the plane of interest
by its outer normal, and the equations yield the stresses on that plane. Thus the stress
matrix at a point in any one coordinate system contains all the information on stresses
acting on other planes through the same point.
2. Stress in a Fluid 29
Example 2.1
The stress matrix at a point P is given by
T
11
T
12
T
13
T
21
T
22
T
23
T
31
T
32
T
33
(
(
(
(
(
=
2 1 3
1 1 2
3 2 1
(
(
(
(
.
Find the stress vector on the plane passing through P and parallel to the plane
whose unit normal is
n =
3
7
i +
6
7
j +
2
7
k.
Solution
We are essentially looking for T
1
, T
2
and T
3
in Eq. (2.28). Substitution of the
components of the stress matrix into Eq. (2.29) yields
T
n1
= T
11
n
1
+ T
21
n
2
+ T
31
n
3
= 2n
1
+ n
2
3n
3
,
T
n 2
= T
12
n
1
+ T
22
n
2
+ T
32
n
3
= n
1
+ n
2
+ 2n
3
,
T
n 3
= T
13
n
1
+ T
23
n
2
+ T
33
n
3
= 3n
1
+ 2n
2
+ n
3
.
These equations may be written in matrix form as
T
n 1
T
n 2
T
n 3
| |
= n
1
n
2
n
3
| |
T
11
T
12
T
13
T
21
T
22
T
23
T
31
T
32
T
33
(
(
(
(
(
=
3
7
6
7
2
7
| |
2 1 3
1 1 2
3 2 1
(
(
(
(
=
6
7
13
7
5
7
| |
.
Substitution into Eq. (2.28) yields the desired stress vector
T =
6
7
i +
13
7
j +
5
7
k.
Example 2.2
The stress matrix is given at the point P in the coordinate system x, y, z by
30 Fluid Mechanics
T
11
T
12
T
13
T
21
T
22
T
23
T
31
T
32
T
33
(
(
(
(
(
=
1 1 1
1 3 2
1 2 2
(
(
(
(
.
a. Is such a matrix possible? If not, make a plausible correction.
b. Are the stresses known now on all the planes passing through P, e.g., on the
planes whose normals are i , j, k, n = in
1
+ jn
2
+ kn
3
?
Solution
a. No, this matrix is not symmetrical. We correct T
31
from 1 to 1. Now the matrix
may be right.
b. Yes, they are known. Inspecting the stress matrix, we write down the stress
vectors on the x, y, z planes, respectively,
T
x
= i + j k,
T
y
= i + 3j + 2k,
T
z
= i + 2 j + 2k.
The stress vector acting on the n plane is
T
n
= i n
1
+ n
2
n
3
( )+ j n
1
+ 3n
2
+ 2n
3
( )+ k n
1
+ 2n
2
+ 2n
3
( ).
Stress in a Fluid at Rest
Here we show that in a stationary fluid the only remaining stress is the pressure
and that this pressure stress is the same in all directions.
A fluid has been defined in Chapter 1 as a continuum, which cannot support a
shear stress while at rest. Thus a fluid at rest admits no shear stress and its stress
matrix, e.g., Eq. (2.24), is a diagonal matrix in all systems of cartesian coordinates:
T
ij
= 0 for i = j .
(2.31)
Furthermore, Eq. (2.29) now reads
T
n1
= T
11
n
1
,
T
n 2
= T
22
n
2
,
T
n 3
= T
33
n
3
.
(2.32)
We claim that because Eq.(2.32) must hold for all cartesian coordinate systems,
2. Stress in a Fluid 31
T
11
= T
22
= T
33
= T
n
,
(2.33)
a result known as Pascal's Law:
In a fluid at rest the pressure at a point is the same in all directions.
Example 2.3
Given Eq. (2.32) and the condition that it holds in all cartesian coordinates, prove
Pascal's law.
Solution
Consider a plane with its outer normal n and the stress vector on it, viewed side-
ways, as shown in Fig. 2.5. Obviously there are no shear stresses and T
n
is parallel to
n. With Eq. (2.32) T
n
becomes
T
n
= iT
n1
+ jT
n2
+ kT
n 3
= in
1
T
11
+ jn
2
T
22
+ kn
3
T
33
= nT
n
,
where
T
n
= nT
n
is due to T
n
being parallel to n.
We now form the scalar product of this ex-
pression with i, with j and with k. These products
result in
T
11
= T
n
, T
22
= T
n
, T
33
= T
n
,
and therefore
T
11
= T
22
= T
33
= T
n
,
and the proof is complete.
Stress in a Moving Fluid
A moving fluid can support shear stress. Consider a fluid that fills the space
between two parallel long plates, as shown in Fig. 2.6. The lower plate is stationary
and the upper one moves at a uniform velocity V under the influence of a weight. The
force pulling the upper plate thus must be counterbalanced by a force equal in
magnitude and opposite in direction, or the upper plate and the weight must accelerate,
which they do not. Consider now a rectangle of fluid Ax wide bounded by the two
plates. We assume the pressure not to depend on x. Hence the shear stress on this fluid
rectangle at the top, where it touches the upper plate, must be balanced by that at the
n
n
T
Figure 2.5 A plane, its outer normal
and its stress vector in a
fluid at rest.
32 Fluid Mechanics
bottom, and we see how this stress transfers to the bottom plate. Furthermore, the same
considerations apply to a block that extends from the top plate to the A-A plane, and
therefore the shear stress on the A-A plane is the same as on the top plate. As the A-A
plane can be selected anywhere in the fluid, one must conclude that the shear stress is
the same on all planes parallel to the plates.
moving top plate
weight
V
y
x
T
yx
T
yx
T
yx
T
yx
A A h
shear stress profile
velocity profile
A x
Figure 2.6 Fluid in shear flow.
It is known from experiments that the fluid adjacent to a solid surface adheres to
it. Hence the fluid adjacent to the upper plate, at y = h, moves with the velocity V,
while that at the lower plate, at y = 0, is at rest. The uniform stress, i.e., force per unit
area, exerted on the viscous fluid by the upper plate, T
yx
, can be expressed by Newton's
law of viscosity, Eq. (1.19):
T
yx
=
du
dy
.
(1.19)
Equation (1.19) may be integrated with the boundary condition of u = 0 at y = 0 to
yield
u =
T
yx
y .
(2.34)
Hence, once the shear stress is known, the velocity at any point in the fluid can be
calculated by means of Eq. (2.34). The shear stress may be eliminated by substituting u
(h ) = V into Eq. (2.34)
V =
T
yx
h ,
(2.35)
resulting in
u =
V
h
y.
(2.36)
2. Stress in a Fluid 33
The flow just obtained is known as Plane Couette Flow or Plane Shear Flow.
Example 2.4
Viscosity of printing ink is measured by first coating a metal plate with the ink
and then passing the plate through a gap between two parallel stationary plates, Fig.
2.7. The velocity of the plate is measured by timing the descent of the weight between
two marks on a yardstick.
Figure 2.7 Measuring viscosity of printing ink.
What is the viscosity of the printing ink if the plate velocity is 1 cm/s and the
length and width of the stationary plates are 20 cm and 10 cm, respectively; the gap on
each side of the moving plate is 0.5 mm, and the weight is 5 N.
Solution
The shear stress applied by the weight on the fluid on both sides of the plate is
T
yx
=
F
2A
=
5
2 0.2 0.1
=125N / m
2
.
The viscosity of the ink is found by substitution into Eq. (2.35)
=
T
yx
h
V
=
125 0.5 10
3
0. 01
= 6. 25 kg / m s = 6250cp.
A Case where the Stress Is Not Constant
In the previous section we considered the case of constant shear in a moving
fluid. In this section we examine a case where the stress in the moving fluid is not
constant.
Consider the two-dimensional plane flow between two stationary plates, as
shown in Fig. 2.8. The flow in this case is induced by a constant pressure drop,
AP/Ax, along the x-axis.
34 Fluid Mechanics
Figure 2.8 Flow between two stationary plates, plane Poiseuille flow.
In a way similar to the shear flow example we consider a rectangular fluid block of
length AL and height 2y, extending from y to +y, as shown in Fig. 2.8. The pressure
difference between its two vertical sides is
AP
L
=
AP
Ax
AL,
and hence the force pushing it to the right is
2 y AP
L
= 2 y
AP
Ax
|
\
|
.
| AL
(note that AP/Ax < 0). This force is balanced by the shear on the upper and lower sides:
T
yx
2AL.
Hence,
T
yx
=
AP
Ax
|
\
|
.
| y ,
(2.37)
and by Newton's law
du
dy
=
A P
A x
|
\
|
.
y .
Integration now yields
u =
1
2
AP
Ax
|
\
|
.
| y
2
+ C,
which with the no-slip boundary condition
u( h) = 0 yields the velocity profile
u =
h
2
2
AP
Ax
|
\
|
.
|
1
y
h
|
\
|
.
|
2
(
( (2.38)
This flow is known as Plane Poiseuille Flow.
2. Stress in a Fluid 35
Inspection of Eq. (2.38) indicates that the highest velocity is at the midplane
between the plates, i.e., at y = 0, and that its magnitude is
u
max
=
AP
AL
|
\
|
.
|
h
2
2
.
(2.39)
Hence, Eq. (2.38) may also be rewritten in terms of u
max
as
u = u
max
1
y
h
|
\
|
.
|
2
(
(
(
.
(2.40)
Example 2.5
The gap between the two plates, shown in Fig. 2.8, is 2h = 20.005 m. The
pressure drop in the fluid between the plates is given by
AP
AL
= 20
N/m
2
m
= 20N/ m
3
.
The fluid between the plates is
a. Water, = 0.001 N
.
s/m
2
.
b. Glycerin, = 1.5 N
.
s/m
2
.
Find the highest velocity in the flow and the shear stress on the plates for each fluid.
Solution
The highest velocity in the flow is calculated from Eq.(2.39).
a. For water:
u
max
= 20
0.005
2
2 0.001
= 0.25 m/s.
b. For glycerin:
u
max
= 20
0.005
2
2 1.5
= 1.66710
4
m/s = 1 cm/min.
The stress at the walls is conveniently obtained from Eq. (1.19) together with Eq.
(2.38) set in the form
T
yx
y=h
=
du
dy
y=h
=
AP
AL
y
y=h
=
AP
AL
|
\
|
.
|
h
Note, the stress on the wall is opposite in sign to the stress in the fluid.
The form just set does not include , and the result is the same for water and
glycerin,
36 Fluid Mechanics
T
yx
= 20 0.005= 0.1 N/ m
2
.
That the result is the same for both fluids is not surprising and is brought by
AP/AL being the same for both. Indeed, choosing the fluid block in Fig. 2.7 to have
y = h makes Eq. (2.38) the same as that used to compute T
yx
on the walls here. In
other words, the pressure-drop forces are eventually supported by the plates.
* The Stress Tensor
Equation (2.30) may be viewed as an operator equation that requires a vector, n,
as an input for performing the operation; its output is also a vector, T
n
. The stress
matrix [T
ij
] is the operator that is given at a point. For any unit vector n at that point a
vector T
n
is generated by the operation indicated in Eq. (2.30). Because this operation
is linear, it is sometimes called a linear vectorvector dependence.
The operator T
ij
is called a tensor, and because it relates to the stress vector T
n
, it
is called the stress tensor. Its components T
ij
are given by the stress matrix, Eq. (2.24).
Equation (2.30) holds quite generally. It describes a physical phenomenon and
therefore does not depend on the coordinates used. Under coordinate transformation
not only do the vectors n and T
n
remain the same, but the form of the relation between
them, Eq. (2.30), must also remain the same. The components of n and T
n
are,
however, coordinate dependent, and therefore the components T
ij
must transform in
such a way that the form of Eq. (2.30) is preserved. In other words, once the stress
tensor components T
ij
are known for a given coordinate system, the stress vector T
n
acting on any plane with its outer normal n may be directly obtained by the use of Eq.
(2.30).
The basic rules for vector and tensor transformations are now derived. The
transformation of a vector is included as an introduction to that of the tensor.
* Transformation of Coordinates Vectors
Consider the vector S given in Fig. 2.9 in terms of the e
i
system:
S= S
i
e
i
,
where the summation convention holds. Let another system
' e
k
be chosen, in which the
same vector assumes the form
2. Stress in a Fluid 37
S= ' S
k
' e
k
.
Because
S= ' S ,
S
i
e
i
= ' S
k
' e
k
.
Scalar multiplication by
' e
k
on both sides results in
' S
k
= S
i
e
i
' e
k
( )
,
(2.41)
which is the transformation rule for vectors. Similarly
S
m
= ' S
n
' e
n
e
m
( )
.
e
e
e
e
e
e
1
1
2
2
3
3
'
'
'
S = S '
Figure 2.9 Components of a vector in two coordinate systems.
* Transformation of Coordinates Tensors
The stress tensor T
ij
yields the stress vector S for the surface with the outer
normal n:
S= e
i
T
ij
(e
j
n) = e
i
T
ij
n
j
=e
k
S
k
.
In the rotated coordinates
' e
k the relation becomes
38 Fluid Mechanics
' S = ' e
m
' T
mn
' n
n
.
However, since S = S', substitution of
( ) ( )
' ' ' '
and of
m i m i n n j j
n n = = e e e e e e
into the expression for
' S
yields
( ) ( )
( )( )
' ' '
' ' '
'
i m i mn n j j
i mn m i n j j
T n
T n
=
=
S e e e e e
e e e e e
Comparison with the expression for S yields
( )( )
' ' '
,
ij mn m i n j
T T = e e e e
(2.42)
which is the rule for the transformation of a tensor component. Similarly
( )( )
' ' '
.
mn ij i m j n
T T = e e e e
* Principal Directions
Of the nine components of the stress tensor the three normal stresses are more
intuitively visualized. Pascal's law, proved in Example 2.3, had been known as an
experimental fact (and hence its being called a "law" rather than a theorem which can
be proved) long before matrices were used. Pascal's law deals with the pressure, which
is intuitively visualized as a negative normal stress. We already know that in fluid at
rest the whole stress tensor is expressible in terms of the pressure only. We also
suspect that in moving fluids the three normal stresses may not be the same, as
otherwise, i.e., had they always been the same, Pascal's law would not be limited to
fluids at rest.
We consider again Fig. 2.5. For a fluid at rest
T
n
= nT
n
= np,
with p being the pressure. In a moving fluid the normal stress on the plane with its
normal n is T
nn
, i.e., the nth component of T
n
.
Now there could be a direction n for which
T
n
= nT
n
.
In other words, we rotate n and with it, of course, the plane, until we find a direction
for which the plane suffers no shear stress. When we find such a direction, we call it
principal:
2. Stress in a Fluid 39
The Principal Directions of a stress tensor are defined as those in which
all shear stresses vanish.
The condition for a principal direction may also be stated as the requirement that
n and T
n
lie on the same straight line, i.e.,
T
n
= n (2.43)
with being a proportionality factor which we already recognized as T
nn
. Thus is the
normal stress in the principal direction, called the principal stress. Substitution into
Eq. (2.30) yields
n
j
= T
ij
n
i
or
T
ij
n
i
n
j
=0,
(2.44)
where T
ij
are the components of the stress tensor in the original coordinate system. We
now rewrite Eq. (2.44) in component form as
T
11
( )
n
1
+ T
21
n
2
+ T
31
n
3
= 0,
T
12
n
1
+ T
22
( )
n
2
+ T
32
n
3
= 0,
T
13
n
1
+ T
23
n
2
+ T
33
( )
n
3
= 0.
(2.45)
This is a system of three linear homogeneous equations in three unknowns n
i
. It has
non-trivial solutions only when the determinant of the coefficients vanishes, i.e.,
T
11
( )
T
12
T
13
T
21
T
22
( )
T
23
T
31
T
32
T
33
( )
= 0.
(2.46)
It is noted that the stress tensor and the determinant of Eq. (2.46) are symmetrical.
Expanding the determinant results in the following cubic equation in :
3
T
11
+ T
22
+ T
33
| |
2
+ T
22
T
33
T
23
2
+ T
33
T
11
T
31
2
+ T
11
T
22
T
12
2
| |
T
11
T
22
T
33
T
12
T
23
T
31
+ T
13
T
21
T
32
T
13
T
22
T
31
T
11
T
23
T
32
T
12
T
21
T
33
| |
= 0
Linear algebra then yields that this cubic equation has three real roots
i
, i = 1,2,3, and
therefore there are three principal directions. Using linear algebra, it can also be shown
that these directions are orthogonal to one another. The rotation needed to reach one of
these principal directions is obtained from
40 Fluid Mechanics
xT
11
+ yT
12
+ z T
13
=
i
x,
xT
21
+ yT
22
+ z T
23
=
i
y,
xT
31
+ yT
32
+ z T
33
=
i
z ,
(2.47)
which may be manipulated to express x/y, y/z and z/x, i.e., the tangents of the angles of
rotation. The particular direction depends on the choice of i, i = 1,2,3.
Because the principal directions are orthogonal to one another, they may be used
as a cartesian system of coordinates. In this system Eq. (2.47) attains the form
T
11
0 0
0 T
22
0
0 0 T
33
(
(
(
(
(
=
1
0 0
0
2
0
0 0
3
(
(
(
(
(
.
(2.48)
The nonvanishing terms in the stress tensor in its principal directions were called
principal stresses. Thus the solutions
i
, i = 1,2,3, have the direct physical inter-
pretation of being the three normal stresses in the principal directions.
Example 2.6
Find the principal stresses and the principal directions for the stress tensor
T
ij
| |
=
3 1 1
1 0 2
1 2 0
(
(
(
(
.
Solution
We first look for the three principal stresses
1
,
2
and
3
.The determinant
equation (2.45) becomes
T
11
( )
T
12
T
13
T
21
T
22
( )
T
23
T
31
T
32
T
33
( )
=
3 1 1
1 2
1 2
= 0.
Expanding, this yields the polynomial equation
( + 2) ( 1) ( 4) = 0,
whose roots give the principal stresses
1
= 2,
2
= 1,
3
= 4.
2. Stress in a Fluid 41
Hence, in terms of the principal stresses our stress tensor assumes the form
T
ij
| |
=
1
0 0
0
2
0
0 0
3
(
(
(
(
(
=
2 0 0
0 1 0
0 0 4
(
(
(
(
.
We now look for three unit vectors n
1
, n
2
and n
3
pointing in the principal
directions.
Letting n
1
be the direction associated with
1
= 2, we find upon substitution into
Eq. (2.45)
5n
1
+ n
2
+ n
3
= 0,
n
1
+ 2n
2
+ 2n
3
= 0,
n
1
+ 2n
2
+ 2n
3
= 0.
The solution is n
1
= 0 and n
2
= n
3
.
Since n
1
is a unit vector,
n
2
2
+ n
3
2
=1,
and we obtain
n
2
= 1 / 2, n
3
= 1 / 2,
such that
n
1
= 0i +
1
2
j
1
2
k.
where i, j, k are the unit vectors in the directions of the x, y, z coordinates.
Similarly, for
2
, Eq. (2.45) yields
2n
1
+ n
2
+ n
3
= 0,
n
1
n
2
+ 2n
3
= 0,
n
1
+ 2n
2
n
3
= 0,
the solution of which leads to
2
1 1 1
,
3 3 3
= n i j k
and for
3
,
n
1
+ n
2
+ n
3
= 0,
n
1
4n
2
+ 2n
3
= 0,
n
1
+ 2n
2
4n
3
= 0,
which yields
42 Fluid Mechanics
n
3
=
2
6
i
1
6
j
1
6
k.
Hence, the unit vectors n
1
, n
2
and n
3
in the principal directions have been obtained.
Example 2.7
The motion of the fluid between two parallel flat plates, with the lower one fixed
and the upper one moving, is described in Figs. 2.6 and 2.10. The stress tensor at point
B is given by
T
xx
T
xy
T
yx
T
yy
(
(
(
=
p V h
V h p
(
(
(
.
Find the principal directions and the principal stresses,
' T
ij
, corresponding to these di-
rections.
T'
T'
xx
yy
B h
y
x
V
Figure 2.10 Velocity between a stationary and a moving plate.
Solution
Equation (2.46) yields for the present case
T
xx
T
xy
T
yx
T
yy
= 0.
Hence
p ( )
2
V / h ( )
2
= 0,
the solution of which yields
2. Stress in a Fluid 43
1
=
V
h
p, ' T
xx
=
V
h
p
and
2
=
V
h
p, ' T
yy
=
V
h
p.
Substitution of
1
and the expressions for the stress tensor into Eq. (2.45) yields
V
h
p
|
\
|
.
|
n
x
= pn
x
+
V
h
n
y
,
V
h
p
|
\
|
.
| n
y
=
V
h
n
x
pn
y
,
with the solution
n
x
= n
y
.
Hence
tano
1
= n
x
/ n
y
; o
1
= 45
o
.
Similarly for
2
one obtains n
x
= n
y
and o
2
= 45
o
.
Example 2.8
The velocity profile in a circular pipe is given by
q
r
= 0, q
u
= 0, q
z
= 2U 1
r
R
|
\
|
.
|
2
(
(
(
,
where U is the average velocity and R is the pipe radius, Fig. 2.11. The x, y and r
coordinates are related by x
2
+ y
2
= r
2
.
Figure 2.11 Velocity profile in a circular pipe.
Measurements and computations yield the following cartesian components for
the stress tensor in this flow:
44 Fluid Mechanics
T
xx
= T
yy
= T
zz
= p, T
xy
= T
yx
= 0,
T
xz
= T
zx
= T
yz
= T
zy
=
4Ur
R
2
,
all at x = y, i.e., at
r =r i + j ( )/ 2, r is the radius vector in the x-y plane.
Write the stress tensor, and find the principal stresses and the principal directions
for r = 0, r = R/2 and r = R.
Solution
For r = 0 the stress tensor is
T
ij
=
p 0 0
0 p 0
0 0 p
(
(
(
(
.
All directions are principal, and all three principal stresses are p.
For r = R/2 :
T
ij
=
p 0 2U/R
0 p 2U/R
2U/R 2U/R p
(
(
(
(
.
The characteristic equation, Eq. (2.46), becomes
p 0 2U/R
0 p 2U/R
2U/R 2U/R p
= 0
or
p ( ) p ( )
2
a
2
| |
a a p ( )
| |
= 0,
where a = 2U/R.
The equation is satisfied by
1
= p, and the remaining equation,
(p )
2
= 2a
2
,
yields
2
= p + 2a,
3
= p 2a.
2. Stress in a Fluid 45
For
1
= p, Eq. (2.44) becomes
0 + 0 an
n
= 0,
0 + 0 an
n
= 0,
an an + 0 = 0,
with the solution n
3
= 0, n
1
= n
2
and
n
1
= i
1
2
j
1
2
,
where n
1
is n for
1
.
For
2
= p + 2a, Eq. (2.32) becomes
2an
1
+ 0 an
3
= 0
,
0 2an
1
an
3
= 0,
an
1
+ an
2
2an
3
= 0
,
with thesolution n
1
=n
2
, n
3
= 2n
1
and
n
2
=
1
2
i + j 2k
( )
,
where n
2
is n for
2
.
For
3
= p 2a, Eq. (2.32) becomes
2an
1
+ 0 an
3
= 0
,
0 + 2an
1
an
3
= 0,
an
1
an
2
2an
3
= 0
,
with thesolution n
1
= n
2
, n
3
= 2n
1
and
n
3
=
1
2
i + j + 2k
( )
,
where n
3
is n for
3
.
The three principal directions have thus been obtained. To clarify the geometry of
the results, it is noted that
i + j = 2 r;
hence n
2
=(r k) / 2 and n
3
= (r +k)/ 2. The
principal directions are set at 45
o
to the direction of the flow and in the u direction.
Note : n
1
n
2
= n
1
n
3
= n
2
n
3
= 0.
Note :
1
+
2
+
3
= 3p = T
11
+ T
22
+ T
33.
The sum of the normal stresses has been conserved.
For r = R: a = 4U/R. The other results are the same as have been obtained for
46 Fluid Mechanics
r = R/2.
* Identical Principal Stresses Pascal's Law
There may be situations where all the principal stresses are identical. In that case
all the directions are principal, i.e., no shear stresses are produced but the stress tensor
is rotated. Furthermore, when no shear stresses appear in any direction, the normal
shear stresses are identical.
A fluid has been defined as a continuum that cannot support shear stress while at
rest. Thus, for a fluid at rest, the form of the stress tensor, in any coordinate system, is
that of Eq. (2.48), i.e., with zeros in the locations reserved for the shear stresses.
Therefore, for a fluid at rest, any coordinate system is principal. The equality of the
normal stresses at a point is a direct conclusion of this property:
1
=
2
=
3
.
(2.49)
The normal stress in a fluid at rest is negative, i.e., compressive, and the term
used for it is pressure. Thus, Eq. (2.49) becomes
p = = const.
(2.50)
The result just obtained has been referred to as Pascal's law:
In a fluid at rest the pressure at a point is the same in all directions.
References
R. Aris, "Vectors, Tensors and the Basic Equations of Fluid Mechanics," Prentice-
Hall, Englewood Cliffs, 1962.
D. Frederick and T.S. Chang, "Continuum Mechanics," Allyn and Bacon, Boston,
1965.
Y.C. Fung, "A First Course in Continuum Mechanics," 2nd ed., Prentice-Hall, Engle-
wood Cliffs, 1977.
2. Stress in a Fluid 47
Problems
2.1 A unidirectional flow between three infinite flat plates is shown in Fig. P2.1.
The two outer plates are stationary, while the midplate moves at a constant
velocity U = 1 m/s, as shown in the figure. The gap between the plates is h =
0.005 m. The fluid has a viscosity of = 2 kg/(m
.
s) and is Newtonian, i.e., it
obeys Newton's law of viscosity, Eq. (1.19).
a. Consider Eqs. (2.34) - (2.36) and calculate the shear stress on the surfaces of
all the plates.
b. Find the force, per unit area of plate, needed to maintain the steady motion
of the midplate.
Figure P2.1
2.2 A two-dimensional flow past a semi-infinite stationary flat plate is shown in
Fig. P2.2. The fluid far from the plate moves with a constant velocity U in a
direction parallel to the plate. The fluid has the viscosity = 2 kg/(m
.
s) and the
density = 1,000 kg/m
3
. The shear stress on the plate is measured as
t
xy
= 0. 332
U
2
Re
x
where Re
x
=
Ux
,
and x is the distance from the leading edge of the plate. This shear integrates
into a force pushing the plate in the flow direction. Find the force, per unit
width of plate, acting on the regions 0 < x < 1 m, 0 < x < 2 m, 0 < x < 10 m.
Figure P2.2
2.3 Consider again the flow between the plates in Problem 2.1. Measurements show
that on all the wetted surfaces,
48 Fluid Mechanics
T
xx
= T
yy
=p = const.
a. Write the two-dimensional stress tensor at a point on the upper surface of
the lower plate.
b. Choose a coordinate system x'-y', which is rotated by t/4 with respect to the
x-y system. Write the two-dimensional stress tensor at a point on the upper
surface of the lower plate in the x'-y' coordinate system.
c. Write again those two expressions for the tensors at a point in the fluid
midway between the lower stationary plate and the moving plate.
2.4 Consider again the two-dimensional flow past a semi-infinite flat plate given in
Problem 2.2. Measurements show that on both wetted surfaces,
T
xx
= T
yy
=p = const.
a. Write the two-dimensional stress tensor at a point x = 1.5 m on the upper
surface of the plate.
b. Choose a coordinate system x'-y', which is rotated by t/4 with respect to the
x-y system, and write the two-dimensional stress tensor at the same point in
the x'-y' coordinate system.
c. Write again those two expressions for the tensors at the point x = 5 m on the
upper surface of the plate.
2.5 The stress tensor at a point is given as
1 4
4 3
(
(
(
.
a. Find the principal directions.
b. Find the direction in which the shear stress is the largest.
c. Can the continuum in which this tensor exists be a fluid?
Hint: a fluid cannot withstand a positive normal stress.
2.6 Three coordinate systems are given as
x
1
, y
1
, z
1
( ); x
2
, y
2
, z
2
( ); x
3
, y
3
, z
3
( ).
System 2 is obtained by turning system 1 by t/4 around the z
1
-axis.
System 3 is obtained by turning system 2 by t/4 around the x
2
-axis.
Express the vectors
R = 5i
1
+ 6j
1
+ 7k
1
and S = i
2
+ 2j
2
+ 3k
2
in all three systems.
2. Stress in a Fluid 49
2.7 Four coordinate systems are given as
x
1
, y
1
, z
1
( ); x
2
, y
2
, z
2
( ); x
3
, y
3
, z
3
( ); x
4
, y
4
, z
4
( ).
System 2 is obtained by turning system 1 by t/4 around the z
1
-axis.
System 3 is obtained by turning system 2 by t/4 around the x
2
-axis.
System 4 is obtained by turning system 3 by (-t/4) around the z
3
-axis.
Express the vectors i = i
1
, j = j
1
, k = k
1
in systems 2, 3, 4.
2.8 a. A solid rod of cross-sectional area A is subjected to a compressive force F,
Fig. P2.8. Assuming that the stress is uniform across the section, find the
shear stress in the plane B-B.
b. A fluid enclosed in a pipe is compressed by two pistons, as shown in the
same scheme as for part a. Find the shear stress on the plane B-B. Why is the
result different from that in part a?
Figure P2.8
2.9 The stress at a point P is given by the matrix
7 0 2
0 5 0
2 0 4
(
(
(
(
.
Find the principal directions, i.e., the directions in which no shear stress exists,
and the stresses in these directions.
2.10 The state of stress at a point (x
o
, y
o
, z
o
) is given by
100 0 0
0 50 0
0 0 100
(
(
(
(
.
Find the stress vector and the magnitudes of the normal and shearing stresses
50 Fluid Mechanics
acting on the plane whose normal is
n =
1
14
i +2j +3k ( ).
Is this material liquid or solid?
2.11 a. Show that in a two-dimensional flow field where the stress tensor is reduced
to
T
ij
=
T
xx
T
xy
T
yx
T
yy
=
A B
C D
,
the two principal directions are perpendicular to one another.
b. Write the tensor in its principal directions and show that it contains no shear
components.
2.12 Show for a two-dimensional field that if the fluid has no shear at all, all direc-
tions are principal.
2.13 Extend the results of Problems 2.11 and 2.12 to three-dimensional flows by
considering pairs from i, j, k.
2.14 For the plane Couette flow find the stresses on the planes with the outer
normals
n
1
=
i + j
2
, n
2
=
i + j
2
.
2.15 For the Poiseuille flow between two plates find the stresses on the planes with
the outer normals
n
1
=
i + j
2
, n
2
=
i + j
2
at the locations y = 0, y = h/2 and y = h.
2.16 A flat plate of 1 m
2
is put inside a wind tunnel, as shown in Fig. P2.16. When
the tunnel is run, the plate is held in place by a force of 50 N, which balances
the drag force exerted by the air on the plate. The intake side of the tunnel is
open to the atmosphere.
a. Find the stress on the two sides of the plate before the tunnel starts running.
b. Find the approximate stress on the surfaces of the plate while the tunnel is
running.
2. Stress in a Fluid 51
Figure P2.16
2.17 A model of a certain airplane wing, which is essentially a flat plate of 1 m
2
, is
put inside a wind tunnel such that its angle of attack is o, as shown in
Fig. P2.17. The angle of attack is defined as the angle by which the plate is
tilted up relative to the incoming flow direction. When the tunnel is run, the
dynamometer balancing the plate shows a force of 50 N in the direction of the
air velocity (drag force) and a force of 500 N in the upward direction (lift
force). The intake side of the tunnel is open to the atmosphere.
a. Find the stress on the two sides of the plate before the tunnel starts running.
b. Find the approximate stress on the surfaces of the plate while the tunnel is
running.
Figure P2.17
52 Fluid Mechanics
53
3. FLUID STATICS
The Equation of Hydrostatics
Hydrostatics is a branch of fluid mechanics that considers static fluids, i.e.,
fluids at rest.
A fluid was defined in Chapter 1 as a continuum that cannot support shear
stresses when at rest. At that stage this definition was quite adequate. Now, how-
ever, one may inquire further: "at rest" - relative to what? The definition of a fluid
is therefore extended now to "at rest relative to any coordinate system." The coor-
dinate system itself may have any velocity or accelerations, and as long as a whole
body of fluid is at rest relative to that coordinate system, it may support no shear
stresses. It is quite important to realize that this coordinate system is not necessar-
ily inertial. Review of Chapter 2 shows that all accelerations and body forces drop
out of the considerations leading to the relations between the stress components.
Therefore, all these relations, including Pascal's law, are valid for a fluid static in
noninertial coordinates too. The momentum equations, (2.13), (2.14) and (2.15),
however, do contain acceleration terms and may, therefore, require some further
considerations on this account.
We start with the simpler case, where the system in which the fluid rests is
inertial. The momentum equations, (2.13), (2.14) and (2.15), contain no shear
terms for the case of the static fluid and become
a
x
=
cT
xx
cx
+ g
x
,
a
y
=
cT
yy
cy
+ g
y
,
a
z
=
cT
zz
cz
+ g
z
.
(3.1)
Pascal's law states
54 Fluid Mechanics
T
xx
= T
yy
=T
zz
=p,
(3.2)
and we obtain
a
x
=
cp
cx
+ g
x
,
a
y
=
cp
cy
+ g
y
,
a
z
=
cp
cz
+ g
z
,
(3.3)
or, using vector notation,
a = Vp + g.
(3.4)
For the special case of a fluid at rest with respect to a static or non-accelerating
frame of reference the acceleration vector is zero. Hence
0 = Vp + g.
(3.5)
Equation (3.5) is known as the equation of hydrostatics. It may be further simpli-
fied for the special case where g is the body force of gravity only and the coordi-
nate system is oriented such that the direction of g coincides with the negative z-
direction, i.e., g= kg. With these simplifications it becomes
cp / cz = g ,
cp / cy = 0,
cp / cx = 0.
(3.6)
The last two expressions in Eq. (3.6) imply that the pressure p is not a function of
either x or y; hence
dp / dz = g,
(3.7)
which may be integrated for any g given as a function of z. The simplest case is,
of course, that of an incompressible fluid for which the specific weight = g is
constant. For this case integration of Eq. (3.7) between some reference position z
o
and z yields
p = p
o
+ z
o
z ( ). (3.8)
Here p depends on z only. An experiment illustrating this fact is shown in Fig. 3.1.
A corollary of Eq. (3.8) states that the fluid in all branches of the apparatus shown
in the figure rises to the same level.
Denoting the elevation difference between z
o
and z by h, Eq. (3.8) becomes
p = p
o
+ h . (3.9)
3. Fluid Statics 55
When the reference surface is the free surface, as in Fig. 3.1, p
o
is the atmo-
spheric pressure. Sometimes one is interested in the pressure excess above that of
the atmosphere. This pressure is called the gage pressure, as distinguished from
the absolute pressure p, and is expressible in terms of the liquid column height:
p
gage
= p p
o
= h.
(3.10)
Figure 3.1 Pressure in an incompressible fluid at rest depends on z only.
Manometers
Manometers are instruments that measure fluid pressure. In the context of this
chapter, however, manometers will denote arrangements that indicate fluid
pressures by the measurement of the heights of fluid columns in them and the
application of Eq. (3.8). We now survey several such arrangements, all shown in
Fig. 3.2.
(a) The Simple (U-Tube) Manometer, Fig. 3.2a
In order to find the pressure p
A
, we write expressions for the pressure p
B
in
terms of the liquid in the left and right legs of the manometer, respectively,
p
B
= p
A
+ z
A
z
B
( )
and
p
B
= p
o
+ z
o
z
B
( ).
56 Fluid Mechanics
Equating both equations for p
B
,
one obtains
p
A
= p
o
+ z
o
z
A
( )= p
o
+h.
(3.11)
This result could be inferred directly by noting that the pressure at z = z
A
is
the same in both legs. Similarly, for Fig. 3.2b one may write
p
o
= p
A
+ h,
(3.12)
p
A
= p
o
h,
(3.13)
thus indicating that p
A
is below the atmospheric pressure by the amount h.
Figure 3.2 Manometers.
3. Fluid Statics 57
(b) The Barometer, Fig. 3.2c
p
A
= p
o
z
A
z
o
( ).
(3.14)
This instrument is first completely filled with a fluid, usually mercury
because of its high density and low vapor pressure. It is then inverted to its shown
position. Thus, p
A
indicates the vapor pressure of the manometric fluid. If one
assumes that p
A
is negligible, then the measurement of z
A
z
o
directly yields the
atmospheric pressure
p
o
= z
A
z
o
( ).
(3.15)
(c) The Inclined Manometer, Fig. 3.2d
Because of the large cross-sectional area of one side of this manometer, z
A
is
practically constant. Instead of measuring z
o
, L is measured, which because of the
inclination results in higher precision.
(d) The Multifluid Manometer, the Differential Manometer, Fig. 3.2e
Computing the pressure for each leg, one obtains
p = p
A
+
23
z
2
z
3
( )
+
12
z
1
z
2
( )
,
p = p
o
+
34
z
4
z
3
( )
+
45
z
5
z
4
( )
+
56
z
6
z
5
( )
.
Elimination of p yields p
A
:
p
A
= p
o
+
56
z
6
z
5
( )
+
45
z
5
z
4
( )
+
34
z
4
z
3
( )
+
23
z
3
z
2
( )
+
12
z
2
z
1
( )
or
p
A
= p
o
+
i i+1 ( )
z
i+1
z
i
( )
i=1
n
. (3.16)
Equation (3.16) also holds for the case shown in Fig. 3.2f. This is so because
the hydrostatic pressure depends only on Az of each fluid and not on the number of
wiggles of the manometer tube. Moreover, if one of the fluids is a gas, the density
of which might be negligible, its contribution to the pressure difference might also
be neglected. The gas column does, however, transmit the pressure without modi-
fying it.
Example 3.1
A liquid of density flows in a tube in which an orifice plate is installed. The
pressure drop across the orifice plate (p
A
p
B
) may be measured by the open
58 Fluid Mechanics
standpipes or by the U-tube manometer, as shown in Fig. 3.3. The density of the
manometric fluid is
m
.
a. Find p
A
p
B
in terms of the manometer reading Ah
m
.
b. Express Ah in terms of Ah
m
.
Solution
a. Equating pressures in both legs of the manometer,
p
A
+ h
2
Ah
m
( )= p
B
+
m
Ah
m
+ h
2
.
Hence
p
A
p
B
= Ah
m
m
( ).
b. For each stand pipe
p
A
= p
o
+ h
A
,
p
B
= p
o
+ h
B
,
p
A
p
B
= h
A
h
B
( )
= Ah .
Comparing this with part a,
Ah = Ah
m
m
( ),
Ah = Ah
m
or
Ah = Ah
m
.
Thus, for a large pressure drop, i.e., large Ah, one must select a heavy manometric
fluid, say mercury, in order to keep the manometer reading Ah
m
within reason. For
small Ah one can amplify Ah
m
by selecting a manometric fluid with a density
m
only slightly higher than .
Example 3.2
A mercury barometer consists of a long glass tube of 4 mm diameter, closed
at the top and immersed in a mercury reservoir of 40 mm diameter, Fig. 3.4. The
barometer was calibrated and was found to give an exact reading at 760 mm
Figure 3.3 Pressure measurement.
3. Fluid Statics 59
mercury. Now it reads 750 mm. What is the atmospheric pressure?
Solution
The atmospheric pressure is approximately
p = h =13,6009.810.750
=100,062 N/ m
2
.
To obtain the exact pressure, it is noted
that when the column of mercury went down in
the tube, it went up in the reservoir. It went
down by Ah =10 mm and therefore had to rise in
the reservoir by AH = Ah (d/D)
2
= 10(4/40)
2
=
0.1 mm. The exact pressure is therefore
p = 13,600 9.81 (760100.1) 10
3
= 100,048.7 N/m
2
.
Note: The outer diameter of the glass tube is
6 mm. Is the "exact" pressure above correct?
Answer: No. The exact rise in the reservoir is
Ah
d
2
D
2
d
o
2
= 10
4
2
40
2
6
2
= 0.102 mm
and therefore,
p=13,6009.81 760100.102 | |10
3
=100, 048.4 N/m
2
.
Equation of Hydrostatics in Accelerating Frames of Reference
Hydrostatics is now extended to accelerating coordinate systems. Consider
the case where a body of fluid moves in such a way that there is no relative motion
between the fluid particles. This is sometimes referred to as a fluid in "rigid body
motion." A coordinate system may be chosen such that this "rigid body" is fixed to
it, i.e., is at rest relative to it. However, this coordinate system is not inertial. As
already stated, Pascal's law holds for this case and no shear stress exists in the
fluid. The only terms left are pressure, and the momentum equations, which de-
scribe this case, take the form of Eqs. (3.3) or (3.4), i.e.,
a = Vp + g
or
Figure 3.4 Mercury barometer.
60 Fluid Mechanics
0 = Vp + ga ( ).
(3.17)
Equation (3.17) is identical to the equation of hydrostatics, Eq. (3.5), except
for the body force term which is now replaced by g a. It applies to a fluid at rest
relative to a coordinate system having an acceleration a. The equation of hydro-
statics in a non-accelerating frame, Eq. (3.4), can thus be considered a particular
case of Eq. (3.17). The following example illustrates the use of Eq. (3.17).
Example 3.3
A cylindrical bucket, originally filled with water
to a level h, rotates about its axis of symmetry with an
angular velocity e, as shown in Fig. 3.5. After some
time the water rotates like a rigid body. Find the
pressure in the fluid and the shape of the free surface,
z
o
= f (r ).
Solution
The gravity and acceleration vectors for this
case are, respectively,
g = e
z
g,
a = e
r
e
2
r ,
and the three components of Eq. (3.17) are
0 =
cp
cr
+ e
2
r,
0 =
1
r
cp
cu
,
0 =
cp
cz
g,
where p/u = 0 implies that p is not a function of u. The other two equations are
integrated to
p =
1
2
e
2
r
2
+ f
1
z
( )
,
p = gz + f
2
r ( ).
The same p must satisfy both expressions; hence
p r, z ( ) = p
o
+
1
2
e
2
r
2
gz + c.
The constant c is eliminated by the use of the boundary condition p(0, h
o
) = p
o
;
Figure 3.5 Rotating bucket.
3. Fluid Statics 61
hence
p r, z ( ) = p
o
+
1
2
e
2
r
2
g z h
o
( )
.
The equation of the free surface z
o
(r) is found from the condition p(r, z
o
) = p
o
, and
thus
z
o
= h
o
+
e
2
r
2
2 g
.
The magnitude of h
o
depends, of course, on the original volume of water in the
bucket:
V = tR
2
h =
0
R
}
z
o
2tr dr = h
o
+
e
2
r
2
2 g
|
\
|
.
|
|
2tr dr .
0
R
}
Integration yields
tR
2
h = tR
2
h
o
+
e
2
R
2
4g
|
\
|
.
|
and finally
h
o
= h
e
2
R
2
4 g
.
Forces Acting on Submerged Surfaces
The only stress present in a static fluid is the normal one, i.e., the pressure,
given by Eq. (3.5). For any given body force g, Eq. (3.5) is considered integrable.
In many cases gravity is the only body force present; hence
g = kg.
For this case Eq. (3.5) assumes the form of Eq. (3.7), and integration results
in Eq. (3.8) or (3.9). Surfaces in contact with a fluid are called submerged surfaces
and must satisfy Eqs. (3.8) and (3.9). The force acting on an element of a sub-
merged surface dS is then
d F = pn dS ,
(3.18)
n being the element's outer normal. The total force acting on the surface S is
F = pndS = p
o
+ h
( )
ndS
S
}
S
}
,
(3.19)
where h is the depth of the fluid below the p = p
o
level. Equation (3.19) is literally
used to compute the details of the forces acting on submerged surfaces. Such
detailed information may be necessary in the design of individual plates in the
sides of boats or in floodgates of dams.
62 Fluid Mechanics
Equations (3.18) and (3.19) are vectorial and may be resolved into their
components. Thus the x-component of the force in Eq. (3.18) is
dF
x
= i dF = p i n ( )dS = pdS
x
.
Here dS
x
is the projection of dS on the x-plane. Similarly from Eq. (3.19)
F
x
= pdS
x
= p
o
+ h
( )
dS
x
S
}
S
}
,
(3.20)
F
y
= pdS
y
= p
o
+ h
( )
dS
y
S
}
S
}
,
(3.21)
F
z
= pdS
z
= p
o
+ h
( )
dS
z
S
}
S
}
.
(3.22)
We first consider a special case of Eq. (3.19), i.e., that of the hydrostatic force
on a plane surface. For this case one may choose a coordinate system such that two
of its axes lie in that plane. In this system the direction of the force coincides with
the axis perpendicular to that plane and just one of the Eqs. (3.20) - (3.22) need be
used.
Force on a Submerged Plane Surface
Plane surfaces, as the one
shown in Fig. 3.6, are repre-
sented by sides of liquid con-
tainers, floodgates of dams, etc.
One side of such a surface is
usually submerged in the liquid
while the other side is exposed to
the atmosphere of p
o
. Thus the
net force exerted on such a sur-
face is due to the excess pressure
above atmospheric.
Let the coordinate system
for this case be chosen such that
the surface coincides with the z-plane, i.e., n dS
z
= k dS and the origin is at a
pressure p = p
o
, Fig. 3.6. Equation (3.19) now simplifies to Eq. (3.22),
F
z
= pdS = hdS
S
}
S
}
.
(3.23)
Substitution of h = y sin o results in
F
z
= sin o ydS = y
c
S sino
S
}
,
(3.24)
Figure 3.6 Force on a plane surface. The cen-
troid of the surface is at C.
3. Fluid Statics 63
F
z
= p
c
S , (3.25)
where y
c
is the y-coordinate of the centroid of the area S, which is defined by
y
c
=
1
S
ydS ,
S
}
(3.26)
and p
c
= y
c
sin o is the hydrostatic pressure at a depth corresponding to y
c
.
Equation (3.25) is particularly useful for surfaces with known centroids.
Thus, for example, the resultant force acting on a circular plate is tR
2
p
c
and its
direction is normal to the plate surface. A resultant single force may be substituted
for the distributed pressure forces acting on the total surface. To represent the
distributed forces correctly, this resultant must
i. equal these forces in magnitude and direction;
ii. have its point of application such that its moment about any axis parallel to
the x-coordinate axis equals the total moment of the distributed forces about
the same axis; and
iii. have its point of application such that its moment about any axis parallel to
the y-coordinate axis equals the total moment of the distributed forces about
the same axis.
Requirement (i) is already satisfied by Eqs. (3.25) and (3.26). With the point
of application denoted (x
F
, y
F
), requirement (ii) determines y
F
, and requirement
(iii) determines x
F
. It is noted that for surfaces symmetrical with respect to the y-
axis, x
F
comes out to be zero. The same, however, does not apply to surfaces
symmetrical with respect to the x-axis, because the pressure field is not symmetri-
cal about the x-axis.
To find this point of application (x
F
,y
F
), also known as the center of pressure,
requirement (ii) implies that moments be taken about the x-axis:
F
z
y
F
= y dF = sino y
2
dS
S
}
S
}
.
(3.27)
The surface integral in Eq. (3.27) is identified as the second moment of the
area, or the moment of inertia of the area with respect to the x-axis, i.e., I
xx
. This
moment of inertia about the x-axis can be related to the moment of inertia about a
parallel x-axis, passing through the centroid of the surface, Fig. 3.6, by the use of a
geometrical relation known as Steiner's Theorem.
Let a coordinate system parallel to the x, y system but with its origin at the
centroid c be x', y', Fig. 3.6. Steiner's theorem then states
I
xx
= y
2
dS = I
' x ' x
+Sy
c
2
.
S
}
(3.28)
Equation (3.27) together with Eqs. (3.28) and ( 3.24) yield the y-coordinate of the
center of pressure,
64 Fluid Mechanics
y
F
=
1
y
c
S
y
2
dS =
I
' x x
y
c
S
+ y
c
S
}
.
(3.29)
Some moments of inertia I
x'x'
are tabulated in standard handbooks and may be used
instead of the integration in Eq. (3.29).
For a plane surface not symmetrical with respect to the y-axis x
F
may be
found in a similar way from
F
z
x
F
= x dF = sino xy dS = sino I
xy
S
}
S
}
.
(3.30)
Steiner's theorem is used again in the form
I
xy
= I
' x ' y
+Sx
c
y
c
.
(3.31)
This yields
x
F
=
1
y
c
S
xydS =
I
' x ' y
y
c
S
+ x
c
S
}
,
(3.32)
where the x-coordinate of the area centroid is defined by
x
c
=
1
S
xdS .
S
}
(3.33)
Example 3.4
a. A vertical rectangular plate AA'B'B is set under water ( = 1,000 kg/m
3
), Fig.
3.7a. Find the resultant force, its direction and its point of application, y
F
.
b. A larger plate is now shown in Fig. 3.7b. Find the resultant and its point of
application.
Solution
a. The centroid of the plate is at C, which is L/2 deep. The force is obtained
from Eq. (3.24) as
F = k
L
2
D L ( )= k 500gDL
2
.
The moment of inertia of the rectangular plate about its centroid is
I
' x ' x
= 2 Dy
2
dy =
2
3
D
L
2
|
\
|
.
|
3
=
1
12
DL
3
.
0
L/ 2
}
3. Fluid Statics 65
Equation (3.29) then yields the y-coordinate of the center of pressure
y
F
=
I
' x ' x
y
c
S
+ y
c
=
1
12
DL
3
L
2
LD
+
L
2
=
2
3
L.
Figure 3.7 Rectangular plate under water.
Note: A diagram of the pressure as a function of depth, drawn along AB
(dashed line in Fig. 3.7a), is triangular. The center of gravity of this triangle is
2/3 L deep. The point of application of the resultant is also 2/3 L deep. Is there a
connection?
b. Equation (3.24) yields the force exerted on the plate,
F = k
L
2
D
L
sino
|
\
|
.
|
= k
500gDL
2
sino
.
Substitution of the moment of inertia and the
y-coordinate
of the centroid of
the area, respectively,
I
' x ' x
=
1
12
DL
3
sin
3
o
, y
c
=
L
2sin o
,
into Eq. (3.29) yields the y-coordinate
of the center of pressure,
y
F
=
1
12
DL
3
sin
3
o
L
2sin o
DL
sin o
|
\
|
.
|
|
+
L
2sin o
=
1
6
+
1
2
|
\
|
.
|
L
sin o
=
2L
3sin o
.
The point of application is again at two-thirds of the plate and at two-thirds of
the maximal depth.
66 Fluid Mechanics
Example 3.5
Two reservoirs, A and B, are filled with water, and connected by a pipe,
Fig. 3.8. Find the resultant force and the point of application of this force on the
partition with the pipe. Use the following dimensions: L = 4 m, H = 3 m, D = 1 m,
b = 3 m.
Figure 3.8 Two reservoirs connected by a pipe.
Solution
To find the force and its point of application on a plate with a circular hole,
we perform the following steps:
Find the force and point of application for the plate without the hole.
Find the force and point of application for a circular plate having the same
dimensions and location as the hole.
The sought force is the difference between that of the rectangular plate and
that of the circle. The combined point of application is obtained by the requirement
that the moment of the resultant force must equal the difference between the
moments of the plate and of the circle.
Let the subscripts 1 and 2 refer to the plate and to the circle, respectively. For
the rectangular plate without the hole
F
1
= L b
L
2
=
L
2
b
2
=
9,810 4
2
3
2
= 235, 440 N.
In Example 3.4 it was shown that the point of application is at 2/3 L. Hence
3. Fluid Statics 67
y
F1
=
2L
3
= 2.667 m.
The force on a circular plate the size of the hole is
F
2
= H
tD
2
4
= 9,810 3
t 1
2
4
= 23,114 N.
For a circular plate
I
' x ' x
=
1
2
I
r
=
1
2
r
2
2tr dr =
tR
4
4
=
tD
4
64
0
R
}
and Eq. (3.29) yields
y
F2
=
tD
4
/ 64
tD
2
H / 4
+ H =
D
2
16H
+3 =3.02 m.
The resultant force on the plate with the hole is
F
12
= F
1
F
2
( )=235, 44023,114=212,326 N.
The point of application is found from
y
F
F
12
= y
F1
F
1
y
F 2
F
2
.
Hence
y
F
=
y
F1
F
1
y
F2
F
2
F
12
=
2.667235,440 3.0223,114
212,326
= 2.63 m.
Example 3.6
a. The inclined triangular wall in Fig. 3.9a with o = 60
o
, L = 4 m, H = 6 m is
immersed in water ( =9,810 N/m
3
). Find the resultant force representing
correctly the effects of the hydrostatic pressure.
b. Repeat part a for the triangle in Fig. 3.9b.
Solution
a. The centroid of a triangle is at the point of intersection of its medians, which
is at one-third of its height. This can be shown analytically by
S y
c
= ydS = y dx
x=0
b
}
dy = by dy
y=0
H
}
,
y=0
H
}
y=0
H
}
68 Fluid Mechanics
S y
c
= L 1
y
H
|
\
|
.
|
ydy = LH
2 1
2
1
3
( )
=
1
6
LH
2
y=0
H
}
=
H
3
LH
2
=
H
3
S .
Figure 3.9 Triangular walls.
Hence, by Eq. (3.24),
F
z
= Sy
c
sin o =
1
6
LH
2
( )
sino =9,810
1
6
46
2
( )
sin 60
o
=203,897 N.
In this case it is easier to obtain I
xx
rather than I
x'x'
; hence
I
xx
= L 1
y
H
|
\
|
.
|
y
2
dy = LH
3
1
3
1
4
( )
=
LH
3
12
= 72 m
4
,
y=0
H
}
and by Eq. (3.27),
y
F
=
I
xx
sin o
F
z
=
9,810 sin60
o
72
203,897
= 3 m.
Similarly,
I
xy
= y xdx
x=0
b
}
dy =
L
2
2
y=0
H
}
1
y
H
|
\
|
.
|
2
y dy =
L
2
2
y=0
H
}
1
2y
H
+
y
2
H
2
|
\
|
.
|
|
ydy
y=0
H
}
=
L
2
H
2
2
1
2
2
3
+
1
4
|
\
|
.
|
|
=
L
2
H
2
24
= 24 m
4
,
and by Eq. (3.30),
3. Fluid Statics 69
x
F
=
I
xy
sino
F
z
=
9,810 sin 60
o
24
203,897
= 1 m .
b. Here we have y
c
= 4 m. Hence
F
z
= Sy
c
sin o =
1
3
LH
2
( )
sino = 9,810
1
3
46
2
( )
sin 60
o
= 407, 794 N.
From part a above
I
xx
a
= 72 = I
' x ' x
+ y
c
a
2
S = I
' x ' x
+412 = I
' x ' x
+ 48,
I
' x ' x
= 24 m
4
.
Equation (3.29) now yields
y
F
=
24
412
+ 4 = 4.5 m.
Similarly, from a above
I
xy
a
= 24= I
' x ' y
+ x
c
a
y
c
a
S = I
' x ' y
+32,
I
' x ' y
a
= 8 m
4
,
I
xy
=
y=0
H
}
y
x=b
0
}
x dx dy =
y=0
H
}
L
2
y
2
2H
2
ydy =
L
2
H
2
2
1
4
=
L
2
H
2
8
72m
4
.
Finally, by Eq. (3.30)
x
F
=
I
xy
sin o
F
z
=
9,810 sin 60
o
72
( )
407, 794
= 1. 5 m.
Note: y
F
for case a is at one-half the depth, and y
F
for case b is at three-
fourths of the depth. Compare this to the rectangular case, Example 3.4, where y
F
came out to be at two-thirds of the depth.
Forces on Submerged Surfaces The General Case
As shown in Fig. 3.10, submerged surfaces divide into three distinct groups,
namely those belonging to
a. completely submerged bodies (S
1
, S
2
);
b. floating bodies (S
3
, S
4
); and
c. submerged boundaries (S
5
, S
6
).
70 Fluid Mechanics
Of these the easiest to analyze is the completely submerged body, because it
is bounded by surfaces of one kind only, i.e., submerged surfaces. We therefore
consider a body of this type first.
Figure 3.10 Submerged bodies.
a. Completely Submerged Bodies (S
1
,
S
2
, Fig. 3.10)
The total force acting on a completely submerged surface, such as S
1
, is given
by Eq. (3.19), which, upon substitution of p from Eq. (3.8), reads
F = pn dS = p
o
+ z
o
z
( )
n dS .
S
}
S
}
(3.34)
The horizontal components of F, F
x
and F
y
are given by Eqs. (3.20) and (3.21),
respectively,
F
x
= pdS
x
= p
o
+ z
o
z
( )
dS
x
,
S
}
S
}
(3.35)
F
y
= pdS
y
= p
o
+ z
o
z
( )
dS
y
.
S
}
S
}
(3.36)
We now proceed to find F
x
by integrating Eq. (3.35). It is convenient to carry
3. Fluid Statics 71
out the integration in two stages: first by performing it at constant z on horizontal
strips of thickness dz and then by integrating over all strips, i.e., on all z. Equation
(3.35) thus becomes
F
x
= p
o
+ z
o
z
( )
all strips
}
dS
x
one strip
}
.
(3.37)
The integral
dS
x
} over a single strip yields twice the projection of the strip
on the x-plane, once with a positive sign as viewed by the happy fish in Fig. 3.10
and once with a negative sign as viewed by the unhappy fish. The integral's total
contribution is therefore nil, and F
x
vanishes. The same result is obtained for F
y
with the conclusion that completely submerged bodies suffer no net horizontal
force.
The argument that led to vanishing horizontal resultants does not apply to the
vertical component. The pressure, p, depends on z and cannot be set outside the
integral over the single strip, as done in Eq. (3.37). For S
1
this vertical force is
easily obtained by the application of Gauss' theorem to Eq. (3.19):
F = pn dS = VpdV = k
dp
dz
dV = k dV = k V
V
}
V
}
V
}
S
}
,
(3.38)
where V is the weight of the fluid displaced by the submerged body. Denoting by
B
the average specific weight of the submerged body, the force applied to the boat
by the hanging rock S
1
is kV
1
(
B1
), while that applied to the concrete block S
6
by the float S
2
is kV
2
(
B2
). The vertical force given by Eq. (3.38) is called
buoyancy force
The center of buoyancy of a submerged body is defined as the point of appli-
cation of the buoyancy force, i.e., the point with respect to which the moment of
the forces exerted on the submerged body by the fluid vanishes. Specifically, for
an arbitrary choice of the origin of the coordinate system the x-coordinate of the
center of buoyancy, x
B
, is found by
x
B
V = x dV
V
}
.
(3.39)
Hence
x
B
=
1
V
x dV
V
}
,
(3.40)
and similarly
y
B
=
1
V
y dV
V
}
.
(3.41)
Let z
b
be defined as
z
B
=
1
V
z dV .
V
}
Then the point ( x
b
, y
b
, z
b
) comes out to be the
72 Fluid Mechanics
center of gravity of the displaced fluid. The center of buoyancy thus coincides with
the center of gravity of the displaced fluid. Considering this point as that at which
the buoyancy force acts results in the buoyancy force correctly representing the
fluid forces as far as forces and moments are concerned.
Example 3.7
A quick and fairly accurate way to measure the density of a solid sample is
to weigh the sample in air and then again in a known liquid, e.g., in water,
Fig. 3.11. Find if the ratio of the weight in water to that in air is G
w
/G
a
= 0.8883.
Figure 3.11 Body balanced in air and in water.
Solution
Let the sample weight in air be G
a
and that in water be G
w
. Then
G
a
= Vg ,
G
w
= Vg V
w
g = Vg
w
( ),
where V
w
g is the buoyancy force. Hence
G
w
G
a
=
w
=1
w
or
=1
G
w
G
a
=10.8883= 0.1117.
Hence,
=
w
/ 0.1117 = 1,000 / 0.1117 = 8,953 kg/m
3
,
3. Fluid Statics 73
and the sample could be made of nickel.
Question: How do you perform this procedure for samples lighter than water?
Answer: Tie a piece of metal to the sample.
Example 3.8
In preparing astronauts for moving on the surface of the moon, they are put
with their space outfits at the bottom of a pool filled with water (
w
= 1000 kg/m
3
).
The volume of a 75-kg astronaut wearing a space outfit is 0.085 m
3
. Gravity on the
surface of the moon is about g/6. The mass of the suit is about 50 kg. Should
astronauts be equipped with weights or with floats to have them feel as if under
moon gravity?
Solution
The force acting on the astronaut on the moon is
F
m
= 75+50 ( )
9.81
6
= 204.4 N.
The force under water is
F
w
= g 75 +50 ( ) 0.085
w
| |
=9.81 75+50 85 | | =392.4 N.
The astronaut needs floats of about 0.02 m
3
(neglecting the floats' own
weight) in order to reduce the force by 392.4 204.4 = 188 N.
b. Floating Bodies (S
3
, S
4
, Fig. 3.10)
Floating bodies differ from completely submerged bodies by not being exclu-
sively enclosed by surfaces in contact with the fluid. Rather, they stick out of the
fluid. Floating bodies are always associated with a free surface of the fluid. Let
this free surface have the coordinate z
o
. The submerged part of the floating body is
now defined as that part of the body bounded by its wetted parts and by the exten-
sion of the z
o
plane into the floating body. That part of a ship, which is below the
water line, is an example of the submerged part of a floating body.
Inspection of the analysis of the completely submerged bodies reveals that the
same analysis may be repeated here for the computation of the horizontal
components of the force, with the conclusion that no horizontal resultant exists for
the floating body.
The vertical force F
z
may be easily calculated using Eq. (3.22):
74 Fluid Mechanics
F
z
= p
o
+ h
( )
dS
z
= p
o
dS
z
S
}
S
}
hdS
z
S
}
= V.
The contribution of p
o
on the upward acting force F
z
cancels out by the downward
acting p
o
on the exposed part of the body. Thus the total force exerted on the float-
ing body by the fluid is still F = k V, as in Eq.(3.38), but with V denoting the
volume of the submerged part only. The buoyancy force obtained is again equal to
the weight of the displaced fluid. This last result for both the completely sub-
merged body and for the floating one is also known as Archimedes' Law.
The center of buoyancy of the floating body is again located at the center of
gravity of the displaced fluid, as can be easily verified by repeating the considera-
tions used for the submerged body.
Example 3.9
Ice at 10
o
C has the density
i
= 998.15 kg/m
3
. A 1,000-ton
(10
6
-kg) spherical iceberg floats at
sea, Fig. 3.12. The salty water has
a density of
S
= 1,025 kg/m
3
. By
how much does "the tip of the
iceberg" stick out of the water?
Solution
The volume of the iceberg is
V =
G
i
i
=
10
6
998.15
= 1001. 9 m
3
,
which corresponds to a sphere of radius R = 6.207 m. The volume of seawater it
must
displace, by Archimedes' law, is
Vw = 10
6
/ 1,025 = 975.6 m
3
.
From Fig. 3.12,
z = R 1 coso
( )
, dz = R sin o do ,
L = R sin o , dV = tL
2
dz = tR
3
sin
3
o do ,
Figure 3.12 Spherical iceberg.
3. Fluid Statics 75
V = tR
3
sin
3
o do = tR
3
1 cos
2
o
( )
sin o d o
0
o
1
}
0
o
1
}
= tR
3
coso
1
+
1
3
cos
3
o
1 ( )
+
2
3
tR
3
= 975. 6 m
3
.
Hence, for R = 6.207 m we obtain
o
1
=144
o
, z
1
=11.229 m.
The height of the part sticking out of the water is
2R z
1
=1.185 m.
For an iceberg that is cubic, the side of the cube is 1,001.9
1/3
= 10.006 m, and the
height of the part sticking out is
10.006
975.6
10.006
2
= 0.262 m.
c. Submerged Boundaries (S5, S6, Fig. 3.10)
The plane surface has already been treated in detail in a previous section. The
surface S
5
in Fig. 3.10 is a more general example of such a boundary. The upper
part of S
5
is completely girdled by horizontal strips, and therefore this part has its
projections on the x- and y-planes canceled and does not contribute to the horizon-
tal force. However, the lower part of S
5
cannot be thus girdled. There are therefore
net horizontal forces acting on S
5
. Each case of such a general submerged surface
must be calculated separately by the use of Eqs. (3.34), (3.35) and (3.36). The only
somewhat simplifying statements that can be made for these surfaces are:
For horizontal components: Equations (3.35) and (3.36) do not change if the
surface S is replaced by its projections, i.e., dS
x
, dS
y
, respectively. Different
surfaces, which have the same projections in a certain horizontal direction,
thus suffer the same horizontal component of the force in that direction.
For vertical components: Equation (3.34) becomes
F
z
= p ds
z
S
}
.
A vertical ray entering S at a particular dS
z
may sometimes pierce S again.
When this happens, the two pierced dS
z
may be counted together in the inte-
gral to yield dF
z
= (Az) dS
z
, where Az is the distance along the ray between
its entry and exit points.
Thus dF
z
becomes just dV, with dV the volume of the cylinder formed by the
76 Fluid Mechanics
base dS
z
and the height Az. When a part of the submerged surface can be com-
pletely separated from the rest of it by a surface formed entirely by such
piercing rays, the resultant vertical force acting on this separated section is the
buoyancy force corresponding to its volume.
Figure 3.13 Submerged surfaces.
Examples for the use of these statements are illustrated in Fig. 3.13.
Example 3.10
Figure 3.14 shows three types of water gates. Assume each gate is 1 m wide
and that the depth of the water is H = 4 m. What is the resultant force of the water
pressure exerted on each gate, and what is its line of application, such that the
correct moment results.
Figure 3.14 Shapes of gates.
3. Fluid Statics 77
Solution
For gate a
p = gh = H 1 coso ( ).
Using Eq. (3.18), we have
d F = p ndS = pnH d o,
where n = r; hence
dF
x
= pH sin o do = H
2
1 coso ( )sin o do,
F
x
= H
2
0
t/2
}
(1coso)sin odo = H
2
1
1
2
| |
=
1
2
H
2
,
z
c
F
x
=
0
t/2
}
z H
2
(1 coso)sin odo = H
3
0
t /2
}
coso cos
2
o
( )
sin odo
= H
3
1
2
1
3
| |
=
1
6
H
3
, z
c
= H / 3,
dF
z
= pH cosodo = H
2
1 coso ( )cosodo,
F
z
= H
2
0
t /2
}
coso cos
2
o
( )
do = H
2
1
t
4
(
(
= 0. 2146 H
2
,
F
z
( )
x
c
( )
= H
3
0
t /2
}
coso cos
2
o
( )
sin o d o = H
3
1
2
1
3
(
(
=
H
3
6
,
( x
c
) = 0.7766 H ,
F = F
x
2
+ F
z
2
=
1
2
H
2
( )
2
+ 0. 2146H
2
( )
2
= 0. 5441H
2
= 0. 5441 9, 810 4
2
= 85, 403 N
and
o
F
= arctan
F
x
F
z
= arctan
1
2
0.2146
= 66. 7
o
.
For gate b, we have from Example 3.4
z
c
=
H
3
,
F = H 2
1
2
H = 0.7071 H
2
=110,986N.
The point of application of that force is at H/3, and its direction is perpendicular to
the inclined plane.
78 Fluid Mechanics
For gate c
p =
1
2
H 1 coso ( ),
dF
x
= p
H
2
sino do =
1
4
H
2
1coso ( )sinodo,
F
x
=
0
t
}
1
4
H
2
1coso ( )sin odo =
1
4
H
2
2 0 | | =
1
2
H
2
,
z
c
F
x
=
1
4
z
0
t
}
H
2
1 coso ( )sin odo=
1
8
H
3
1+ coso ( )
o
t
}
1 coso ( )sino do
=
1
8
H
3
0
t
}
1 cos
2
o
( )
sin odo =
1
8
H
3
2
2
3
| |
=
1
6
H
3
.
Hence
z
x
=
z
x
F
x
F
x
=
H
3
.
The force acting upward, F
z
, is found by Archimedes' law as the buoyancy force
acting on half the cylinder:
F
z
= V =
tH
2
1
4 2
|
\
|
.
|
|
= 0.3927H
2
.
We also know from solid mechanics that the center of gravity of half a circle is at
x
c
=
4
3t
r =
2
3t
H = 0.2122 H ,
F = F
x
2
+ F
z
2
=
1
2
H
2
( )
2
+ 0.3927H
2
( )
2
= 0. 6358H
2
= 0. 6358 9, 810 4
2
= 99, 792 N
and finally
o
F
= 360 arctan
F
x
F
z
= 308
o
.
Note that in both cases a and c there is no moment about the center of the cylinder.
Example 3.11
A float valve is schematically shown in Fig. 3.15. In its closed position the
sphere is pressed by the water pressure onto the valve seat. To open the valve,
force F must be applied to overcome the water pressure and the valve weight G.
3. Fluid Statics 79
Once open, the valve must stay open until the flush tank is emptied. It is kept open
by the buoyancy forces overcoming its weight G. Given the radius of the ball R,
find bounds for permissible G and required F.
Figure 3.15 Float valve.
Solution
The weight of the valve G should be less than that of the water it displaces, i.e.,
G <
4
3
tR
3
w
.
The applied force F should be
F > G + h R sino ( )
o
t/ 2
}
2tR coso Rsin odo
= G + 2tR
2
sinocoso R sin
2
o coso
( )
do
o
t/2
}
or
F > G+ tR
2
h
2
3
R
( )
.
For R = 3 cm, h = 30 cm,
G <
4
3
tR
3
w
=
4
3
t 0.03
3
9,810 =1.11N
and
F > G + tR
2
h
2
3
R
( )
= 1.11+ 9,810t 0.03
2
0.3
2
3
0.03
( )
= 8.88 N .
Hydrostatic Stability
A completely submerged body or a floating one has its buoyancy force acting
through its center of buoyancy in the positive z-direction. It also has its weight
acting through its center of gravity in the negative z-direction.
80 Fluid Mechanics
The body can remain static, i.e., not move up or down only when these two
forces are equal in magnitude. Furthermore, it must have its center of buoyancy
and its center of gravity on the same vertical line or it cannot be in equilibrium,
i.e., it will roll. When this equilibrium exists, however, it can still be stable, un-
stable or neutral. Static stability is important both for structures that are apparently
static, such as dams, floating cranes, buoys, etc., and for those which move at
noticeable speeds, such as ships or submarines. It serves as a necessary condition
in the design of the structures and also as an approximation for their dynamic
stability near equilibrium.
At static equilibrium the moment of the couple of forces, buoyancy and
weight is zero, because they act along the same vertical line, e.g., Fig. 3.16a,c.
When the considered body rolls with the angle o, the relative positions of the
center of gravity, G, and the center of buoyancy, B, change, while the forces them-
selves remain vertical.
A moment may appear, which tends to further increase the roll angle o, in
which case the situation is called unstable; or the moment may tend to decrease o
and diminish the roll, in which case the situation is denoted stable. When no
moment appears, a neutral equilibrium is inferred. The submarine with its cylin-
drical cross section in Fig. 3.16c,d is in stable equilibrium. It is seen at once that a
Figure 3.16 A submarine and a sailboat with center of gravity below center of buoyancy.
3. Fluid Statics 81
submarine, while underwater, can be in stable equilibrium only if its center of
gravity is below its center of buoyancy. Indeed, such a submarine would correct its
position even from a roll of o = 180
o
, assuming that nothing has been shifted
inside it while rolling.
This manifestation of absolute stability seems very attractive and some
sailboats are designed this way (Fig. 3.16a,b), mainly boats that must withstand
very large moments caused by high masts and large sails. Unfortunately, when a
wave strikes from the side a boat so designed, the boat rolls through large angles,
because the wave's momentum is transferred to it quite above its center of gravity.
For this reason most boats and all ships have their center of gravity in the vicinity
of the water line, where waves usually strike. This results, however in the center of
gravity being above the center of buoyancy. Such a boat and a stable one is shown
in Fig. 3.17.
Figure 3.17 A stable boat.
The point M in Fig. 3.17 is obtained by the intersection of the line of action of
the instantaneous buoyancy force with the straight line passing through the center
of gravity and the center of buoyancy of the structure at equilibrium. The length
GM is called the metacentric height, and one notes at once that as long as GM > 0
(i.e., M is above G), the o range of stable equilibrium has not been exceeded.
Neglecting dynamical effects, this means that the ship will not capsize.
Historically the line BG was provided by the mast, and a shipmaster who
knew his ship could estimate the metacentric height at different roll angles. This
criterion, however, is as good as any other and is still used by naval architects.
Some more examples of stable and unstable equilibria are shown in Fig. 3.18.
82 Fluid Mechanics
Figure 3.18 Stable and unstable equilibria.
Example 3.12
The rectangular block of balsa wood in Fig. 3.19 has the dimensions ab2c
and the density
b
. It is put in water of density
w
. Find the returning moment of
the block as it rolls with small angles.
Given: a = b = 4 m, 2c = 2 m,
b
= 500 kg/m
3
,
w
= 1000 kg/m
3
.
Also given from trigonometric considerations,
QB =
5
6
sino +
2
3
sin otan
2
o ~
5
6
o,
where B is the center of buoyancy.
Solution
The weight force of the block is
W =
b
gV =
b
g a b(2c),
which is counterbalanced by the buoyancy force
F =
W
g a b s,
where s is its submerged part. Because
W
= 2
b
, s = c, and the block sinks by c =
1 m into the water. The buoyancy force is therefore
F =
W
ga b c =1,000 9.806 4 41= 156,900N.
3. Fluid Statics 83
When the block tilts at an angle o, the returning moment is
M
o
= F QB = 156, 900
5
6
o = 130, 750o Nm.
Figure 3.19 Rolling block.
Problems
3.1 What is the pressure at point A in Fig. P3.1?
Figure P3.1 Multifluid manometer.
3.2 The inclined manometer shown in Fig. P3.2 is filled with water that reaches
mark B on the inclined leg when the pressure tap is open to the atmosphere.
84 Fluid Mechanics
A pressure p is applied and the water level rises by L = 100 mm. If d = 2
mm and D = 100 mm, what is the pressure p? What is the error introduced
by not measuring the water level drop in the wide leg?
L
d
30
o
D
p
B
p
o
Figure P3.2
3.3 The pressure difference between
points 1 and 2 in the water pipe
shown in Fig. P3.3 is measured by
the manometer shown, where
H = 3 m and h = 0.2 m. The pres-
sure at point 2 is 1.5 bars. What is
the pressure at point 1? Which
way does the pump pump?
Figure P3.3
3.4 A kerosene line is fitted with a dif-
ferential manometer, Fig. P3.4. Find
the absolute pressure at points A
and B, given
p
o
=100 kPa,
water
=1000 kg/m
3
,
kerosene
= 900 kg/m
3
,
mercury
= 13600 kg/m
3
.
3.5 All known materials behave like
fluids at pressure higher than
10,000 - 20,000 bars. Assuming the
upper solid layers of the earth to
have a density of 2.5 g/cm
3
, estimate the thickness of the earth's crust that
Figure P3.4
3. Fluid Statics 85
still behaves like solid. Estimate its thickness below the bottom of the sea,
assuming sea depth of 10 km.
3.6 A mercury barometer gave a true reading of H
1
mm
Hg. A quantity of fluid B was then injected into the
bottom of the vertical barometer tube. The B fluid
rose on top of the mercury, and the situation became
that of Fig. P3.6. What is the vapor pressure of the B
fluid?
Data:
H
1
= 760 mm, H = 660 mm, h = 100 mm,
Hg
= 13,600 kg/m
3
,
B
= 850 kg/m
3
.
3.7 The reservoir in Fig. P3.7 has a gate with h = 1 m,
b = 0.75 m, B = 1.5 m, H = 6 m. What is the resultant
force acting on the gate? What is its point of application? What should be
the angle of the slope of the walls, o, to make the line of action of the
resultant force on the wall, R, pass through C?
c
H
b
B
h
o
C
R
Figure P3.7
3.8 What is the resultant of the forces of the fluid acting on the gate AB in Fig.
P3.8? What is the moment acting at point A? (Give the answer per unit
width.)
Figure P3.8
Figure P3.6
86 Fluid Mechanics
3.9 A rectangular water channel is 4 m wide and has a circular gate as shown in
Fig. P3.9. The gate is kept closed by a weight at point E. Neglect the weight
of the gate itself and find what weight just keeps the gate closed.
Figure P3.9
3.10 A tank is divided into two parts by a vertical wall, Fig. P3.10. On each side
of the wall there is a different fluid. A conical stopper plugs an opening in
the wall.
a. What is the resultant horizontal force exerted on the stopper by the fluids?
b. What is the resultant vertical force exerted on the stopper by the fluids?
c. What is the moment of all the forces that act on the stopper with respect
to its midpoint at the center of the wall?
Figure P3.10
3.11 Assume air to satisfy pv = RT, R = 0.287 kJ/kg
.
K, all through the atmosphere,
and the atmosphere to be isothermal at 300
o
C. Find the height below which
lies 99% of the mass of the atmosphere.
3. Fluid Statics 87
3.12 A spherical balloon with a diameter of 5 m floats in the air at a height of
200 m. The atmospheric pressure at sea level is 100 kPa. Assume the atmo-
sphere to be an ideal gas satisfying p = RT, R = 287 J/kg K at the constant
temperature T = 300 K. Find the total mass of the balloon.
3.13 The balloon in Fig. P3.13 has a
diameter of 10 m and is filled with
hydrogen at 0.1 MPa and 300
K.
Neglect the weight of the balloon
itself and find what force it applies
to the basket?
What is its force when the balloon is
filled with helium?
Air, hydrogen and helium satisfy the
perfect gas equation, pv = RT.
Figure P3.13
3.14 The float shown in Fig. P3.14 is
used to control the specific density
of brine with = 1100 kg/m
3
. It is
desired that variations of 1% in the
specific density manifest them-
selves by variations of 1 cm in the
height of the dry part of the float.
For d = 4 mm, what is the weight
of the float?
Figure P3.14
3.15 A cylinder with a bottom section made of lead (
lead
= 11,370 kg/m
3
) is
filled with water, as shown in Fig. P3.15. The total volume of the cylinder is
15.7 liters and its weight together with the water is 16.5 kg. The cylinder is
placed in a solar pond. The pond density varies as = 1+0.01y g/cm
3
( y is
in m). Find the vertical location of the cylinder.
Figure P3.15
88 Fluid Mechanics
3.16 A wooden cylinder ( = 700 kg/m
3
)
is shown in Fig. P3.16. Calculate
the forces and moments that act on
the cylinder (per unit length). Will
the cylinder rotate? Would it if it
were an elliptical cylinder? Assume
that the fluid does not flow from
the tank through the seal.
Figure P3.16
3.17 An open tank shaped as a truncated cone is filled with oil to its top. The
tank is suspended from a rope and is partly immersed in water, as shown in
Fig. P3.17. A piece of wood floats in the oil, as shown. Calculate the force
F exerted by the rope on the tank assuming the weight of the empty tank is
negligible.
Figure P3.17
3.18 The diving bell in Fig. P3.18 is lowered
into the water. Its weight in air is
W = 4,800 kg, and L = 2 m, D = 10 m,
d = 2 m. The air inside the bell is com-
pressed isothermally, i.e., pv = const.
The atmospheric pressure is p
o
= 10
5
N/m
2
. Neglect the volume of the metal
of which the bell is made.
How high will the water level rise
inside the bell? Will the bell sink by
itself? Once sunk, will the bell float by
itself?
Figure P3.18
3. Fluid Statics 89
3.19 A steel bell (
s
= 8,000 kg/m
3
) is shown
in Fig. P3.19. The weight of the bell in
air is W = 8,000 kg. Its inside height is
3 m, and it contains 6 m
3
of air. The bell
is lowered into the sea until its bottom is
10 m below water level. The air inside
the bell is assumed to be compressed
isothermally.
a. Find the tension in the cable which
holds the bell.
b. A window of a 60
o
circular arc sector
is fitted in the bell as shown in the
figure. Find what force acts on the
window.
3.20 A man named Archimedes had a 100-g sphere which looked like gold. He
balanced the sphere on a scale using a cube of gold, as shown in Fig. P3.20,
and then took a bath. Being absent-minded, he took the scale with him into
the water. Now he suspects the sphere to be gold-plated silver. How much
silver should he add beside the sphere to have the scale balance in water?
gold
= 19,300 kg/m
3
,
silver
= 10,500 kg/m
3
.
Figure P3.20
3.21 A boxlike barge is shown in Fig. P3.21. When loaded, D = 1 m, B = 5 m and
C = 3.5 m. The barge rolls to the angle o = 15
o
. Assuming the load to be
well secured, what is its righting moment? Assuming the load to be like
water, with a free surface, as shown by b in the figure, what is the righting
moment? How is the last result modified if the barge had a vertical partition
dividing it into two equal halves?
Figure P3.19
90 Fluid Mechanics
Figure P3.21
3.22 What is the restoring torque of the body that is shown in Fig. P3.22? Sketch
the torque as a function of u, M = M(u). Assume all other information that
you need.
Figure P3.22
3.23 A safety gate is constructed as shown in Fig. P3.23. The width of the gate is
1 m, b = 1 m, h = 4 m. Find the moment the water exerts on the gate. At
what water height h
1
will the gate open?
h
b
b
p
p
o
o
1
F
= 50,000 N
water
.
h
Figure P3.23 A safety gate.
3. Fluid Statics 91
3.24 A rectangular tank containing oil accelerates at a = 0.2 g, as shown in Fig.
P3.24. Once the oil is again at rest (with respect to what?), what is o?
(Ignore the manometer inside the oil.)
Figure P3.24
3.25 The tank in Fig. P3.24 accelerates at constant a. From the reading of the
mercury manometer, what is o? What is a? What will the manometer show
when the experiment is repeated without the oil?
3.26 A pipe inclined at an angle o = 30
o
is
closed at the bottom, as shown in
Fig. P3.26. It is filled with water and
then rotated with e = 8 rad/s.
What is the pressure, p
B
, at the bottom
of the pipe?
What is the pressure, p
o
, at the axis of
rotation?
3.27 A cart on wheels accelerates freely,
because of gravity, on a slope,
Fig. P3.27. Neglecting friction, find at
what angle | would a fluid rest,
relative to the cart, for o = 30
o
,
o = 50
o
.
Figure P3.26
92 Fluid Mechanics
Figure P3.27
3.28 A cylindrical tank is filled with oil, as shown
in Fig. P3.28. A U-tube filled with mercury is
fitted in the tank as shown. The tank and the
oil rotate as a solid body about the vertical
axis. Determine the angular speed of the tank.
Given:
oil
= 900 kg/m
3
,
M
= 13,600 kg/m
3
.
3.29 A circular hollow top, Fig. P3.29, is filled
with water. A small hole is made at point A.
The top spins about its vertical axis with an
angular velocity of 10 rad/s; H = 1 m,
R = 0.75m.
Find the pressure on the internal surfaces, 1,
2 and 3.
Find the forces acting on the seams con-
necting the flat plate 1 with the cylindrical
shell 2 and that with the conical shell 3.
Figure P3.29
3.30 A cylindrical bell, Fig. P3.30, has a mass of 1,500 kg and an inner volume of
6 m
3
while just touching the water surface. The bell inner height is H = 3 m.
The bell is pushed downward with the initial speed q m/s, and the air inside
is assumed to be compressed isothermally, i.e., with pV = const. Neglect the
volume of the metal from which the bell is made, and consider the hydrody-
namic resistance of the water to the motion of the bell an additional margin
of safety. Find the maximal value of q which still permits the bell to float
back by its own buoyancy.
Figure P3.28
3. Fluid Statics 93
H = 3 m
q
V = 6 m
3
H= 3 m
q
V= 6 m
3
Figure P3.30
3.31 Figure P3.31 shows a chamber designed to test underwater windows. The
windows can be set in three orientations as shown: horizontal (A), vertical
(B) and at an angle o to the vertical (C). The windows themselves are plane
and have the shapes shown in Fig. P3.31a - e. Find the total force exerted on
the windows, its direction and its point of application. Solve the problem for
all the windows and for each of the three orientations.
a. b. c. d. e.
Figure P3.31 A chamber for testing windows.
94 Fluid Mechanics
3.32 Repeat Problem 3.31 for the case of the window
given in Fig. P3.32. Note, that while for the
shapes in Fig. P3.31 the force and its point of
application may be computed either by direct
integration or by making use of the shape's
moment of inertia together with Steiner's theo-
rem, for the shape of Fig. P3.32 direct integra-
tion is extremely difficult.
3.33 A research minisubmarine is designed to operate at a depth of 500 m in the
sea. The density of seawater is = 1050 kg/m. The glass windows of the
submarine are round, with a diameter of 0.3 m, and flat, to prevent optical
distortion. The submarine operates at all orientations, i.e., the windows may
look downward, sideways or in any other direction. Find for what force and
for what moment must the windows be designed.
3.34 Considerations of strength led engineers to consider spherical shells made
of glass for the windows of the submarine in Problem 3.33. The wet part of
the windows has the shape of a section of a sphere - all variations are con-
sidered, from a flat plane to half a sphere. Find the forces and the moments
for the designs of these windows.
3.35 A cylindrical tank containing wa-
ter ( = 1000 kg/m), oil ( = 850
kg/m), some ice and a piece of
wood is set on a scale as shown in
Fig. P3.35. At the bottom of the
tank there is a circular hole with
the diameter of 0.3 m blocked by a
plug. The bottom area of the tank
is 2 m
2
.
a. Find the force needed to keep
the plug in the hole.
b. Find the weight of the tank as
shown by the scale. Neglect
the weight of the empty tank.
3.36 A wharf is protected against being rammed by a ship by a bumper connected
to a float as shown in Fig. P3.36. When no ship touches the bumper, the
force in the chain is 20 kN.
Figure P3.32
Figure P3.35
Oil
Water
Ice
Wood
Scale
e
1 m
2 m
3. Fluid Statics 95
1m
0.6m
1m
Ship
Pivot
Wharf
6m
Chain
Figure P3.36
a. Find the maximal resistance force the bumper can offer.
b. Find the amount of energy absorbed by the bumperfloat system before
this maximal force appears.
c. Find the amount of energy taken from the moving ship as it pushes in the
bumper by 0.4 m.
96 Fluid Mechanics
97
4. FLUIDS IN MOTION INTEGRAL
ANALYSIS
Moving fluids are subject to the same laws of physics as are moving rigid
bodies or fluids at rest. The problem of identification of a fluid particle in a
moving fluid is, however, much more difficult than the identification of a solid
body or of a fluid particle in a static fluid. When the fluid properties, e.g., its
velocity, also vary from point to point in the field, the analysis of these properties
may become quite elaborate.
There are cases where such an identification of a particle is made, and then
this particle is followed and the change of its properties is investigated. This is
known as the Lagrangian approach. In most cases one tries to avoid the need for
such an identification and presents the phenomena in some field equations, i.e.,
equations that describe what takes place at each point at all times. This differential
analysis, which is called the Eulerian approach, is the subject of the next chapter.
There exist, however, quite a few cases where there are sufficient restrictions
on the flow field, such that while the individual fluid particles still elude identifi-
cation, whole chunks of fluid can be identified. When this is the case, some useful
engineering results may be obtained by integrating the relevant properties over the
appropriate chunk. This approach is known as integral analysis.
In most cases where integral representation is possible, there are walls
impermeable to the fluid. The moving fluid which cannot pass through these walls
must flow along them, and while doing so its location is better kept under control.
We call the volume enclosed by these walls a control volume.
We begin this chapter with the definition of a control volume and with rela-
tions between thermodynamic systems and control volumes. We then present the
Reynolds transport theorem and apply it to obtain integral relations of conservation
of mass and Newton's second law of motion for a control volume.
98 Fluid Mechanics
Thermodynamic Systems and Control Volumes
A very useful concept in integral analysis is that of the control volume, which
consists of an enclosure with well-defined impermeable walls and well-defined
openings. The walls may deform and the openings may change dimensions, but as
long as they remain well defined, the enclosure may still serve as a control volume.
The control volume may move in space and may even accelerate. It is the choice of
a proper control volume which determines whether a particular problem can be
treated successfully by integral methods.
The three physical laws most important in fluid mechanics are conservation
of mass, Newton's second law of motion, and the first law of thermodynamics. All
three laws are formulated for a thermodynamic system which is classically defined
as a specified amount of matter with well-defined boundaries. In a moving fluid
such a thermodynamic system will, in general, continuously change its location
and shape and will be hard to follow.
The object of the general considerations presented here is to reformulate two
of the three laws mentioned above, i.e., for mass and momentum, in terms conve-
nient to apply for control volumes. This control volume formulation is done by the
application of the Reynolds transport theorem, which states what takes place when
a thermodynamic system moves through a control volume.
Reynolds Transport Theorem
Let a thermodynamic system be chosen inside the fluid, Fig. 4.1. At the time t
it occupies the volume V. Let B be some general extensive property of the thermo-
dynamic system inside the volume V and let b be its specific value, that is, the
value of B per unit mass; then
B = b dV .
V
}
(4.1)
The theorem deals with the rate of change of B with time. We are interested in this
because for conservation of mass B is identified with the total mass of the thermo-
dynamic system, and therefore its rate of change in time vanishes, and for B identi-
fied with, say, the x-component of the total momentum of the system, Newton's
second law demands that the rate of change in time of this component equals the x-
component of the resultant force acting on the system.
We note that B belongs to the system and that the system may change its
location and its shape in time. Had our system been a rigid body, say a steel ball,
4. Fluids in Motion Integral Analysis 99
the notation dB/dt might have sufficed because it would be clear that we talked
about the steel ball. However, for a fluid system it is useful to emphasize that the
rate of change of B is sought while following the system. This is done by writing
DB/Dt, which is called the material
derivative of B.
Thus the rate of change of the property B is
DB
Dt
=
D
Dt
bdV .
V
}
(4.2)
For example, conservation of mass states that a given system cannot change
its mass. This may be expressed by substituting B = m and b = 1 in Eq. (4.2),
leading to
Dm
Dt
=
D
Dt
dV = 0
V
}
.
(4.3)
In Eq. (4.2) the domain of the integration, V, depends on time because the
considered thermodynamic system changes its location and its shape. These
changes are not arbitrary. They are determined by the flow field, and the velocity
vector is expected to appear in the expressions for the rate of change of B.
Figure 4.1 Thermodynamic system moving in velocity field.
100 Fluid Mechanics
Consider the system of fluid inside the volume V(t) shown in Fig. 4.1, which
has the extensive property B as defined by Eq. (4.1). The property B may depend
on time and location. We look now for the rate of change of B, as expressed by
Eq.(4.2), while the system moves with the flow field. This rate of change is DB/Dt.
In Fig. 4.1 we see the system at the time t and at the time t + At. Since our
considerations are for the limit of vanishing At, there is always a region common to
both locations of the system, as designated by A in Fig. 4.1. As we see, at the time t
the system occupies the volume A + A', while at t + At it extends over A + A". The
common part, A, is, of course, the same at t and at t +At, but the specific property b
in it may be different because b depends also on time. Thus the rate of change of B
becomes
( ) ( ) ( ) ( ) ( ) ( )
" '
0 0
lim lim
A A A A
t t
B t t B t B t t B t t B t B t DB
Dt t t
A A
+A +A + +A
= =
A A
or
( ) ( ) ( ) ( )
" '
0 0
lim lim .
A A A A
t t
B t t B t B t t B t DB B
B
Dt t t t
A A
+A +A c
= + = +
A A c
(4.4)
The first term on the right-hand side of Eq. (4.4) gives the change of B at
constant position. This result is what would be expected had the fluid been at rest.
This term is just
cB
ct
=
c
ct
V
}
bdV =
V
}
c
ct
b ( )dV ,
(4.5)
where the order of the differentiation and the integration may be interchanged
because V is fixed in space during both operations.
The second term, B , expresses the contribution due to the flow of the fluid.
Thus the total change in the property B of the system that originally occupied
regions A + A' is due to the time change of B inside A plus the flux of part of the
system with its property B into region A" minus the flux that has left region A'.
Referring to Fig. 4.1, it is noted that the volumetric flow rate of fluid, dQ,
passing through a differential area dS is the product of the area and the velocity
component normal to the area:
dQ = q
n
dS = q dS coso =q ndS .
(4.6)
A positive
q n
denotes outflow, i.e., a flow from A to A", while a negative
q n
denotes inflow, i.e., flow from A' to A. The flux through dS is
4. Fluids in Motion Integral Analysis 101
dB bdQ b dS = = q n
and the total contribution
( ) ( )
" '
0
lim
A A
S t
B t t B t
B b dS
t
A
+A
= =
A
}
q n (4.7)
Thus the total rate of change of B for a system, DB/Dt , is
DB
Dt
=
cB
ct
+ bqndS ,
S
}
(4.8)
DB
Dt
=
c
ct
bdV + bq ndS
S
}
V
}
(4.9)
and by Eqs. (4.1) and (4.5)
D
Dt
b ( )dV
V
}
=
c
ct
b ( )dV + bq ndS .
S
}
V
}
(4.10)
Equations (4.8) - (4.10) state the fact that the rate of change of B for a given
material is due to the change of B inside a volume V(t) plus the net outflow of B
through the boundaries of the volume. Equations (4.9) - (4.10) are known as the
Reynolds transport theorem.
Suppose we have a control volume with the volume V. Equation (4.9) now
states that the rate of change of the extensive property B of the thermodynamic
system, which just fills the control volume, is obtained by the rate of change of B
inside the control volume plus the flux of the property B from the control volume.
This flux can take place, of course, only through the openings of the control
volume.
Control Volume Analysis of Conservation of Mass
In many practical cases, the details of the flow field are not known. Such
cases arise either because these details cannot be obtained or because they are not
really required. Still, an overall inventory of the mass for a given volume is de-
sired. This information can be obtained by an integral mass balance performed for
the control volume using the Reynolds transport theorem.
Let B in Eq. (4.9) be the mass of the system located inside the control volume
at the instant of consideration. The mass per unit mass, to be substituted for b, is
just l, and Eq. (4.9) becomes
Dm
Dt
=
c
ct
dV + q n dS .
S
}
V
}
(4.11)
Conservation of mass requires that the system has the same mass, m, at all times,
thus
102 Fluid Mechanics
Dm
Dt
= 0,
which upon substitution into Eq. (4.11) yields
0 =
c
ct
dV + qn dS .
S
}
V
}
(4.12)
Let the surface S which encloses the thermodynamic system be split into
S = S' +S
o
,
where S' is where the system actually touches the walls of the control volume,
through which there is no flow, and S
o
denotes the openings of the control volume.
Obviously
q n dS
S
}
= 0
and Eq. (4.12) becomes
c
ct
dV = qn dS .
S
o
}
V
}
(4.13)
Equation (4.13) is general and holds for deformable, moving and accelerating
control volumes, provided that the velocity q is always taken relative to the differ-
ential surface n dS. Equation (4.13) may be rewritten in a slightly different form by
noting that
dV = m
cv
V
}
,
where m
cv
is the total mass inside the control volume. Hence
cm
cv
ct
= q n dS .
S
o
}
(4.14)
The surface integral in Eq. (4.14) stands for the total mass flowrate leaving
through the control surface. Thus, Eq. (4.14) indicates that the rate of mass reduc-
tion inside the control volume equals the total rate of mass leaving through the
control surface.
Both Eqs. (4.13) and (4.14) require the details of the velocity field to enable
integration over the control surface. However, in many cases there is just a finite
number of well-defined openings in the control volume, e.g., several pipes feeding
and emptying a water tank, and the integration over the surface excluding these
openings contributes nothing. When the flow through the cross section of each of
these openings is uniform, i.e., (q
n
)
i
= const, Eqs. (4.13) and (4.14) simplify to
c
ct
dV
V
}
= q nA ( )
i
i=1
(4.15)
4. Fluids in Motion Integral Analysis 103
or
cm
cv
ct
= q
n
A ( )
i
i=1
,
(4.16)
where the area S
o
now consists of the sum of A
i
, and where q
.
n = q
n
is the compo-
nent of the velocity vector normal to the corresponding area A
i
.
Example 4.1
A solid-fuel rocket has a nozzle with a critical cross-sectional area of
A
2n
= 0.01 m
2
, Fig. 4.12 The average velocity of the gas flowing through the critical
section is 880 m/s, relative to the rocket, and its density is 1.4 kg/m
3
. As the rocket
starts its vertical motion up at sea level, its mass is 300 kg. The solid fuel is used
up after 20 s. Find the mass of the rocket after 20 s.
Solution
The given gas velocity is relative to the critical cross section. Therefore,
Eq. (4.16) yields
cm
cv
ct
= q
n
A
c
= 1.4 880 0.01= 12.32kg / s
or
dm
cv
= q
n
A
c
dt = 12.32 dt .
Integration yields
m
cv
20
=m
cv
0
+ 12.32 ( )
0
20
}
dt =300 +20 12.32 ( ) = 53.6kg .
Conservation of Mass under Steady State Conditions
An important special case of Eqs. (4.15) and (4.16) is that of steady state, i.e.,
time independent flow. At steady state there is no change with time of any property
at any point inside the control volume, that is, /t = 0. Thus, for example, the
pressure, p, at a given point remains constant at steady state, although there may be
a pressure drop, Ap, between two given points. This pressure drop, again, does not
change with time at steady state. For steady state Eq. (4.16) simplifies to
i
V
i
A
i
= 0
i=1
(4.17)
and for incompressible flows, where = const,
104 Fluid Mechanics
V
i
A
i
= 0
i=1
. (4.18)
In Eqs. (4.17) and (4.18), A
i
is the cross-sectional area normal to V
i
. The terms in
the sums of Eqs. (4.17) and (4.18) should be taken algebraically with their proper
signs, positive for outgoing flows and negative for incoming flows.
There are many examples of control volumes to which Eqs. (4.17) and (4.18)
apply and which have only two openings. Such examples are pipe sections,
nozzles, pumps, compressors, turbines, etc. Here the mass flowrate, m, through
the control volume is constant and Eq. (4.17) becomes
1 1 1 2 2 2
. V A V A m = =
(4.19)
Equations (4.17), (4.18) and (4.19) can be modified to apply to cases where
(q)
i
is not uniform provided that some average mass velocity
G
i
= (q)
i
can be
found. For the average values
G
i
= q
( )
i
=
1
A
i
A
i
}
q n dA .
(4.20)
Equation (4.17) becomes
q
n
( )
i
A
i
= 0
i=1
.
(4.21)
For constant density flows, can be canceled out, and the average velocity,
V
i
, used in Eq. (4.18) is defined in a way similar to Eq. (4.20),
V
i
=
1
A
i
q n dA
A
i
}
.
(4.22)
It is noted that in many practical situations it is easier to evaluate or to measure
these mean velocities than to obtain velocity distributions over the openings. In
many cases the easiest quantity to measure is the mass flux,
. m VA =
Example 4.2
A cylindrical water tank, shown in Fig. 4.2,
has a cross section of 5 m
2
. Water flows out of
the tank through a pipe of 0.2 m diameter with
an average velocity of q
1
= 6 m/s. Water flows
into the tank through a 0.1-m-diameter pipe with
an average velocity q
2
= 12 m/s. Define a control
volume and find at what rate the water level
rises in the tank.
Figure 4.2 Water tank and piping.
4. Fluids in Motion Integral Analysis 105
Solution
A convenient control volume is chosen as bounded by the entrance and the
exit to the pipes, by the wetted part of the tank walls and by the free water surface.
From Eq. (4.16)
cm
ct
|
\
|
.
|
cv
= A
i
q
ni
or
cV
ct
|
\
|
.
|
cv
= A
i
q
ni
= A
1
q
n1
+ A
2
q
n 2
( )
=
t
4
0.2
2
6 0.1
2
12
( )
= 0.094 m
3
/s.
Here q
n2
= 12 m/s is negative because the velocity at the inflow points is in a
direction opposite to that of the outer normal to the surface. The rate of change in
the water level, z/t, is found from
cV
ct
=
c( Az )
ct
= A
cz
ct
= 5
cz
ct
=0.094 m
3
/ s
and finally
cz
ct
=
0.094
5
= 0.0188m/ s = 1.88 cm/ s.
Hence, the water level is reduced at a rate of 1.88 cm/s.
Example 4.3
Water flows in a pipe of diameter d
1
= 0.05 m with an average velocity
V
1
= 0.2 m/s and enters a sprinkler, Fig. 4.3. When the flow starts, the sprinkler ac-
celerates from e = 0 to e = 4 rad/s. The sprinkler pipe ends with two nozzles, each
of diameter d
= 0.005 m. Find the exit speed of water relative to the nozzle, V
2
.
Figure 4.3 Water sprinkler.
106 Fluid Mechanics
Solution
The acceleration is not relevant at all, because the velocity sought is relative
to the nozzle. Assuming that half of the inflow leaves through each nozzle,
Eq. (4.18) yields
V
2
=
V
1
2
|
\
|
.
|
d
1
d
2
|
\
|
.
|
|
2
= 0.1
0.05
0.005
|
\
|
.
|
2
=10 m / s.
Example 4.4
A two-dimensional water flow
divider is shown in Fig. 4.4. The flow in
through opening A varies linearly from 3
m/s to 6 m/s and is inclined with an angle
o = 45
o
to the opening itself. The area of
the A opening is 0.4 m
2
/m, that of B is 0.2
m
2
/m, and that of C is 0.15 m
2
/m. The
flow at B, normal to the opening varies
linearly from 3 m/s to 5 m/s, and the flow
at C, normal to the opening, is approxi-
mately constant. Find q
c
.
Solution
The flow is incompressible, and from Eq. (4.14)
A
}
q n dA = q
A
A
A
cos45
o
=
(3+ 6)
2
0.4 0.7071= 1.273 m
3
ms,
B
}
q n dA = q
B
A
B
=
(3+ 5)
2
0.2 = 0.8 m
3
ms.
Hence,
q
C
A
C
= 1.273 0. 8 = 0. 473 m
3
m s,
q
C
= 0. 473 / 0.15 = 3.153 m / s.
The Momentum Theorem for a Control Volume
The forces exerted by moving fluids on rigid bodies are important in many
fields of engineering. Examples of such forces are those acting on airplane wings,
hulls of ships, turbine blades, pipe bends and weather vanes. These forces can be
computed either by integrating the stresses at the boundaries or by an integral
Figure 4.4 Flow divider.
4. Fluids in Motion Integral Analysis 107
approach utilizing directly Newton's second law of motion. In many cases the
integral approach, which relies on control volume analysis, is easier to use.
Newton's second law of motion applied to a thermodynamic system of fluid
states that the sum of all the forces acting on the system, i.e., body and surface
forces, equals the rate of change of the momentum of this system:
F
B
+ F
S
=
D
Dt
q
V
}
dV .
(4.23)
Substitution of the body force, F
B
, and the surface force, F
S
, in Eq. (4.23) results in
a momentum equation of the form
g dV
V
}
+ T dS
S
}
=
D
Dt
q
V
}
dV , (4.24)
where g is the general body force per unit mass and T is the stress at the system
boundary.
Consider now the x-component of this momentum equation, still written for
the thermodynamic system. For this case Eq. (4.24) becomes
g
x
dV
V
}
+ T
x
dS
S
}
=
D
Dt
u dV
V
}
(4.25)
in which u is the x-component of the velocity vector, q = iu + jv + kw.
We would like to apply the Reynolds transport theorem to the integral on the
right-hand side of Eq. (4.25). Before doing this, it is noted that Newton's second
law of motion requires that the coordinate system in which Eqs. (4.23), (4.24) and
(4.25) hold be an inertial system, i.e., the vector q = iu + jv + kw is given in an
inertial coordinate system. It is also noted that q
n
= q
.
n which appears in the
Reynolds transport theorem, e.g., in Eqs. (4.9) and (4.13), is the velocity relative to
the control volume. Finally, we want a momentum theorem that applies to acceler-
ating control volumes as well as to ones which may be considered inertial systems.
Therefore, to remove any possible confusion, the notation q
r
.
n will be used to
remind one that this velocity is relative to the control volume, while q elsewhere is
absolute, given in an inertial coordinate system.
The Reynolds transport theorem, Eq. (4.9), is now applied to the property
B = mu and b = u, in Eq. (4.25), resulting in
D
Dt
u dV =
c
ct
u dV
V
}
+ u(q
r
n) dS
S
}
V
}
.
(4.26)
Comparison of Eqs. (4.25) and (4.26) yields another form for the x-component of
the momentum equation,
108 Fluid Mechanics
g
x
dV + T
x
dS
S
}
=
c
ct
u dV
V
}
+ u(q
r
n) dS .
S
}
V
}
(4.27)
On the right-hand side of Eq. (4.27) the first term stands for the instantaneous
rate of change of the total momentum enclosed inside the volume V, as it is now
(hence the partial /t), while the second term indicates the instantaneous flux of
momentum through the surface S, which encloses V. These two operations are
control volume operations. Thus Eq. (4.27) makes control volume analysis appli-
cable for Newton's second law of motion.
The y- and z-components of the momentum equation are obtained in a manner
similar to that of the x-component and have the forms
V
}
g
y
dV +
S
}
T
y
dS =
c
ct
V
}
vdV +
S
}
v q
r
n
( )
dS ,
(4.28)
V
}
g
z
dV +
S
}
T
z
dS =
c
ct
V
}
wdV +
S
}
w q
r
n
( )
dS .
(4.29)
The three components of the integral momentum equation, Eqs. (4.27), (4.28)
and (4.29), may now be combined back into a vectorial expression,
V
}
gdV +
S
}
T dS =
c
ct
V
}
qdV +
S
}
q q
r
n ( ) dS . (4.30)
Equation (4.30) is known as the Integral Momentum Theorem for a control
volume. This theorem states that the rate of change of momentum in the control
volume and the outflow of momentum through the control surfaces are balanced
by the action of the body force and by the surface stress on the fluid.
Equation (4.30) holds for any control volume, whether stationary, moving or
deforming. Again it is emphasized that the velocity vector q
r
in the expression for
the differential volumetric flowrate (q
r
.
n) dS should be taken relative to the sur-
face dS, and for a moving or deforming control volume it must be taken relative to
the moving surface.
The surface forces acting on the fluid in the control volume consist of those
exerted by adjacent fluid layers and those acting on the fluid by the solid bound-
aries.
As before, the surface S is split into
S = S' +S
o
,
where S' is the contact surface between the fluid inside the control volume and the
walls of the control volume and S
o
is the area of the openings. At the walls
q
r
n = 0,
and therefore the momentum flux term applies to the openings only, i.e.,
4. Fluids in Motion Integral Analysis 109
q(q
r
n)dS
S
}
= q q
r
n ( )dS
S
o
}
. (4.31)
The surface forces in Eq. (4.30) can also be split into those acting on the solid
walls and those felt over the openings
TdS =
' S
}
TdS +
S
o
}
TdS
S
}
.
(4.32)
The integral over the solid walls equals the total force applied to the fluid system
by these walls of the control volume; let this integral be denoted R, i.e.,
S '
}
TdS = R.
(4.33)
Then, by Newton's third law, +R is the force applied by the fluid to the surface S'
of the control volume.
Combining Eqs. (4.30) and (4.33), an expression for R is obtained,
R =
S
o
}
T dS +
V
}
gdV
c
ct
V
}
q dV
S
o
}
q q
r
n
( )
dS .
(4.34)
It is customary to resolve the stress T on the surfaces of the openings into
normal stress, i.e., pressure, and into shear stress:
S
o
}
TdS =
S
o
}
pn dS + t
S
o
}
dS .
(4.35)
In many cases the last integral which represents shear force on the openings is
very small and may be neglected. In such cases Eq. (4.34) simplifies to
R =
S
o
}
pn dS +
V
}
g dV
c
ct
V
}
q dV
S
o
}
q q
r
n
( )
dS .
(4.36)
Equation (4.36) is used to calculate R, which is the resultant of all the forces
exerted by the fluid passing through the control volume on the inner walls of the
control volume. This equation is known as the momentum theorem for a control
volume.
There are situations in which Eq. (4.36) can be further simplified. One such
situation is where the last integral on the right, i.e., momentum flux through the
openings, may be evaluated by the use of mean values for the velocities. In such a
case we substitute
S
o
}
q q
r
n
( )
dS =
i=1
k
q q
r
n
( )
A
| |
i
,
(4.37)
where the velocities on the right-hand side are mean velocities, A
i
are the opening
110 Fluid Mechanics
areas and the summation is over all openings, say k. Indeed, when no details of the
flow are known, the average velocities are more readily available because their
measurements involve total flow rates only.
Another rather common simplifying situation is where the engineering device
may be classified as a single-inputsingle-output (SISO) device. In such a case
there are only two terms in the summation on the right-hand side of Eq. (4.37). The
two last simplifications, when applicable, change Eq. (4.36) into
R = p
1
nA ( )
1
p
2
nA ( )
2
+mg
c
ct
mq ( )
1
q
1
q
1r
n
( )
A
1
2
q
2
q
2r
n
( )
A
2
,
(4.38)
where average pressures have also been assumed.
Finally, some cases may be one-dimensional. We assign the coordinate x to
the relevant direction and rewrite Eqs. (4.36) and (4.38) in their x-component
terms, i.e., as scalar equations. Hence, Eq.(4.36) becomes
R = p dS
n
S
o
}
+
V
}
g
x
dV
c
ct
V
}
u dV u u
r
( )
dS
n
S
o
}
,
(4.39)
where dS
n
is positive or negative according to the sign of (i
.
n), and u is the veloc-
ity in the x-direction. Equation (4.38) simplifies for a one-dimensional SISO
device to
R =p
1
A
1n
p
2
A
2n
+mg
x
c
ct
mu ( )
1
u
1
(u
1r
)A
1n
2
u
2
(u
2r
)A
2n
,
(4.40)
where the signs of A
1n
and A
2n
are those of (i
.
n
1
) and (i
.
n
2
), respectively, and, of
course, all u values are algebraic, i.e., must have their correct signs. Figure 4.5
depicts a one-dimensional single-inputsingle-output control volume together with
the various terms of Eq. (4.40).
Figure 4.5 One-dimensional single-inputsingle-output control volume.
Example 4.5
Water enters a pipe bend with a uniform velocity, q
1
= 5 m/s. Referring to the
bend shown in Fig. 4.6a, the other known data are
p
1
= 3.37510
5
Pa, d
l
= 0.1 m,
4. Fluids in Motion Integral Analysis 111
p
2
= 1.510
5
Pa, d
2
= 0.05 m.
The size of the bend is such that it contains 4 kg of water between its two
flanges. Find the force, R, exerted on the bend by the water passing through it.
Figure 4.6 Pipe bend and forces.
Solution
The bend is a stationary SISO device; hence Eq. (4.38) may be used. The
control volume is conveniently chosen to consist of the bend itself and the two
planes of the flanges. The flow of the water is incompressible, and Eq. (4.21)
yields
q
1
A
1
= q
2
A
2
,
or
q
2
= q
1
d
1
d
2
|
\
|
.
|
2
=
0.1
0.05
|
\
|
.
|
2
= 20 m/s.
For this uniform flow SISO device Eq. (4.38) yields, for an incompressible
fluid of density ,
R = p
1
nA
1
( )
p
2
nA
2
( )
+ mg q
1
q
1r
n
( )
q
2
q
2r
n
( )
= p
1
A
1
i ( ) p
2
A
2
j ( )+ mg j ( )+ iq
1
2
A
1
+ jq
2
2
A
2
= 2650.72i + 294.52j 39.24j +196.35i + 785.40j.
The total force, R, exerted on the bend by the fluid is thus
112 Fluid Mechanics
R= 2847.07i +1040.68j ( ) N.
The various forces involved are shown schematically in Fig. 4.6b.
Example 4.6
In Example 4.1 the cross-sectional area of the rocket nozzle at the exit is
0.0144 m
2
, and the speed of the exhaust gas is u
e
= 1350 m/s, relative to the nozzle.
The pressure of the gas at the exit is p
e
= 1.110
5
Pa.
The rocket is fixed on a horizontal test bench and fired. Find R
H
, the horizon-
tal component of the force which the gas flowing through the rocket applies to it.
Figure 4.7 Rocket on test bench.
Solution
The coordinate system selected is fixed to the bench. The case is one-dimen-
sional and Eq. (4.40) reduces to
R = p
e
A
e
+u
e
(
e
u
e
A
e
).
Conservation of mass, Eq. (4.19), requires (see Example 4.1)
e
u
e
A
e
=
c
u
c
A
c
=
m = 1.4880 0.01=12.32kg / s.
Thus,
5
( ). 1.1 10 0.0144 1350 12.32 18.216 N.
e e e e e e
R p A u u A = + = + =
Other Forces on the Control Volume
Equations (4.34) (4.40) yield only R, i.e., the force exerted on the control
volume by the fluid passing through it. For a control volume moving in space, with
no atmosphere around it and where all body forces are negligible, R is indeed the
4. Fluids in Motion Integral Analysis 113
only force which the control volume suffers. However, in most cases control
volumes are located within the atmosphere and are subject to body forces. To
obtain the total force which acts on a control volume two modifications are
required: one which accounts for the environment pressure, and another which
deals with body forces.
Body force modifications are considered first. Let the mass of the control
volume itself, i.e., exclusive of the fluid inside it, be m
o
. With the general body
force still denoted by g this modification becomes
F
g
=m
o
g.
(4.41)
The environment atmosphere comes into contact with the outside surface of
the control volume. The outside surface, exclusive of the openings, is denoted by
S", which may be quite different from S', the inner surface of the walls. The stress
at the openings is already accounted for in R. In general the modification due to
the stress on S" is
F
S "
=
S "
}
T dS ,
where T is the stress on the outside surface S".
In many cases the approximation T = pn holds, and then
F
S"
=
S"
}
pn dS .
We note that S" + S
o
, where S
o
represents the openings, forms a closed surface, and
we also note that with p
o
constant,
p
o
n dS = p
o
n dS = 0,
} }
on any surface. Therefore
S"+S
o
}
p
o
n dS =0,
S"
}
p
o
n dS = p
o
n dS ,
S
o
}
and the modification due to the stress becomes
F
S"
=
S"
}
pn dS =
S
o
}
p
o
n dS = p
o
i
(nA)
i
,
(4.42)
where the summation is over all the openings.
We denote the total force which acts on a control volume by F, and then
F = R + F
g
+ F
S"
= R + m
o
g+
S
o
}
p
o
n dS
(4.43)
Substitution of Eq. (4.36) into Eq. (4.43) results in an expression for the total
force, F, which acts on the control volume,
114 Fluid Mechanics
F =
S
o
}
p p
o
( )
n dS +
V
}
g dV + m
o
g
c
ct
V
}
q dV
S
o
}
q q
r
n
( )
dS .
(4.44)
For the special case of a one-dimensional single-inputsingle-output control
volume Eq. (4.40) may be modified to yield the total force as
F =( p
1
p
o
) A
1n
( p
2
p
o
) A
2n
+(m +m
o
)g
x
c
ct
mu ( )
1
u
1
u
1r
A
1n
2
u
2
u
2r
A
2n
,
(4.45)
where the signs of the various terms are determined as in Eq. (4.40).
Example 4.7
Consider the pipe bend in Example
4.5. The mass of the steel walls of the
bend is 6 kg, and the atmospheric pres-
sure is 10
5
Pa. Find the total force which
acts on the bend, i.e., that force that the
bolts at the bend connections must sup-
port.
Solution
From Eq. (4.41),
F
g
m
o
g = 69.81 ( )j = 58.86j.
By Eq. (4.42),
F
S"
= ip
o
A
1
jp
o
A
2
= i 10
5
0.07854
( )
j 10
5
0.001963
( )
= 785.40i 196.35j ( ).
Thus, Eq. (4.43) yields
F = R +F
g
+F
S"
= i 2,847.07785.40
( )
+ j 1,040.68 196.558.86
( )
= 2,061.67i + 785.47j
( )
N.
Example 4.8
We return to the rocket examples, Example 4.1 and Example 4.6. From
Example 4.1 we know that the mass of the rocket and the fuel together is
m + m
o
= 30012.32t kg
and that
Figure 4.8 More forces on pipe bend.
4. Fluids in Motion Integral Analysis 115
cm
ct
= 12.32 kg/s.
We also know that the cross-sectional area of the rocket nozzle at the exit is
0.0144 m
2
, and the speed of the exhaust gas there is u
e
= 1350 m/s, relative to the
nozzle. The atmospheric pressure around the rocket is p
o
= 10
5
Pa.
a. The rocket is fixed on a horizontal test bench and fired, as in Example 4.6.
The pressure of the gas at the exit is p
e
= 1.1 10
5
Pa. The test bench is held in
place by a dynamometer. Find the force read on the dynamometer.
b. Measurements in a wind tunnel show that when the wind speed is 200 m/s the
pressure at the exit section A
e
of the unfired rocket is p
e
= 0.9 10
5
Pa, and the
total resistance force which keeps the rocket in place is F = 16,776 N. The
rocket is then attached to an airplane which flies at 200 m/s and then releases
the rocket. The rocket engine starts at the moment of release. Find whether
the rocket accelerates.
c. The rocket is set with its nose upward and fired. Neglect the aerodynamic
resistance and find its acceleration.
Figure 4.9 Total force on a rocket on a test bench.
Solution
a. We choose to solve this part anew, i.e., not using the results of Example 4.6.
Defining the rocket as a control volume, we apply Eq. (4.45) with A
1n
= A
e
,
p
1
= p
e
and u
1
= u
1r
= u
e
; and obtain the force read on the dynamometer
F = ( p
e
p
o
) A
e
+ u
e
(
e
u
e
A
e
) = (1.11) 10
5
0.0144+12.32 1,350 = 16,776 N.
Note that there is a vertical component of the body force, downward, of
m + m
o
( )g = 30012.32t ( ) 9.81= 2,943120.86t .
b. We start by looking for the total resistance force acting on S", i.e., the outer
surface of the rocket not including its opening. Since the pressure at the exit
section A
e
pushes the rocket forward, the total resistance force becomes
116 Fluid Mechanics
F
S
"
=16,776 + A
e
0.910
5
=16,776+0.0144 0.910
5
=18,072N.
Figure 4.10 Unfired rocket in wind tunnel.
Next we assume the rocket to proceed with the constant speed of release,
200 m/s. If all forces acting on the rocket sum up to zero, it would mean that
indeed this uniform motion is realized. A coordinate system convenient now
is one which moves with the rocket. In this coordinate system R comes out to
be exactly the same as in Example 4.6, i.e., R = 18,216 N. Equation (4.43)
now gives the net force on the rocket as
F
net
= R + F
' ' S
=18,21618,072 =144 N.
The rocket accelerates forward.
Figure 4.11 Fired rocket in flight.
c. The sought acceleration is that of the whole rocket, i.e., that of the fuel and
the structure together. Let this time-dependent acceleration be a. Let the
coordinate system be chosen stationary with x pointing upward. This is also a
one-dimensional case, and Eq. (4.45) states
F = p
e
p
o
( )
A
e
m +m
o
( )
g
c
ct
mu ( ) +
e
u
e
u
( )
u
e
A
e
,
4. Fluids in Motion Integral Analysis 117
where u is the instantaneous velocity of the
rocket, m is the mass of the fuel and m
o
is the
mass of the structure of the rocket:
( ) mu u
um m um ma
t t
c c
= + = +
c c
( ) ( )
hence, with ,
e e e
e o e o e
m u A
F p p A m m g ma mu
=
= +
By Newton's second law
( ) ( )
o e o e o e
F m a p p A m m g ma mu = = +
Thus,
( ) ( ) ( )
o e o e o e
m m a p p A m m g mu + = +
From Example 4.1,
m
o
+m = 30012.32t ,
12.32 m t =
and
5
0.1 10 0.0144 12.32 1,350 16,776
9.81.
300 12.32 300 12.32 300 12.32
a g
t t t
= + =
The velocity of the rocket may also be obtained as
u = a dt
0
t
}
=
16, 776
12. 32
ln
300
300 12. 32t
|
\
|
.
|
9.81t .
This result, of course, is accurate only as long as aerodynamic resistance may
still be neglected, i.e., for small velocities and hence for small t values.
Example 4.9
A single-engine jet airplane flies at the speed of V = 280 m/s. The atmo-
spheric pressure is p
a
= 10
5
Pa. The well-designed air intake of the engine has a
cross-sectional area of A
1
= 0.1 m
2
. The exit cross section of the jet nozzle has the
area of A
2
= 0.3 m
2
. The gas pressure at the exit is p
2
= 1.110
5
Pa. The fuel-to-air
ratio in the engine is 1 : 20. The specific density of the air at the inlet is = 1.2
kg/m
3
. The gas speed at the exit relative to the airplane is U = 650 m/s. What is the
horizontal force transferred by the bolts connecting the engine to the airplane
body?
Figure 4.12 Vertical rocket.
118 Fluid Mechanics
Figure 4.13 Jet engine and control volume.
Solution
Let a control volume be specified as consisting of the jet engine, Fig. 4.13.
Now the jet engine is assumed to be of constant mass and to move in the x-direc-
tion with a constant speed. Moreover, in this case p
1
= p
o
. The horizontal force
transferred to the airplane body can be found from Eq. (4.45), which for this case
simplifies to
F =( p
2
p
o
)A
2n
1
u
1
u
1r
A
1n
2
u
2
u
2r
A
2n
.
In our case the intake air moves with respect to the control volume with a speed
V; hence,
u
1
=u
1r
= V, A
1n
= A
1
,
1
u
1
A
1n
=
1
VA
1
=
m
1
and
u
2r
= U , A
2n
= A
2
.
As the density of the exhaust gas is not given, we find the exit mass flowrate by
using continuity, Eq. (4.17), and by accounting for the added mass of the fuel in
the exit stream,
1 2 2
1.05
n
u A m m = =
Substitution into Eq. (4.45) yields the force that the engine applies in the x-direc-
tion to the airplane body,
( )
( ) ( ) ( )
2 2
5 5
( ) 1.05
0.3 1.1 10 1 10 1.2 0.1 280 1.05 650 280
3.000N+13,524N 16,524N
o
F A p p m U V = =
=
= =
4. Fluids in Motion Integral Analysis 119
Example 4.10
The ancient Romans used a water-hammer pump, Fig. 4.14, to raise water
above the level of their water sources, as they occurred in nature. They let water
run in pipe A from the source. Once steady state was reached, they blocked the
pipe with a wedge W. The water in pipe A had to decelerate, rising in pipe B up to
the height h into an overflow service tank C. Then they used pipe D to lead the
water to their houses.
Find h, approximately, as a function of the length, L, of pipe A, and of the
velocity, u
o
, of the water in it at steady state, i.e., before the wedge is inserted.
Neglect flow losses due to friction, and assume all the pipes to be of the same
diameter.
Figure 4.14 Roman water-hammer pump.
Solution
Let the pipe section between points E and F, Fig. 4.14, which is L meters
long, be chosen as a control volume. Let the average velocity in this pipe at steady
state and just before the wedge is inserted be u
o
. The pressure above the water in
pipe B is atmospheric, and so is, approximately, the pressure at point E. The
difference of the water level, h', between the water source and point F is needed to
accelerate the water to the velocity u.
Once the wedge is inserted, conservation of mass requires that as long as
there is any flow, the water must rise in pipe B.
Denote the water level in pipe B by h. Then
120 Fluid Mechanics
dh
dt
= u.
Since the control volume can have forces applied along its axis by shear
stresses only, and since these are to be neglected, R
x
= 0 and Eq. (4.36) yields
0 =
S
o
}
pn dS +
V
}
g dV
|
\
|
.
|
|
x
c
ct
V
}
q dV
S
o
}
qudS
= A
A
p
o
p
F
( )+ 0 LA
A
du
dt
0
p
F
p
o
= L
du
dt
.
Let p
F
p
o
be measured using a water manometer, i.e., the height of a water
column, and let this manometer be pipe B. This is an approximation because h
depends now on time. Then
p
F
p
o
=gh =L
du
dt
.
One more differentiation yields
dh
dt
=
L
g
d
2
u
dt
2
.
Substitution of dh/dt = u leads to
d
2
u
dt
2
+
g
L
u = 0.
The solution of this linear differential equation yields
u = u
o
cos
g
L
t
|
\
|
.
|
and
h = u
0
t
}
dt = u
o
L
g
sin
g
L
t
|
\
|
.
|
|
,
which attains its maximum at
g
L
t
|
\
|
.
|
|
=
t
2
, i.e., t
max
=
t
2
L
g
.
Hence
h
max
= u
o
L
g
.
Examples are
4. Fluids in Motion Integral Analysis 121
u
o
= 3 m/s
u
o
= 5 m/s
L (m) 100 50 100 50
h
max
(m)
9.5 6.7 16.0 11.2
h'
(m) 0.5 1.3
Here h' is the height needed to accelerate the flow to u
o
. Its calculation is explained
in Chapter 7.
Note: A simpler argument can be offered, although with the same level of approx-
imation, by the use of the Bernoulli equation derived in Chapter 7. The
Romans, however, did not know about this future glorious son of their land,
Bernoulli, while their engineers appreciated very well the effect of inertia.
Just regard their battering rams and note the similarity between a horizontally
moving beam stopped abruptly by a wall and a horizontally moving water
column stopped abruptly by a wedge.
Example 4.11
The nozzle shown in Fig. 4.15 ejects water at the rate of Q m
3
/s. The cross-
sectional area of the water jet is A m
2
. The jet meets a moving vane, B, flows along
its surface and leaves it with the same relative velocity as at the point of their
meeting. The vane moves away from the nozzle with the speed V. The pressure
everywhere is atmospheric. You are an observer moving with the vane. Find the
force exerted on the vane by the water jet; find the power transmitted and the
velocity V at which this power becomes maximal.
Figure 4.15 Water jet and moving vane.
122 Fluid Mechanics
Solution
To solve the problem, we use Eq. (4.40) for a control volume which moves
with the vane at the constant speed of V.
For a fluid of constant density at steady state and a pressure uniform every-
where in the control volume, Eq. (4.40), which gives the force that the jet exerts on
the vane, simplifies to
R = u
1
u
1r
A
1n
+u
2
u
2r
A
2n
| |
.
In our case
A
1n
= A
2n
= A .
The velocity of the jet as it emerges out of the nozzle is U = Q/A, and this
velocity is assumed to be conserved until the jet meets the vane. The jet velocity
relative to the vane at point 1 is
u
r1
= U V .
The relative velocity at point 2, where the flow leaves the vane, is
u
r2
= U V ( ).
As long as the jet velocity remains constant, continuity requires that its cross-sec-
tional area be conserved. Thus
u
r1
A
1n
= U V ( )A = U V ( )
Q
U
, u
r2
A
2n
= U V ( )
Q
U
.
Now, in the inertial coordinate system moving with the vane
u
1
=u
r1
=U V , u
2
=u
r2
U V ( ).
Substitution into Eq. (4.40) yields
R =2
Q
U
U V ( )
2
.
The power transmitted is
P = RV = 2
Q
U
V U V ( )
2
.
This power has its maximum when dP/dV = 0, or
2Q
U
U V ( )
2
4 U V ( )
QV
U
= 0,
which leads to
V =U / 3 .
4. Fluids in Motion Integral Analysis 123
Example 4.12
Repeat Example 4.11 for an observer standing on the ground, i.e., stationary
with respect to the nozzle.
Solution
The control volume is again chosen to move with the vane at the constant
speed V. The velocity of the jet as it emerges out of the nozzle is U = Q / A, and this
velocity is assumed to be conserved until the jet meets the vane. The velocities
relative to the vane are still
u
r1
=U V, u
r2
= U V ( ).
As long as the jet velocity remains constant, continuity requires that its cross-
sectional area be conserved. Thus, once again,
u
r1
A
1n
= U V ( )A = U V ( )
Q
U
, u
r2
A
2n
= U V ( )
Q
U
.
Now, in the inertial coordinate system fixed to the ground,
u
1
=U, u
2
= U V ( )+V =2V U.
Equation (4.40), again without the gravity and the pressure terms, yields
R = u
1
u
1r
A
1n
+ u
2
u
2r
A
2n
| |
= U U V
( )
Q
U
+ 2V U
( )
U V
( )
Q
U
(
(
= 2
Q
U
U V
( )
2
.
The power transmitted is, again,
P = RV = 2
Q
U
V U V ( )
2
,
which again has its maximum at V = U / 3.
Angular Momentum Theorem for a Control Volume
Many applications of fluid mechanics are through the use of rotating
machines, such as turbines, turbo pumps, turbo compressors, propellers, etc. A
natural and convenient choice of the control volume for such a machine is the
machine itself. Due to the large number of such applications, it is helpful to
prepare a form analogous to Eq. (4.36), which is directly applicable to rotating
124 Fluid Mechanics
control volumes. This form is known as the angular momentum theorem.
It is noted that the need for relations directly applicable to rotating coordi-
nates exists in solid mechanics too, and analogous forms exist there. The use of
these relations, which are valid for a thermodynamic system, simplifies the
construction of the angular momentum theorem.
Reference is made to the control volume shown in Fig. 4.1. A center is
chosen in the plane of the figure. This center may be any point, but once chosen, it
must remain fixed. The radius vector from this center to any element of mass in the
thermodynamic system is denoted r.
The angular momentum of the thermodynamic system shown there at the time
t is
r q dV ( )
V
}
.
It is noted that this expression for the angular momentum may be obtained by
cross multiplication of r by q in the expression for the linear momentum: Explic-
itly, since the angular momentum of the thermodynamic system is the sum of the
angular momenta of the mass elements of the system, this cross-multiplication may
be done before integration, i.e., under the integral sign. Proceeding to do this in
Eq. (4.36), it becomes
M
R
= r TdS
S
o
}
+ r g dV ( )
c
ct
r q ( )
| |
dV r q ( ) q
r
n
( )
dS
S
o
}
V
}
V
}
(4.46)
where M
R
is the reaction moment that the fluid exerts on the inner walls of the
control volume.
Because the center from which r stems must be fixed, the direct application
of Eq. (4.46) is limited to rotating systems with a fixed center of rotation. This
fixed center is then the origin for the radius vector r.
When the system is not rotating, any point may be chosen as a center, and
Eq. (4.46) may then be used to calculate moments about that point.
Example 4.13
A small water turbine rotates at n = 200 rpm. The water enters at the center,
at the rate of 2 kg/s, and leaves at the circumference with the relative velocity
q
r
= 17.2 m/s, and with the angle | = 25.8
o
between the relative velocity q
r
and the
circumferential velocity q
c
, Fig. 4.16. Note that | is determined by the geometry
of the vanes of the turbine. The radius of the turbine rotor is r = 0.2 m. Find the
4. Fluids in Motion Integral Analysis 125
absolute velocity at the exit, q
o,
the moment of the turbine and the power.
Solution
q
c
= er =
2tn
60
0.2 = 20.94 0.2 = 4.2 m / s,
q
o
= q
c
+ q
r
.
Using the cosine theorem, Fig. 4.16,
q
o
2
= q
c
2
+ q
r
2
2q
c
q
r
cos| = 4.2
2
+17.2
2
2 4.2 17.2cos25.8
o
,
leading to
q
o
=13.54 m/ s .
By the sine theorem
sin 180
o
o
( )
=
q
r
q
o
sin| =
17.2
13.54
sin25.8
o
and
o = 33.6
o
.
Neglecting air friction, Eq. (4.46) yields
Figure 4.16 Water turbine.
M = r q
o
q
r
n ( ) dS = 1000
S
o
}
0.2 13.54 coso 17.2 Asin |,
but
q
r
n ( )A =100017.2Asin| = 2 kg/s.
Hence,
M=20.213.54cos33.6
o
=4.5Nm,
and the power is
Power = |e
.
M = 20.94 4.5 = 94.2 Nm/s = 94.2 W.
Example 4.14
For the pipe bend considered in Example 4.5, Fig. 4.6, what is the moment
about point A just at the upper edge of flange number 1 for r
12
= 0.15 m?
126 Fluid Mechanics
Solution
Using Eq. (4.42),
r T dS
}
= d
1
/ 2
( )
A
1
p
1
p
2
( )
+ r
12
A
2
p
2
p
o
( )
= 0. 05 0. 05
2
t
( )
3.375 1. 0 ( )10
5
+ 0.15 0. 025
2
t
( )
1. 5 1. 0 ( )10
5
= 108. 00 Nm
and
r q
( )
q
r
n
( )
dS
A
o
}
=
d
1
2
196.35+ r
12
785.4
= 0.05196.35+ 0.15 785.4 = 127.63 Nm.
The mass of the bend together with the water within it is 6 kg, and the contri-
bution of gravity to the moment is thus
mgr
12
= 6 9.810.15 = 8.83 Nm.
The total moment is
M = 108.00 + 127.63 8.83 = 226.8 Nm.
Its direction is counterclockwise.
Example 4.15
Consider again the sprinkler of Example 4.3, Fig. 4.3. The angle of its short
nozzle relative to the radial direction is 45
o
, and the projection of the arm is
r = 0.4 m.
a. Find the moment acting on the sprinkler when the water is just turned on,
neglecting friction and pressure effects.
b. Find the friction moment at steady state, given e = 4 rad/s.
c. Suppose friction disappeared altogether. Find now the angular velocity of the
sprinkler.
Solution
The velocity triangles for the sprinkler are schematically shown in Fig. 4.17,
with
q
o
absolute velocity of water,
q
c
circumferential velocity of sprinkler, and
q
r
relative velocity of water with respect to sprinkler.
q
o
= q
c
+ q
r
.
The mass flux through the two arms of the sprinkler is
4. Fluids in Motion Integral Analysis 127
2
0.005 0.2 1000 0.393 kg/s.
4
m
t
= =
a. Equation ( 4.46 ) yields
( )( ) ( )
o
o r o
A
ds m = =
}
M r q q n r q
q
c
= 0, q
r
= 10 m/s,
Figure 4.17 Velocity triangle for sprinkler.
b.
q
c
= er = 4 0. 4 = 1. 6m/ s, q
r
= 10 m/ s,
q
o
2
= q
r
2
+ q
c
2
2q
r
q
c
cos45
o
, q
o
= 8. 94m/ s,
sin | = q
c
q
o
( )
sin 45
o
, b = 7. 27
o
, 45
o
b = 37. 73
o
,
o
0.4 8.94 sin37.73 2.19,
2.19 0.86 Nm.
f
m
= =
= = =
o
r q
M M
At steady state the moment of friction just balances the moment of the water.
c. As friction disappears, so does the moment of the water. In this case r q
o
= 0,
or | = 45
o
, which requires
q
c
= q
r
/ 2 = 7.07m/ s=er,
e = 7.07/0.4 = 17.68 rad/s = 169 rpm.
128 Fluid Mechanics
Correction Factor for Average Velocity in Momentum
Theorem
To use control volume considerations for the conservation of mass, the inte-
gral in Eq. (4.14) of the mass flux through the opening must be computed, i.e.,
q n ( ) dS
A
o
}
.
To use control volume considerations for the momentum theorem, both the
linear and the angular one, the integrals in Eq. (4.36) , i.e.,
q q
r
n
( )
dS
A
o
}
,
and in Eq. (4.46) , i.e.,
r q ( ) q
r
n
( )
dS
A
o
}
,
must be evaluated.
In many practical cases it seems desirable to use average velocities, rather
than the detailed velocity profiles, which may not be available. This possibility,
however, raises a new problem. Let the term average, as applied here to velocities,
mean that in considerations of conservation of mass, both integration of the
detailed velocity profile and calculations based on the average velocities yield
exactly the same mass flux. Obviously, conservation of mass then needs no correc-
tion factors at all. But suppose that the same average velocities are used to calcu-
late the flux of momentum into the control volume. Are the results thus obtained
the same as those obtained by integrating the velocity profiles? In general the
results are different, and those results obtained using average velocities must be
multiplied by correction factors to yield the correct results.
Let the correction factor for the momentum be denoted |
M
, and let its evalua-
tion be restricted to cases where q, q
r
and n are parallel.
The momentum flux in terms of average velocities is
A
i
i
q
1
2
,
(4.47)
with
q
i
being the average velocity, while the flux obtained by integration is
q q
r
n ( )dS
A
o
}
= q
2
dS
A
o
}
. (4.48)
Hence the correction factor for the momentum flux becomes
4. Fluids in Motion Integral Analysis 129
|
M
=
1
i
A
i
q
i
2
q
2
dS
A
}
.
(4.49)
The correction factors obtained in Examples 4.16 and 4.17 can be used to estimate
errors when the exact velocity profiles are not known.
Example 4.16
The velocity profile in laminar flow through a circular pipe is
2
max
1
r
w w
R
(
| |
=
( |
\ .
(
Find the momentum correction factor for this flow.
Solution
We first find for this flow the average velocity, i.e., the volumetric flowrate Q
per unit cross-sectional area A,
w =
Q
A
=
1
tR
2
= wdA =
1
tR
2
= w
max
1
r
R
|
\
|
.
|
2
(
(
2tr dr
0
R
}
A
}
or
w =
w
max
2
.
The correction factor is obtained from Eq. (4.49),
|
M
=
1
tR
2
w
max
2
|
\
|
.
|
2
w
max
2
1
r
R
|
\
|
.
|
2
(
(
(
2
2tr dr =
4
3
0
R
}
.
Example 4.17
The velocity profile in turbulent flow through a circular pipe is
w = w
max
1 r / R | |
1/ 7
.
Find the correction factor for the momentum for this flow.
130 Fluid Mechanics
Solution
The average velocity is here
w =
Q
A
=
1
tR
2
w dA =
1
tR
2
w
max
1
r
R
|
\
|
.
|
1
7
2tr dr =0.817w
max
0
R
}
A
}
.
The correction factor is found from Eq. (4.49),
|
M
=
1
t R
2
0.817w
max
( )
2
w
max
2
1
r
R
(
(
2
7
2tr dr =1. 020
0
R
}
.
Hence, for turbulent flow the correction factor can usually be neglected.
Problems
4.1 A system of pipes is shown in Fig. P4.1. All pipes have the same diameter
d = 0.1 m. The average flow velocity in pipe number 1 is q
1
= 10 m/s to the
right. In pipe 2, q
2
= 6 m/s to the right. Find q
3
. Can there be a flow with q
2
to the left? What is q
3
then?
Figure P4.1 Three pipes.
4.2 The average air velocity in the intake duct of an air conditioner is 3 m/s. The
intake air temperature is 35
o
C. The air comes out of the air conditioner at
20
o
C and flows through a duct, also at 3 m/s. The pressure is atmospheric at
10
5
Pa everywhere. What is the ratio of the cross-sectional areas of the
ducts?
Assuming that some water has condensed from the air in the air conditioner,
how does it affect the velocity in the outlet duct?
4.3 A pressure vessel is equipped with a small piston of area A
S
and a large
piston of area A
L
, Fig. P4.3. The vessel is filled with oil, and a force F is
applied to the small piston. Neglect friction and differences of heights of oil
and show that conservation of mass and conservation of energy lead to
4. Fluids in Motion Integral Analysis 131
G = F( A
L
/A
S
). This is the principle of the hydraulic lift. For F = 10 N and
A
L
/A
S
= 1,000, find G.
Figure P4.3 Hydraulic lift.
4.4 Compressed air is introduced into the ballast tank of a submarine and drives
the water out of there with the rate of 2 m
3
/s when the submarine is at the
depth of 10 m. The atmospheric pressure is 10
5
Pa. What is the flowrate of
the ejected water when the same mass flux of compressed air is introduced
at the depth of 100 m? Can there be a situation in which, because of depth
and downward motion of the submarine, the rate of water ejection becomes
zero or even negative?
4.5 A steam boiler is fed with water at the rate of 1 kg/s. It supplies steam at at-
mospheric pressure and 105
o
C at the same rate. The steam speed in the pipe
is 10 m/s. What is the pipe diameter?
4.6 A water container has a hole in its side,
Fig. P4.6. The water flows through this
hole with the velocity of q
h
= 5 m/s. The
effective size of the hole is A = 0.1 m
2
.
The pressure around the container is
atmospheric. What horizontal force
acts on the container because of the
hole?
4.7 A firefighter hose ends with a nozzle, as shown in Fig. P4.7. Measurements
show A
F
= 0.01 m
2
, A
E
= 0.0025 m
2
, q
F
= 8 m/s, p
F
= 578,000 N/m
2
,
p
E
= 98,000 N/m
2
and so is the atmospheric pressure. The hose is flexible
and the nozzle is connected to it by a flange at G. What are the magnitude
and direction of the forces acting on the flange? If the hose is not held,
which way will it move?
Figure P4.6 Container with hole.
132 Fluid Mechanics
Figure P4.7 Fireman hose.
4.8 A jet of fluid of absolute velocity q
i
emerges from a nozzle and hits a
moving vane and then turns, Fig. P4.8. The speed of the fluid leaving the
vane relative to the vane, q
er
, equals the speed of the fluid hitting the vane
relative to the vane, q
ir
. The vane moves at the absolute speed V
v
. Derive
expressions for the force that the fluid applies to the vane and for the power
developed. Find V
v
for which the force is the largest and one for which the
power is maximized.
Figure P4.8 Jet, vane and velocities.
4.9 A nozzle moves with the velocity V
n
, Fig. P4.9. It ejects a jet which has the
velocity q
er
, relative to the moving nozzle. The nozzle can swivel to form
different angles | between V
n
and q
er
. Find an expression for the force driv-
ing the nozzle in the direction of V
n
, and the power obtained by this force.
Are there V
n
values which maximize this force, or this power, as in Problem
4.8? Are there | values which do this?
4. Fluids in Motion Integral Analysis 133
4.10 The nozzle in Problem 4.9 is fixed
to the circumference of a rotating
disk, such that it still moves with
V
n
. Find the turning moment and
the power.
4.11 The sprinkler of Examples 4.3 and
4.15 (Figs. 4.3 and 4.17) is set
inside a high-pressure water tank
such that now water enters
through the nozzles at its arms and
exits through its central pipe. The water flow through this central pipe is at
0.6 m/s.
a. Which way does the sprinkler turn?
b. What is the maximum power obtained at 60 rpm?
4.12 A rocket is released from a flying vehicle at a horizontal velocity of 300 m/s.
The rocket engine ignites and sends backward a horizontal jet of gas at a
relative velocity of 280 m/s. The flux of the gas is 3 kg/s, and the mass of
the rocket is 30 kg. The flight takes place at such a height that atmospheric
pressure and aerodynamic resistance are negligible.
a. What is the acceleration of the rocket 1 s after release?
b. What is the velocity of the rocket's jet relative to the ground?
4.13 A ram-jet airplane flies horizontally with a velocity of V = 600 m/s, which is
also the velocity with which it sees air coming into its engine. Fuel is burnt
at the rate of 0.12 kg/kg air. The combustion gas leaves the nozzle at q
r
=
700 m/s relative to the airplane. The air intake area is 0.25 m
2
, and the
nozzle exit area is 0.75 m
2
. The density of the air is
i
= 1.2 kg/m
3
. Find the
driving force and the power used.
4.14 A steel pipe of d = 0.150 m, wall thickness of 6 mm and 1,000 m long is
used to supply water. The water flows at 10 m/s, and the nominal stress
allowed in the pipe wall is 5,000 N/cm
2
(these pipes get rusty). To prevent
high stresses due to water hammer, the valve at the end of the pipe must not
be closed faster than at a certain rate. What is the minimal time required to
close the valve?
4.15 What are the estimated correction factors for Problems 4.6, 4.7, 4.8, 4.9,
4.10, 4.13 and 4.14 ?
Figure P4.9 Moving nozzle and jet.
134 Fluid Mechanics
4.16 The flow rate in the water turbine of Example 4.13 is doubled, to 4 kg/s.
The moment remains the same. Find q
o
, n and the power of the turbine.
Hint: | remains the same, and | q
r
is doubled.
4.17 A floating anchor is a device used in lifeboats to keep the nose of the boat
directed against the waves. It is made of heavy cloth and has the shape of a
cone, with holes at both ends, Fig. P4.17. It is tied to the rear of the boat and
is dragged underwater by the boat, which is itself dragged by the waves and
the wind. The water is thus forced into the anchor at its larger opening and
comes out at the narrow end. Experiments show that the pressure at the exit,
i.e., the narrow end, is slightly below the hydrostatic pressure for this depth,
that the pressure at the wide end is above hydrostatic by about 0.4V
2
,
where V is the speed of the anchor relative to the water, and a good assump-
tion is that the water does not leave the anchor at a relative speed higher
than V.
For d
i
= 1 m, d = 0.4 m, estimate the resistance force of this anchor to rela-
tive speeds of 1 m/s, 3 m/s, 10 m/s, 15 m/s, 20 m/s.
Figure P4.17 Floating anchor.
4.18 A shallow water boat is propelled by a pump as shown in Fig. P4.18. The
inlet pipe, A, has a diameter of 0.4 m, and the average velocity in it is 10
m/s. The outlet pipe, B, has the diameter of 0.2 m. The centers of both pipes
are 1.5 m below the water level and the pressures at these centers are
approximately hydrostatic. Find the force that the pump and its pipes apply
to the boat.
An inventor who did not study fluid mechanics but had such a boat modified
it by taking the outlet pipe 1.5 m into the air, as shown in configuration C in
Fig. P4.18. Is this a good idea?
4. Fluids in Motion Integral Analysis 135
Figure P4.18 A boat with a pump.
4.19 A boat moves in the water at a speed of 10 m/s propelled by a pump which
takes in water at the front (point 1) through a suction pipe with the diameter
of D
1
= 0.25 m and discharges the water at the rear (point 2) through a
discharge pipe with the diameter D
2
= 0.2 m; H
1
= 2 m, H
2
= 2.5 m.
The pressures p
1
and p
2
are the same as the hydrostatic pressures outside the
boat, and the relative velocity V
1
is that of the boat, i.e., V
1
= 10 m/s.
a. Find the total driving force applied to the boat (i.e., the force which is
transferred through the bolts which connect the pump to the boat).
b. Someone suggests to connect a short converging cone at point 2 such
that the exit of the water becomes through an opening of D
3
= 0.15 m.
The pump operates as in a. Will this increase or decrease the speed of
the boat?
c. Someone suggests to connect a short diverging cone at point 2 such that
the exit of the water becomes through an opening of D
4
= 0.25 m. The
pump operates as in a. Will this increase or decrease the speed of the
boat?
Figure P4.19 A boat propelled by a pump.
4.20 The pressure at the entrance to pipe 1 in Problem 4.1 is 200,000 Pa, and that
at the exit of pipes 2 and 3 is 100,000 Pa, which is also the atmospheric
pressure. Find what force is needed to keep the pipe structure from moving.
136 Fluid Mechanics
4.21 A jet of water has the diameter of 0.04 m and
the average velocity of 8 m/s. The water hits a
stationary flat vane, as shown in Fig. P4.21.
The water is assumed to spread at the point of
impact with cylindrical symmetry, and its
velocity is conserved. The pressure every-
where outside the water jet is atmospheric,
and so it is inside the jet well before it hits the
vane. Find the force acting on the vane and
the power extracted by the vane.
4.22 The vane in Problem 4.21 is tilted by the
angle t/4, Fig. P4.22. The water still con-
serves its velocity after hitting the vane.
Now, however, conservation of momen-
tum in the direction tangent to the vane
decides how the flow is divided at the
point of impact. Assume the flow two-di-
mensional and find the force acting on
the vane and the power extracted by the
vane.
4.23 The vane in Problem 4.21 now recedes
from the water jet with the speed of 2
m/s, as shown in Fig. P4.23. The water
now conserves its velocity relative to the
vane after hitting it. Find the force acting
on the vane and the power extracted by
the vane.
4.24 The vane in Problem 4.22 now recedes
from the jet with the speed of 2 m/s, Fig.
P4.24. The water still conserves its veloc-
ity relative to the vane after hitting it. But
now conservation of relative momentum in
the direction tangent to the moving vane
determines how the flow is divided at the
point of impact. Assume the flow two-di-
mensional and find the force acting on the
vane and the power extracted by the vane.
Figure P4.21
Figure P4.22
Figure P4.23
Figure P4.24
4. Fluids in Motion Integral Analysis 137
4.25 A water container is open at its top. It is fitted with an
inlet pipe controlled by valve A and with an outlet
pipe and valve B, as shown in Fig. P4.25. When a
valve is opened, the flow in the corresponding pipe is
1 m
3
/s at the mean velocity of 10 m/s. When both
valves are closed, the container transfers to the floor a
force of 100,000 N. Find what forces are transferred
to the floor a short time after:
a. Valve A is opened.
b. Valve B is opened.
c. Both valves are opened.
4.26 The angle between the course of a sailboat and the wind is o + |. The
velocity of the wind is 15 m/s. The sail has an area of 20 m
2
and is set with
the angle | to the wind direction, as shown in Fig. P4.26. Assume that the
wind is completely turned by the sail so as to become tangent to its back-
ward direction and that the speed of the wind relative to the sail is con-
served. The density of the blowing air is 1.15 kg/m
3
, and the speed of the
boat is 0.5 m/s.
a. Find the force vector the mast transfers to the boat. Note that only the
component in the sailing direction of this force vector serves to advance
the boat.
b. Find | which makes this forward component the largest.
c. For o + | = 60
o
find numerical values.
o
|
D
ir
e
c
tio
n
o
f
S
a
il
in
g
W
in
d
S
a
i
l
Figure P4.26 A sailboat.
Figure P4.25
138 Fluid Mechanics
4.27 Referring to Problem 4.26, show that a sailboat can sail upwind, i.e., when
o + | < t/2. For a given wind, show that the greatest speed that the boat may
achieve is sometimes not directly downwind, i.e., not for o + | = t.
4.28 A pipe bend is shown in Fig. P4.28. The
inner diameter at the entrance flange,
point A, is 0.5 m, and that at the outlet,
point B, is 0.25 m. The mean velocity of
the water at point A is 5.0 m/s. The
manometric pressure in the water at point
A is 187,500 Pa, and the atmospheric
pressure, which is also the pressure at
point B, is 100,000 Pa. The mass of the
pipe section between points A and B is estimated as 10% of that of the body
of water inside the pipe. For bend angles of o = t/6 and o = t/4, find the
mass of water in the bend section which makes the force in the bolts of the
flange at A just horizontal, i.e., there is no vertical component. Does this
mean that there are no moments at point A?
4.29 For the dimensions, pressures and velocities given in Problem 4.28 find the
angle o for which the horizontal force in the bolts of flange A becomes the
largest. Is there an angle o for which this horizontal force vanishes? Is there
an angle for which the bend section pushes to the left against the water
supply pipe?
4.30 The pipe which supplies the water to the
bend in Problems 4.28 and 4.29 is made
from some flexible material and looks
as in Fig. P4.30. The bend angle is
o = 0, i.e., no bend, just a cone. Now the
bend and the flexible supply pipe must
be held to prevent their motion. Find
the force necessary to hold them in
place.
4.31 Figure P4.31 shows a water sprinkler with dissimilar arms. The inlet pipe
has the diameter of 0.02 m, and the mean velocity of the water there, point
A, is 15 m/s. The outlets are nozzles with diameters of 0.003 m. The length
of the longer arm, AB, is 0.3 m, and that of the shorter one, AC, is 0.2 m.
The sprinkler is designed to run at 120 rpm. The water is assumed to divide
equally between the two arms, and the power obtained from the rotor is used
Figure P4.28
Figure P4.30
4. Fluids in Motion Integral Analysis 139
to drive the sprinkler over the field. Find the angles o
1
and o
2
by which the
nozzles must be set to maximize the power obtained. Calculate this power.
o
1
2
o
A
B
C
Figure P4.31 Asymmetric sprinkler.
4.32 A propeller has a diameter of 1.0 m, and when it rotates, it spans a disk,
sometimes referred to as an actuating disk. It is designed for a nominal
operation where a stream of air at atmospheric pressure enters the actuating
disk at 400 km/h and leaves it at 750 km/h, still at atmospheric pressure.
a. Find the thrust of the propeller operating on an airplane flying at
400 km/h.
b. Find the power delivered by it to the airplane.
4.33 A ventilation scoop, Fig. P4.33, is used to venti-
late compartments in ships. One such scoop has
an opening with the diameter of 1 m. The winds
the ship is expected to encounter are not faster
than 100 m/s. Find the upward force for which
the scoop must be designed.
4.34 A cheap blower has the vanes in its rotor
made of flat strips of metal, Fig. P4.34. The
air enters centrally and comes out at the
circumference, with its velocity relative to
the rotor tangent to the vanes. The angle
between the tangent to the vane at the cir-
cumference and the radius is o = t/6. The
inlet conduit has a diameter of r = 0.04 m,
and the velocity of the air there is 20 m/s.
The outer diameter of the rotor is 0.10 m,
and the width of the rotor at the outlet is
0.01 m. The rotor rotates at 3000 rpm. The
density of the air may be taken as constant,
at 1.1 kg/m
3
. Find the power needed to run
the blower.
Figure P4.33
Figure P4.34
140 Fluid Mechanics
4.35 A round jet of water comes straight up from a nozzle in
a water fountain. The jet diameter as it comes out of the
nozzle is 0.002 m, and its velocity there is 20 m/s. A
little boy places a small glass sphere in the jet and en-
joys seeing it balance there, Fig. P4.35. The glass
sphere has a mass of 0.01 kg. Find the diameter of the
water jet just before it hits the glass sphere.
4.36 A light airplane is used to spray cotton fields. The
spray nozzles are directed toward the rear of the air-
plane and the spray comes out at a rate of 100 kg/s and
at a speed of 20 m/s relative to the airplane.
a. Draw the velocity vectors of the spray relative to
the airplane and as seen by an observer on the
ground.
b. Find the thrust added to the airplane by the spray.
c. Find the velocity and the thrust when the nozzles are directed downward.
Figure P4.35
141
5. FLUIDS IN MOTION DIFFERENTIAL
ANALYSIS
Differential Representation
Two conditions must be satisfied to make the integral analysis considered in
the previous chapter useful: The sought information must be such that no details
are required; and enough information must be known a priori to supply numerical
values for all the symbols in the equations of the integral analysis. In many situa-
tions details are meaningful and necessary, and in very many cases, particularly
engineering ones, no a priori information is given.
The difficulty mentioned in Chapter 4, that of identification of fluid particles,
cannot be alleviated any more by an integral approach and the concept of the field
must be introduced.
The field point of view, which is traditionally called the Eulerian approach,
uses the same properties associated with fluid particles, e.g., density, velocity,
temperature, etc., but assigns them to the field. One then considers the field den-
sity, i.e., the density at a particular point (x,y,z ) and at a particular time, or one
considers the field velocity or the field temperature. There is then a flow field, with
all its properties at each point and for any time, and the problem of identification
seems to have been successfully circumvented.
There is still, however, an intermediate step necessary before field analysis
can be applied: the basic laws dealing with mass and momentum, which are formu-
lated for well-defined, and therefore identifiable, thermodynamic systems, must be
translated into expressions which contain field concepts only. An analogous step
has been taken in going from thermodynamic systems to control volumes in
Chapter 4. The next step is, therefore, to go from macroscopic thermodynamic
system considerations to differential relations of the field properties, i.e., to obtain
the differential equations of the field.
The most important field properties are the density, the velocity, the pressure
142 Fluid Mechanics
and the temperature, all of which become now dependent variables, with the
independent variables being the coordinates x, y, z and the time t. The differential
equations of the field now contain these dependent and independent variables and
express the basic laws of physics.
The two differential equations sought are that which expresses conservation
of mass and the other which represents Newton's second law of motion.
We note that both physical laws involved contain rates of change: the rate of
change of the mass of a thermodynamic system is nil, and the rate of change of the
momentum of a thermodynamic system equals the sum total of the forces acting on
the system. We also note that the rates of change in both cases must be evaluated
"while following the system."
Now, such problems have already been addressed in the integral analysis,
using Reynolds transport theorem. We would therefore like to use some results
already obtained for control volumes. We observe that in the integral analysis all
the examples involved nonfluid geometries, such as pipes, airplanes, rockets, etc.,
which conveniently served as control volumes. To apply our integral results to
regions of pure fluid, we must find there natural control volumes. We therefore
start our analysis by showing that indeed the flow field itself provides all the
necessary control volumes. We then take up the concept of the material derivative,
briefly mentioned in Chapter 4, and define it using field terms. Once this has been
achieved, we are ready to derive the differential equations that express the laws of
conservation of mass and change of momentum.
Streamlines, Stream Sheets and Stream Tubes
The velocity in a continuum constitutes a vectorial field, which is defined
everywhere in the fluid, at any moment.
A sufficiently smooth vectorial field admits field lines, defined as continuous
lines tangent to the field vector everywhere. The velocity field lines, i.e., those
lines tangent everywhere to the velocity vector, are called streamlines. For steady
flows the streamline pattern is also steady. In time-dependent flows the streamlines
change in time. Streamlines can intersect only at points of zero velocity, as
otherwise a finite velocity vector would have more than one direction at a point.
Also, because the velocity vector is tangent to the streamline, there is no velocity
component normal to it and, therefore, no flow through it.
To enhance the geometric interpretation of the concept of streamlines, two
auxiliary concepts are introduced: pathlines and streaklines.
A pathline is a line traced by a particle in the fluid. A small particle, such as a
dust speck moving in a clear fluid, can be photographed using a movie camera.
The time history of a single fluid particle obtained this way is the particle pathline.
5. Fluids in Motion Differential Analysis 143
By seeding a fluid with many small particles, a large number of pathlines can be
recorded simultaneously.
A streakline is the locus of all the particles which have passed through a
particular point. A thin hollow needle, held in place in the fluid domain, slowly
releasing a continuous stream of dye, would cause all particles passing at the point
of the needle to be dyed. These particles then form a dyed streakline in the fluid.
In steady flows both pathlines and streaklines coincide with streamlines, and
this is their main usefulness: they both can be used to make streamlines visible. We
now return to streamlines and remember that because the field has the velocity
vector defined everywhere, each and every point has a streamline passing through
it.
a.
b.
c.
System at time t
System at time t + A t
Figure 5.1 a. Stream sheet formed by streamlines intersecting a curve.
b. Stream tube formed by streamlines intersecting a closed curve.
c. Thermodynamic system moving through a stream tube.
Consider a general curve drawn in the fluid domain, Fig. 5.1a. This curve is
chosen not to be a streamline itself; hence it is continuously intersected by
streamlines. All of these intersecting streamlines constitute a stream sheet, which
144 Fluid Mechanics
is a three-dimensional surface with no flow through it. When the originally drawn
curve is closed, i.e., forms a ring, the stream sheet forms a tube with no flow
through its walls, called a stream tube, Fig. 5.1b. The concepts of streamlines,
stream sheets and stream tubes are thus very useful in the description of flows. It is
noted that because there is no flow through the "walls" of a stream tube, it can
conveniently serve as a wall of a control volume. The flow field is thus full of such
partitions available for use whenever necessary. In particular consider the ther-
modynamic system shown in Fig. 5.1c. It is shown at the time t and at the time t +
dt, and because its motion has been affected by the velocity field, it has moved in a
stream tube, i.e., through a control volume.
The Material Derivative
Let two problems be considered simultaneously: that of a steel ball located at
the point (x, y, z ) and that of a similar "sphere of fluid" located at the same point
in another set of coordinates. Let the rates of change in time of the densities of
both systems be investigated. The investigator of the steel ball will probably use
the notation d/dt to describe the rate of change of the ball density. There is no
doubt whatsoever as to which density is considered, and even when several balls
take part in the process, the addition of a subscript, i.e. (d/dt)
i
, will suffice. The
investigator of the moving fluid, however, cannot use this notation without some
additional qualification. A simple d/dt could mean the rate of change of density at
the point (x
1
,
y
1
, z
1
) brought about because at the time t + At it is occupied by a
fluid different from that of the time t; or it could mean the time rate of change of
the density of the small sphere of fluid which was resident at (x
1
,
y
1
, z
1
) at time t
but went elsewhere at t + At. Both possibilities are meaningful, but with quite
different meanings.
The notation adopted in fluid mechanics is such that when a rate of change at
a point is considered, /t is used, while for a rate of change observed while
following the fluid system (system in the thermodynamic sense, i.e., the same
particles), the notation becomes D/Dt. Thus /t is the rate of change of the
density at a point, and it vanishes for time independent flows. The term D/Dt is
the rate of change of the density of a thermodynamic system; this rate need not
vanish for time-independent flows.
The differences between the various derivatives can be explained in a more
formal manner as follows: Consider a fluid particle moving with a local velocity:
q = iu + jv + kw , (5.1)
and let the change of the property b = b (x,y,z,t) of the particle be investigated.
5. Fluids in Motion Differential Analysis 145
The change in b with time and position may be expressed as
db =
cb
ct
|
\
|
.
| dt +
cb
cx
|
\
|
.
| dx +
cb
cy
|
\
|
.
|
|
dy +
cb
cz
|
\
|
.
| dz ,
(5.2)
and the rate of change of b in time, db/dt , is
db
dt
=
cb
ct
|
\
|
.
| +
cb
cx
|
\
|
.
|
dx
dt
+
cb
cy
|
\
|
.
|
|
dy
dt
+
cb
cz
|
\
|
.
|
dz
dt
.
(5.3)
Now
dx
i
/ dt is the rate of change in the coordinate x
i
of the particle and is there-
fore equal to q
i
. This leads to the definition of the material derivative, i.e., the
derivative taken while following the fluid motion,
Db
Dt
=
cb
ct
+ u
cb
cx
+ v
cb
cy
+ w
cb
cz
,
(5.4)
which may also be written, using vector notation, as
Db
Dt
=
cb
ct
+ q Vb.
(5.5)
For the case of a static fluid, Eq. (5.5) becomes
Db
Dt
=
cb
ct
(5.6)
while for time-independent flows it takes the form
Db
Dt
= q Vb = u
cb
cx
+ v
cb
cy
+ w
cb
cz
.
(5.7)
Equation (5.5) is quite general and holds in any system of coordinates. Its detailed
form, using cartesian coordinates, is Eq. (5.4). Analogous forms in other systems
of coordinates may be obtained by either coordinate transformations, from the
cartesian to the desired ones, starting with Eq. (5.4), or by the direct operation of
b V q
in the new coordinates. Thus Eq. (5.5) in cylindrical and spherical coordi-
nates, respectively, becomes
Db
Dt
=
cb
ct
+ q
r
cb
cr
+ q
u
cb
rcu
+ q
z
cb
cz
(5.8)
and
Db
Dt
=
cb
ct
+ q
r
cb
cr
+ q
u
cb
rcu
+ q
|
cb
r sinu c|
.
(5.9)
146 Fluid Mechanics
The basic laws used in thermodynamics and mechanics are formulated for
thermodynamic systems. When these laws are applied in fluid mechanics, the point
of view must still be that of the system, i.e., while following the particle, and the
operator D/Dt is used. This operator is defined by Eq. (5.5), in which b may be any
scalar property.
The material derivative may also be defined for a vector quantity. Formally
one may write for a vector A
DA
Dt
=
cA
ct
+ q V ( )A (5.10)
in which the operator ( ) V q is taken literally as meaning: perform the scalar prod-
uct of the vector q and the operator V in the particular coordinate system and use
the resulting terms as an operator on the vector written to the right of the paren-
theses. The form of Eq. (5.10) is quite similar to that of Eq. (5.5), expressing the
material derivative of a scalar. However, b V q in Eq. (5.5) is a common vectorial
operation; whereas ( ) V q A had to be specifically defined here.
For the particular case of A = q, Eq. (5.10) can also be written in terms of
common vectorial operations, yielding an expression for the material derivative of
the velocity vector of a fluid particle,
Dq
Dt
=
cq
ct
+
1
2
V qq ( ) q Vq ( ). (5.11)
In cartesian coordinates this vector may be written as
Dq
Dt
=
cq
ct
+ u
cq
cx
+ v
c q
cy
+ w
c q
cz
.
(5.12)
In dynamics of solids the time rate of change of the velocity of a particle,
taken while following the particle, i.e., the material derivative, has the simple
interpretation of the acceleration vector of the particle. Considerations of Newton's
second law of motion, applied to a system, will show this interpretation to be
indeed valid in fluid dynamics too, i.e., to give the correct acceleration vector in
the field.
The operator D/Dt has already been used in the formulation of the Reynolds
transport theorem, in Chapter 4, with the same interpretation, i.e., that of while
following the system. We note, however, that there it has been operating on
extensive quantities, i.e., on integrals over the thermodynamic systems. Here it
applies to local properties, and its detailed forms are therefore different.
Example 5.1
The price of fruit, P, is given by
P = C + Ax Bt ,
5. Fluids in Motion Differential Analysis 147
where x is the distance from its place of growth and t is the time that has elapsed
since it was picked. Find the rate at which the price of fruit changes while it is
being transported at a velocity v.
Solution
The rate of change in P is
DP
Dt
=
cP
ct
+
dx
dt
cP
cx
= B +
dx
dt
A.
In our case dx/dt = v and thus, DP/Dt = -B + vA. For the special case when no
transportation takes place, v = 0 and
DP
Dt
=
cP
ct
= B
while for fruit which does not spoil easily, B ~ 0 and DP/Dt = vA.
Conservation of Mass The Equation of Continuity
The law of conservation of mass has already been presented in a form appli-
cable to a control volume, as Eq. (4.12), which may be rewritten as
0 =
c
ct
dV + q n dS .
S
}
V
}
(5.13)
Application of the divergence theorem to the surface integral
q n dS
S
}
= V q ( )dV
V
}
(5.14)
transforms Eq. (5.13) into
c
ct
+ V q ( )
(
( dV = 0
V
}
.
(5.15)
Equation (5.15) must be satisfied by any thermodynamic system chosen in the
domain of the fluid. In other words the integration region V is arbitrary. The inte-
gral of Eq. (5.15) can vanish over any arbitrary region inside a given domain only
if its integrand vanishes at all points in that domain; hence
c
ct
+V q ( )= 0.
(5.16)
Equation (5.16) is known as the equation of continuity. It is the differential form of
148 Fluid Mechanics
the law of conservation of mass, written in terms of the flow field.
Equation (5.16) is now rewritten in detail in the three most commonly used
coordinate systems.
In cartesian coordinates:
c
ct
+
c u
( )
cx
+
c v
( )
cy
+
c w
( )
cz
= 0.
(5.17)
In cylindrical coordinates:
c
ct
+
1
r
c rq
r
( )
cr
+
1
r
c q
u
( )
cu
+
c q
z
( )
cz
= 0.
(5.18)
In spherical coordinates:
c
ct
+
1
r
2
c r
2
q
r
( )
cr
+
1
rsinu
c q
u
sinu
( )
cu
+
1
r sinu
c q
|
( )
c|
= 0.
(5.19)
In some particular cases the equation of continuity assumes simpler forms,
given here in cartesian coordinates.
a. Time-independent flows steady flows, /t = 0:
V q ( ) = 0,
(5.20)
c u
( )
cx
+
c v
( )
cy
+
c w
( )
cz
= 0.
b. Incompressible flows, = const:
V q = 0,
(5.21)
cu
cx
+
cv
cy
+
cw
cz
= 0.
Example 5.2
Using the equation of continuity, show that for any specific property b the
Reynolds transport theorem, Eq. (4.10), may be put in the form
D
Dt
b dV =
Db
Dt
dV .
V
}
V
}
Solution
Using the form of Eq. (4.10) of the Reynolds transport theorem,
5. Fluids in Motion Differential Analysis 149
D
Dt
b ( )dV
V
}
=
c b
( )
ct
dV + bqndS
S
}
V
}
,
we apply the divergence theorem to the surface integral on the right-hand side of
the equation and obtain
bq n dS
S
}
= V bq ( )dV .
V
}
Substitution into the Reynolds transport theorem results in
D
Dt
b
( )
dV =
c b ( )
ct
+ V bq
( )
(
(
dV
V
}
V
}
or
D
Dt
bdV =
cb
ct
+ qVb
|
\
|
.
| +b
c
ct
+ V q ( )
|
\
|
.
|
(
(
(
dV
V
}
V
}
.
Now the first term inside the integral on the right-hand side of the equation yields
cb
ct
+ q Vb
|
\
|
.
=
Db
Dt
while the equation of continuity, Eq. (5.16), makes the second term vanish,
b
c
ct
+ V q ( )
|
\
|
.
| = 0.
Hence
D
Dt
b dV =
Db
Dt
dV .
V
}
V
}
Example 5.3
Air flows at steady state in a square duct of constant cross section, Fig. 5.2.
Measurements at two points, 1 and 2, which are 30 m apart indicate uniform
velocities of u
1
= 30 m/s and u
2
= 130 m/s, respectively. The flow is assumed one
dimensional. Is the flow compressible?
Solution
The flow is one dimensional and time independent. Equation (5.20) thus
becomes
150 Fluid Mechanics
d u
( )
dx
= 0 or
du
dx
+ u
d
dx
= 0.
Since du/dx is different from zero so must be dp/dx and the flow is compressible.
Furthermore d u ( ) / dx = 0 implies u = const. Hence,
1
=
u
1
u
2
=
30
130
= 0. 23 .
Example 5.4
Water flows in a square duct of constant cross section, Fig. 5.2. Measurements
show that v = w = 0. A man who did not study fluid mechanics claims that "because
of friction," he expects the water to slow down along the duct. Is he right?
Figure 5.2 Duct flow.
Solution
Water is incompressible. Thus Eq. (5.21) applies for this case. Because the
flow is in the x-direction only, v = w = 0. Hence Eq. (5.21) simplifies to
cu
cx
= 0, u = const,
which implies that u u(x). Therefore, the velocity of the fluid does not change in
the x-direction. On the other hand, the velocity may depend on y or z. Indeed,
integration of the continuity equation, given above for this case, leads in the most
general case to
u = u y, z ( ) .
Newton's Second Law of Motion
Newton's second law of motion states that the rate of change of the momen-
tum of a thermodynamic system equals the sum total of the forces acting on this
system. Thus, the change in the momentum of any thermodynamic system subject
5. Fluids in Motion Differential Analysis 151
to both body forces and surface forces, as defined in Chapter 1, is
D
Dt
v
}
qdV =
v
}
g dV +
s
}
T dS ,
(5.22)
where g is a general body force per unit mass and T is the stress at the system
boundary.
Newton's second law has already been applied to a small cube in Chapter 2,
resulting in the three components of the momentum equations (2.13), (2.14),
(2.15), i.e.,
a
x
= g
x
+
c T
xx
c x
+
c T
yx
c y
+
c T
zx
c z
,
(2.13)
a
y
= g
y
+
c T
xy
cx
+
c T
yy
cy
+
c T
zy
cz
,
(2.14)
a
z
= g
z
+
c T
xz
c x
+
c T
yz
c y
+
c T
zz
c z
,
(2.15)
where a
i
are the three components of the acceleration vector of the small cube,
which we now suspect to be given by Eqs. (5.11) or (5.12).
Equation (5.22) holds in inertial coordinate systems only. In such systems the
cartesian unit vectors may be considered constant. The unit vectors in cartesian
coordinates do not undergo differentiation and behave like constant multipliers.
Let the x-component of Eq. (5.22) be first considered:
D
Dt
V
}
u dV =
V
}
g
x
dV + T
nx
S
}
dS .
(5.23)
The Reynolds transport theorem may now be applied to the left-hand side of this
equation (see Example 5.2), to yield
D
Dt
V
}
u dV =
V
}
Du
Dt
dV =
V
}
c u
c t
+ u
c u
c x
+ v
c u
c y
+ w
c u
c z
(
(
(
dV .
(5.24)
The stress term T
x
inside the surface integral of Eq. (5.28) is now rewritten in terms
of its components as given by Eq. (2.29) to yield
.
yx
xx zx
nx xx yx zx
S S V
T
T T
T dS T T T dS dV
x y z
c
c c
c c c
(
( = + + = + +
(
} } }
i j k n
(5.25)
where the divergence theorem has been used again.
Substitution of Eqs. (5.24) and (5.25) into Eq. (5.23) yields
152 Fluid Mechanics
cu
ct
+u
cu
c x
+ v
cu
c y
+ w
cu
c z
g
x
(
(
(
c T
xx
c x
+
cT
yx
c y
+
cT
zx
c z
(
(
(
V
}
dV = 0.
(5.26)
But because Newton's second law applies to any thermodynamic system, V is
arbitrary and for the integral to vanish the integrand must vanish at any point of the
domain of integration, i.e.,
c u
ct
+ u
c u
c x
+ v
c u
c y
+ w
c u
c z
(
(
(
= g
x
+
cT
xx
c x
+
c T
yx
c y
+
c T
zx
c z
.
(5.27)
Similarly for the y - and z-components
c v
c t
+ u
c v
c x
+v
c v
c y
+ w
c v
c z
(
(
(
= g
y
+
c T
xy
c x
+
c T
yy
c y
+
c T
zy
c z
,
(5.28)
c w
c t
+ u
c w
c x
+ v
c w
c y
+ w
c w
c z
(
(
(
= g
z
+
c T
xz
c x
+
c T
yz
c y
+
c T
zz
c z
.
(5.29)
Equations (5.27) - (5.29) relate rates of change of momentum to body forces
and to stresses existing in the fluid. It is enlightening to compare these equations
with Eqs. (2.13) - (2.15). The progress we made in attaining the Reynolds transport
theorem and the concept of the control volume made it possible to express the
acceleration, which in Eqs. (2.13) - (2.15) was just a letter a, in field terms that are
meaningful and correct. We have, however, paid a heavy price: Newton's second
law is linear, and Eqs. (2.13) - (2.15) look linear, but Eqs. (5.27) - (5.29), which
represent this law in field terms, are nonlinear.
As they stand, Eqs. (5.27) - (5.29) cannot be solved because they contain too
many dependent variables. Therefore, some additional relations connecting
stresses and velocities must be added to make this system of equations complete.
Newton's Law of Viscosity
Consider the fluid between the lower stationary plate and the upper moving
plate, shown in Fig. 1.4. Experiments show that the force per unit area required to
move the upper plate is proportional to the velocity of this plate, V, and inversely
proportional to the gap between the parallel plates, h:
F
A
V
h
.
(5.30)
The force per unit area exerted on the upper plate is equal to shear stress T
yx
applied to the fluid by the upper plate. The proportionality constant that converts
Eq. (1.17) into an equality is called the dynamic or absolute viscosity . Thus
5. Fluids in Motion Differential Analysis 153
T
yx
=
V
h
.
(5.31)
Equation (5.31) may be generalized for the case of two adjacent layers of fluid
separated by a distance dy, both moving parallel to the x-direction with the veloc-
ities u and u + du, respectively. Here the shear stress exerted by one layer of fluid
on the other is proportional to its velocity relative to that of the other layer and
inversely proportional to the distance
between them. Hence
T
yx
=
du
dy
.
(5.32)
Equation (5.32) is known as Newton's law of viscosity, and a fluid obeying it
is called a Newtonian fluid. Equation (5.32) states that in unidirectional flow the
shear stress in a Newtonian fluid is directly proportional to the transverse velocity
gradient, du/dy, also known as the rate of shear strain or the rate of shear
deformation.
There is no obvious reason why real fluids should obey Newton's law of
viscosity, Eq. (5.32). As a matter of fact, there are more fluids that do not obey Eq.
(5.32) than those that do. Such fluids are called non-Newtonian. Fortunately, the
three most abundant fluids, air, water and petroleum, obey Newton's law of
viscosity quite closely. Typical non-Newtonian fluids are paints, polymer solutions
and melts, blood and many liquid food products, such as soups, jellies, etc.
*Analysis of Deformation
Equation (5.32), which defines a Newtonian fluid, can be applied to unidi-
rectional flows only. However, the definition of a Newtonian fluid as one in which
the stress depends linearly on the rate of deformation may be generalized to three-
dimensional flows. To obtain this general relationship, the rate of deformation has
to be considered in some more detail. We start by noting two points:
a. A fluid has been defined as a continuum which cannot support shear stress
while at rest with respect to any coordinate system. Because coordinate
systems can move and rotate, a fluid whose only motion is like that of a rigid
body does not suffer shear stresses. As already shown, the stress tensor for
such motions reduces to a diagonal matrix, with the three equal normal
stresses, i.e., the pressure, occupying the diagonal. The magnitude of this
pressure, however, does depend on the fluid motion. The diagonal terms in
the stress tensor thus seem to contain a part which does not vanish in a rigid-
body-like motion, i.e., with a zero rate of deformation. To exhibit shear
154 Fluid Mechanics
stresses, the fluid must undergo a motion which cannot look like rest to any
observer. We call such a motion deformation, and its rate, the rate of defor-
mation. The shear terms in the stress tensor are taken as proportional to these
rates of deformation; the normal stresses are also modified, with the modifi-
cations proportional to the rates of deformations. These proportionality rela-
tions are the extension of the Newtonian fluid definition to three-dimensional
flows.
b. Suppose our analysis is finished, and we have those desired relations between
stress and deformation. When our system of coordinates rotates the stress
components transform as tensor components should. However, they must
remain proportional to the deformation. We should therefore expect the
deformation to transform accordingly. One way to satisfy this requirement is
for the deformation to be a tensor, a symmetrical one with its principal direc-
tions coinciding with those of the stress tensor. Indeed, in such a case the
transformation of a stress component would be just the transformation of a
deformation component multiplied by the proportionality constant.
Figure 5.3 System undergoing deformation.
With these two points in mind we consider a fluid system that undergoes
deformation as shown in Fig. 5.3. In order to describe this deformation, we first
seek the motion of point P relative to C. The difference between the x-components
of the velocity u at points P and C which are very close together is
du =
cu
cx
dx +
cu
cy
dy +
cu
cz
dz .
(5.33)
5. Fluids in Motion Differential Analysis 155
In general, for the i-th velocity component,
dq
i
=
cq
i
cx
j
dx
j
.
(5.34)
The motion of P relative to C depends, therefore, on the nine components dq
i
/dx
j
and may be written as
d q =
cu
cx
cu
cy
cu
cz
cv
cx
cv
cy
cv
cz
cw
cx
cw
cy
cw
cz
d r.
(5.35)
The relative motion described by Eqs. (5.33) - (5.35) results from the
combined effects of rotation and deformation. A rotating rigid body which does
not deform still exhibits relative motion between its points. To eliminate the effect
of rotation, we subtract from the relative motion of Eq. (5.35) the part that corre-
sponds to a rigid body rotation.
j dx
- dy
dq
rot
i dy
v
x
dx
C
P
dr = i dx + j dy
u
y
Figure 5.4 Rigid body rotation in x-y plane.
Similar relations can be obtained for e
x
and e
y
:
To obtain this rotation, we consider Fig. 5.4 which describes a rigid body
156 Fluid Mechanics
rotating in the x - y plane. The rotational velocity dq
rot
is given by
d q
rot
= i
cu
cy
dy + j
cv
cx
dx .
(5.36)
Rotation in the x - y plane may also be described as e = ke
z
, and thus
dq
rot
=edr = ie
z
dy + je
z
dx . (5.37)
Comparing Eqs. (5.36) and (5.37), one obtains
e
z
=
cu
cy
, e
z
=
cv
cx
,
(5.38)
which, upon addition and division by 2, attains the more convenient form
e
z
=
1
2
cv
cx
cu
cy
|
\
|
.
|
|
.
(5.39)
e
x
=
1
2
cw
cy
cv
cz
|
\
|
.
|
|
,
(5.40)
e
y
=
1
2
cu
cz
cw
cx
|
\
|
.
| .
(5.41)
With the additional notation e
xy
= e
z
, e
yz
= e
x
and e
zx
= e
y
, we obtain
e
ij
=
1
2
c q
j
cx
i
c q
i
cx
j
|
\
|
.
|
. (5.42)
The subscript ij denotes the plane in which the component of rotation is defined.
Using this notation, the rotational part of Eqs. (5.34) and (5.35) may be put as
dq
rot
=
0 e
12
e
13
e
21
0 e
23
e
31
e
32
0
dr.
(5.43)
We are now in a position to subtract the rotational relative velocity dq
rot
from
the total relative motion, dq, Eq. (5.35). We thus obtain the relative motion caused
by deformation, dq
def
, and the tensor which gives the rate of deformation c c.
5. Fluids in Motion Differential Analysis 157
Hence,
d q
def
= dq dq
rot
=
cu
cx
cu
cy
cu
cz
cv
cx
cv
cy
cv
cz
cw
cx
cw
cy
cw
cz
dr
1
2
0
cu
cy
cv
cx
|
\
|
.
|
|
cu
cz
cw
cx
|
\
|
.
|
cv
cx
cu
cy
|
\
|
.
|
|
0
cv
cz
cw
cy
|
\
|
.
|
|
cw
cx
cu
cz
|
\
|
.
|
cw
cy
cv
cz
|
\
|
.
|
|
0
d r
=
cu
cx
1
2
cu
cy
+
cv
cx
|
\
|
.
|
|
1
2
cu
cz
+
cw
cx
|
\
|
.
|
1
2
cv
cx
+
cu
cy
|
\
|
.
|
|
cv
cy
1
2
cv
cz
+
cw
cy
|
\
|
.
|
|
1
2
cw
cx
+
cu
cz
|
\
|
.
|
1
2
cw
cy
+
cv
cz
|
\
|
.
|
|
cw
cz
d r = e d r.
c c,
.
d r. (5.44)
The rate of deformation tensor, c c, is noted to be symmetrical, as we have
anticipated. The components of this rate of deformation tensor may be put in the
form
c
ij
=
1
2
c q
i
cx
j
+
c q
j
cx
i
|
\
|
.
|
|
.
(5.45)
This rate of deformation must now be related to the stress in the fluid. The relation
has to be consistent with two premises: It must reduce to the elementary definition
of a Newtonian fluid for unidirectional flow, Eq. (5.32), and it must simplify to the
hydrostatic stress equation for the case of no deformation, Eq. (3.2). The relations
that indeed satisfy these requirements are presented in the following section.
Newtonian Fluids
The definition of a Newtonian fluid, which so far related to one-dimensional
relations of the form of Eq. (5.32), may now be extended to three-dimensional
relations using the so-called rate of deformation tensor, Eq. (5.45), i.e.,
c
ij
=
1
2
c q
i
cx
j
+
c q
j
cx
i
|
\
|
.
|
. (5.45)
We, therefore, redefine the Newtonian fluid as one that satisfies
T
ij
= po
ij
+ 2c
ij
(5.46)
158 Fluid Mechanics
in which the Kronecker delta o
ij
equals unity for i = j and vanishes for i j. The
stress T
ij
thus reduces to its pressure terms, Eq.(3.2), for vanishing strain c
ij
.
Equation (5.46) is written in detail in cartesian coordinates as
T
xx
= p + 2
c u
c x
, T
yy
= p + 2
cv
cy
, T
zz
= p + 2
cw
cz
,
T
xy
=
c u
c y
+
c v
c x
|
\
|
.
|
|
, T
xz
=
cu
cz
+
cw
cx
|
\
|
.
| , T
yz
=
cv
cz
+
cw
cy
|
\
|
.
|
|
.
(5.47)
Equation (5.46) suggests a definition for the pressure in a moving fluid.
Addition of the three normal components of the stress matrix results in
T
xx
+T
yy
+T
zz
= p 1+1+1 ( )+2Vq =3p.
(5.48)
The
V q
term vanishes for incompressible flows. Hence, the thermodynamic pres-
sure may be defined for an incompressible fluid as the average normal stress:
p =
T
xx
+ T
yy
+ T
zz
3
.
(5.49)
It is customary to separate out the pressure terms from the total stress, Eq. (5.46).
The remainder of T
ij
, i.e., the part expressing the deviation of the stress from pure
pressure, is called the deviatoric stress and is usually denoted by t
ij
. Thus
T
ij
= po
ij
+ t
ij
(5.50)
and
t
ij
= 2c
ij
.
(5.51)
Equation (5.50) is rewritten in tensor form as
T T = P P + t t (5.52)
where P P is the diagonal tensor
P P
=
p 0 0
0 p 0
0 0 p
(5.53)
Equation (5.50) is used to modify the momentum equations, Eqs.(5.27) (5.29), to
cu
ct
+ u
cu
cx
+ v
cu
cy
+ w
cu
cz
(
(
=
cp
cx
+ g
x
+
ct
xx
cx
+
ct
yx
cy
+
ct
zx
cz
, (5.54)
c v
c t
+ u
cv
cx
+ v
cv
cy
+ w
c v
c z
(
(
=
cp
cy
+ g
y
+
ct
xy
cx
+
ct
yy
cy
+
ct
zy
cz
, (5.55)
cw
ct
+ u
cw
cx
+ v
cw
cy
+ w
cw
cz
(
(
=
cp
cz
+ g
z
+
ct
xz
cx
+
ct
yz
cy
+
ct
zz
cz
,
(5.56)
5. Fluids in Motion Differential Analysis 159
which may be put in the symbolic more compact form
Dq
Dt
= Vp +g +V t.
(5.57)
Equations (5.54) - (5.57) are quite general. For Newtonian fluids the stress
components are given by Eq. (5.51). Where the relations between t
ij
and c
ij
are
nonlinear, the fluid is non-Newtonian, but Eqs. (5.54) - (5.57) still hold. Equation
(5.51) expresses a fundamental relation between stress and rate of strain and is
independent of the coordinate system used. The expressions for the stress and the
rate of strain components in several coordinate systems are now written down:
In cartesian coordinates
q = iu + jv + kw
c
xy
=
1
2
c u
c y
+
c v
c x
|
\
|
.
|
|
, c
yz
=
1
2
cv
c z
+
c w
cy
|
\
|
.
|
|
, c
xz
=
1
2
c u
c z
+
c w
c x
|
\
|
.
|
|
,
t
xy
=
c u
c y
+
c v
c x
|
\
|
.
|
|
, t
yz
=
cv
c z
+
c w
cy
|
\
|
.
|
|
, t
xz
=
c u
c z
+
c w
c x
|
\
|
.
|
|
,
c
xx
=
c u
c x
, c
yy
=
c v
c y
, c
zz
=
c w
c z
,
t
xx
= 2
c u
c x
, t
yy
= 2
c v
c y
, t
zz
= 2
c w
c z
.
In cylindrical coordinates
q = e
r
q
r
+e
u
q
u
+e
z
q
z
c
ru
=
1
2
1
r
c q
r
c u
+ r
c
cr
q
u
r
|
\
|
.
|
(
(
(
, t
ru
=
1
r
c q
r
c u
+ r
c
cr
q
u
r
|
\
|
.
|
(
(
(
,
c
rz
=
1
2
c q
r
c z
+
c q
z
c r
(
(
(
, t
rz
=
c q
r
c z
+
c q
z
c r
(
(
(
,
c
uz
=
1
2
c q
u
c z
+
1
r
c q
z
c u
(
(
(
, t
uz
=
c q
u
c z
+
1
r
c q
z
c u
(
(
(
,
(5.59)
c
rr
=
c q
r
c r
, c
uu
=
1
r
cq
u
cu
+
q
r
r
|
\
|
.
, c
zz
=
c q
z
cz
,
t
rr
= 2
c q
r
c r
, t
uu
= 2
1
r
cq
u
cu
+
q
r
r
|
\
|
.
, t
zz
= 2
c q
z
cz
.
160 Fluid Mechanics
In spherical coordinates
q = e
r
q
r
+ e
u
q
u
+ e
|
q
|
c
Ru
=
1
2
1
R
c q
R
cu
+ R
c
c R
q
u
R
|
\
|
.
(
(
, t
Ru
=
1
R
c q
R
cu
+ R
c
cR
q
u
R
|
\
|
.
(
(
,
c
R|
=
1
2
1
R sin u
c q
R
cu
+ R
c
c R
q
u
R
|
\
|
.
(
(
, t
R|
=
1
R sin u
c q
R
cu
+ R
c
cR
q
u
R
|
\
|
.
(
(
,
c
u|
=
1
2
1
R sin u
c q
u
c |
+
sinu
R
c
cu
q
|
sinu
|
\
|
.
(
(
, t
u|
=
1
R sin u
c q
u
c |
+
sin u
R
c
c u
q
|
sin u
|
\
|
.
(
(
,
(5.60)
c
RR
=
c q
R
cR
, t
RR
= 2
c q
R
c R
,
c
uu
=
1
R
cq
u
cu
+
q
r
R
|
\
|
.
|
, t
uu
= 2
1
R
cq
u
c u
+
q
r
R
|
\
|
.
|
,
c
||
=
1
R sin |
c q
|
c|
+
q
R
R
+
q
u
cot u
R
|
\
|
.
|
, t
||
= 2
1
R sin |
cq
|
c|
+
q
R
R
+
q
u
cot u
R
|
\
|
.
|
.
The NavierStokes Equations
Equations (5.58) - (5.60) may be used to eliminate the stress components
from the differential momentum equations (5.54) - (5.56). The result is
cu
ct
+ u
cu
cx
+ v
cu
cy
+ w
cu
cz
|
\
|
.
|
= g
x
cp
cx
+
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
,
cv
c t
+ u
cv
cx
+ v
cv
cy
+ w
cv
cz
|
\
|
.
|
= g
y
cp
cy
+
c
2
v
cx
2
+
c
2
v
cy
2
+
c
2
v
cz
2
|
\
|
.
|
,
cw
ct
+ u
cw
cx
+ v
cw
cy
+ w
cw
cz
|
\
|
.
|
= g
z
cp
cz
+
c
2
w
cx
2
+
c
2
w
cy
2
+
c
2
w
cz
2
|
\
|
.
|
.
(5.61)
These momentum equations are called the NavierStokes equations. They
constitute a system of three nonlinear second order partial differential equations.
Together with the continuity equation they form a set of four equations which is
complete for incompressible Newtonian flows, i.e., in principle they are sufficient
to solve for the four dependent variables p, u, v, and w.
The NavierStokes equations require for their solution initial conditions as
well as boundary conditions. The proper boundary conditions for the velocity on a
5. Fluids in Motion Differential Analysis 161
rigid boundary are
q
n
= q
t
= 0,
(5.62)
where q
n
is the normal component of the velocity relative to the solid boundary,
and q
t
is its tangential component. These conditions are also termed the no-pene-
tration (q
n
= 0) and no-slip (q
t
= 0) viscous boundary conditions. When the region
occupied by the fluid is not closed, i.e., the fluid is not completely confined,
additional conditions are still required on some surfaces which completely enclose
the domain of the solution. These may represent some real physical surfaces or
they may be chosen quite arbitrarily, provided the velocity on them is known.
The pressure, which is also a dependent variable, requires boundary condi-
tions too. More is said on this point later, e.g., in Examples 5.11 - 5.14.
We now proceed to express the NavierStokes equations in other coordinate
systems. To do this, we first write these equations in their vectorial form as
Dq
Dt
= Vp +g V(Vq) = Vp +g +V
2
q.
(5.63)
The first form, i.e., the one containing V(Vq), is a standard vectorial form for
the incompressible NavierStokes equations and can be shown to be correct by
expansion in cartesian coordinates. The other form is symbolic and must be taken
literally, i.e., as if V
2
is the Laplacian operator applied to the velocity vector in
cartesian coordinates. ExpandingV(Vq) in cylindrical polar coordinates and
using the equation of continuity, V
.
q = 0 for incompressible fluids, we obtain
c q
r
c t
+ q
r
c q
r
c r
+ q
u
c q
r
rc u
+ q
z
c q
r
c z
q
u
2
r
|
\
|
.
|
|
= g
r
cp
cr
+
c
cr
1
r
c
c r
(rq
r
)
|
\
|
.
|
|
+
1
r
2
c
2
q
r
cu
2
+
c
2
q
r
cz
2
2
r
2
cq
u
cu
(
(
(
,
(5.64)
c q
u
c t
+ q
r
c q
u
c r
+ q
u
c q
u
rcu
+ q
z
c q
u
c z
+
q
r
q
u
r
|
\
|
.
|
|
= g
u
+
c p
rc u
+
c
c r
1
r
c
c r
(rq
u
)
|
\
|
.
|
|
+
1
r
2
c
2
q
u
cu
2
+
c
2
q
u
c z
2
+
2
r
2
c q
r
cu
(
(
(
,
(5.65)
c q
z
ct
+ q
r
c q
z
c r
+ q
u
c q
z
rcu
+ q
z
c q
z
c z
|
\
|
.
|
|
= g
z
c p
c z
+
1
r
c
cr
r
c q
z
c r
|
\
|
.
|
|
+
1
r
2
c
2
q
z
cu
2
+
c
2
q
z
c z
2
(
(
(
.
(5.66)
162 Fluid Mechanics
Repeating the process for spherical coordinates, we obtain
c q
R
c t
+ q
R
c q
R
c R
+
q
u
R
c q
R
cu
+
q
|
R sin u
cq
R
c |
q
u
2
+ q
|
2
R
|
\
|
.
|
|
= g
R
c p
c R
+
1
R
2
c
c R
R
2
c q
R
c R
|
\
|
.
|
|
+
1
R
2
sin u
c
c u
sin u
c q
R
cu
|
\
|
.
|
(
(
(
+
1
R
2
sin
2
u
c
2
q
R
c |
2
2q
R
R
2
2
R
2
c q
u
c u
2q
u
cotu
R
2
2
R
2
sin u
c q
|
c|
(
(
(
,
(5.67)
c q
u
c t
+ q
R
c q
u
c R
+
q
u
R
c q
u
c u
+
q
|
Rsin u
c q
u
c |
q
R
q
u
R
q
|
2
cotu
R
|
\
|
.
|
|
= g
u
1
R
c p
c u
+
1
R
2
c
c R
R
2
c q
u
c R
|
\
|
.
|
|
+
1
R
2
sin u
c
cu
sin u
c q
u
cu
|
\
|
.
|
(
(
(
+
1
R
2
sin
2
u
c
2
q
u
c|
2
+
2
R
2
cq
R
cu
q
u
R
2
sin
2
u
2 cosu
R
2
sin
2
u
cq
|
c|
(
(
(
,
(5.68)
c q
|
ct
+ q
R
c q
|
cR
+
q
u
R
c q
|
cu
+
q
|
Rsin u
c q
|
c|
+
q
|
q
R
R
+
q
u
q
|
cotu
R
(
(
(
= g
|
1
R sinu
cp
c|
+
1
R
2
c
cR
R
2
cq
|
cR
|
\
|
.
|
|
+
1
R
2
sin u
c
cu
sinu
cq
|
cu
|
\
|
.
|
|
(
(
(
+
1
R
2
sin
2
u
c
2
q
|
c |
2
q
|
R
2
sin
2
u
+
2
R
2
sin
2
u
cq
r
c|
+
2 cosu
R
2
sin
2
u
cq
|
c|
(
(
(
.
(5.69)
The Euler Equations
Substitution of = 0 in the NavierStokes equations (5.61) - (5.69) reduces
them to a form called the Euler equations:
Dq
Dt
= gVp.
(5.70)
Historically the Euler equations were formulated earlier than the Navier
Stokes equations and were considered an approximation. It is noted that the Euler
equations are of the first order and cannot in general satisfy both boundary
conditions Eq. (5.62). It is therefore formally concluded that the Euler equations
do not form a good approximation near a rigid boundary. Far from a boundary, and
where ~ 0 is a fair estimate, they have an important role as approximations and
are generally easier to solve than the full NavierStokes equations. The Euler
equations are further considered in later chapters.
5. Fluids in Motion Differential Analysis 163
Solutions for Two-dimensional Flows The Stream Function
Solutions of the NavierStokes equations result in velocity vectors, q, and
pressures, p, which satisfy both the momentum equations and the continuity
equation. Given such a combination, [q , p], one can check whether it constitutes a
solution by substitution into the equations. How to find such a solution is another
matter, and any general step leading toward this goal is useful. For two-
dimensional flows it is possible to eliminate the continuity equation from the
system of equations by using only functions which satisfy the continuity equation.
This elimination is a formal step toward a solution, and the functions which affect
this elimination are the stream functions.
A flow is defined as two dimensional when its description in cartesian coor-
dinates shows no z-component of the velocity and no dependence on the z-
coordinate. Such a flow can be described in the z = 0 plane, with the velocity
vector and the streamlines lying in this plane. Furthermore, z = C planes, which are
parallel to the z = 0 plane, show a flow pattern identical to that in the z = 0 plane.
The z = 0 plane is therefore called the representative plane.
Figure 5.5 shows a representative plane for a two-dimensional flow, with four
streamlines denoted by the letters A,B,C,D. The whole pattern may be shifted in
the z-direction parallel to itself. Thus the streamlines also represent stream sheets,
i.e., barriers which are not crossed by the flow. The mass flux entering at the left,
between, say, streamlines A and B must therefore come out at the right side
without change. Because the distance between the two streamlines accommodating
this mass flux seems in the drawing to increase, the mass flux per unit cross
section, q, must decrease from left to right. There is therefore some relation
between the convergence and divergence of streamlines and the vector q.
Furthermore, because stream sheets are not crossed by the flow, each sheet repre-
sents a certain mass flux per unit depth of stream sheet taking place "below it," i.e.,
flowing between it and some particular stream sheet representing zero flux. This
mass flux is called the stream function and is denoted by .
Let the stream function corresponding to streamline C in Fig. 5.5 be kg/m.
Let streamline D be close to it and have the stream function + d. Obviously,
d = dy ( ) u
( )
= dx ( ) v
( )
.
From which follows
u =
c
cy
, v =
c
cx
.
(5.71)
164 Fluid Mechanics
A
B
C
D
y
x
P (x , y)
M(r , u )
dy
dx
r du
u
u
- q
r
q
r
dr
v
u
Figure 5.5 Two-dimensional representative plane and streamlines.
Using plane polar coordinates in the representative plane and letting
B
=
A
+ d ,
d = rd u ( ) q
r
( )
= dr
( )
q
u
( )
,
from which follows
q
r
=
1
r
c
cu
, q
u
=
c
cr
.
(5.72)
It is noted that substitution of Eq. (5.71) or of Eq. (5.72) in Eq. (5.17) satis-
fies the continuity equation identically. Indeed, conservation of mass which gen-
erated the equation of continuity is also the basis for the derivation of Eqs. (5.71)
and (5.72).
For incompressible flows appears in the definition of the stream function as
a constant coefficient, and so it does in Eqs. (5.71) and (5.72). It is customary to
drop this constant for incompressible flows and to define as a measure of the
volumetric flowrate, i.e., to have the dimensions of [m
3
/(m
.
s)]. The analog of
Eqs. (5.71) and (5.72) for incompressible steady flows is
u =
c
cy
, v =
c
cx
(5.73)
and
5. Fluids in Motion Differential Analysis 165
q
r
=
1
r
c
cu
, q
u
=
c
cr
,
(5.74)
and the relations, of course, satisfy Eq. (5.21) identically.
It is important to realize that the use of stream functions is equivalent to the
inclusion of the continuity equation in the considerations. Crudely speaking, in
steady two-dimensional flows the use of stream functions is equivalent to "one
differentiation and one substitution"; this is so because further reference to one
first order differential equation, the continuity equation, becomes unnecessary, but
the remaining equations increase their order by 1. It is also helpful to review the
manner in which this equivalence has been established. Conservation of mass has
led to both the continuity equation and the stream function, and therefore their
mathematical consequences must be consistent.
Finally it is noted that the tangent to the streamline,
d s = i dx + jdy + kdz ,
is parallel to the velocity vector,
q = iu + jv + kw,
which implies that their components are proportional to one another, i.e.,
dx
u
=
dy
v
=
dz
w
(5.75)
or, for two-dimensional flows, v dx u dy = 0. Thus there is some function which
is conserved on the streamlines. One may check whether the form written above is
a total differential, i.e., whether
cv
cy
=
cu
cx
,
which is indeed satisfied by continuity. Denoting this total differential as d, we
obtain
d = v dx + u dy = 0 on streamlines.
This is an alternative way to introduce the concept of the stream function.
Example 5.5
A two-dimensional source of intensity Q, shown in Fig. 5.6, is defined as a
singular point out of which a fluid flows at the constant rate of Q kg/s
.
m, dis-
tributed symmetrically in all angular directions.
Write the equation of continuity in cartesian and in polar coordinates and
166 Fluid Mechanics
choose the convenient form for this case. Find the velocity field and the stream
function. Assume incompressible flow.
x
y
r
u
Figure 5.6 Two-dimensional source flow.
Solution
The flow is independent of time, and the continuity equation becomes
V
.
q = 0.
In cartesian coordinates
V q =
cu
cx
+
cv
cy
+
cw
cz
= 0.
In polar coordinates
V q =
1
r
c
cr
rq
r
( )
+
1
r
c
cu
q
u
+
c
cz
q
z
= 0.
For two-dimensional flow w = q
z
= 0. Because of the angular symmetry, q
u
= 0.
The polar form, therefore, is the more convenient, retaining one term only:
d
dr
rq
r
( )
=0, rq
r
=const.
The intensity of the source Q is related to q
r
by Q = 2trq
r
; hence, q
r
=Q/2tr. In
polar coordinates
1
r
c
cu
= q
r
.
Integration yields
=
Qu
2t
+ f (r).
Also
c
cr
= ' f = q
u
= 0.
Hence f is a constant which just determines the location of u = 0. The stream
function is constant along radii.
5. Fluids in Motion Differential Analysis 167
Example 5.6
Parallel flow at a constant velocity U is shown in Fig. 5.7. The flow is for-
mally defined by
u = U, v = 0.
.
Find the stream function for this case in cartesian and in polar coordinates.
U
y
x
r
u
u
q
r
q
Figure 5.7 Parallel flow at constant velocity U.
Solution
Substitution of u = U into the definition of the two-dimensional stream
function, Eq. (5.73) yields
c
cy
= U ,
which upon integration leads to
= Uy + f (x),
where f is found by the substitution of v = 0 into Eq. (5.73),
v =
c
cx
= ' f = 0, f =const.
Hence, the cartesian stream function is given by
= Uy + C.
Since y = r sinu, the polar form becomes
= Ur sin u + C.
One may try to first express the velocity in polar form and then obtain . The polar
decomposition of U, shown in Fig. 5.7, yields the velocity components
q
r
=U cosu, q
u
=Usin u.
168 Fluid Mechanics
Substitution of the radial velocity component into Eq. (5.74) yields
c
cu
= Ur cosu,
= Ur sinu f r
( )
.
The function f is obtained from the tangential velocity component
c
cr
=Usinu ' f =q
u
,
leading to f' = 0, and f = const; and the same result has been obtained.
Example 5.7
What flow results of the superposition of source flow and parallel flow?
Solution
Addition of the stream functions of source flow and parallel flow, obtained in
Examples 5.5 and 5.6, respectively, yields
=
Q
2t
u +Ur sinu.
Making use of Eq. (5.74) leads to
q
r
=
1
r
c
cu
=
Q
2tr
+ U cosu,
q
u
=
c
cr
= U sin u.
For very large r the flow approaches that of the parallel flow, while for very small
r it becomes the source flow. The flow is sketched in Fig. 5.8. As seen in the
figure, at point A the streamline splits and the velocity vector seems to have four
different directions. This can happen only if the velocity vanishes there, because
only a vector of zero length may have several directions without indicating a
contradiction. Such a point of zero velocity is denoted a stagnation point.
Indeed, for
u = t, i.e., on = Q/ 2 ,
q
u
= 0 , and
q
r
=
Q
2tr
U .
Thus for
r = Q /( 2tU ) ,
q
r
= 0, and the stagnation point has been obtained.
5. Fluids in Motion Differential Analysis 169
Figure 5.8 Superposition of source flow and parallel flow.
Example 5.8
The flow between two parallel flat plates, with the lower one fixed and the
upper one moving, as shown in Fig. 2.6, is given by
u = V
y
h
, v = w = 0.
Does the flow satisfy the continuity equation? Find the stream function for this
flow.
Solution
The continuity Eq. (5.21) states
cu
cx
+
cv
cy
+
cw
cz
= 0
and is therefore satisfied by this flow. Now
u = c / cy, the flow is two-dimen-
sional and hence
c
cy
= y
V
h
, = y
2
V
2h
+ C.
Example 5.9
The steady flow between two parallel plates shown in Fig. 2.6, and consid-
ered in Example 5.8, has the velocity vector
170 Fluid Mechanics
q = iu = V
y
h
.
Find Dq/Dt.
Solution
From Eq. (5.10)
Dq
Dt
=
c q
ct
+ u
cq
cx
+ v
cq
cy
|
\
|
.
= 0.
Example 5.10
The source flow shown in Fig. 5.6, and considered in Example 5.5, has the
velocity vector
q = e
r
q
r
=
e
r
r
Q
2t
|
\
|
.
| .
Find Dq/Dt .
Solution
From Eq. (5.10)
Dq
Dt
=
c q
c t
+ q
c q
r
c r
+ 0
(
(
= 0 +
Q
2t
|
\
|
.
2
e
r
r
c
c r
1
r
|
\
|
.
=
e
r
r
3
Q
2t
|
\
|
.
2
.
Or, another way, using Eq. (5.11),
Dq
Dt
=
cq
ct
+
1
2
V q q ( )q Vq ( ),
where
V q =
e
r
r
e
u
e
z
r
c
cr
c
cu
c
cz
Q
2tr
0 0
= 0
5. Fluids in Motion Differential Analysis 171
and
q q =
Q
2t
|
\
|
.
|
2
1
r
2
.
Hence,
Dq
Dt
=
1
2
V q q ( ) =
1
2
e
r
c
c r
q q ( ) =
Q
2t
|
\
|
.
|
2
e
r
r
3
.
The Pressure in the Momentum Equations
We already know that the momentum equations are nonlinear. On the other
hand the relations between the stream function and the velocity components, Eqs.
(5.71) - (5.74), are linear, and so is the continuity equation. Having two reasonable
flow fields, such as the source flow, Example 5.5, and the parallel flow, Example
5.6, we have attempted superposition in Example 5.7, which has resulted in a flow
field which is indeed the superposition of those two fields. The continuity equation
is, obviously, satisfied. However, because of the nonlinearity of the momentum
equation, the pressure gradient, obtained from the substitution of the combined
velocity field, is not the superposition of the pressure gradients corresponding to
the two individual fields. This point is considered in the following examples.
Example 5.11
Consider the source flow discussed in Example 5.5. The velocity vector
found there has been obtained without the use of the NavierStokes equations.
Substitute this velocity vector into the equations and evaluate the pressure field.
Solution
The velocity vector for the source flow is
q = e
r
q
r
=e
r
Q
2tr
.
The form convenient in this case is the expression of the NavierStokes
equations in cylindrical coordinates, Eqs. (5.64) - (5.66), which yield
q
r
cq
r
cr
|
\
|
.
| =
cp
cr
+g
r
+
c
cr
1
r
c
cr
rq
r
( )
|
\
|
.
| ,
172 Fluid Mechanics
0 =
cp
cu
+g
u
,
or taking g = 0,
Q
2
(2t)
2
1
r
3
=
dp
dr
.
Integration yields
p = p
o
Q
2
8t
2
1
r
2
1
r
0
2
(
(
(
.
The NavierStokes equations are satisfied, and we now know the resulting
pressure field. This flow can exist only if the obtained pressure is possible. An
acceptable boundary condition may be
p = p
= const at r ,
which then implies
p = p
Q
2
8t
2
1
r
2
.
We also note that in the solution for the pressure there is no trace of the
viscosity. This pressure, therefore, also satisfies the Euler equation (5.70).
Example 5.12
Consider the parallel flow, Example 5.6. Check whether the velocity vector
obtained there satisfies the NavierStokes equations.
Solution
The velocity vector obtained is
q = iU, and we choose, therefore, the carte-
sian form of the NavierStokes equations, Eq. (5.61). These yield
0 = g
x
cp
cx
,
0 = g
y
cp
cy
, or, neglecti ng g, p = const .
An acceptable boundary condition for the pressure is
p = p
= const at r .
Again we find that the same velocity and pressure also satisfy the Euler equation.
5. Fluids in Motion Differential Analysis 173
Example 5.13
Consider the superposition of parallel flow and source flow, Example 5.7.
Find what pressure comes out when the combined velocity vector is substituted in
the NavierStokes equations.
Solution
The combined velocity vector is
q = e
r
q
s
+e
u
q
u
,
q
r
=
Q
2tr
+U cosu ,
q
u
= U sin u.
Substitution in Eqs. (5.64) and (5.65), with g neglected, yields
Q
2tr
+U cosu
|
\
|
.
|
Q
2tr
2
+ Usin u
U
r
sin u
U
2
r
sin
2
u
(
(
(
=
cp
cr
+
U
r
2
cosu
U
r
2
cosu + 2
U
r
2
cosu
(
( ,
Usin u
1
r
U cosu
1
r
Q
2tr
+ U cosu
|
\
|
.
|
U sin u
(
(
(
=
cp
rcu
+
U
r
2
sin u +
U
r
2
sin u
2
r
2
U sin u
(
(
,
or
Q
2
(2t)
2
r
3
QU
2tr
2
cosu =
cp
cr
,
QU
2tr
2
sin u =
1
r
cp
cu
.
The first equation may be integrated with respect to r to yield
=
Q
2
(2t)
2
2r
2
+
QU
2tr
cosu + f (u ) .
Differentiation with respect to u, division by r and substitution into the second
equation yield
QU
2tr
2
sinu +
1
r
df
du
=
QU
2tr
2
sin u,
or
174 Fluid Mechanics
df
du
= 0
and
f = const.
Thus finally,
p = p
o
Q
2
8t
2
r
2
+
QU
2tr
cosu
(
(
(
,
which is rather different from the superposition of the pressures obtained in
Examples 5.11 and 5.12. We remember, however, that the NavierStokes equa-
tions are not linear, and there is no reason for the superimposed pressures from
Examples 5.11 and 5.12 to equal this pressure.
An acceptable boundary condition for the pressure is
p = p
= const at r .
Again we note that the viscous terms do not appear here and that the solution
satisfies the Euler equation too.
Example 5.14
Consider the shear flow of Example 5.8. Find what pressure is obtained when
this flow is substituted in the NavierStokes equations.
Solution
The velocity vector for the shear flow is
q = iu = iV
y
h
, v = w = 0.
Equation (5.61) states, with g neglected,
cp
cx
=
cp
cy
= 0 ;
hence, p is constant.
The NavierStokes equations are satisfied, and so are the Euler equations.
5. Fluids in Motion Differential Analysis 175
Problems
5.1 Find the two-dimensional flow field described by
a.
= U (y x ) (Parallel flow)
b.
=Ur
2
(Rigid body rotation)
c.
= Qu /( 2t) (Sink flow)
d.
= Qu /(2t) (Source flow)
e.
=Uy
2
(Shear flow)
Find the x- and y-components of the velocity field and draw, free-hand, lines
of flow. Are these streamlines? Pathlines? Streaklines?
5.2 Check explicitly if all the flows in Problem 5.1 satisfy the continuity equa-
tion. Suppose one flow did not satisfy the continuity equation. Is this
possible?
5.3 Do all the flows in Problem 5.1 satisfy the NavierStokes equations? How
do you check this? Are there sufficient boundary conditions? If not, add the
missing ones.
5.4 A function F = F(x,y) is continuous and has at least three partial derivatives.
Can such a function always be considered a stream function? Does it
necessarily satisfy the equation of continuity? The NavierStokes equa-
tions? Try some polynomials of various orders.
5.5 A two-dimensional source of intensity Q is located at (0, 0). A sink of the same
intensity is located at (5, 0). Find the velocity field and sketch the
streamlines. What are the shapes of the streamlines?
5.6 Consider superposition of parallel flow and the sourcesink combination of
Problem 5.5. How does the flow look? Can the flow field so obtained be
considered a flow around a rigid oval body? Explain.
5.7 The flow in a round pipe is given by
w = w
max
1
r
R
|
\
|
.
|
2
(
(
(
.
Check if this flow satisfies the continuity equation and find the pressure
distribution.
5.8 For the flow in the round pipe given in Problem 5.7:
176 Fluid Mechanics
a. Find the shear stress at the wall and the total shear force at the circum-
ference. Compute the pressure gradient necessary to balance this force.
b. Substitute the velocity vector in the NavierStokes equations and obtain
the pressure gradient. Compare with a above.
5.9 An elliptical pipe has the inner contour
x
2
a
2
+
y
2
b
2
=1 .
The velocity in the pipe is suggested as
q = kw, with
w = w
max
1
x
2
a
2
y
2
b
2
|
\
|
.
| .
Does it satisfy the continuity equation? The NavierStokes equation? The
viscous boundary conditions? Find the pressure distribution in the flow.
5.10 A circular pipe, Problems 5.7 and 5.8, and an elliptical pipe, Problem 5.9,
have the same w
max
. Find relations between a, b and R such that the longi-
tudinal pressure drops are the same.
5.11 A can of milk completely full with milk such that there is no air bubble in it
is set on a turntable and rotated with e. After some time the milk rotates like
a solid body. Write the velocity vector in the milk and check whether it
satisfies the continuity equation, the NavierStokes equation and the viscous
boundary conditions. Find the pressure distribution.
5.12 A square can half filled with water is set on a turntable and rotated with e.
Once it reaches solid body rotation, does the velocity field satisfy continuity,
momentum and boundary conditions?
5.13 Does any velocity field which looks stationary to any observer (not neces-
sarily in an inertial coordinate system) satisfy continuity, momentum and
boundary conditions?
5.14 A two-dimensional source of strength Q = 4 m
3
/s
.
m is located at the origin,
point (0;0), and a sink of the same strength is located at point (5;0). Also
given are points A (-3;1), B (-3;-1) and C (0;4).
a. Calculate the volumetric flow [m
3
/ m
.
s] between points A and B, between
points A and C and between points B and C.
b. A parallel flow,
q = iu = 9 i [m / s],
5. Fluids in Motion Differential Analysis 177
has been added to the flow field. Find the stagnation points. Find the
velocity at the point (2.5; 2.5).
5.15 The stream function in a certain region of a flow may be approximated by
= 4 x
2
y
2
( )
.
a. Find the velocity vector and the pressure distribution in this region.
b. The flow is viewed by an observer who moves with the velocity V = 3i.
Find the velocity vector of the field, as seen by this observer.
Note that the moving system of the observer is also an inertial one.
5.16 Find the shear stress distribution in the region described in Problem 5.15.
5.17. The velocity field in a certain region of a flow is approximated by
q = iy j x .
Find the pressure distribution and the stream function in this region.
5.18 Find the shear stress distribution in the region described in Problem 5.17.
5.19 A rectangular conduit has the sides 0.05 m by 0.025 m and is oriented such
that its sides are parallel to the z-axis, Fig. P5.19. Water flows through the
conduit with a velocity which may be approximated by
q = kw = 3ksin
tx
0.05
sin
t y
0.025
.
Figure P5.19 Flow in a rectangular channel.
a. Is the flow one, two or three dimensional?
b. Are the viscous boundary conditions satisfied by this approximation?
c. Is the continuity equation satisfied by this approximation?
d. Find the shear stresses on the sides of the conduit.
178 Fluid Mechanics
e. Find the approximate pressure distribution in the flow. Note that the
shear forces on the sides must be balanced by the pressure forces.
5.20 The side of the conduit in Fig. P5.19 located at x = 0.05 m now slides in the
direction of the z-axis with the velocity 2 m/s. The velocity field is now
approximated by
q = kw = k 2sin
tx
0. 05
sin
ty
0. 025
+ 2
x
0. 05
sin
ty
0. 025
|
\
|
.
|
.
a. Find the shear stress distribution on the sides of the conduit and the
approximate pressure distribution.
b. Is the continuity equation satisfied by this approximation?
c. Are the viscous boundary conditions satisfied?
5.21 The velocity in a certain region of a flow field is approximated by
q =iu =i 4siny ( ).
a. Is the flow two-dimensional?
b. Find the stream function there, the pressure distribution and the shear
stress distribution.
c. Draw the streamlines and the velocity distribution.
179
6. EXACT SOLUTIONS OF THE
NAVIERSTOKES EQUATIONS
The NavierStokes equations, obtained in the previous chapter, are nonlinear
and submit to no general method of solution. Each new problem must be carefully
formulated as to geometry and proper boundary conditions, and then some scheme
of attack may be chosen, with the hope of reaching a solution. In most cases, all
attempts to obtain an exact solution fail, and approximate solutions must suffice. In
a few cases, exact solutions can be found. Some of these solvable cases are
presented now as examples.
Figure 6.1 Parallel plates and coordinates.
Flows between Parallel Plates
Two parallel plates with the gap d between them in the y-direction are shown
in Fig. 6.1. The plates and the incompressible fluid between them are assumed to
extend very far in the x-direction, and the flow field is considered two-
dimensional, i.e., nothing depends on the z-coordinate. The upper plate has the
180 Fluid Mechanics
velocity iU, where U may be a function of time. The lower plate is stationary. The
velocity of the fluid,
u u = + q i j
, where ( ) ( ) , , , , , x y t x y t u u u u = = , must satisfy
the boundary conditions
u x,0,t
( )
= 0,
(6.1)
u x,d,t
( )
= U t
( )
,
(6.2)
v x,0,t ( )= v x,d, t ( )= 0
(6.3)
The continuity equation, Eq. (5.21), and the NavierStokes equations, Eqs. (5.61),
for this case are
cu
cx
+
cv
cy
= 0
,
(6.4)
cu
ct
+ u
cu
cx
+ v
cu
cy
|
\
|
.
=
cp
cx
+ g
x
+
c
2
u
cx
2
+
c
2
u
cy
2
|
\
|
.
| ,
(6.5)
cv
ct
+ u
cv
cx
+ v
cv
cy
|
\
|
.
=
cp
cy
+g
y
+
c
2
v
cx
2
+
c
2
v
cy
2
|
\
|
.
|
(6.6)
The boundary conditions, Eq. (6.3), together with the assumption that the
plates and the fluid extend very far in the x-direction, bring to mind the possibility
that perhaps the flow is unidirectional, i.e.,
v x, y,t
( )
= 0
(6.7)
Equation (6.7) is not an assumption. Rather it is an intuitive guess that we pursue
until we either find a solution or become convinced that it leads to no solution, in
which case we mark it as an unsuccessful trial.
With Eq. (6.7), the continuity equation, Eq. (6.4), becomes
cu
cx
= 0, (6.8)
which upon integration yields
u = u y,t
( )
Hence, the velocity u depends on y and t only. Whatever the velocity profile is at
some x-coordinate, it repeats itself for other x values. Such a flow is denoted fully
developed. In going along a constant y line, we find the same u values, i.e., no
further development. With Eqs. (6.7) and (6.8), Eqs. (6.5) and (6.6) become
cu
ct
=
1
cp
cx
+ g
x
+ v
c
2
u
cy
2
,
(6.9)
6. Exact Solutions of NavierStokes Equations 181
0 =
1
cp
cy
+ g
y
(6.10)
At this point, the NavierStokes equations, (6.9) and (6.10), may be
simplified by combining the gravitational force g with the effect of the pressure p.
We define the distance from some reference plane upward, i.e., in the direction
opposed to that of gravity, as
h = h x, y, z
( )
and let
g = g Vh (6.11)
or
g
x
= g
ch
cx
, g
y
= g
ch
cy
, g
z
= g
ch
cz
.
For example, if h coincides with the coordinate y in Fig. 6.1, implying that g is
directed downward, one would obtain from Eq. (6.11)
g
y
= g, g
x
= g
z
= 0.
Substitution of Eq. (6.11) into Eq. (5.63) yields for the case of constant and g
( ) ( )
D
p gh
Dt
= V + V V
q
q
(6.12)
One may now define a modified pressure
P = p + gh
(6.13)
thus simplifying Eq. (6.12) to
Dq
Dt
= VP V Vq ( )
(6.14)
and Eqs. (6.9) and (6.10) to
cu
ct
=
1
cP
cx
+ v
c
2
u
cy
2
(6.15)
0 =
cP
cy
(6.16)
The modified pressure, P, defined by Eq. (6.13), is most useful in hydrody-
namics. In a fluid at rest, the modified pressure is constant throughout, and a vari-
ation in P indicates that the fluid is moving. However, the use of this modified
pressure is limited to cases in which is constant and the absolute pressure p does
not affect the boundary conditions. Thus, it may be used in any internal flow of an
182 Fluid Mechanics
incompressible fluid. In flows with free surfaces, the use of the static pressure may
be more convenient.
Elimination of g from the equations makes the solutions independent of the
orientation of gravity. Thus, the solution for a vertical tube is the same as for a
horizontal tube provided both flows are subject to the same modified pressure
gradient, VP.
Returning to our problem, we note from Eq. (6.16) that P does not depend on
y. Now let Eq. (6.15) be rearranged,
c
2
u
cy
2
cu
ct
=
cP
cx
.
(6.17)
In this form, the right-hand side of the equation does not depend on y, while its
left-hand side does not depend on x. Thus, both sides can depend only on t, i.e.,
cP
cx
= F(t)
(6.18)
Equation (6.18) indicates that a necessary condition for a fully developed flow is a
uniform pressure gradient all through the fluid, i.e.,
cP
cx
=
AP
Ax
= F(t)
(6.19)
We now proceed to solve Eq. (6.15) under various sets of boundary conditions.
Example 6.1
A fluid of density = 1,000 kg/m
3
flows between horizontal parallel plates,
Fig 6.2a. The x- and y-coordinates are as shown, and the respective dimensions are
x
1
= 6 m, x
2
= 10 m, x
3
= 14 m, d = 1.5 m.
The pressure is measured at points 1, 2 and 3, as shown, and the measurements are
p
1
= 170,000 Pa, p
2
= 145,285 Pa, p
3
= 150,000 Pa.
Can the flow be fully developed? Find the modified pressure at points 1, 2 and 3.
Solution
Let points 1 and 3 have the height zero. The modified pressures are
P
1
= p
1
= 170, 000 Pa,
P
2
= p
2
+ gh = 145, 285+ 1, 000 9.811. 5 = 160, 000 Pa,
P
3
= p
3
= 150,000 Pa.
6. Exact Solutions of NavierStokes Equations 183
The pressure gradient is
AP
Ax
|
\
|
.
|
12
=
10,000
4
= 2,500
Pa
m
=
AP
Ax
|
\
|
.
|
23
.
As shown, the pressure gradient is uniform. Hence, the flow may be fully devel-
oped.
Figure 6.2 Flow between parallel plates.
a. Horizontal. b. Inclined by 30
o
.
Example 6.2
The channel of Example 6.1 is now tilted by 30
o
, as shown in Fig. 6.2b. Once
again x
4
= 6 m, x
5
= 10 m , x
6
= 14 m, with x measured along the channel axis. The
pressure at point 4 is p
4
= 130,000 Pa. The flow is fully developed and the same
flowrate as in Example 6.1 is measured. Find p
5
and p
6
.
Solution
Let point 4 have the height zero. Then
P
4
= p
4
= 130,000 Pa.
To have the same flowrate, the modified pressure must have the same gradient as
in Example 6.1. Therefore
P
5
= P
4
+2,500 4 = 140,000 Pa,
P
6
= P
5
+10,000= 150,000 Pa.
Hence
p
5
= P
5
gh
5
=140, 0001, 000 9.81 4sin30
o
+1. 5cos30
o
( )
=107, 636 Pa,
184 Fluid Mechanics
p
6
= P
6
gh
6
=150, 000 1, 000 9. 81 8sin30
o
= 110, 760 Pa.
Time-independent Flows
Let us first consider time-independent flows. For this case Eq. (6.19) be-
comes
cP
cx
=
AP
Ax
= F = const.
(6.20)
and Eq. (6.17) yields
AP
Ax
=
c
2
u
cy
2
. (6.21)
The velocity u is not a function of x, and this equation can be integrated directly to
yield the general solution
u =
AP
Ax
d
2
2
y
d
|
\
|
.
2
+ C
1
y
d
|
\
|
.
+ C
2
(6.22)
using the boundary conditions
y = 0, u = 0,
(6.23)
y = d, u = U,
One obtains
u =
AP
Ax
|
\
|
.
|
d
2
2
y
d
y
d
|
\
|
.
|
2
(
(
(
+ U
y
d
|
\
|
.
|
.
(6.24)
The special case of U = 0 results in
u =
AP
Ax
|
\
|
.
d
2
2
y
d
y
d
|
\
|
.
2
(
(
,
(6.25)
which is known as plane Poiseuille flow, while the case of AP/Ax = 0 results in
u = U
y
d
,
(6.26)
which is known as shear flow or plane Couette flow.
Several velocity profiles given by Eq. (6.24), including the plane Poiseuille
and the plane Couette flows, are shown in Fig. 6.3.
6. Exact Solutions of NavierStokes Equations 185
U
U
a. b. c. d.
U
Figure 6.3 Parallel flow, Eq. (6.24):
a. Plane Poiseuille flow. b. Shear flow
c. Flows with AP/Ax< 0. d. Flows with AP/Ax> 0.
Example 6.3
Find the maximum velocity, the net flowrate and the average velocity for
plane Poiseuille flow.
Solution
The velocity profile is given by Eq. (6.25) as
u =
AP
Ax
|
\
|
.
|
d
2
2
y
d
y
d
|
\
|
.
|
2
(
(
(
.
This expression attains its maximum at y = d/2. The maximum velocity, u
o
, is there-
fore
u
o
=
AP
Ax
|
\
|
.
|
d
2
8
.
The velocity distribution may be expressed in terms of u
o
as
u = 4u
o
y
d
y
d
|
\
|
.
|
2
(
(
(
.
The flowrate per unit channel width, Q, is now obtained,
Q = u dy = 4u
o
y
d
y
d
|
\
|
.
|
2
(
(
(
dy =
2
3
u
o
d
0
d
}
0
d
}
,
and by substituting the value of u
o
, this becomes
Q =
AP
Ax
|
\
|
.
|
d
3
12
.
186 Fluid Mechanics
The average velocity is defined as the flowrate divided by the distance between the
plates, d, i.e.,
u =
Q
d
=
2
3
u
o
.
Example 6.4
Find the net flowrate and the average velocity for plane Couette flow.
Solution
From Eq. (6.26),
y = U
y
d
,
Q = u dy
0
d
}
= U
y
d
dy
0
d
}
=
U d
2
,
u =
Q
d
=
U
2
.
Example 6.5
The viscosity of glycerin at 10
o
C is 2 Ns/m
2
. A gap of 0.05 m between two
plates is filled with glycerin. One plate is at rest and the other moves in the
x-direction with the velocity U = 2 m/s. Because of an adverse pressure gradient,
there is no net flow in the gap (see Fig. 6.3d). Find the pressure gradient.
Solution
The equation that governs this flow, Eq. (6.21), is linear, and therefore
superposition is permissible. Thus, the flowrate for the flow under the combined
effect of pressure gradient and plate movement is the sum of flowrates for plane
Poiseuille flow (Example 6.3) and plane Couette flow (Example 6.4):
Q =
AP
Ax
|
\
|
.
|
d
3
12
+
Ud
2
.
For no net flow, Q = 0, and the required pressure gradient is
AP
Ax
=
6U
d
2
=
6 22
0.05
2
= 9,600Pa/ m.
6. Exact Solutions of NavierStokes Equations 187
Note: The expression for the flowrate may be alternatively obtained by the inte-
gration of the velocity, Eq. (6.24), over the flow gap.
Time-dependent Flow The Rayleigh Problem
Let us now return to Eq. (6.15) but consider flows for which cP/cx vanishes.
Equation (6.15) now assumes the form
cu
ct
= v
c
2
u
cy
2
. (6.27)
Consider an infinite flat plate with an infinite domain of fluid on its upper
side, Fig. 6.4. The fluid and the plate are at rest. At the time t = 0 the plate is
impulsively set into motion with the velocity U and continues to move at that
speed. Thus the initial condition and the boundary conditions are, respectively,
u y , 0 ( )= 0,
u , t
( )
= 0,
u 0 , t
( )
= U, for t > 0.
(6.28)
This is known as the Rayleigh Problem, and again a solution is sought, in which v
= 0 everywhere and u satisfies Eqs. (6.27) and (6.28).
y
x
Fluid
Figure 6.4 Rayleigh flow.
To obtain this solution, we try to find a new independent variable q in the
form
q =Byt
n (6.29)
and then to express u as a function of this single variable. Such a method of solu-
tion is called a similarity transformation. Thus if u = u(q), then
cu
ct
=
du
dq
cq
ct
=
n
t
q
du
dq
, (6.30)
188 Fluid Mechanics
cu
cy
=
1
y
q
du
dq
,
(6.31)
c
2
u
cy
2
=
1
y
2
q
2
d
2
u
dq
2
.
(6.32)
Substitution into Eq. (6.27) yields
n
t
du
dq
=
v
y
2
q
d
2
u
dq
2
(6.33)
or
d
2
u
dq
2
n
q
y
2
vt
|
\
|
.
|
|
du
dq
= 0.
(6.34)
If u is to be expressed as a function of q only, no y or t terms should remain in the
differential equation, Eq. (6.34). We therefore choose q such that the combination
y
2
/t is proportional to q
2
and select a convenient B such that Eq. (6.34) becomes
d
2
u
d q
2
+ 2q
du
dq
= 0.
(6.35)
Comparison with Eq. (6.34) requires
q =
y
2 vt
, n =
1
2
, B =
1
2 v
.
The boundary conditions in terms of q are
u = U at q = 0,
u = 0 at q .
(6.36)
Equation (6.35) may be rewritten as
u "
u '
= 2q ,
which is integrated to
lnu' = q
2
+ lnC
1
or
2
1
.
du
C e
d
q
q
=
Another integration yields
u = C
1
e
q
2
dq + C
2
}
.
(6.37)
6. Exact Solutions of NavierStokes Equations 189
The boundary conditions, Eqs. (6.36), then give
u = U 1
2
t
e
q
2
dq
0
q
}
(
( ,
(6.38)
which can also be written in terms of the error function as
u = U 1 erf q | | = U 1 erf
y
2 vt
|
\
|
.
|
(
(
(
.
(6.39)
Obviously, for a given q, u is uniquely determined. However, there are infinite
pairs of y and t which give the same q, all those satisfying
q = y 2 vt ( )
. The same
u is thus obtained for all points on the
y 2 vt ( )= const curve in the y-t plane. In
this sense all these points are similar, and hence the term similarity solution.
Example 6.6
The kinematic viscosity of air at 25
o
C is v
a
= 1.510
-5
m
2
/s, and that of water
is v
w
=10
6
m
2
/s. A large plate immersed in water is suddenly set into motion at
U = 5 m/s, and the resulting flow is similar to that of the Rayleigh problem.
How long will it take a water layer 1 m away from the plate to reach the
velocity of 2 m/s? Repeat for 3 m away and for air.
q 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
erf(q) 0.117 0.223 0.329 0.428 0.520 0.604 0.678 0.742 0.797 0.843
_______________________________________________________________________
q 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
erf(q) 0.880 0.910 0.934 0.952 0.966 0.976 0.984 0.989 0.993 0.995
Table 6.1 Short table of the error function.
Solution
Equation (6.39) states
u =U 1erf q | |
where q = y / 2 vt
( )
. In our case u = 2 and U = 5. Hence,
erf (q) = 1
u
U
= 0.6
190 Fluid Mechanics
and from Table6.1, q =0.6 = y / 2 vt
( )
. Therefore
t
w
=
y / q ( )
2
4v
=
1/ 0.6
( )
2
410
6
= 6.9410
5
s =193 h.
For air, the same yields
t
a
= t
w
v
w
v
a
= 0. 067t
w
= 12. 9 h.
For 3 m,
t
w
3
=t
w
1
3
2
=1,737h and t
a
3
=t
a
1
3
2
=116.1h.
Steady Flow in a Round Tube
We consider a round tube with the z-coordinate as its axis of symmetry. The
tube is filled with an incompressible fluid and extends very far in the _ z-direc-
tions. The boundary conditions are
q = 0 at r = R, (6.40)
and there is complete circular symmetry about the z-axis.
As in the case of flows between parallel plates, we start by trying the possi-
bility
q
r
= q
u
= 0 everywhere.
(6.41)
The continuity equation in polar cylindrical coordinates reads
V q =
1
r
c
cr
(rq
r
) +
c
r cu
q
u
+
c
cz
q
z
= 0,
(6.42)
and by Eq. (6.41)
c q
z
c z
= 0.
Hence
q
z
= w =w r ,u ( )
(6.43)
However, because of the assumed complete axisymmetry of the flow,
w = w r ( ) only, (6.44)
w does not change with z and the velocity profile is the same everywhere. This
flow is, therefore, also fully developed. The NavierStokes equations (5.64) -
6. Exact Solutions of NavierStokes Equations 191
(5.66) become for this case of steady flow
0 =
cP
cr
,
(6.45)
0 =
1
r
cP
cu
,
(6.46)
0 =
cP
cz
+
r
c
cr
r
cw
cr
|
\
|
.
| ,
(6.47)
where P p gh = + is the modified pressure. The first two equations indicate that P
is a function of z only, and in a manner similar to the flow between parallel plates,
one may show that
cP
cz
= const=
AP
Az
.
(6.48)
Thus, the flow equation becomes
1
AP
Az
=
1
r
c
cr
r
cw
cr
|
\
|
.
| ,
(6.49)
which upon integration results in a general solution,
w =
1
4
AP
Az
|
\
|
.
| r
2
+ C
1
+ C
2
lnr .
(6.50)
For a flow in a tube of radius R, with the boundary condition of Eq. (6.40) and
finite velocity at the tube centerline, one obtains
w =
AP
Az
|
\
|
.
|
R
2
4
1
r
R
|
\
|
.
|
2
(
(
(
.
(6.51)
The minus sign indicates that the flow is in the direction of decreasing modified
pressure.
The centerline velocity w
o
is obtained by letting r = 0,
w
o
=
AP
Az
R
2
4
,
(6.52)
and combining Eqs. (6.51) and (6.52) results in
w = w
o
1
r
R
|
\
|
.
|
2
(
(
(
.
(6.53)
The volumetric flowrate may now be obtained from
Q = w(r)2tr dr =
0
R
}
w
o
1
r
R
|
\
|
.
|
2
(
(
2tr dr
0
R
}
, (6.54)
192 Fluid Mechanics
which upon integration yields
Q =
1
2
tR
2
w
o
(6.55)
or in combination with Eq. (6.52)
Q =
tR
4
8
AP
Az
|
\
|
.
| . (6.56)
Equation (6.56) is known as the Poiseuille equation.
The average velocity
w is found from Eq. (6.54) or Eq. (6.55) as
w =
Q
A
=
1
2
tR
2
w
o
tR
2
=
w
o
2
(6.57)
or
w =
AP
Az
R
2
8
. (6.58)
Example 6.7
A heavy oil of viscosity = 1.5 Ns/m
2
flows in a pipe whose diameter is
d = 0.01 m, with a mean velocity of
w =0.1 m/s. Find the pressure gradient. Find
the shear stress at the pipe wall, and show that the pressure drop is balanced by the
shear forces.
Solution
The pressure gradient is found from Eq. (6.58) ,
AP
Az
=
8w
(d / 2)
2
=
8 0.11. 5
0. 005
2
= 48, 000
N/ m
2
m
.
The wall shear stress is found from Eq. (5.59) and Eq. (6.53),
c
rz
=
1
2
cq
r
cz
+
cq
z
cr
|
\
|
.
| =
1
2
cw
cr
= 2w
o
r
2R
2
r=R
=
w
o
R
From Eq. (5.51)
t
rz
= 2c
rz
=
2w
o
R
.
The total shear force per unit length of pipe is
F
s
= 2tRt
rz
=4tw
o
= 8tw = 8t 1.50.1= 3.77N.
The pressure drop over this length, i.e., for 1 z A = , is just 48,000 N/m
2
, and the
6. Exact Solutions of NavierStokes Equations 193
force exerted by this pressure drop is
F
p
=
t
4
d
2
AP =
t
4
0.01
2
48,000 = 3.77N = F
s
.
Flow in an Axisymmetric Annulus
Starting with the same assumptions as in the considerations of the Poiseuille
flow, Eq. (6.50) is obtained for the flow in a cylindrical annulus too. The boundary
conditions for the annulus are, however, different. As in the flow between plates
let us consider first a flow for which there is no axial pressure drop. The inner
cylinder does slide axially relative to the outer one, which is assumed to be at rest.
This kind of motion may be encountered in dies for coating of wire with plastic
material.
The boundary conditions for this case are
w = 0 at r = R
o
,
w = V
i
at r = R
i
,
(6.59)
and with AP / Az = 0, Eq. (6. 50) yields the velocity
w = V
i
ln r / R
o
( )
ln R
i
/ R
o
( )
,
(6.60)
which is the analog of the velocity distribution in plane shear flow.
The volumetric flowrate is obtained as
Q = w(r)2tr dr =
1
2
tR
o
2
V
i
1 R
i
/ R
o
ln R
o
/ R
i
( )
2
R
i
R
o
|
\
|
.
|
2
(
(
(
R
1
R
o
}
. (6.61)
Next, let us consider a case where both cylinders are stationary but there is an
axial pressure drop. The boundary conditions are now
w = 0 at r = R
o
,
w = 0 at r = R
i
.
(6.62)
For this case Eq. (6.50) yields the velocity distribution,
( ) ( )
( )
( )
2 2 2 2
ln / 1
.
4 ln /
o
o o i
i o
r R P
w R r R R
z R R
(
A
| |
=
( |
A
\ .
(6.63)
which is the analog of the plane Poiseuille flow.
The maximum velocity is obtained at
dw/ dr = 0, i.e.,
194 Fluid Mechanics
2r
R
o
2
1
R
i
R
o
|
\
|
.
|
|
2
(
(
(
(
1/ r
ln R
o
/ R
i
( )
|
\
|
.
|
|
= 0
or
r
R
o
|
\
|
.
|
max
2
=
1 R
i
/ R
o
( )
2
2ln R
o
/ R
i
( )
, (6.64)
which can now be substituted in Eq. (6.63) to give the maximum velocity, w
o
,
w
o
=
AP
Az
|
\
|
.
|
R
o
2
4
1
1 R
i
/ R
o
( )
2
2ln R
o
/ R
i
( )
1 ln
1 R
i
/ R
o
( )
2
2ln R
o
/ R
i
( )
(
(
(
|
\
|
.
|
|
. (6.65)
Again, the volumetric flowrate is obtained as
( )
( )
2
2
4
4
1 /
( )2 1
8 ln /
o
i
R
i o
o i
o o i R
R R
P R R
Q w r r dr
z R R R
t
t
| |
(
| | A
| | |
= =
| |
|
A
\ .
\ .
|
\ .
}
(6.66)
and the average velocity comes out as
( )
( )
( )
2 2
2
2 2
1 /
1 .
8 ln /
i o o i
o o i
o i
R R Q Q P R R
w
A z R R R
R R
t
(
| | A
| |
= = = + (
| |
A
\ .
( \ .
(6.67)
Superposition of the solutions Eqs. (6.60) and (6.63) yields flows corre-
sponding to combinations where one of the cylinders moves and an axial pressure
gradient exists. Such flows are schematically shown in Fig. 6.5.
U U U
R
R
o
R
i
R
o
R
i
a. b. c. d. e.
Figure 6.5 Velocity profiles in pipe and annulus flows.
a. Poiseuille flow. b. Poiseuille annulus flow.
c. Annulus shear flow. d. Shear plus axial pressure gradient.
e. Shear minus axial pressure gradient.
6. Exact Solutions of NavierStokes Equations 195
Example 6.8
An experimental system consists of a long tube of radius R
o
filled with glyc-
erin. The pressure gradient along the tube AP/Az = dP/dz is given, and the flow is
fully developed. The temperature in the glycerin at the center of the tube is sought,
and a suggestion is made to stretch a thin wire along the tube axis, as shown in Fig.
6.6. It is claimed that since the wire is very thin, the flow field is only slightly
modified by the presence of the wire. Assess the modification.
Figure 6.6 Tube with wire at its axis.
Solution
With the wire in the middle, the flow is an annular flow. To assess the modi-
fication caused by the wire, we consider the ratio of maximum velocity of the
annular flow to that of the tube flow and the ratio of the average velocity of
annular flow to that of the tube flow.
We use the subscript p for the pipe without the wire and w for the tube with
the wire, i.e., for the annulus. The ratio of maximum velocities is found by the
division of Eq. (6.65) by Eq. (6.52) resulting in
w
w
w
p
|
\
|
.
|
|
max
= 1
1 R
i
/ R
o
( )
2
2ln R
o
/ R
i
( )
1 ln
1 R
i
/ R
o
( )
2
2 ln R
o
/ R
i
( )
(
(
(
(
.
The ratio of the average velocities is found from Eqs. (6.67) and (6.58),
w
w
w
p
= 1+
R
i
R
o
|
\
|
.
|
|
2
1 R
i
/ R
o
( )
2
ln R
o
/ R
i
( )
.
Numerically computed values of the location of the maximum velocity,
Eq.(6.64), values of (w
w
/w
p
)
max
and values of w
w
/w
p
are given in Table 6.2 for
several ratios of R
i
/R
o
. As seen from the table, the modification of the flow is not
small.
196 Fluid Mechanics
0.01 0.32949 0.65038 0.78297
0.02 0.35744 0.60936 0.74488
0.05 0.40803 0.53503 0.66953
0.10 0.46365 0.45456 0.58005
0.20 0.54611 0.34093 0.44352
0.30 0.61475 0.25434 0.33417
0.50 0.73553 0.12664 0.16798
0.70 0.84554 0.04516 0.06013
0.90 0.94956 0.00500 0.00667
R
i
R
o
r
R
|
\
|
.
|
max
w
w
w
p
|
\
|
.
|
max
w
w
w
p
|
\
|
.
|
mean
Table 6.2 Numerical values for w
max
and w
mean
for annular flow in
comparison to flow in a pipe at equal pressure gradient.
Example 6.9
A metal wire of radius R
i
= 2 mm is pulled vertically upward with the speed
V
i
through a long pipe of radius R
o
= 5 mm, Fig. 6.7. The gap between the wire and
the pipe is filled with molten plastic material of density and viscosity . As the
wire comes out of the pipe, it carries on its surface a layer of plastic, which cools
and solidifies. The thickness of the solid layer is 0.1 mm. Find the speed with
which the wire is pulled upward.
Solution
The net flowrate of the molten plastic between the wire and
the pipe must supply the coating. The volumetric flowrate of the
0.1-mm-thick solidified plastic coating moving upward with the
wire at a speed V
i
is
Q = AV
i
,
where A is the cross-sectional area of the annulus formed by the
plastic coating,
( )
2
4 2 6 2
10 1.288 10 m .
i i
A R R t
(
= + =
(
This flow must equal the sum of the flow induced by the motion of the wire
alone, Eq. (6.61), and of the flow downward due to gravity. The latter can be
obtained from Eq. (6.66) written for a flow under the influence of a modified pres-
Wire coating.
6. Exact Solutions of NavierStokes Equations 197
sure drop AP/ Az. In our case of constant pressure p
o
, we have
P = p
o
+gh
and
AP
Az
= g
Ah
Az
= g.
This expression for pressure drop is substituted into Eq. (6.66), which is combined
with Eq.
(6.61) to yield the expression for the flowrate induced by the motion of
the wire coupled with the effects of gravity,
( )
( )
2
2
2 4
2 2
0
0
1 /
1 /
4 2 1 .
8
ln ln
i
i o i i o i
i
o i o o o
i i
R R
R R R V R gR R
Q AV
R V R R R
R R
t
( | | | |
( | |
(
| | | |
( | |
= =
| |
( | |
| | | |
\ . \ .
( | |
| |
( \ . \ . \ . \ .
The term containing
V
i
is extracted, leading to
V
i
gR
o
2
=
1
R
i
R
o
|
\
|
.
|
4
1 R
i
/ R
o
( )
2
(
2
ln R
o
/ R
i
( )
8
1 R
i
/ R
o
2 ln(R
o
/ R
i
)
R
i
R
o
|
\
|
.
|
2
A
tR
o
2
(
(
(
and substitution of R
i
= 0.002 m, R
o
= 0.005 m, A = 1.28810
-6
m
2
gives
V
i
gR
o
2
= 0.16915
or
V
i
=
0.16915gR
o
2
.
For = 1,000 kg/m
3
, g = 9.81 m/s
2
and = 2 N
.
s/m
2
we obtain
V
i
= 0.0207 m/s ~ 2.1 cm/s.
Flow between Rotating Concentric Cylinders
In this case, we again consider the annular region between two cylinders, as
in the previous case. This time, however, the cylinders rotate with the angular
velocities, O
i
and O
o
, and the boundary conditions are
198 Fluid Mechanics
q
r
= q
z
= 0, q
u
= R
i
O
i
at r = R
i
,
q
r
= q
z
= 0, q
u
= R
o
O
o
at r = R
o
.
(6.68)
We consider the cylinders being long and the flow steady, resulting in
/ 0 z c c = and / 0 t c c = . We assume q
r =
q
z
= 0 everywhere, and obtain from the
continuity equation
c
cu
q
u
= 0.
(6.69)
Hence, q
u
can depend on r only.
The NavierStokes equations (5.64) (5.66) with the modified pressure,
Eq. (6.13), substituted become for this case
1
r
q
u
2
|
\
|
.
|
=
dP
dr
,
(6.70)
d
dr
1
r
d
dr
rq
u
( )
(
( = 0.
(6.71)
with the solution of Eq. (6.71) for the velocity profile
q
u
= Ar +
B
r
.
(6.72)
The integration constants are evaluated with the help of the boundary conditions
given in Eq. (6.68), resulting in the velocity profile
( ) ( )
( )
2
2
2
/
1 /
i o i o i o i
i o
R R r R r
q
R R
u
O O O O
(
+
=
(6.73)
This is known as Couette flow
Couette flow is the only example in this chapter where some nonlinear term
remained in the equations at the solution stage. It is noted that the part actually
solved, Eq. (6.71), is still linear and that the nonlinearity appears only in Eq. (6.70),
and even then as a given function, i.e., the solution, Eq. (6.72), is now substituted
into Eq. (6.70), which is then solved for the modified pressure P by direct
integration.
For the special case where the outer cylinder is stationary (O
o
= 0) while the
inner cylinder is rotating with an angular velocity O
I
, Eq. (6.73) simplifies to
( ) ( )
( )
2
2
2
/
.
1 /
i i o o
i o
R R r R r
q
R R
u
O
=
(6.74)
An instrument consisting of an outer stationary and inner rotating cylinder is
called a Couette viscometer and is used to measure the viscosity of liquids.
6. Exact Solutions of NavierStokes Equations 199
Example 6.10
Two long concentric cylinders have the radii
R
i
and R
o
. The fluid between
them has a density and a viscosity . Consider the following cases:
a. O
i
= 0, O
o
= O
b. O
I
= O O
o
= 0
c. O
i
= O / 2 O
o
= O / 2
Find the velocity distribution, the pressure distribution and the moments acting on
both cylinders in cases a, b and c.
Solution
a. For O
i
= 0, O
o
= O Eq. (6.73) simplifies to
q
u
= O
r R
i
2
/ r
1 R
i
/ R
o
( )
2
.
The pressure distribution is given by Eq. (6.70) as
( )
( )
2 2 4 3
2
2
2
2 / /
.
1 /
i i
i o
r R r R r
dP q
dr r
R R
u
O
(
+
= =
(
and
( )
4 4
2 2
2 2
1 2
2
2ln
2
.
1 /
i
i
i i
i o
r R r
R
R r R
P P
R R
O
(
(
= +
(
The shear stress at, say, the inner surface, is, Eq. (5.51),
t
ru
= 2c
ru
= r
d
dr
q
u
r
|
\
|
.
|
r =R
i
=
2O
1 R
i
/ R
o
( )
2
,
and the shear moment per unit height of cylinder becomes
M = 2tR
i
2
t
ru
=
4tOR
i
2
R
o
2
R
o
2
R
i
2
.
The same moment acts, of course, on the outer cylinder.
200 Fluid Mechanics
b. For O
I
= O, O
o
= 0, the velocity distribution is given by Eq. (6.74):
( ) ( )
( )
2
2
2
/
.
1 /
i o o
i o
R R r R r
q
R R
u
O
=
The pressure distribution is
( ) ( )
( )
4
2 2 4 3
2
2
2
/ 2 / /
.
1 /
i o o o
i o
R R r R r R r
dP q
dr r
R R
u
O
(
+
= =
(
and
( )
4 2 4 2
4
2 2
2 2
1 2
2
1 1 1 1
2 ln
2 2 2 2
.
1 /
i o o o
i
o i i i i i
i o
R r R R R r
R
R R R r R R R
P P
R R
O
(
| | | | | | | | | |
+ (
| | | | |
( \ . \ . \ . \ . \ .
= +
(
The shear stress at, say, the outer surface is, Eq. (5.51),
t
ru
= 2c
ru
= r
d
dr
q
u
r
|
\
|
.
|
r = R
o
=
2O R
i
/ R
o
( )
2
1 R
i
/ R
o
( )
2
,
and the shear moment per unit height becomes
M = 2tR
o
2
t
ru
=
4tOR
i
2
R
o
2
R
o
2
R
i
2
for both inner and outer cylinders.
c. For O
i
= O / 2, O
o
= O / 2, Eq. (6.73) yields
( )
( )
2
2
2
/ 1
.
1 /
i o o
i o
r R R R r
q
R R
u
O
| |
(
+
|
\ .
=
The pressure distribution is
( )
2
2 2
2 4
2
3
2
2
2
2
1 1
1 /
i i i i
o o
i o
R R R R
r
R r R r
dP q
dr r
R R
u
O
| |
( (
| | | |
|
+ + + ( (
| |
|
( ( \ . \ .
\ .
= =
(
and
6. Exact Solutions of NavierStokes Equations 201
P = P
i
+
O
2
r
2
R
i
2
2
|
\
|
.
|
|
R
i
R
o
|
\
|
.
|
2
+1
(
(
(
2
2R
i
2
R
i
R
o
|
\
|
.
|
2
+1
(
(
(
ln
r
R
i
+
R
i
4
2
1
R
i
2
1
r
2
|
\
|
.
|
|
1 R
i
/ R
o
( )
2
The shear stress at, say, the inner radius is
t
ru
=
2O
1 R
i
/ R
o
( )
2
,
and the shear moment for unit height becomes
M = 2tR
o
2
t
ru
=
4tOR
i
2
R
o
2
R
o
2
R
i
2
.
Note that the moments come out to be the same in all three cases, but not the
pressures. An explanation for this is that rotating coordinate systems can be found
in which cases b and c are kinematically reduced to case a. However, the dynamics
in the rotating system requires the inclusion of appropriate D'Alambert forces,
which modify the pressure.
References
R.B. Bird, W.E. Stewart and E.N. Lightfoot, "Transport Phenomena," Wiley, New
York, 1960.
W.F. Hughes, "An Introduction to Viscous Flow," Hemisphere, Washington, DC,
1979.
H. Schlichting, "Boundary-Layer Theory," 7th ed., McGraw-Hill, New York, 1979,
Chapter 5.
Problems
6.1 The space between two long parallel plates (Fig. P6.1) is filled with a fluid
of viscosity = 910
3
dyn
.
s/cm
2
. The upper plate moves with a velocity of
3 m/s and the lower plate is stationary. What is the shear stress distribution
in the fluid?
202 Fluid Mechanics
y
x
U= 3 m/s
h = 2 cm
Figure P6.1
6.2 Coating of electric wire with insulating material is done by drawing the wire
through a tubular die as shown in Fig. P6.2. The viscosity of the coating
material is 100 poise. Simplify the flow equations for this case and calculate
the force F required to draw the wire.
Figure P6.2
Wire coating.
6.3 A laminar layer of glycerin slides down on a
semi-infinite vertical wall, Fig. P6.3.
a. Calculate the shear stress on the wall and
the flowrate of the glycerin.
b. What will be the answers to part a if the
wall leans at an angle o ?
6.4 An instrument for measuring viscosity con-
sists of a rotating inner cylinder and station-
ary outer cylinder as shown in Fig. P6.4. The
inner cylinder rotates at 3,600 rpm and the
viscosity of the fluid is (1) 10 poise, (2) 100
poise.
a. What is the moment acting on the outer
cylinder in the two cases?
b. What is the efficiency of this instrument
as a hydraulic transmission of moment?
Find the moment as a function of the rpm
of the outer cylinder and the power
transmitted.
Figure P6.4 Viscometer.
o
Glycerin
= 1.26 g/cm
3
= 10 poise
Figure P6.3
Fluid
100 mm
10 mm
20 mm
6. Exact Solutions of NavierStokes Equations 203
6.5 A long pipe bend, Fig.P6.5, has the inner diameter of 0.025 m and is 2 m
long. A solution of sugar in water, which has
the same density as water but whose
viscosity is ten times that of water, flows
through the bend. Measurements show that
the pressure distribution along the pipe is the
same as that in a straight pipe of the same
dimensions. At point B the pressure is the
outside pressure, i.e., atmospheric. The
flowrate is 6 kg/s. Neglect gravity and cal-
culate the forces transmitted through flange A.
Figure P6.5 Pipe bend
6.6 A very wide, shallow layer of water is approximately two-dimensional. The
bottom is a rigid plate with a small inclination angle o , Fig. P6.6, and the
constant water depth, measured vertically, is h. The flow is assumed fully
developed.
a. Write the simplified form the NavierStokes equations assume in this
case. Write the boundary conditions at the bottom and the top of the
layer.
b. Solve the equation and obtain the velocity profile and the mean velocity.
c. Find the shear stress at the bottom.
o
Figure P6.6 Shallow water layer.
6.7 A two-dimensional air bearing consists of an upper plate A that is 2L m
wide and 1 m long and a lower plate B of the same size but with a slot in its
middle, Fig. P6.7. Air is forced into the slot through a series of pipes C and
comes out at the edges. As a result, the upper plate, which carries a load F,
is raised to the height h. Assuming laminar flow find the air supply pressure
necessary to support F, and find the supply rate necessary to maintain the
height h. Both depend on L, of course.
.
A
B
i
j
M
204 Fluid Mechanics
2L
A
B
h
C
Figure P6.7 Air bearing.
6.8 The distance between two parallel plates in a two-dimensional flow, Fig.
P6.8, is d = 0.05 m. The gap between the plates extends to infinity and is
filled with a fluid whose density is = 1,000 kg/m
3
and whose viscosity is
= 500 poise. The lower plate is stationary and the upper one moves in the x-
direction with the velocity of 1 m/s.
y
x
U= 1 m/s
d = 0.05m
Figure P6.8 Stationary and moving plates.
a. If p/x = 0, find the mass flux between the plates.
b. What p/x is necessary to make the net mass flow vanish?
c. What p/x is necessary to have the same mass flux as in a, but in the
other direction?
d. Find p/x needed to double the mass flux of part a.
6.9 The servomechanism in Fig. P6.9 consists of a piston, A, equipped with
rods, B, moving inside a cylindrical sleeve, C. The servomechanism is filled
with oil and a gear pump maintains a pressure difference Ap between the
two sides of the piston. Once started, the piston motion is approximately at a
constant speed V and it moves until its side is flush with the sleeve edge,
i.e., the total length of its travel from one edge to the other is L
2
L
1
. The
time required for this motion is At = (L
2
L
1
)/V and the force which the
servomechanism must put out is F.
6. Exact Solutions of NavierStokes Equations 205
=
Figure P6.9 Servomechanism.
All the dimensions shown in Fig. P6.9 are given, and so are At, F, and the
viscosity of the oil, . Find the pressure difference Ap that the pump must
maintain and the volumetric flux of oil, Q, it must supply. You may assume
fully developed flow in the gap between the piston and its cylindrical sleeve
and also that (d
2
d
1
)/d
1
<< 1, i.e., the gap need not be considered as an
annulus but may be taken as that between two plates.
Note that the force applied to the piston by the oil consists of two parts:
pressure on the flat surface and shear on the curved surface. The shear force
is to the right for small V but to the left for large V.
6.10 Repeat Problem 6.9 without the assumption (d
2
d
1
)/d
1
<< 1. The gap must
now be considered an annulus.
6.11 With reference to Problems 6.9 and 6.10:
d
1
= 60 mm, d
2
= 63 mm, d
3
= 10 mm,
L
1
= 80 mm, L
2
= 120 mm, F = 2000 N.
= 20 poise, At = 5 s.
Compute Ap [Pa] and Q [m
3
/s] for the gear pump, using once the simpler
analysis of Problem 6.9 and once the more elaborate one of Problem 6.10.
Compare the results and decide whether the approximation of Problem 6.9
is sufficient for (d
2
- d
1
)/d
1
= 0.05.
F
d
3
p
A
B
V
L
2
pump
p + A
p
C
L
1
d
2
d
1
206 Fluid Mechanics
6.12 The viscosity of a given fluid is = 210
-3
kg/m
.
s. The fluid flows in a pipe
that has a diameter of 0.025 m and is 20 m long. The average flow velocity
is 1 m/s. The pipe opens to the atmosphere. Calculate the flowrate and the
pressure at the entrance to the pipe.
6.13 It is suggested to change the pipe described in Problem 6.12 and use two
smaller pipes for the same total flowrate and the same average velocity in
each pipe, i.e., 1 m/s.
a. Calculate the required pressure drop in these pipes.
b. Now, repeat the calculations for a pressure drop along the pipes, which
is equal to that of Problem 6.12, and find the average velocity and the
flowrate in the two pipes.
6.14 The same amount of fluid as in Problem 6.12 must now be supplied through
a 0.0125-m-diameter pipe.
a. Calculate the pressure drop in this pipe and compare your result with
that of Problem 6.12.
b. Note that for a constant flowrate, the pressure drop is proportional to the
diameter of the pipe raised to the power n. Find n.
c. Also note that for a constant pressure drop the flowrate is proportional to
the diameter of the pipe raised to the power s. Find s.
6.15 A fluid flows in a pipe that has a diameter D. The same fluid flows in the
gap between two parallel flat plates. The size of the gap is also D. The same
pressure gradient exists in both systems. Calculate the ratio between the
flowrate of the fluid in the pipe and the flowrate between the plates, per
width of D.
6.16 The pipe and the plates of Problem 6.15 are now used such that the flowrate
in the pipe is the same as that between the plates, per width D. Calculate the
ratio between the power needed to pump the fluid through the pipe and that
needed to pump between the plates.
6.17 The velocity distribution in a fully developed flow through a rectangular
square duct with the sides D is, approximately,
q = kw = kCsin
tx
D
sin
ty
D
,
where C is a constant.
a. Calculate the flow through this duct and compare with those in the pipe
and the plates as in Problem 6.15, for the same pressure gradients.
b. Calculate the power needed to pump the fluid and compare with the pipe
6. Exact Solutions of NavierStokes Equations 207
and the plates as in Problem 6.16.
6.18 A concentric cylinder viscometer is shown in
Fig. P6.18. The height of the inner cylinder is
0.15 m, and its diameter is D
i
= 0.10 m. The inner
diameter of the outer cylinder is D
o
= 0.11 m. The
viscometer is used to measure viscosities in the
range of = 210
-3
kg/m
.
s. The outer cylinder is
stationary and the inner one turns at the rate of
3,000 rpm
a. Find the range of the moments measured.
b. It is necessary to measure viscosities in the
range of = 0.2 kg/m
.
s but to have the range
of moments measured unchanged. One pos-
sibility is to decrease the number of revolu-
tions of the inner cylinder. Calculate this new
number.
c. Another possibility is to keep the turning rate
but to decrease the diameter of the inner
cylinder. Calculate this new diameter.
6.19 The viscometer described in Problem 6.18, before the suggested modifica-
tions are made, is used to measure the viscosity of water at room tempera-
ture. The inner cylinder of the viscometer is suddenly set to rotate at 3,000
rpm Assume that even at the transient stage of the flow there exist velocities
in the angular direction only and that for very short times the flow field
between the two cylinders resembles the solution of the Rayleigh flow. Find
the time at which the velocity at r = 0.0501 m is 90% of that of the surface of
the inner cylinder. Now, without using again the error function, but rather
using the similarity properties of the Rayleigh flow, find times at which this
velocity appears at r = 0.0502 m, at r = 0.0503 m and at r = 0.0504 m.
6.20 The pressure gradient in a laminar flow in a pipe is given by Eq. (6.58). An
engineer wants to express the modified pressure difference between two
points along a pipe by an equation of the form
P
1
P
2
= C
f
L
D
w
2
2
|
\
|
.
|
|
,
where L is the distance between the two points, D is the pipe diameter,
w is
the mean velocity and C
f
is a "friction coefficient." Using Eq. (6.58), find an
Figure P6.18
0.15 m
D
1
D
2
208 Fluid Mechanics
expression for C
f
.
6.21 Two flat plates are set as shown in Fig.
P6.21. The lower plate is stationary
and the upper plate moves to the right
at the velocity of 2 m/s. The pressure at
point 1 is p
1
= 100,000 Pa, and the
pressure at point 2, which is 1 m
further along the plates, is p
2
, as
shown. The viscosity of the fluid be-
tween the plates is = 210
-2
kg/m
.
s.
a. Find p
2
for which the net flow between the plates is nil.
b. Find p
2
for which the flow per unit width is 0.10 m
3
/m
.
s.
c. Find p
2
for which the flow per unit width is ( 0.10)
m
3
/m
.
s.
6.22 A long cylinder with its lower side closed is shown in
Fig. P6.22. The cylinder contains a fluid with the vis-
cosity of = 310
-3
kg/m
.
s. A piston in the form of a
long cylinder moves into the fluid at the rate of 1 m/s.
The flow between the cylinder and the piston is assumed
fully developed.
When flow starts, the length of the inserted part of the
piston is 0.60 m, and the length it can still travel is
0.40 m.
a. Find the velocity profile in the gap between the
piston and cylinder.
b. Find all the points where the velocity of the fluid is
zero.
c. Find the shear stress and shear rate at the cylinder
wall.
d. Find the force pushing the piston.
6.23 A viscometer is designed along the general form of the configuration shown
in Figure P6.22, and is used to measure viscosities in the range
0.001 - 0.02 kg/m
.
s.
a. As a first step in the design use the dimensions of Fig. P6.22 and calcu-
late the time needed for the piston to travel 0.1 m, provided it is pushed
down with a force of 10.0 N.
b. It is desired that the times to be measured are of the order of 10.0 s.
Suggest convenient weights to be used to push the piston down for the
Figure P6.21
Figure P6.22
d
V = 2 m/s
1
2
p
1
2
p
d = 0.01 m
0.1m
0.098
piston
cylinder
fluid
6. Exact Solutions of NavierStokes Equations 209
range of viscosities measured.
6.24 A shock absorber is shown in Fig. P6.24. The viscosity of
the fluid is 110
-3
kg/m
.
s. Assume the flow is laminar and
fully developed. Find the characteristic behavior of this
shock absorber, i.e., the force that resists the motion of the
piston as a function of the speed of the piston.
6.25 Water flows at the rate of 0.05 m
3
/s through the annular gap
with an inner radius of 0.10 m and an outer radius of
0.12 m. The length of the annulus is 100 m. Assume the
flow laminar and fully developed.
a. Find the pressure drop along the annulus.
b. Find the magnitude and location of the maximum
velocity in the annular gap.
6.26 Careful measurements show that there is a 0.003m eccentricity in the
annulus described in Problem 6.25, i.e., there is a distance of 0.003 m
between the centers of the inner cylinder and the outer one. The flow in this
eccentric annulus is still 0.05 m/s and is still assumed fully developed. An
experienced engineer suggests that calculations for a concentric annulus
having an inner radius of 0.1 0.003 m and an outer radius of 0.12 m or an
inner radius of 0.10 m and an outer radius of 0.12 0.003 m, both yield
upper bounds to the pressure drop. He also suggests that calculations for a
concentric annulus having an inner radius of 0.1 0.003 m and an outer
radius of 0.12 m or an inner radius of 0.1 m and an outer radius of 0.12 +
0.003 m both yield lower bounds to the pressure drop. Find the better upper
and lower bounds to the pressure drop.
6.27 The vertical concentric annulus in Fig. P6.27 has
an inner radius of 0.100 m and an outer radius of
0.141 m. The outer cylinder is stationary and the
flow in it is upward. The wall of the inner cylin-
der is very thin and there is a downward flow
inside the inner cylinder. The same fluid flows in
the inner cylinder and in the annulus. Neglect the
weight of the thin-walled inner cylinder, and find
the ratio between the two flow rates necessary to
keep the inner cylinder floating, i.e., such that the
shear force on the inside of its walls is just bal-
anced by that on the outside.
Figure P6.24
Figure P6.27
0.081
0.08
V
0.2 m
0.1
0.141
210 Fluid Mechanics
6.28 A gas turbine can operate using
several different fuels. It is desirable
to measure the density of the fuel
while it is flowing in a pipe to the
combustion chamber. A scheme to do
this is shown in Fig. P6.28. The fuel
has the viscosity and the density .
Find expressions for the pressure
drop between points 1 and 2 and
between points 3 and 4. Show that
indeed
p
1
p
2
( )+ p
4
p
3
( )= 2gL,
i.e., the density may be thus obtained from pressure drop measurements,
though the viscosity and the velocity of the fluid are not known.
6.29 The pipe shown in Fig. P6.28 has a diameter of 0.025 m; the lengths
between points 1 and 2 and between points 3 and 4 are 2 m each. The pipe
carries fuel with a viscosity of 210
-3
kg/m
.
s and a density of 900 kg/m
3
at a
velocity of 3 m/s. Assuming fully developed laminar flow, find the pressure
drops between points 1 and 2 and between points 3 and 4.
6.30 A very long pencil of 1 cm in diameter is centrally located in a pipe, Fig.
P6.30. Water flows in the pipe at the rate of 0.005 m
3
/s. The pencil density
is the same as that of water. The pencil is at first held in place, and then it is
released.
a. Neglect the effects of the ends of the pencil and find its acceleration at
the moment of release.
b. Find the velocity of the pencil once it has reached steady motion. Esti-
mate the error made by not taking into account the effect of the edges in
the calculation of the acceleration.
d 0.03 m
Figure P6.30
Figure P6.28
6. Exact Solutions of NavierStokes Equations 211
6.31 Figure P6.31 shows three flat plates. The upper and lower ones are station-
ary, and the middle one can move in its own plane. The x-wise pressure
gradient in the fluid above the moving plate is +300 Pa/m, and that in the
fluid below is 300 Pa/m.
Find the velocity of the midplate.
Figure P6.31.
212 Fluid Mechanics
213
7. ENERGY EQUATIONS
Bernoulli's Equation
To obtain a clear physical appreciation of the various terms that appear in the
energy equation, we begin our discussion by considering an idealized frictionless
case.
It seems that the notion of conservation of mechanical energy was first
inferred from the integration of the equations of motion of solids, idealized for no-
friction situations. Newton's equation
md
2
x / dt
2
= F
x
can be integrated by multi-
plication with
dx / dt =V to become
m
dx
dt
d
2
x
dt
2
=
1
2
m
d
dt
dx
dt
|
\
|
.
2
= F
x
dx
dt
or
2
,
2
x
V
md F dx
| |
=
|
\ .
which can be interpreted as the change in the kinetic energy being equal to the
work of the external force. The extension to three-dimensional motions and to
forces that correspond to potentials, hence the appearance of potential energy, is
not difficult. The inclusion of rotational motions is slightly more complicated, but
still leads to the same result, i.e., that the sum of the kinetic and potential energies
of a rigid body changes by exactly the amount of work performed on the body.
We use this result of classical mechanics as a guideline and begin by rewrit-
ing the Euler equation, i.e., the equations of motion for frictionless flows. To sim-
plify the analysis we consider only time-independent flows, and Eq. (5.70) with its
left-hand side expanded using Eq. (5.11) becomes
2
1 1
.
2
q p
V V = V + q q g (7.1)
214 Fluid Mechanics
We further restrict the flow to be incompressible. Compressible flows are consid-
ered later.
Let the body force g be expressible as
g = g Vh . (6.11)
Hence, h is the elevation potential, and the negative sign simply means that we
want objects to fall "down." Equation (7.1) may now be written as
q V q =
1
Vp gVh
1
2
Vq
2
or
q V q = V
1
2
q
2
+
p
+ gh
(
(
. (7.2)
The expression in the square brackets is known as the Bernoulli polynomial. We
first seek the rate of change of this expression along a streamline: Let s be the unit
vector tangent to a streamline. The scalar product of s with Eq. (7.2) is
s q V q ( ) = s V
1
2
q
2
+
p
+ gh
(
(
=
c
cs
1
2
q
2
+
p
+ gh
(
(
,
(7.3)
where /s means, of course, the directional derivative, i.e., the rate of change of
the expression in the square brackets in the direction of s.
The vector s is parallel to q and therefore perpendicular to ( ) V q q . Hence
( ) 0 V = s q q
,
and therefore
c
cs
1
2
q
2
+
p
+ gh
(
(
= 0, (7.4)
or
q
2
/ 2+ p/ + gh = B does not change along s. Since s lies on a streamline, B is
conserved along streamlines, i.e., for two points on the same streamline,
1
2
q
1
2
+
p
1
+ gh
1
=
1
2
q
2
2
+
p
2
+ gh
2
.
(7.5)
In engineering practice the equation is sometimes divided by g, resulting in
7. Energy Equations 215
p
1
+ h
1
+
q
1
2
2g
=
p
2
+ h
2
+
q
2
2
2g
, (7.6)
where all terms have the dimensions of length. Engineers use the expressions
pressure head, static head and velocity head to denote these various terms,
respectively. Equation (7.6) is known as Bernoulli's equation.
Returning to Eq. (7.3), we now let s have the direction of V = q , the so-
called vorticity vector. Because ( ) V = q q q is also perpendicular to , the
same argument that led to the conservation of B on streamlines now leads to its
conservation along vorticity lines, i.e., along the field lines tangent to every-
where. Streamlines and vorticity lines intersect and form surfaces, called Bernoulli
surfaces. Each point on a Bernoulli surface can be reached from any other point on
the same surface by going along streamlines and vorticity lines. Because
Bernoulli's polynomial is conserved along both these families of lines, it is
conserved on Bernoulli surfaces.
So far Bernoulli's equation (7.6) has been shown to hold on Bernoulli
surfaces only. For such cases the equation is referred to as the weak Bernoulli
equation. There are flow fields in which Bernoulli's polynomial has the same value
everywhere, i.e., not only on Bernoulli's surfaces. This situation is said to satisfy
the strong Bernoulli equation. The two most important flows satisfying the strong
Bernoulli equation are those for which:
a. V q = 0 in the whole domain of flow. Equation (7.2) then becomes
2
0
2
q p
B gh
(
V = V + + =
(
(7.7)
and, obviously, B is conserved in the whole domain.
b. Every point in the domain of interest can be traced back along a streamline to
another domain, where the strong Bernoulli equation holds. The flow field in
this other domain may be trivially simple in many cases, e.g., at a great
distance in front of an approaching airplane. Because of the weak Bernoulli
equation B values are conserved along streamlines; and since all the stream-
lines in the domain of interest originate in the other domain, where B values
are one and the same, the streamlines carry with them their B values into the
new domain.
Mathematically, Bernoulli's equation is an intermediate integral of the
equations of frictionless motion, i.e., it is an algebraic relation between the
dependent variables. It is a useful tool in calculations of flows, and even when
the flow in general is not frictionless, the equations can still serve as a good
216 Fluid Mechanics
approximation in many engineering situations.
Example 7.1
Figure 7.1 shows an airplane wing, which is to be analyzed for a constant
velocity flight. Show that once the velocity field around the wing is known, the
strong Bernoulli equation holds, which may then be used to calculate the pressure
on the wing and hence the lift force of the wing.
Figure 7.1 Wing in a flow field.
Solution
All streamlines in the vicinity of the wing started very far in front of it. A
stationary observer located very far in front of the wing sees its environment at
rest, with its pressure the atmospheric one. For him B = p = p
o
= const, and the
strong Bernoulli equation holds. A moving observer, sitting on the wing, sees the
vicinity of the first observer moving toward him with the velocity of the wing, V.
For the second observer the Bernoulli polynomial on all streamlines very far in
front is the same, being B =p
o
/ +V
2
/2. Thus all these streamlines carry with them
their Bernoulli polynomials all over the vicinity of the wing, which therefore
satisfies the strong Bernoulli equation. The pressure near the wing, where the
velocity is q, can therefore be obtained as
( )
2 2
2
o
p p V q
= +
Example 7.2
Figure 7.2 shows a device called a venturi tube, which consists of a
converging pipe section followed by a diverging section. The device is used to
measure the flowrate of a fluid in a pipe. The velocity of the fluid in section 1 is
lower than that in section 2, which results in a pressure difference between the
sections, according to the Bernoulli equation. This pressure difference is measured
and used to calculate the flow rate. Derive expressions for velocity and mass
flowrate to be used for the venturi tube.
7. Energy Equations 217
Solution
Neglecting friction, we write
Bernoulli's equation (7.6):
p
1
+
v
1
2
2g
=
p
2
+
v
2
2
2 g
,
where, for simplicity, we have as-
sumed
h
1
= h
2
.
Because of continuity,
v
2
= v
1
A
1
A
2
,
and hence
p
1
p
2
=
2
v
2
2
v
1
2
( )
=
2
v
2
2
1
A
2
A
1
|
\
|
.
|
|
2
(
(
(
(
=
2
v
2
2
1
d
2
d
1
|
\
|
.
|
|
4
(
(
(
(
,
or
v
2
=
2
p
1
p
2
( )
1 d
2
/ d
1
( )
4
.
and
m =
t
4
d
2
2
v
2
=
t
4
d
2
2
2 p
1
p
2
( )
1 d
2
/ d
1
( )
4
.
Example 7.3
Old train engines operated without condensers and had to replenish the water
in the boiler rather frequently. One way to do this without having to stop the train
was to lower a water scoop, as shown in Fig. 7.3, into an open water tank laid
beside the railway.
For a train moving to the left with the speed V
1
and a water tank, which
permits a 50 m run with the scoop lowered, find how much water can be taken into
the train and what resistance force is applied to the train.
1
2
d
2
V 1
d
1
Figure 7.2 Venturi tube.
218 Fluid Mechanics
o
2
2
1
z = 0
V
1
o
1
H
Figure 7.3 Water scoop.
Solution
An observer sitting on the scoop observes a sheet of water of thickness o
1
entering the scoop with the relative velocity V
1
to the right. The average height of
the layer is o
1
/2 from the bottom of the scoop, and its average pressure is
p
o
+ o
1
/ 2 , p
o
being the atmospheric pressure. The Bernoulli equation, (7.6), may
now be written for a streamline connecting points 1 and 2:
p
o
+ o
1
+
V
1
2
2g
=
p
o
+ H
o
2
2
|
\
|
.
|
+
V
2
2
2 g
.
Hence the exit velocity from the scoop V
2
, relative to the observer, is given by
V
2
2
=V
1
2
2H o
2
2o
1
( )g
,
which is less than
V
1
2
. Thus, because of continuity,
1
2 1 1
2
V
V
o o o = >
and the water layer leaving the scoop is thicker. We assume, however, that o is
much smaller than H and that V
2
may be approximated by
V
2
2
= V
1
2
2gH.
For a scoop b meters wide, the amount of water taken is 50o
1
b, which takes the
train t = 50/V
1
seconds to fill.
To find the force transferred by the scoop to the train, we may use the integral
form of the momentum theorem, Eq. (4.45),
F
s
= ib o
1
V
1
2
+ o
2
V
2
2
| | .
The pressure terms cancel out because we assume the water to follow the wet back
7. Energy Equations 219
side of the scoop. This is not quite correct, but the error involved is not more than
the mean hydrostatic pressure there, i.e.,
bo
1
2
.
Still, this is not the resistance force applied to the train. The water layer
leaving the scoop, Fig. 7.3, enters a tank where it settles to the train velocity. Thus
the momentum
ibo
2
V
2
2
is destroyed in this tank, and the tank transfers to the train
a force
F
T
= i bo
2
V
2
2
.
The resistance force is therefore
F
R
= F
S
+ F
T
= ibo
1
V
1
2
.
A man who has not studied fluid mechanics reasons as follows: As the train
moves ahead 1 m it scoops bo1 kg water. This water has been at rest, but once
scooped, it has the kinetic energy
b o
1
V
1
2
/ 2
. This kinetic energy is imparted
while the train moves 1 m, hence the resistance to its motion must be
b o
1
V
1
2
/ 2
,
i.e., half of the result obtained above!
A student of fluid mechanics tells him that he has forgotten the kinetic energy
lost in the tank and also the higher water level in the train, by about H.
Static, Dynamic and Total Pressures
Consider a streamline in a flow along which the Bernoulli polynomial
B =
1
2
q
2
+
p
+ gh
is conserved, i.e., for points 1 and 2 on the streamline Eq. (7.5) holds,
1
2
q
1
2
+
p
1
+ gh
1
=
1
2
q
2
2
+
p
2
+ gh
2
.
(7.5)
Let the contributions of gh
1
and gh
2
be so small that they may be ignored. Now
suppose a point 2 is found with q
2
= 0. The Bernoulli equation now reads
p
2
= p
1
+
1
2
q
1
2
.
Relating to point 1 we see that by stopping the flow the pressure there, p
1
, may be
increased to p
2
by the amount
q
1
2
/ 2
. We now define the following nomenclature:
1
2
q
1
2
is the dynamic pressure at point 1;
p
1
is the static pressure there;
p
2
is the total pressure, or the stagnation pressure, at point 1.
220 Fluid Mechanics
All these terms describe point 1, and given q
1
and p
1
, their magnitude may be
computed. Their values do not depend on whether point 2 is physically realized,
i.e., on whether there really exists a stagnation point on the streamline. Point 2 has
been used just to indicate the reason for choosing the names dynamic, static and
total pressures. In many engineering applications one finds these terms useful, and
it is important to note that they arise from the Bernoulli equation; hence they imply
the Bernoulli equation approximation.
Example 7.4
A Pitot tube, Fig. 7.4, is used to measure the
speed of a fluid moving relative to the tube. It
consists of an inner tube with its open end directed
against the flow located inside a sleeve with holes,
H, Fig. 7.4. The sleeve generators are parallel to
the flow and the holes' axes are thus perpendicular
to the flow direction. The pipe and the sleeve are
connected to a manometer. Find the relation
between the pressure difference measured by the
manometer and the speed of the fluid. The flow is
incompressible.
Solution
We choose point 1 far upstream from the Pitot tube, on the streamline that
eventually hits the center of the hole in the inner tube, point 2. This streamline
must have a stagnation point there. The Bernoulli equation, (7.5), reads for this
case
p
1
+
q
1
2
2
=
p
2
and indeed p
2
is the total pressure at the stagnation point.
Now at point H, close to point 2 but on a streamline that bypasses the sleeve,
the pressure p is still p
1
, because this streamline is not affected by the presence of
the tube. The strong Bernoulli equation is assumed to hold, and the fluid on this
second streamline does not change its velocity as it moves parallel to the sleeve.
The holes in the sleeve thus measure p
1
, the static pressure. Therefore,
q
1
=
2
p
2
p
1
( )
.
1
H
q
1
To
manometer
2
Figure 7.4 Pitot tube.
7. Energy Equations 221
Let the pressure difference be measured by a manometer in which a manometric
fluid of density
m
is used, and let the manometer show a reading Ah
m
. Then from
Example 3.1
p
2
p
1
=
m
gAh
m
,
and therefore, the velocity at point 1, q
1
, is related to the manometer reading, Ah
m
,
by
q
1
= 2 g Ah
m
.
Extension to Compressible Flows
Let Bernoulli's equation (7.4) be now extended to apply to compressible
flows. This is a mathematical step and requires no additional physical laws. To
obtain Eq. (7.2), we have used
1
Vp = V
p
|
\
|
.
|
|
,
(7.8)
which holds for incompressible flows only. In order to retain the gradient form for
variable density, we define a pressure function, F, which we expect to satisfy
1
Vp = VF .
(7.9)
With this function the Bernoulli polynomial becomes just
B =
1
2
q
2
+ gh + F
(7.10)
with all the previous results intact.
We now recall the Leibnitz rule for differentiation of an integral,
c
cx
f ( )d = f g
( )
cg
cx
c
g( x)
}
and also
V f
( )
d
c
g( x)
}
= f g ( )Vg.
Comparison of this rule with Eq. (7.9) suggests at once
222 Fluid Mechanics
F =
d
( )
=
dp
p ( )
=
dp
,
p
o
p
}
p
o
p
}
p
o
p
}
(7.11)
where in the last term on the right is considered a function of p only.
Generally for compressible fluids the density is a function of both the pres-
sure and the temperature. However, there are some important flows where the
density may be expressed as a function of the pressure only. These satisfy the form
= p ( ).
(7.12)
Examples for such flows are the isentropic flow ( p/
k
= const) and the isothermal
flow (p/ = const). Flows in which the density depends on the pressure only are
called barotropic flows. Obviously, Eq. (7.11) and with it Eq. (7.10) hold for
barotropic flows only.
The most important family of barotropic flows is that of the isentropic flows,
and for this case the pressure function has a direct interpretation. The first and
second laws of thermodynamics, combined, state that
T ds = di
1
dp, (7.13)
which for isentropic (ds = 0) processes becomes
di =
1
dp .
(7.14)
The differential of the enthalpy function, di in Eq. (7.14), is a complete differen-
tial and therefore dp / is also a complete one. Isentropic flows are therefore
barotropic, with no restriction to ideal gas behavior. Furthermore, from Eqs. (7.11)
and (7.14),
F =
dp
= di = i i
o
i
o
i
}
p
o
p
}
.
(7.15)
Hence the pressure function is just the thermodynamic enthalpy, i.
Isentropic flows serve as very good approximations for many real flows, and
Eq. (7.15) combined with Eq. (7.10) give the form
B = i +
1
2
q
2
+ gh = i
T
,
(7.16)
where the term total enthalpy, i
T
, is used instead of B.
Returning to Eq. (7.11) we emphasize that the identification of the pressure
function with enthalpy is a particular case, although a most important one,
restricted to isentropic flows. In general
dp =
dp
d
|
\
|
.
|
d =
d
dp
|
\
|
.
1
d
(7.17)
7. Energy Equations 223
and (d/dp) is just the compressibility of the fluid multiplied by its density, as
given by Eq. (1.28). The more general interpretation of F for compressible flows
is, therefore, the elastic energy of the fluid, which for isentropic flows equals the
change in its enthalpy.
Example 7.5
A tank filled with fluid has a hole in its bottom. The pressure on the inside,
just near the hole, is 1.510
5
Pa. The outside pressure is 10
5
Pa. Assuming the
fluid flow out of the tank to be isentropic, find its average speed
a. for a fluid that behaves like water,
b. for a fluid that behaves like air.
Solution
Let subscript i denote in and o denote out; then:
a. For water, from Eq. (7.5)
1
2
V
o
2
+
p
o
=
p
i
,
5
2 10 0.5
2 10 m/s.
1,000
i o
o
p p
V
| |
= = =
|
\ .
b. For air, from Eq. (7.16)
1
2
V
o
2
+ i
o
= i
i
,
V
o
= 2 i
i
i
o
( ).
Now
i
i
i
o
= di =
dp
,
p
o
p
i
}
i
o
i
i
}
p
i
i
k
=
p
o
o
k
=
p
k
and
dp
=
p
i
i
k
k
k2
d.
Thus
224 Fluid Mechanics
dp
=
p
o
p
i
}
p
i
1
k
k
k2
d =
k
k 1
p
i
1
k
i
k1
o
k1
| |
i
}
=
k
k 1
p
i
i
p
o
o
(
(
(
=
k
k 1
RT
i
1
p
o
p
i
|
\
|
.
|
|
k1
k
(
(
(
(
(
.
Let T
i
= 300
K. Then
1
0.4
1.4
2 2 1.4 1
1 287 300 1 257m/s.
1 0.4 1.5
k
k
o
o i
i
k p
V RT
k p
(
(
| |
| |
(
(
= = =
| |
(
(
\ .
\ .
(
Example 7.6
The hole at the bottom of the tank in Example 7.5 is fitted with an ideal
engine, through which the fluid expands isentropically from the inside pressure to
the environment. Find the work obtained from the transfer of 1 kg of fluid through
the engine.
Solution
From thermodynamics we know that for an isentropic process:
w
io
= i
i
i
o
.
a. For water:
i
i
= e
i
+ p
i
/
i
, i
o
= e
o
+ p
o
/
o
,
where e is the internal thermal energy of the water. The engine is ideal and
we do not expect e to change. Also does not change. Thus,
W
io
= i
1
i
o
=
p
i
p
o
o
=
1
p
i
p
o
( )
=
1
1,000
0.510
5
= 50 J.
b. For air: From Example 7.5,
W
io
= i
i
i
o
=
k
k 1
RT
i
1
p
o
p
i
|
\
|
.
|
|
k 1
k
(
(
(
(
(
.
Thus
7. Energy Equations 225
W
io
=
1. 4
0. 4
287 300 1
1
1. 5
|
\
|
.
|
0. 4
1. 4
(
(
(
(
= 32, 963 J.
In Example 7.5 the same energies were manifested as the kinetic energies of
the emerging fluids. We may check this by
W
iob
W
ioa
=
32, 963
50
= 659. 3 =
V
ob
V
oa
|
\
|
.
|
|
2
=
257
10
|
\
|
.
|
2
= 659.3,
which indeed comes out correct.
Modifications of the Bernoulli Equation for Conduit Flow
a. Average Velocities
The results obtained so far for the Bernoulli equation are strictly applicable to
streamlines only. In conduit flow, where the velocity generally varies between
streamlines, it becomes more convenient to use the average velocity for the whole
cross section. The situation is analogous to that of using the average velocity in the
application of the integral momentum theorem, in Chapter 4, and indeed the
difficulty is resolved in the same manner as in Chapter 4, i.e., by the introduction
of an energy correction factor |
E
.
Let a conduit have the velocity distribution q and let the total flow rate of
mechanical energy through its cross section, A, be computed:
p
+
q
2
2 g
+ h
|
\
|
.
|
q ndA = const.
A
}
(7.18)
Because of the Bernoulli equation, the same value is obtained for all cross
sections, A. The mass flowrate through A is just
q n dA = V A = const.
A
}
(7.19)
The average mechanical energy per unit mass is obtained by dividing
Eq. (7.18) by Eq. (7.19). This average is also conserved along the conduit.
Let all streamlines be approximately parallel to the conduit axis. For this case
p/ + h is the same over any given cross section and therefore is not modified by
the averaging process. The kinetic energy term, however, is modified, and the
resulting form is
226 Fluid Mechanics
p
+ h +|
E
V
2
2g
=const.
(7.20)
The kinetic energy correction factor |
E
is obtained from
|
E
=
1
AV
3
q
3
dA
A
}
.
(7.21)
For a uniform velocity profile |
E
=1, while for a parabolic profile, as in laminar
pipe flow, Eq. ( 6.53), |
E
= 2. Thus all applications of Bernoulli's equation to
streamlines may be extended to conduits, provided the correction factor, |
E
, is
used. In many cases |
E
may be taken as 1, e.g., in turbulent flows.
Example 7.7
Find how would the results in Example 7.2 be modified when the flow in the
pipe is the viscous Poiseuille flow, i.e., has a velocity distribution as given by Eq.
(6.51).
Solution
In Example 7.2, V
1
and V
2
are the average velocities. From Eqs. (6.53) and
(6.57) with q the velocity,
q = 2q 1
r
R
|
\
|
.
|
2
(
(
(
.
and Eq. (7.21) reads
|
E
=
1
tR
2
q
3
2t8q
3
1
r
R
|
\
|
.
|
2
(
(
(
(
3
r dr
0
R
}
=
16
R
2
r 3
r
3
R
2
+ 3
r
5
R
4
r
7
R
6
(
(
(
dr
0
R
}
= 16
1
2
3
4
+
3
6
1
8
(
(
= 2.
Bernoulli's equation now reads
p
1
+ |
E
V
1
2
2 g
=
p
2
+|
E
V
2
2
2g
.
Hence
( )
( )
2 4
2 1
1
.
1 /
i o
p p
V
d d
and
7. Energy Equations 227
( )
( )
2 2 2 4
2 1
,
1 /
i o
p p
m V A A
d d
= =
i.e., 2 times less than the mass flowrate in the original example.
b. Friction Head and Work Head
Another correction required in conduit flows is for friction. Friction is
defined in thermodynamics, and one of its pronounced effects is to reduce the
mechanical energy of the system. Since Bernoulli's equation is a balance of
mechanical energies, scaled to have the dimensions of length ("head"), careful
measurements always show the downstream total head to be deficient with respect
to the upstream total one. This deficiency is exactly the amount of mechanical
energy dissipated by friction, per unit mass, i.e., scaled to the dimension of length,
and denoted friction head loss
h
f
.
To balance the equation, h
f
must be included:
p
1
+ |
E
q
1
2
2g
+ h
1
=
p
2
+ |
E
q
2
2
2 g
+ h
2
+ h
f
.
(7.22)
The numerical evaluation of h
f
is considered in a later chapter. When h
f
is small,
Eq. (7.22) is approximated by Eq. (7.6), the unmodified Bernoulli equation.
Suppose the flow passes through a pump. A pump increases the mechanical
energy of the fluid, and this increase per unit mass, i.e., scaled to the dimension of
length, is called the work head h
w
. The relation for two points separated by a pump
thus becomes
p
1
+|
E
q
1
2
2g
+h
1
+h
w
=
p
2
+|
E
q
2
2
2g
+h
2
+h
f
,
(7.23)
where for negligible friction the h
f
term may be dropped.
It is noted that the h
w
term is not the power required to run the pump, but
rather that part of the power, which actually reaches the fluid as work. The ratio
between the two is denoted the hydraulic efficiency of the pump.
Example 7.8
Find an expression for the friction head, h
f
, for Poiseuille flow in a circular pipe.
228 Fluid Mechanics
Solution
From Eq. (6.56),
Ap =
32
D
2
V L .
Hence,
h
f
=
Ap
=
32
D
2
L
V
g
.
Example 7.9
Water is being discharged from a large
tank open to the atmosphere through a vertical
tube, as shown in Fig. 7.5. The tube is 10 m
long, 1 cm in diameter, and its inlet is 1 m below
the level of the water in the tank. Find the
velocity and the volumetric flowrate in the pipe,
assuming:
a. Frictionless flow.
b. Laminar viscous flow.
c. A turbine is connected at the tube outlet,
point 2. Find the maximum power obtain-
able under assumptions a and b, above.
Solution
a. We select a streamline connecting the top surface, point 1, with the discharge,
point 2. These two points have atmospheric pressure, and Eq. (7.6) simplifies
to
v
1
2
2 g
+ h
1
=
v
2
2
2 g
+ h
2
.
Neglecting the velocity v
1
at the tank surface, we obtain
v
2
= 2gh
1
= 2 9.8 11 =14.7m/s,
and the flowrate is
Q = Av =
t
4
0.01
2
v
2
3,600= 0.2827v
2
m
3
/hr = 4.16m
3
/h.
b. Bernoulli's equation between points 1 and 2, including friction and velocity
variation across the stream tube, Eq. (7.22), is
10 m
h = 0
h = 11 m
V
1m
1
2
Figure 7.5 Flow from a water tank
through a vertical tube.
7. Energy Equations 229
h
1
= |
E
v
2
2
2g
+ h
f
.
Assuming laminar flow (|
E
= 2) we substitute h
f
for the Poiseuille flow,
Eq. (6.56) (see Example 7.8),
h
f
=
Ap
=
32v
2
L
D
2
g
and obtain
gh
1
= v
2
2
+
32vL
D
2
v
2
.
For water at 20
o
C, v = 10
-6
m
2
/s; hence
9.8 11= v
2
2
+
32 10
6
10
0. 01
2
v
2
or
v
2
2
+3.2v
2
107.8 = 0,
resulting in
v
2
= 8. 9 m / s
and
Q = 0. 2827 8. 9 = 2. 516 m/ s.
c/a. For this case the modified Bernoulli equation becomes
h
1
= |
E
v
2
2
2g
+ h
f
+ h
w
.
For negligible friction and |
E
= 1
h
w
= h
1
v
2
2
2 g
and the power of the turbine is
P =
m gh
w
= Q h
w
= h
w
tD
2
4
v
2
.
Substitution of h
w
gives P as a function of v
2
:
P =
tD
2
4
v h
1
v
2
2g
|
\
|
.
|
|
.
To find the maximum power, we differentiate P with respect to v and equate
the derivative to zero,
230 Fluid Mechanics
cP
cv
=
t D
2
4
h
1
3
2 g
v
2
|
\
|
.
|
|
= 0 ,
resulting in
v =
2
3
gh
1
=
2
3
9.811= 8.48m/s
and
h
w
= h
v
2
2 g
= 11
71. 9
2 9.8
= 7. 33m.
Hence, the maximum power is
P
max
= 9.810
3
7.33
t
4
0.01
2
8.48= 47.8W.
c/b. The work head including frictional effects is
h
w
= h
1
|
E
v
2
2
2g
32vL
gD
2
v
and the power is
P =
tD
2
4
v h
1
|
E
v
2
2
2g
32vL
gD
2
v
|
\
|
.
|
|
.
Now the power is maximal when
cP
cv
=
tD
2
4
h
1
3|
E
v
2
2
2g
64vL
gD
2
v
|
\
|
.
|
|
= 0,
v
2
+
64vL
3D
2
v
gh
1
3
= 0,
v
2
+
6410
6
10
310
4
v
9.810
3
= 0,
v
2
+2.133v 32.7 = 0,
v = 4. 75 m / s.
Hence
h
w
= 11
4. 75
2
9.8
3210
6
10
9.8 0.01
2
4.75= 7.15m
and
P
max
= 9.810
3
8.37
t
4
0.01
2
4.75= 26.14W.
7. Energy Equations 231
Example 7.10
Equation (7.23) is the Bernoulli equation for conduit flow modified to in-
clude friction and mechanical work. Compare this equation with the first law of
thermodynamics written for a control volume applicable to this case and evaluate
the friction head h
f
in terms of the appropriate thermodynamic variables.
Solution
We write the first law of thermodynamics for a control volume at steady state
having heat and work interactions with the surroundings and a fluid entering the
control volume at point 1 and leaving at point 2 as
( )
1
2 2
2 1
2 2 1
[ ( )]
2
x
q q
Q W m i i g h h
= + +
where m is the mass flowrate of the fluid, Q the heat input and W the work
output. Rewriting this equation per unit mass flow and rearranging yields
i
1
+
q
1
2
2
+ gh
1
W
m
= i
2
+
q
2
2
2
+ gh
2
+
Q
m
.
. .
.
.
The work output, W , may be rewritten in terms of the work head input, h
w
, as
w
W mgh =
Also, the enthalpy, i, is given by
i = e + p / ,
where e is the internal energy of the fluid. Hence, the first law of thermodynamics
takes on the form
( )
2 2
1 1 2 2
1 2 2 1
.
2 2
w
p q p q Q
gh gh gh e e
m
(
+ + + = + + +
(
Dividing this equation by g renders it in terms of "heads,"
2 2
1 1 2 2 2 1
1 2
.
2 2
w
p q p q e e Q
h h h
g g g mg
(
+ + + = + + +
(
.
Comparing this with Eq. (7.23), we obtain an expression for the friction head h
f
,
( )
2 1
1
.
f
Q
h e e
g m
(
=
(
Hence, the mechanical energy of the fluid "lost" as friction is "found" in the inter-
nal energy of the fluid and in the heat transferred to the surroundings.
232 Fluid Mechanics
Low Pressures Predicted by Bernoulli's Equation
Consider the pipe and the reservoir shown in Fig. 7.6. We assume the height
of the reservoir above the pipe, H, to be large and neglect friction in the pipe, i.e.,
h
f
= 0.
Bernoulli's equation applied between points 1 and 3 yields
p
o
+
v
1
2
2
+ gH =
p
o
+
v
3
2
2
.
d
H
D
1
2
3
p
o
p
o
Figure 7.6 Contraction in a pipe.
We assume a very wide reservoir, hence v
1
= 0, and obtain
v
3
2
= 2 gH,
which is not new (see Example 7.9). Now we apply Bernoulli's equation again, but
this time between points 2 and 3:
p
2
=
p
o
+
v
3
2
2
v
2
2
2
.
Because of continuity,
v
2
2
= v
3
2
D / d
( )
4
,
and therefore,
p
2
=
p
o
v
3
2
2
D
d
|
\
|
.
4
1
(
(
=
p
o
gH
D
d
|
\
|
.
4
1
(
(
. (7.24)
Equation (7.24) indicates that the pressure at point 2 may be forced to
arbitrarily low values, including negative values, by the manipulation of either D/d
or H.
From physics we know that gases cannot have negative pressures and that
fluids boil once the pressure becomes lower than their vapor pressures. We con-
7. Energy Equations 233
clude, therefore, that Eq. (7.24), and with it Bernoulli's equation, may not be
applied once they predict negative or even saturation pressures. Such flow con-
ditions require considerations that are too elaborate to be included in a basic text
such as this.
Example 7.11
A dentist uses a suction device as shown in Fig. 7.7. Water from the main line
flows in the larger pipe, which changes its diameter from D
to d and back to D.
Eventually the water flows out to the sink, at point 3. The pressure in the main
water line, which feeds the device, is induced by a water tower 5 m high. The
atmospheric pressure is p
o
= 10
5
Pa. The vapor pressure of water at 30
o
C is
4246 Pa, and to ensure smooth operation, the pressure must never drop below
5,000 Pa. Find the highest D/d ratio.
2
3
suction
p
o
D
d
2
Figure 7.7 Dentist's suction device.
Solution
Point 1 is chosen at the top of the water tower. Repeating the analysis, which
has led to Eq. (7.24), we obtain
gh +
p
o
=
p
o
+
v
3
2
2
, v
3
2
= 2gh,
p
2
= p
o
gh
D
d
|
\
|
.
4
1
(
(
,
and for the lowest value of p
2
= 5,000 Pa
234 Fluid Mechanics
5, 000 =100, 0001, 000 9. 81 5
D
d
|
\
|
.
|
4
1
(
(
(
,
leading to
D
d
|
\
|
.
max
=1.309.
References
R.B. Bird, W.E. Stewart and E.N. Lightfoot, "Transport Phenomena," Wiley, New
York, 1960, Chapter 10.
W.F. Hughes, "An Introduction to Viscous Flow," Hemisphere Publishing,
Washington, DC, 1979, Chapter 4.
R.H. Sabersky, A.J. Acosta and E.G. Hauptman, "A First Course in Fluid Mechan-
ics," 2nd ed., Macmillan, New York, 1971, Chapter 3.
Problems
7.1 A dentist uses a suction device as shown in Fig. P7.1. Water from the line
flows in the larger pipe, which changes its diameter from D to d and back to
D; eventually the water flows out to the sink. The pressure in the main water
line, before the water enters the device, is p
1
= 1.510
5
Pa. Find the suction
pressure for d/D = 0.8.
D d
Figure P7.1 Dentist's suction device.
7. Energy Equations 235
7.2 Before installing the water pipes in his new house, a man measured the
pressure in the main water line near the house and found it to be p
L
= 10
6
Pa.
He then installed pipes with an inner diameter of 2 cm. Find the maximum
rate of water supply he may expect.
7.3 A pump is used to raise water into a reservoir, Fig. P7.3. Both the water
source and the reservoir have free surfaces, i.e., are open to the atmosphere.
Water is raised at the rate Q = 3 m
3
/min to the height h = 10 m. The average
velocity of the water in the pipe is 8 m/s. Find the power required to run the
pump and the value of the various terms of the Bernoulli polynomial at the
entrance and the exit of the pipe and also at points in the water source and in
the reservoir far away from the pipe.
7.4 A Pitot tube, Fig. 7.4, is used to measure the speed relative to a moving
fluid. The details of the Pitot tube are described in Example 7.4. For
p
2
= 1.510
5
Pa and p
1
= 10
5
Pa, find the flow velocity when the moving
fluid is
a. water, b. air.
Note that for air it is not enough to know the dynamic pressure
p
D
= p
2
p
1
.
Figure P7.3 Pump raising water.
7.5 A water turbine consists of an outer shell, propeller and diffuser as shown in
Fig. P7.5. At point 1 the diameter is l m, the velocity of the water is 30 m/s
and the pressure is 250 kPa. At point 2 the diameter is 2 m, while at point 3
it is 3 m. The flow losses in the turbine are about 25% of the power of an
ideal turbine.
a. What is the power of an ideal turbine?
236 Fluid Mechanics
b. What is the actual power?
c. What is the ideal and actual power without the diffuser?
d. If the propeller rotates at 900 rpm, what is the moment acting on the
outer cover and what is the direction of this moment relative to that of
the propeller rotation?
Figure P7.5 Water turbine.
7.6 A two-dimensional hollow body in the shape of an "igloo" rests on the
ground, Fig. P7.6. The curved part has a diameter D and a length L = tD/2.
There is a wind of speed V= 40 km/h. Assuming that the average wind
velocity along the curved part is given by
V
L
= VL / D = V t / 2,
calculate the lift on the igloo.
Figure P7.6 Igloo in cross wind.
7. Energy Equations 237
7.7 A large tank of water has a hole at its
side, h = 4 m below the water level.
The hole diameter is d = 0.05 m,
Fig. P7.7. Neglect friction and find
the rate of the water flow out of the
tank.
Note: The answer is not
A 2gh; the
shape of the stream tube coming out
of the tank is not cylindrical, but
rather curved. This constriction of
the cross section is called vena con-
tracta.
7.8 Find the correction factors,
|
M
for the use of the average velocity in the momentum theorem,
|
E
for the use of the average velocity in the Bernoulli equation,
for the flow in a rectangular duct of sides a and b. Assume the velocity to be
approximated by
w = w
o
sin
tx
a
|
\
|
.
|
sin
ty
b
|
\
|
.
|
.
7.9 Water at 20
o
C is flowing through a 10-cm pipe at 1.5 m/s. At one point, the
static pressure is 135 kPa gauge, and the friction head between this point
and a second point 10 m below the first is 3.5 m. Find the static pressure at
the second point.
Now suppose the second point is 10 m above the first point; can the flow
take place?
7.10 Oil at 20
o
C (viscosity 900 cp, specific gravity
0.95) is to flow by gravity through a 1-mile-
long pipeline from a large reservoir to a lower
station. The difference in levels between the
tanks is 14 m. The discharge rate required is
1.6 m
3
/min. Calculate the diameter of the steel
pipe necessary for these conditions.
7.11 A tank 1.25 m deep contains water, Fig. P7.11.
The inlet end of the siphon is h = 1
m below
the surface. Friction losses amount to 10 kPa.
How far below the surface must the outlet end
be placed to give a velocity of 0.75 m / s?
Figure P7.7 Vena contracta.
Figure P7.11
Tank with siphon.
238 Fluid Mechanics
7.12 A fluid flows in a pipe that has a sudden increase in its cross section, Fig.
P7.12. Assume that the pressure in the wider section right after the jump p
2
retains its old value, p
1
. Then choose a control volume containing sides 1
and 3 of the jump point, Fig. P7.12, and use conservation of mass and the
momentum theorem to find the new average velocity. Find the head loss
across the jump.
Figure P7.12 Sudden increase in cross section and control volume.
7.13 An inventor who did not study
fluid mechanics suggests to put a conical funnel in a fast-flowing river and
to place a turbine rotor at the funnel's apex, Fig. P7.13. Because of conti-
nuity, he claims, the ratio of the river flow speed to the speed at the rotor is
the inverse of the funnel cross sec-
tional areas. Thus, in principle, any
speed can be attained at the apex.
"Not so fast," says a student of fluid
mechanics, "the maximum attainable
speed is quite finite." Find this finite
speed and explain what happens
when the area ratio becomes larger
than that corresponding to the
highest possible velocity ratio.
7.14 Water flows in a rectangular channel
open to air at p
o
, Fig. P7.14. At section 1 the width of the channel is b
1
, the
height of water h
1
and its average velocity q
1
. At another section the channel
width is b
2
, while the pressure at the surface is still p
o
. Find q
2
and h
2
. Find
conditions for h
2
> h
1
and for h
2
< h
1
. Identify situations when the Bernoulli
equation cannot hold between the two sections.
Hint: h
2
< 0.
Figure P7.13 Turbine with guide funnel.
7. Energy Equations 239
Figure P7.14 Sections of a rectangular channel.
7.15 A stationary nozzle ejects a jet of water with the mass-flowrate of 200 kg/s
and with the velocity u
1
= 20 m/s, Fig. P7.15. The water jet hits a plane vane
with the angle o = 30
o
between the jet axis and the plane surface. The vane
recedes in the direction of its normal with the speed v = 6 m/s. The
environment pressure is constant and gravitation and viscosity effects are
negligible.
a. Choose a control volume and define it carefully.
b. Find the division of the jet mass-flowrate on the vane: the flux to the left
and the flux to the right, i.e., how thick are o
2
and o
3
.
c. Find the force applied by the jet to the vane.
Figure 7.15
7.16 A water turbine operates between two reservoirs open to the atmosphere, as
shown in Fig. P7.16. All friction losses are negligible, except at the pipe
leading to the lower reservoir, which is terminated, with a sudden enlarge-
ment of the cross section; and the total velocity head there, i.e., V
2
2
/2, is lost.
Find the maximum power obtainable by this turbine.
240 Fluid Mechanics
7.17 The pressure in a main water line is
410
5
Pa, its diameter is 0.150 m
and the mean velocity of the water
is 10 m/s. The atmospheric pressure
is 10
5
Pa. The main line is 1 m
below ground level. An industrial
washing machine that is located 3
m above ground level has a peak
water demand of 0.020 m
3
/s.
Find the diameter of the pipe lead-
ing from the main line to the ma-
chine.
7.18 The water level in a reservoir is 20
m above ground level. The top of
the tank is open to the atmosphere.
Water is supplied to a field at
ground level using a pipe with a
diameter of 0.012 m. Neglecting
friction, find the maximum flowrate that can be supplied. To increase the
supply the pipe is changed to another having a diameter of 0.019 m. Find
the new flowrate.
7.19 A small length of the old pipe, that of 0.012 m diameter in Problem 7.18, is
laid under a wall. In changing the old pipe for the new one with the larger
diameter it has been suggested to change the pipe section between the
reservoir and up to the wall, to change the pipe starting on the other side of
the wall, but to leave the old pipe section under the wall. It has been argued
that since this old piece is very short, its effect is negligible. Find whether
this effect is indeed negligible.
7.20 The mean velocity of water in a main line is 10 m/s, and the water pressure
there is 310
5
Pa. The water then passes through a booster pump and is
distributed into houses located 100 m above the main line. When a tap is
opened in a house, the water comes out at the speed of 20 m/s. The atmo-
spheric pressure is 10
5
Pa. Friction head losses between the main line and
the houses are estimated as 5 times the velocity head at the houses. The
booster pump overall efficiency is 0.8. Find the power needed to run the
booster pump.
Figure P7.16
7. Energy Equations 241
7.21 A two-dimensional fluid jet
hits a curved vane, which re-
cedes from the jet, Fig. P7.21.
Show that the thickness of the
water layer, as it follows the
contour of the vane surface, is
constant.
7.22 Water comes out of a 25-m-
high water tower, open at the
top, and flows through a 0.15-
m-diameter, 600-m-long pipe.
The water is then distributed
through 0.025-m-diameter pipes, one of which is 100 m long, Fig. P7.22.
Measurements of pressure drops along the horizontal pipes yield P
A
P
B
=
1,000 Pa, and P
D
P
E
= 2,000 Pa. The mean velocity of the flow in the
0.15-m-diameter pipe is V = 0.15 m/s.
Assume that at a given velocity the friction losses in a pipe are proportional
to the pipe's length, and find the pressure at point F.
20 m
600 m 100m
25 m
D E
C
B A
G
F
d = 0.15 m
d = 0.025 m
10m
10 m
Figure P7.22
7.23 Measurements show that friction losses in straight pipes are proportional to
L/d, where L is the pipe's length and d is its diameter, and approximately
proportional to V
2
, where V is the mean velocity in the pipe. In the system
of Fig. P7.22 all pipes are closed except the 0.15-m-diameter pipe and the
0.025-m-diameter pipe designated by the letters DEF. Thus the whole flow
Vane
b
b
Jet
Figure P7.21
242 Fluid Mechanics
is now through these two pipes, along the path ABCDEF. At point F the
pipe is open to the atmosphere. Find the flowrate.
7.24 Another pipe of 0.025 m diameter, identical to that along DEF in Fig. P7.22,
is opened, such that the water coming out at point C is equally divided
between the two parallel 0.025-m-diameter pipes, both of which are open to
the atmosphere. Find the flowrate.
7.25 It is necessary to double the flowrate of the water under the conditions of
Problem 7.23. To do this, a booster pump is installed at point B in Fig.
P7.22. The booster pump overall efficiency is 80%. Find the power needed
to run the booster pump.
7.26 The booster pump designed in Problem 7.25 is used under the conditions of
Problem 7.24. Its motor draws the same power as obtained in Problem 7.25.
Find the flowrate under these operating conditions.
7.27 Given a laminar viscous flow in a circular pipe. Find the friction head,
expressed as length, as a function of the fluid properties and of the velocity.
7.28 Water is poured form a jar into a bottle. The water stream is assumed to
have a circular cross section, and its flowrate is 0.0005 m
3
/s. The diameter
of the neck of the bottle is 0.02 m. Find how high must the jar be held above
the bottle to have the water jet enter the bottle.
7.29 Figure P7.29 shows a conical
nozzle connected at the end of a
pipe. The mean velocity in the
pipe is 10 m/s. The wide opening
of the nozzle is 0.150 m diameter
and the narrow opening is 0.050 m
diameter. At the exit to the at-
mosphere the water pressure is
atmospheric.
a. Find the mean velocity of the
water at the exit. Find the distribution of the pressure inside the cone.
b. Find the distribution of the axial force along the walls of the cone.
7.30 A production process consists of lowering a ceramic form into a bath of hot
molten plastic, and then taking the form, now coated with a plastic layer, out
to cool. The plastic solidifies as it cools. Finally the plastic is cut along
Figure P7.29
7. Energy Equations 243
predetermined lines, while still on the form, using a high-speed thin water
jet. A surface stress, i.e., a pressure of 1,000 kPa, is required in order to cut
the plastic, while a pressure of 10,000 kPa results in pitting on the ceramic
surface.
a. Find the necessary velocity of the water jet.
b. Find the pressure needed to produce the jet.
c. Find the maximal pressure and jet velocity, which still do not harm the
ceramic form.
244 Fluid Mechanics
245
8. SIMILITUDE AND
ORDER OF MAGNITUDE
Dimensionless Equations
All the equations derived until now are dimensional, i.e., their various terms
have physical dimensions. These terms represent numerical values, which thus
depend on the system of physical units adopted. While there are some arguments
why a particular system of units should be preferred, e.g., the SI system, still this
choice is just a convention and is highly arbitrary. It seems, therefore, that by an
orderly and preplanned removal of this arbitrariness one could gain some addi-
tional information. Such a process would result in Dimensionless Equations.
There are several goals for using dimensionless equations; two of the most
important ones are similitude and order of magnitude. While these two goals are
quite different from one another, both require an intermediate step consisting of
essentially the same algebraic procedure, which leads to dimensionless equations.
We choose to show this procedure first.
Let us rewrite the continuity equation, (5.21), and the momentum equations,
(5.27) - (5.29), for some incompressible flow, adding "stars" to all dependent and
independent variables. The role of these stars is to denote dimensional variables,
such that when these same variables appear later without stars they are considered
dimensionless:
cu *
cx *
+
cv *
cy *
+
cw *
cz *
= 0
(8.1)
cu *
ct *
+ u*
cu *
cx *
+ v*
cu *
cy*
+ w *
cu *
cz *
|
\
|
.
|
|
= g
x
*
cp*
cx *
+
c
2
u *
cx*
2
+
c
2
u *
cy *
2
+
c
2
u *
cz *
2
|
\
|
.
|
|
,
(8.2)
246 Fluid Mechanics
cv*
ct *
+ u *
cv*
cx*
+ v*
cv*
cy*
+ w *
cv*
cz *
|
\
|
.
|
|
= g
y
*
cp *
cy*
+
c
2
v*
cx*
2
+
c
2
v *
cy *
2
+
c
2
v *
cz *
2
|
\
|
.
|
|
,
(8.3)
cw*
ct *
+ u *
cw*
cx *
+ v*
cw*
cy *
+ w *
cw*
cz *
|
\
|
.
|
|
= g
z
*
cp*
cz *
+
c
2
w *
cx *
2
+
c
2
w *
cy *
2
+
c
2
w*
cz *
2
|
\
|
.
|
|
.
(8.4)
We now assume that some characteristic physical quantities have been
defined and attempt to express the dimensional starred variables in terms of these
characteristic quantities. The problem of how to define the characteristic quantities
will be considered later, because their definitions are quite different for similitude
and for order of magnitude. Here it suffices to assume that such characteristic
quantities exist and that they are dimensional. Thus, if t is the characteristic time,
which is measured in, say, seconds, then
t =
t *
t
, t* = t t.
(8.5)
The first of these equations is really a definition of the dimensionless time t.
Similarly, a characteristic length L, which may be measured in meters, satisfies
x* = Lx , y* = Ly , z* = Lz ,
(8.6)
meaning that we choose to measure length in units of L meters. A characteristic
velocity V, measured in m/s, satisfies
u* =Vu, v*= Vv, w*= Vw
(8.7)
and a characteristic pressure t satisfies
p* = t p
. (8.8)
We also choose the constant of gravity g as the characteristic body force, i.e.,
g
x
*
= gg
x
. The starred variables, Eqs. (8.5) - (8.8), are now substituted in Eqs. (8.1)
- (8.4), resulting in the disappearance of stars from these. Equation (8.1) thus
becomes
V
L
cu
cx
+
cv
cy
+
cw
cz
|
\
|
.
|
|
= 0,
(8.9)
and Eq. (8.2)
V
t
cu
ct
+
V
2
L
u
cu
cx
+ v
cu
cy
+ w
cu
cz
|
\
|
.
|
|
(
(
(
= gg
x
t
L
cp
cx
+
V
L
2
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
.
(8.10)
We now select some term in each equation and divide the whole equation by
the group of physical constants and characteristic quantities that precedes it. The
8. Similitude and Order of Magnitude 247
term now stands alone and is dimensionless. Because the equations we use are
always dimensionally homogeneous, all other terms in the equation have also
become dimensionless, and so must be the groups composed of the characteristic
quantities that appear as multipliers in the equation. Thus Eq. (8.9) divided by V/L
becomes
cu
cx
+
cv
cy
+
cw
cz
= 0.
(8.11)
Equation (8.10) divided by V
2
/L becomes
L
tV
|
\
|
.
|
cu
ct
+ u
cu
cx
+ v
cu
cy
+ w
cu
cz
(
(
(
=
Lg
V
2
|
\
|
.
| g
x
t
V
2
|
\
|
.
|
|
cp
cx
+
VL
|
\
|
.
|
|
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
.
(8.12)
The dimensionless groups that have emerged in the equations are recognized
as
L
tV
= O the time number,
t
V
2
= Eu the Euler number,
VL
=
VL
v
= Re the Reynolds number,
and
V
2
gL
= Fr
2
, Fr is the Froude number.
Using these dimensionless numbers, Eqs. (8.11) - (8.12) are rewritten in their
dimensionless forms as
cu
cx
+
cv
cy
+
cw
cz
= 0
(8.13)
O
cu
ct
+ u
cu
cx
+ v
cu
cy
+ w
cu
cz
=
1
Fr
2
g
x
Eu
cp
cx
+
1
Re
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
.
(8.14)
The characteristic physical quantities are now hidden in the dimensionless
numbers, and Eqs. (8.13) and (8.14) may be used for both similitude and order of
magnitude analysis. The selection of the characteristic quantities is considered now
separately for each objective.
248 Fluid Mechanics
Similitude
We begin by considering a particular example of steady unidirectional verti-
cal film flow shown in Fig. 8.1a, the x-component of the dimensionless Navier
Stokes equations. Equation (8.14) becomes in this case
0 =
1
Fr
2
g
x
+
1
Re
d
2
u
dy
2
.
With the coordinates used in Fig. 8.1a, g
x
= 1, and the equation takes the form
d
2
u
dy
2
+ N = 0,
(8.15)
with
N =
Re
Fr
2
=
L
2
g
Vv
.
The expression for N contains two physical
constants, the constant of gravity, g [m/s
2
],
and the kinematic viscosity of the fluid,
v [m
2
/s]. It also contains the characteristic
length, L [m], and the characteristic velocity,
V [m/s]. These must be selected now, and we
choose them to be (see Fig. 8.1a)
L = o ,
the thickness of the film;
V =
1
o
u dy
o
o
}
,
the average fluid velocity.
The boundary conditions that the velocity, u, must satisfy are
u = 0 at y = 0,
du
dy
= 0 at y = 1.
(8.16)
The first boundary condition is recognized as the no-slip viscous boundary
condition, Eq. (5.62). The second boundary condition expresses the fact that there
is zero shear at the gasliquid interface.
o
y
g
a. Plane wall
b. Elliptical conduit
x
Figure 8.1 Steady film flow on
vertical wall.
8. Similitude and Order of Magnitude 249
The solution of Eq. (8.15) with the boundary conditions of Eq. (8.16) is
u = N
y
2
2
+ y
|
\
|
.
|
,
which relates the dimensionless velocity to the dimensionless y-coordinate. Thus,
for example, at the free edge of the film, i.e., at y = 1, u = N/2; the velocity there is
N/2 times the average velocity.
Let two flows have the same N value. These flows need not be identical, yet
they have the same velocity distribution. We express this result by stating that a
condition for the similarity of the velocity distributions in two such flows is that
they have the same similarity parameter N . This similarity permits one to exper-
iment with a model and apply the results to real life flows.
Equations (8.15) and (8.16) have one similarity parameter. We note that when
an equation is written in its dimensionless form, the similarity parameters can be
identified by simple inspection, i.e., by listing the various dimensionless groups
that have appeared.
When looking for similitude, i.e., for similarity conditions, it is necessary to
first have the equations set in their dimensionless form. As already seen, one must
start with the definition of the characteristic quantities. At this point it seems help-
ful to consider an additional example.
Consider Eq. (8.17), which is recognized as the same problem with the flow
taking place on the inside surface of an elliptical conduit, Fig. 8.1b:
c
2
u
cy
2
+
c
2
u
cz
2
= N =
Re
Fr
2
|
\
|
.
. (8.17)
The boundary conditions are of the form u = 0 on the solid wall, and u/n = 0
at the free surface. Although Eq. (8.17) is also one-parametric, conditions for
similarity here must be formulated more carefully. Here similarity of geometry is
also a necessary condition, which has its implication when the characteristic
quantities are chosen. In this example the cross section of the conduit is elliptical.
For two cases to be similar, they must both have elliptical cross sections, with
similar ellipses, i.e., they must have geometrical similarity, with all corresponding
lengths having the same ratio. They must have similar boundary conditions, i.e.,
the same boundary conditions applied on corresponding surfaces on both
boundaries. And also they must have the same similarity parameters.
At this point it is clear that the characteristic length may be chosen as the
small axis of the ellipse, or as the large one, provided it is chosen in the same way
for all cases.
This conclusion is now generalized: The similarity of two problems requires
similarity of geometry, similarity of boundary conditions, and corresponding
equality of all similarity parameters. The characteristic physical quantities used to
construct the similarity parameters can be chosen arbitrarily, provided they are
250 Fluid Mechanics
well defined and correspond to the same geometrical locations in both cases.
Another way to state this is that two problems, A and B, are similar, if their
dimensionless formulations contain no clue as to whether it is problem A or prob-
lem B that has led to the formulation.
Example 8.1
Steamships use sea water to cool their condensers. The intake of the sea water
may be designed as shown in Fig. 8.2. Sea water streams into the intake nozzle,
accelerates there into the pump, and proceeds into the condenser. The designer
idea is that at the cruising speed of the ship the pump would run idly, with the
required pumping power supplied by the main motion of the ship, i.e., the relative
kinetic energy of the incoming water. On the other hand, at other speeds the pump
must do some pumping. A model is tested in a water tunnel. Find conditions for
similarity.
Pump
Sea water
intake
Nozzle
To
condenser
1
Figure 8.2 Cooling water intake.
Solution
Obviously, the model must be geometrically similar to the ship's system.
Using the subscript s for the ship's system and M for the model, similarity requires
that
Re
s
= Re
M
, Eu
s
= Eu
M
.
Therefore: V
s
L
s
/v
s
= V
M
L
M
/v
M
, and because both run in water, V
s
L
s
= V
M
L
M
, or
V
M
= V
S
L
S
L
M
.
The characteristic pressure t is chosen as the pressure difference between a point
just before the pump, point 1 in the figure, and the hydrostatic pressure at the
equivalent depth outside the ship. Thus t = p
1
- p
o
, and
2 2
,
S M
S M
V V
t t
=
8. Similitude and Order of Magnitude 251
or
t
M
= t
S
V
M
2
V
S
2
= t
S
L
S
L
M
|
\
|
.
|
|
2
.
During the test the pump of the model must be run such that this condition on the
pressure is satisfied.
Hidden Characteristic Quantities
There are cases where several similarity parameters do appear in the dimen-
sionless equations but some of the characteristic physical quantities do not occur in
the formulation of the problem. These missing characteristic quantities must
therefore be defined using other quantities. Under such circumstances the actual
number of parameters that must be kept similar may be smaller than the apparent
one. An example of such a case is the incompressible flow in a cylindrical conduit.
z
Figure 8.3 Cylindrical conduit.
Consider a cylindrical conduit of arbitrary cross section, with the cylinder
generators parallel to the z-axis, Fig. 8.3. Let the cross-sectional area of the cylin-
der be A
c
and its circumference be C
c
. A characteristic length may be defined as
*
D
= 4A
c
/C
c
, and a characteristic velocity may be chosen as the average velocity in the
cylinder,
V =
1
A
c
w dA
A
c
}
.
The dimensionless z-component of the time-independent momentum equation
(8.14) is
u
cw
cx
+ v
cw
cy
+ w
cw
cz
= Eu
cP
cz
+
1
Re
c
2
w
cx
2
+
c
2
w
cy
2
+
c
2
w
cz
2
|
\
|
.
|
|
(8.18)
where P is the modified pressure. There are two additional equations, for the x-
and y-components, which are not written. The important thing to notice here is the
*
This characteristic length is called the hydraulic diameter. The number 4 appears just to make it the reg-
ular diameter for a circular cylinder.
252 Fluid Mechanics
appearance of the two similarity parameters:
Re =
DV
v
, Eu =
t
V
2
.
It seems that for two cylinders of similar cross section the flows would come out to
be similar when
Re
1
= Re
2
, Eu
1
= Eu
2
.
We notice, however, that while the characteristic length and the characteristic
velocity emerged rather naturally, the statement of the problem does not suggest a
particular characteristic pressure. We know that there is going to be a pressure
gradient along the pipe, but its magnitude is not known a priori. We therefore de-
fine the characteristic pressure t in terms of the characteristic velocity
t =
1
2
V
2
.
(The
1
2
is put there just for consistency with the Bernoulli equation. Everything
works out without the
1
2
too). Now for similarity,
Re
1
= Re
2
,
D
1
V
1
v
1
=
D
2
V
2
v
2
.
(8.19)
The Euler numbers, however, come out now to be
Eu
1
=
t
1
V
1
2
=
1
2
, Eu
2
=
t
2
V
2
2
=
1
2
(8.20)
and they are automatically the same, i.e., we do not have to do anything physical to
make them the same.
Because the dimensionless continuity equation (8.11) contains no additional
similarity parameters and the x and y momentum equations contain the same
parameters as does the z-component equation, we have now complete similarity
between the two flows, provided Eq. (8.19) holds.
The knowledge of the details of the flow in one cylinder gives us the details
of the flow in the second one. Solving for or measuring the pressure drop along
one cylinder, say, C
o
times the characteristic pressure (the only unit to measure
dimensionless pressures), supplies at once the same information for the second
cylinder. Far from the zone of entrance to the cylinder we expect this pressure drop
to become constant per unit length (i.e., per characteristic length measured along
the cylinder axis
*
), and indeed that same constant, say C
f
, must apply to both
flows. Thus, for a section of length L
*
= DL we expect the dimensionless pressure
drop to be C
f
L and the dimensional one to be
*
Nothing important changes if the characteristic length is taken as the largest diameter of the cross-section
and the characteristic velocity as the largest velocity. One finds a variety of values to choose from, all of
which are well defined.
8. Similitude and Order of Magnitude 253
AP = C
f
Lt = C
f
L
*
D
V
2
2
,
(8.21)
for all similar cylinders. Since Eu = 1/2 holds for all these cylinders C
f
depends on
Re only, i.e., C
f
= f (Re) , and can be obtained, for instance, by a set of experiments
using one particular cylinder. The results apply to all similar cylinders provided C
f
is taken for the same Reynolds numbers as in the tested cylinder.
Example 8.2
A duct has a rectangular cross section, with the sides 1 m by 0.5 m. It is to be
used to transfer water at the rate of Q = 2 m
3
/s. It is proposed to construct a smaller
model first and to find experimentally what pump is required for the full-scale
duct. The pump available for the experiment with the model in the laboratory
delivers Q
M
= 0.4 m
3
/s water. Find the dimensions of the model. Experiments with
the model just found yield a pressure gradient of AP
M
= 30,000 Pa/m. Find the
pressure gradient in the full-scale duct. Note that D and V in Eq. (8.21) are
dimensional.
Solution
Let the characteristic length for the duct be chosen as its hydraulic diameter,
D =
4A
C
=
4 (10.5)
21+20.5
=0.667m.
The characteristic velocity is chosen as the mean velocity,
V =
Q
A
=
2
10.5
= 4 m/s.
The Reynolds number for the flow in the duct is then
Re =
V D
v
=
2.667
v
.
Now for the model
D
M
=
4A
M
C
M
, V
M
=
Q
M
A
M
,
Re
M
=
V
M
D
M
v
=
4Q
M
/ C
M
v
,
and for similitude,
254 Fluid Mechanics
Re
M
= Re,
4Q
M
C
M
=
4 0.4
C
M
= 2.667,
or
C
M
= 0.6 m.
Similitude also requires similarity of geometry. Let the larger side of the
model be b; then the smaller side is b/2, and
C
M
= 2(b + b / 2) = 3b = 0.6. The model
is, therefore, rectangular, with the sides 0.2 m and 0.1 m. Also
V
M
=
Q
M
A
M
=
0.4
0.2 0.1
= 20 m/ s,
and
D
M
=
4A
M
C
M
=
4 (0. 2 0.1)
0. 6
= 0.133 m.
To evaluate the pressure drop, we conveniently use Eq. (8.21). For the model, with
L
*
= 1 m, we find
AP
M
= 30,000 = C
f
1
0.133
1,000
2
20
2
;
hence C
f
= 0.02.
For the full-scale duct C
f
is the same. With L
*
= 1 m we find
AP = 0.02
1
0.667
1,000
2
4
2
= 240 Pa.
The pressure gradient in the main duct is 240 Pa/m.
Example 8.3
An important step in the preparation of frozen chickens is just after cleaning,
when the chicken is moved through a blast tunnel where very cold air is blown on
it to chill it very fast. Larger chickens stay in the tunnel longer than smaller ones.
Assuming all chickens similar, find similarity rules and deduce how long
should a chicken of each size stay in the tunnel. The differential equation govern-
ing the cooling process is given as
cT
*
ct
*
= o V
*
2
T
*
,
where T
*
is temperature, t
*
is time and o [m
2
/s] is the thermal diffusivity.
Solution
Let the characteristic length, L, be some well-defined length of the chicken
(e.g., its width), and the characteristic time, t, be the time it has to stay in the
8. Similitude and Order of Magnitude 255
tunnel. Then
t* = t t, x*, y*, z* = x, y, z ( )L,
V*
2
= V
2
/ L
2
,
and the equation becomes
cT
ct
= FoV
2
T , where Fo=
ot
L
2
is theFourier number.
Thus conditions for similarity for two chickens are
Fo
1
= Fo
2
,
o
1
t
1
L
1
2
=
o
2
t
2
L
2
2
.
Let the mass of a chicken be m and its density be . Then, for geometrically similar
chickens,
m
1
m
2
=
1
2
L
1
L
2
|
\
|
.
|
3
.
Assuming o
1
= o
2
and
1
=
2
, the condition for the equality of the Fourier
numbers becomes
t
1
t
2
=
L
1
L
2
|
\
|
.
|
|
2
=
m
1
m
2
|
\
|
.
|
|
2/ 3
.
A chicken which is twice the mass of another should stay 2
2/3
= 1.587 times
the time of stay of the other.
Example 8.4
The force of air resistance to a train moving at the speeds of 3, 10 and 25 m/s
is to be obtained by the use of a model. The linear dimension of the model is 1/5 of
that of the train. The model is run in a wind tunnel, in air. Find the corresponding
air speeds in the wind tunnel. Find the relations between the forces measured on
the model and those, which will act on the full-size train.
Solution
The governing equations are the continuity equation and the NavierStokes
equations. The similarity dimensionless numbers are, therefore, the Reynolds
number and the Euler number.
For the Reynolds number,
256 Fluid Mechanics
Re
T
= Re
M
,
V
T
L
T
v
T
=
V
M
L
M
v
M
.
Let the characteristic length be the height, H, and assume v
T
= v
M
. Then L
T
= H
T
,
L
M
= H
M
and
V
M
= V
T
H
T
H
M
= 5V
T
.
The corresponding speeds in the wind tunnel are
V
M1
= 3 5 =15m / s; V
M2
= 10 5 = 50 m/ s; V
M3
= 25 5= 185m/ s.
For the Euler number
Eu
T
= Eu
M
,
t
1
1
V
1
2
=
t
2
2
V
2
2
.
We assume
1
=
2
, and since the problem has no typical characteristic pressure,
we choose
t =
1
2
V
2
, i.e., t
T
=
1
2
T
V
T
2
, t
M
=
1
2
M
V
M
2
.
With this the condition for the equality of the Euler numbers is satisfied automati-
cally, because
Eu
T
=
1
2
, Eu
M
=
1
2
.
To find relations between the forces, we need the dimensionless forces, which are
equal in similar problems. We define a characteristic force by
= t A =
1
2
V
2
A.
Thus the dimensionless forces, F, relate to the dimensional ones, F*,
F
*
= F = F
1
2
V
2
A ,
i.e.,
F
T
*
= F
T
1
2
T
V
T
2
A
T
, F
M
*
= F
M
1
2
M
V
M
2
A
M
.
We assume
T
=
M
. Also
A
T
/ A
M
= L
T
2
/ L
M
2
. At similarity F
T
= F
M
. Equality of
the Reynolds numbers gives
V
M
/ V
T
= L
T
/ L
M
.
Hence, substitution yields
F
T
*
= F
M
*
.
The force acting on the model is the same as that acting on the train. Note the high
speeds of the air flowing past the model.
8. Similitude and Order of Magnitude 257
Order of Magnitude
The previous section deals with similitude and shows how solutions of
particular cases can be extended and applied to whole families of similar flows.
However, there are many varieties of flows that a fluid dynamicist is compelled to
consider. For many of them no exact solutions are known, experiments are not
practical and approximate methods must be used.
A way to affect an approximate solution is to identify in the equations some
terms, which are quite smaller than the other terms. One may try then to drop the
smaller terms and to solve the truncated equation, obtaining, hopefully, an
approximation. Two distinct questions arise in this procedure. The first one is the
correct estimate of the order of magnitude of the various terms in the equations.
The second question is whether dropping a small term in the equation results in a
small error in the solution.
The first question is considered under the heading of the proper choice of the
characteristic quantities; the second comes under perturbations but is still closely
connected with the characteristic quantities chosen.
Here, again, the equations are written in their dimensionless form, but now
we make an effort to choose the characteristic quantities such that all terms
containing dependent variables are of order 1. The order of magnitude of each
term in the dimensional equation is reproduced in the dimensionless equation by
the dimensionless parameters associated with the term.
The dimensionless groups, i.e., the similarity parameters, become now the
order of magnitude parameters. The following example further illustrates this
point. Let Eq. (8.2) be rewritten, with the order of magnitude of each term indi-
cated by a number written below it; for simplicity we assume v = w = 0 and
consider flows in which body forces are negligible:
cu
*
ct
*
+ u
*
cu
*
cx
*
=
1
cp
*
cx
*
+ v
c
2
u
*
cx
*2
+
c
2
u
*
cy
*2
+
c
2
u
*
cz
*2
|
\
|
.
|
|
.
(8.22)
O(1) O(100) O(1,000) O(1,000)
In its dimensionless form, as in Eq. (8.14), Eq. (8.22) becomes
O
cu
ct
+ u
cu
cx
= Eu
cp
cx
+
1
Re
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
.
(8.23)
Assuming success in our effort to enforce order 1 on all terms containing depen-
dent variables, we have
258 Fluid Mechanics
cu
ct
= O 1
( )
, u
cu
cx
= O 1
( )
,
cp
cx
= O 1
( )
,
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
= O 1
( )
.
(8.24)
and since the uu/x term has no coefficient, the whole equation must have been
rescaled with respect to it, i.e., divided by 100. Hence, now,
O = O 1 / 100
( )
= O 0. 01
( )
,
Eu = O 1, 000 / 100
( )
= O 10
( )
,
1 / Re = O 1, 000 / 100 ( ) = O 10 ( ).
(8.25)
Equation (8.22) describes a certain physical phenomenon i.e., it is a physical truth.
This physical truth cannot be changed by nondimensional formulation, and
therefore the relative magnitude of the four terms in Eq. (8.23) must be the same as
in Eq. (8.22). In trying to make all dependent variable groups of order 1, the
relative correct order of magnitude is transferred to the dimensionless parameters,
Eqs. (8.25). At this point it seems that an approximate solution to either Eq. (8.22)
or Eq. (8.23) may be obtained neglecting the u/t term on the left-hand side.
Estimates of the Characteristic Quantities
Until now we have taken the order of magnitude of the various terms in the
equation as known. We must remember, however, that the object of the whole
procedure is to simplify the equations, and therefore we must have a method to
estimate these orders of magnitude a priori.
The major difficulty here is the selection of quantities that are really charac-
teristic of the problem and thus have the correct order of magnitude. This is by no
means an automatic procedure, and clearly, the criterion of "well defined but
otherwise arbitrary," which applies in our similarity considerations, completely
fails here. The proper selection depends on the understanding and the information
the selector has on the physics of the problem, and there are examples of
disagreement on this selection. There are, however, a few helpful indicators for a
reasonable selection, which can be presented rather generally.
i. The investigated phenomena must be sensitive to variations in the selected
characteristic quantity. Consider, for instance, the flow field at point A, which
is at a distance L from a submerged body, which is an ellipsoid of axes 2a, 2b,
2c, Fig. 8.4. Under certain circumstances, e.g., for large L, the flow at the
point of consideration is quite insensitive to variations in a, b or c. Under
these conditions a, b or c are not characteristic lengths and should not be
8. Similitude and Order of Magnitude 259
chosen as such. A possible reasonable choice for this case is the distance L
between the ellipsoid and the considered point; we assume here that the flow
at least "feels" the presence of the body, through the q = 0 boundary condition
on the rigid body surface.
ii. When scaling a derivative, say u/y, a fair scaling is Au/Ay in the region of
interest. Although u/y may vary in this region, there exists at least one point
where u/y = Au/Ay, and since only one scaling is permitted, it might as well
be this one. The characteristic velocity in such a case is
u
2
u
1
and the
characteristic length
y
2
y
1
, 2 and 1 being subscripts for the limits of the
region.
L
2a
2b
A
Figure 8.4 Submerged ellipsoid.
iii. A second derivative is scaled as a first derivative of a derivative, and so on to
higher derivatives.
iv. When no characteristic quantity is apparent, such a quantity may be con-
structed using other characteristic quantities and physical coefficients,
provided the constructed combination has the right dimensions and satisfies
the sensitivity test.
An example of such a situation has already appeared in incompressible pipe
flow, where the characteristic pressure had to be constructed in terms of V
2
.
Example 8.5
The flow between a stationary inner cylinder and an external rotating one is
given by the differential equation (6.71):
d
2
q
u
*
dr
*2
+
1
r
*
dq
u
*
dr
*
1
r
*2
q
u
*
= 0
(6.71)
with the boundary conditions,
260 Fluid Mechanics
q
u
*
= 0 at r
*
= R
i
,
the inner cylinder;
q
u
*
= e R
o
at r
*
= R
o
,
the outer cylinder.
Using order of magnitude considerations, find an approximation to this flow
for small dimensionless gaps, c, between the cylinders, where
c =
R
o
R
i
R
i
.
Solution
With rules (ii) and (iii) in mind we denote the gap between the two cylinders
by o,
o = R
o
R
i
,
and select it as the characteristic length; the difference of the velocities between
the two cylinders,
V = q
u
*
R
o
q
u
*
R
i
= eR
o
,
is chosen as the characteristic velocity. Now rules (ii) and (iii) yield
dq
u
*
dr
*
=
eR
o
o
dq
u
dr
,
d
2
q
u
*
dr
*
2
=
eR
o
o
2
d
2
q
u
dr
2
,
where we keep our convention that starred variables are dimensional and non-
starred ones are not. The scaling of r
*
itself is done by R
i
, because r
*
= R
i
. Hence,
r
*
= R
i
r =
R
o
R
i
( )
c
r =
o
c
|
\
|
.
| r .
Substitution of all the dimensional scaled quantities into the differential equation
(6.71) yields
eR
o
o
2
|
\
|
.
|
d
2
q
u
dr
2
+
c
o
|
\
|
.
|
1
r
eR
o
o
|
\
|
.
|
dq
u
dr
c
o
|
\
|
.
|
2
1
r
2
eR
o
( )
q
u
= 0,
or through division by
eR
o
o
2
,
d
2
q
u
dr
2
+ c
1
r
dq
u
dr
c
2
1
r
2
q
u
= 0 .
Now for c << 1 the approximation becomes
8. Similitude and Order of Magnitude 261
d
2
q
u
dr
2
= 0 ,
with the dimensional solution
q
u
= eR
o
r R
i
o
|
\
|
.
|
;
and when c is not very small compared to 1, but still c
2
<< 1, the approximation
becomes
d
2
q
u
dr
2
+
c
r
dq
u
dr
= 0 ,
with the dimensional solution
q
u
= e R
o
ln(r / R
i
)
ln(R
o
/ R
i
)
~ e R
o
r R
i
o
|
\
|
.
|
3R
i
r
3R
i
R
o
|
\
|
.
|
.
Both approximations may now be compared with the exact solution
q
u
= eR
o
r R
i
o
R
o
r
|
\
|
.
|
r + R
i
R
o
+ R
i
|
\
|
.
|
|
,
obtained from Eq. (6.73) for O
i
= 0, O
o
= e .
*The Concept of Perturbations
Let a differential equation have the form
F y, x
i
, o ( ) = 0,
(8.26)
where o is some parameter, x
i
are independent variables and y is the dependent
variable. Equation (8.26) may also mean that the partial derivatives y/x
i
and
higher ones do appear. The equation is assumed to be accompanied by sufficient
boundary conditions, and the uniqueness and existence of the solution are also
implied.
Let the solution of this equation for a given o, say o = 0, be known at a
particular point, i.e., at a particular set of x
i
values. The dependent variable y at this
point is, of course, also known. The whole problem may be reformulated and
solved anew for different values of the parameter o, and y may now be considered
a function of o at that particular point.
Let this y (o) have a series expansion in o, around the given value o = 0. Let
this expansion hold for a whole region of points, i.e., for a continuous region of x
i
values. Thus
262 Fluid Mechanics
y x
i
, o ( ) = y
o
x
i
, 0 ( )+ o y
1
x
i
, 0
( )
+ o
2
y
2
x
i
, 0 ( )+
.. .
(8.27)
and y
o
(x
i
, 0) must satisfy Eq. (8.26) with o = 0:
F y
o
, x
i
, 0 ( ) = 0.
(8.28)
Equation (8.27) is a series solution of Eq. (8.26) around the point o = 0. The
solution at that point, Eq. (8.28), is obtained by a "perturbation" in the original
equation (8.26), i.e., by setting o = 0. A solution in the form of Eq. (8.27) is there-
fore called a perturbation solution.
In some difficult cases the series is terminated after y
o
(x
i
, 0), simply because
the evaluation of additional terms is not practical. Then y
o
(x
i
, 0) is used as an
approximation to y (x
i
, o) provided o is small. This approximation is said to have
been obtained by perturbation in the original equation (8.26). The equation is
considered "perturbed" by the substitution of zeros for small order of magnitude
terms, and the smaller this perturbation is, the better the approximation.
*Singular Perturbations
The solution of a differential equation satisfies, by definition, both the equa-
tion and the boundary conditions. The number of the boundary conditions to be
satisfied depends in general on the order of the equation.
The process of perturbation as described in the previous section has the effect
of deleting some terms from the equation. It may happen that the deleted term is
the highest derivative and the equation left after the perturbation is of a lower
order. The solution of this new equation cannot satisfy all the boundary conditions.
Therefore, near a boundary it is simply wrong and cannot serve even as an
approximation. Far from a boundary it may still be an approximation. However,
near the boundary the perturbation process must be modified and performed very
carefully, so as to result in an equation of the same differential order as the
original one. This new process is called Singular Perturbation.
Singular perturbation, i.e., perturbation in the highest derivative of the equa-
tion, is even more complicated than the regular perturbation considered before. It
will therefore be used here only in the particular cases necessary to obtain the
boundary layer equations. Still, the concept is quite general and is used in the
solution of differential equations.
The Boundary Layer Equations
Consider the dimensionless NavierStokes equations written out for two-
dimensional time-independent flows:
8. Similitude and Order of Magnitude 263
u
cu
cx
+ v
cu
cy
= Eu
cp
cx
+
1
Re
c
2
u
cx
2
+
c
2
u
cy
2
|
\
|
.
|
|
,
(8.29)
u
cv
cx
+ v
cv
cy
= Eu
cp
cy
+
1
Re
c
2
v
cx
2
+
c
2
v
cy
2
|
\
|
.
|
|
(8.30)
together with the continuity equation
cu
cx
+
cv
cy
= 0 .
(8.31)
In many practical cases these equations cannot be solved analytically, and
approximate formulations become very useful.
Inspection of Eqs. (8.29) - (8.31) reveals that extreme values of the Reynolds
number may indeed lead to such approximations:
Very small Re enhance the relative importance of the second derivative
terms, making them dominant. Then the nonlinear terms may be neglected. This
situation is considered in Chapter 9.
Very large Re tend to make the second derivative terms negligible. This
changes the NavierStokes equations into the Euler equations, already mentioned
as Eq. (5.70). This direction is followed in Chapter 10. However, the Euler
equation is of the first order and cannot satisfy the two boundary conditions
q
n
= 0, q
t
= 0
(5.62)
on rigid boundaries. The Euler equations, therefore, cannot be the correct approx-
imation near a rigid boundary. Thus close to a boundary the perturbation associ-
ated with Re is a singular perturbation. The resulting approximation, called
the boundary layer equations, are further treated in Chapter 11. Their derivation,
however, being a singular perturbation, follows here.
We consider again Eqs. (8.29) - (8.31). We look for the perturbed form of the
equations for Re , and our region of interest is in the vicinity of rigid
boundaries. It is impossible to use here a regular perturbation where Re is
simply substituted in the equations, which leads to the Euler equations. The two
boundary conditions do not permit this. A singular perturbation must therefore be
performed.
We assume the rigid surface to be generally flat, e.g., a flat plate in parallel
flow. Let a coordinate system be chosen as in Fig. 8.5, with y normal to the rigid
surface. The x-coordinate points along the surface and coincides with the direction
of the velocity, U, at some distance far from the rigid surface.
It seems reasonable to choose U as the characteristic velocity and because v,
the kinematic viscosity of the fluid, is known, the Reynolds number is almost
determined; we still need only to select the characteristic length.
264 Fluid Mechanics
x
o
U
u
y Y
Figure 8.5 Boundary layer flow.
The problem we are faced with is what relations must exist between 1/Re and
c
2
u cx
2
+ c
2
u cy
2
( ) such that for Re and at very small y this second
derivative term is not completely lost.
No help comes from the
c
2
u cx
2
= 0
term. On the rigid surface u = 0 because
of the no-slip boundary condition. For flat surfaces
cu cx =c
2
u cx
2
=0
because u
is constant, i.e., u = 0, for all x. Thus this term vanishes even without Re . We
must therefore direct our efforts at
1
Re
c
2
u
cy
2
= 0.
We note that for a fixed c
2
u cy
2
,
lim
Re
1
Re
c
2
u
cy
2
|
\
|
|
.
(
(
= 0 .
Hence, to keep
1 Re c
2
u cy
2
( )
finite as Re ,
c
2
u cy
2
cannot be held constant
but must also become infinitely large. In other words, when we consider a series of
flow fields, quite similar to one another except that each flow has a Reynolds
number larger than the flow which has just preceded it, the flow must also have
larger
c
2
u cy
2
values; as Re , say because U , then
c
2
u cy
2
must also
diverge.
To find the rate at which this second derivative must diverge, we refer to the
sections on order of magnitude and estimates of the characteristic quantities and
find, under rules (ii) and (iii),
c
2
u
cy
2
= O
U
o
2
|
\
|
.
| ,
where o is the thickness of the layer near the boundary over which u changes from
8. Similitude and Order of Magnitude 265
U
o
to zero. We expect now that with increasing Re this o-thick layer becomes
thinner and thinner, so as to make
lim
Re
1
Re
c
2
u
c y
2
|
\
|
.
|
|
(
(
(
= lim
Re
1
Re
U
o
2
(
(
finite, i.e., of order 1. Inspection of this relation reveals that Re
.
o
2
or o
.
Re
1/2
must
remain finite. Because o does not appear explicitly in the equations we proceed as
follows: Let a new y-coordinate, say Y, be defined:
Y = y Re
1
2
, y = Y Re
1
2
.
(8.32)
This makes
1
Re
c
2
u
cy
2
=
c
2
u
cY
2
,
which no longer depends on the Reynolds number, and thus does not vanish. The
layer o over which u decreases from U
o
to 0 is very thin, i.e., of order Re
-1/2
. The
whole change in the velocity takes place in this narrow layer near the boundary,
called the boundary layer.
To find the order of magnitude of v inside the boundary layer, we substitute
y =Y Re
1/2
in the continuity equation (8.31) to obtain
cu
cx
+ Re
1/ 2
cv
cY
= 0,
and to keep this equation meaningful, i.e., to have both terms of the same order, we
must define
V =vRe
1/ 2
, v =V Re
1/2
(8.33)
to make it
cu
cx
+
cV
cY
= 0
(8.34)
The y-component of the velocity, v, is of order Re
-1/2
.
We now proceed to substitute Y and V in Eq. (8.30), which becomes
Re
1
u
c V
c x
+ V
c V
c Y
|
\
|
.
|
|
= Eu
c p
c Y
+ Re
1
Re
1
c
2
V
c x
2
+
c
2
V
cY
2
|
\
|
.
|
|
.
(8.35)
For large Re, Eq. (8.35) yields the order of magnitude of p /Y as
( )
1
Eu =O Re ,
p
Y
c
c
(8.36)
and for
Re
this term vanishes. Hence, the pressure does not depend on y and
266 Fluid Mechanics
therefore does not vary across the boundary layer. Equation (8.30) reduces to
c p
c y
= 0.
(8.37)
The same substitution in Eq. (8.29) yields
u
cu
cx
+ V
cu
cY
= Eu
cp
cx
+ Re
1
c
2
u
cx
2
+ Re
c
2
u
cY
2
|
\
|
.
|
|
.
(8.38)
For
Re
,
Re
1
(c
2
u cx
2
) vanishes, and because
cp / cy also vanishes,
cp cx = dp dx . Equation (8.38) thus becomes
u
cu
cx
+ v
cu
cy
= Eu
dp
dx
+
1
Re
c
2
u
cy
2
(8.39)
which together with Eqs. (8.29) and (8.37) constitute the boundary layer equa-
tions. In dimensional form the boundary layer equations are
u
cu
cx
+ v
cu
cy
=
1
dp
dx
+ v
c
2
u
cy
2
,
(8.40)
cp
cy
= 0,
(8.41)
cu
cx
+
cv
cy
= 0
(8.42)
together with the boundary conditions
u = v = 0 at y = 0,
(8.43)
u =U at yRe
1/ 2
,
(8.44)
where
y Re
1/ 2
means the outer edge of the boundary layer.
Buckingham's t Theorem
In a previous section similitude has been motivated by the need to apply
results obtained for a model to a similar real-life situation. Similitude has more
applications, and as an example consider the determination of the drag force acting
on a sphere moving in a liquid. At this stage we do not know how to solve this
case, so we may turn to experiments. We need data for several combinations of
sphere diameters, fluid viscosities and sphere velocities, and therefore a certain
range of the Reynolds number is chosen and k experiments are performed, say
k = 10. Had we not known that the Reynolds number was a similarity parameter for
this case, we would have had to choose a range of velocities, a range of viscosities
and a range of diameters; and to come out with the same coverage we would have
8. Similitude and Order of Magnitude 267
had to perform k
3
experiments, 1,000 of them. The difference between 10 and
1,000 is very significant, and similarity parameters have now gained the additional
role of information concentrators; in this example the concentration is a hundred-
fold. An engineer therefore views similarity parameters as quite desirable. The
engineer knows how to get them from the governing equations, as already shown.
But is there a way to hunt for them when even the governing equations are not
completely available? The Buckingham's t theorem indicates such a procedure.
Imagine there exist two sheets of paper. On one an experienced engineer has
written down the correct set of governing equations, in their dimensionless form.
The engineer therefore sees right away the various similarity parameters, which
are, of course, dimensionless numbers. Unfortunately we cannot read the paper. On
the second sheet we have listed down all the physical quantities,
Q
j
, j = 1, 2, . . . m ,
that might affect the phenomena. Now we start a procedure: each physical quantity
Q
j
is raised to some power n
j
, and a product is formed:
t = Q
1
n
1
Q
2
n
2
...
Q
j
n
j
...
=
j =1
m
[
Q
j
n
j
(8.45)
If all n
j
are adjusted such that t comes out to be dimensionless, we consider the
procedure successful. There may be many combinations of n
j
values, including
some n
i
= 0 values, that satisfy the rules of this game. But note: all the similarity
parameters on the experienced engineer's paper are reproduced by the procedure.
This is so because all relevant physical quantities are presumed listed. This game
may also produce some dimensionless combinations, which are not on the engi-
neer's list. These extraneous parameters are bothersome but not dangerous because
the experiments would prove them irrelevant. What should worry one is the
chance of missing some important similarity parameter. To avoid such a mishap, it
is important to know the largest number of dimensionless parameters that the
procedure can produce.
Suppose some of the physical quantities Q
j
, which are dimensional, include
the dimension of time, in the form
t
a
i
. Then for Eq. (8.45) to be dimensionless, i.e.,
for t to cancel out of the equation, we require that
a
1
n
1
+
a
2
n
2
+
. . .
=
j = 1
m
a
j
n
j
= 0 .
(8.46)
Suppose some of the Q
j
contain the dimension of length, in the form L
b
i
; then
again
b
1
n
1
+
b
2
n
2
+
. . .
=
j = 1
m
b
j
n
j
= 0 .
(8.47)
Thus each additional dimension contained in some of the Q
j
produces an additional
268 Fluid Mechanics
linear equation similar to (8.46) or (8.47).
Let the total number of physical quantities, and therefore of n
j
, be m, and the
total number of independent dimensions be r ; then there are r linear equation for
the m variables n
j
. Now a set of r linear equations for m unknowns has in general
m r + 1 sets of linearly independent solutions. Inspection of Eqs. (8.46) and (8.47)
reveals that all the equations are homogeneous, and therefore
n
1
= n
2
=
. . .
= n
j
=
. . .
= n
m
= 0
is also a solution. This trivial solution yields no similarity parameter and therefore
may be ignored. The number of the remaining solutions is therefore m r, which is
the t theorem. Thus, when we find m r linearly independent solutions, and
therefore have a list of m r dimensionless groups, we may be assured that all the
similarity parameters from the wise engineer's paper are included in our list.
It is noted that products, quotients and powers of dimensionless groups are
not counted as new groups, and this is a direct result of the requirement that the
solutions for the n
j
combinations be linearly independent.
Thus, Buckingham's t theorem is indeed a powerful tool, which facilitates the
establishment of similitude even when the governing equations are not known.
Like most powerful tools it must be handled with care. The main pitfall in its
application is the omission, due to ignorance, of some physical quantities, which
affect the phenomenon. It is noted that in a situation where even the governing
equations cannot be formulated, ignorance is not a harsh word but rather a call for
caution. An experienced engineer does use the t theorem when necessary, but
whenever the governing equations are available they would be preferred as the
source of the similarity parameters.
Example 8.6
A sphere submerged in a liquid is released and sinks down, or floats up. Find
the similarity parameters of the problem.
Solution
The following quantities are presumed to affect the phenomenon:
D[ m ] the diameter of the sphere;
s
[kg / m
3
] the density of the sphere material;
f
[kg / m
3
] the density of the fluid;
V [ m / s ] the instantaneous velocity of the sphere;
[ kg / m
.
s ] the viscosity of the fluid;
g [ m / s
2
] the acceleration of gravity.
8. Similitude and Order of Magnitude 269
The number of physical quantities is m = 6, and the number of involved
dimensions is r = 3. We may therefore expect three dimensionless parameters.
Rather than plow through the steps leading to the formal solution of the set of
equations, let us try to guess these parameters. This approach permits one to use
some physical intuition, and the Buckingham t theorem reveals when one may
stop
guessing because enough parameters have been found.
Let our guess be
DV
f
|
\
|
.
|
,
f
s
( )
D
2
g
V
|
\
|
.
|
|
,
f
s
s
|
\
|
.
|
and the reader may check that indeed all three are dimensionless.
A clue to the selection of the list of physical quantities is also a clue to the
guess of the similarity parameters:
As the sphere moves, the fluid undergoes acceleration and viscous effects,
and the ratio between the relevant terms is the Reynolds number, i.e., the first
parameter. The second parameter presents the ratio between buoyancy and viscous
forces, and the third one gives the ratio between buoyancy and the sphere's own
acceleration.
The sphere does accelerate, and when the time, t [ s ], during which acceler-
ation is important is within the range of interest another dimensionless group
emerges
gDt
|
\
|
.
|
;
and when a glass container of diameter B is used in the experiments, the dimen-
sionless group D/B must be included.
References
G. Birkhoff, "Hydrodynamics, a Study in Logic, Fact and Similitude," Dover, New
York, 1955.
J.D. Cole, "Perturbation Methods in Applied Mathematics," Blaisdell, Waltham,
MA, 1968.
W.J. Duncan, "Physical Similarity and Dimensional Analysis," E. Arnold & Co.,
London, 1953.
270 Fluid Mechanics
M. Holt, "Dimensional Analysis," in V.L. Streeter (ed.), "Handbook of Fluid
Dynamics," McGraw-Hill, New York, 1961.
S.J. Kline, "Similitude and Approximation Theory," McGraw-Hill, New York,
1965.
H.L. Langhaar, "Dimensional Analysis and Theory of Models," Wiley, New York,
1951.
L.I. Sedov, "Similarity and Dimensional Methods in Mechanics," Academic Press,
New York, 1959.
M.D. Van Dyke, "Perturbation Methods in Fluid Mechanics," Parabolic Press,
Stanford, CA, 1975.
Problems
8.1 A model of a car is 1/4 of the length of the car itself. The car runs between
50 and 100 km/h. It is suggested to test air resistance to the motion of the
car by running the model (a) in an air tunnel, (b) in a water tunnel. Find the
running speed in each tunnel and the interpretation of the measured resis-
tance forces.
8.2 Complete similarity between a ship model and a full-size ship requires the
same Reynolds and Froude numbers. Suppose that
Re = LV/v = 1110
8
and Fr
-1
= gL / V
2
= 33.25.
Assuming a model 1/100 the size of the ship, can you design an experiment
having full similarity? If not, why? Could you design a full-similarity exper-
iment for a submarine?
8.3 Find conditions for similarity for time-independent flows between parallel
plates: shear flows, plane Poiseuille flows and combinations of the two.
Find rules for the interpretation of the results.
8.4 Find conditions for similarity between Rayleigh flows. Find rules for the
interpretation of the results.
8.5 Find conditions for similarity between annulus flows where the inside
cylinder does not have the same center as the outer one. Find rules for the
interpretation of the results.
8. Similitude and Order of Magnitude 271
8.6 Find conditions for similarity between Couette flows for two rotating
cylinders. Find rules for the interpretation of the results. Repeat for non-
concentric cylinders.
8.7 For the flow between parallel plates, using the differential equation (6.21)
and the boundary conditions, Eq. (6.23), find conditions under which the
shear flow part can be neglected and conditions for which the Poiseuille
flow part can be neglected. Check your results using Eq. (6.24).
8.8 A plane shear flow has been demonstrated by the instructor setting the upper
plate suddenly into motion at its full velocity. Find times for which the flow
field is approximated by the Rayleigh flow and times for which the plane
shear flow is a good approximation.
8.9 For the Couette flow between concentric rotating cylinders find conditions
for which the rotation of the inner cylinder may be ignored. Do the same for
the outer cylinder.
8.10 A cylindrical drum full of fluid is suddenly set into rotation at the angular
velocity e. Find times and regions of flow where the phenomenon is
approximated by the Rayleigh flow. Find times for which rigid body rota-
tion is a good approximation.
8.11 It has been suggested to construct a cylinder with a bullet head on one side
and with a thin soft plastic fish-tail on the other. A top view of the cylinder
is shown in Fig. P8.11. The cylinder is made of soft iron, and when placed
in a periodic magnetic field, it wiggles its tail and moves forward. By
making it hollow, it can have neutral buoyancy. The idea was to let it swim
inside blood vessels, for medical purposes. Find the similarity parameters
for this magnetic fish.
Figure P8.11 Cylinder with tail fin.
8.12 Assume you have not seen the solution of the Rayleigh problem: the
suddenly accelerated flat plate. Find its similarity parameters using Buck-
ingham's t theorem. Check the similarity parameters with those obtained
from the differential equation, and note the clue they give you toward the
272 Fluid Mechanics
exact analytical solution.
8.13 In Example 8.3 the physical quantities that affect the cooling rate of the
chicken are
D [ m ] its size,
[ kg/m
3
] its density,
c [ kJ/kg
o
C] its specific heat,
k [ w/m
o
C] its thermal conductivity,
h [ w/m
2 o
C] the heat convection coefficient,
(assumed infinite in Example 8.3 but not here),
t [ s ] the time it stays in the tunnel.
Find the similarity parameters of the problem using Buckingham's t theo-
rem.
8.14 A submerged body is designed to move in oil at the speed of 2 m/s. A model
1/8 the size of the body is run in water and yields a drag force of F
d
= 300
N.
For oil
1
= 880 kg/m
3
,
1
= 0.082 Pa
.
s.
For water
2
= 998 kg/m
3
,
2
= 0.082 Pa
.
s.
a. At what speed must the model be run?
b. What is the drag force on the submerged body?
8.15 In fighting forest fires, specially equipped airplanes release a body of water
above the fire. As it falls down, the water is broken into smaller and smaller
drops. Once these drops reach a certain size, they do not break any more.
The details of this phenomenon are to be investigated experimentally. Using
Buckingham's t theorem, find the similarity parameters of this
phenomenon.
8.16 An airplane wing is tested in a
wind tunnel, Fig. P8.16. The pur-
pose of the test is to obtain drag
forces and lift forces acting on the
wing at various flight speeds.
Drag forces are those acting on the
wing in the direction of the
velocity of the oncoming air,
while lift forces act in the
direction perpendicular to that velocity. These forces are measured on the
V
Lift
Drag
Support
Figure P8.16
8. Similitude and Order of Magnitude 273
model. Translate them into the forces, which will act on the full-size wing.
8.17 One way to drill holes in sand, in order to
set posts, is to push a water hose, with the
water running, into the sand, Fig. P8.17.
Find the similarity parameters of this
method.
8.18 While a fuel tank in a ship is used, it must
have a free surface. As the ship rolls, the
fuel motion inside the tank is rather com-
plex. The fuel applies forces to the ship and
these forces are to be measured experi-
mentally, using a model. Find the similarity
parameters for these experimental forces and the way to translate them into
the full-size ship forces.
8.19 A model of a ship's propeller is first tested in a water tunnel, and then it is
assembled on a ship's model and tested again. Find the similarity parameters
for the propeller alone and for the assembled propeller. Formulate the
translation of the model experimental results into the real propeller
performance.
8.20 A weather balloon has a diameter of 2 m and is to be used in air at 20
o
C. To
find the drag force on the balloon, an
experiment was conducted in which a
0.02-m-diameter sphere was held in
water moving with the velocity of 10 m/s.
The drag force on the sphere was
measured as 6.5 N. Find the speed of the
wind past the balloon, which corresponds
to the experiment. Find the drag force on
the balloon at this wind velocity.
8.21 A sorting machine for olives is shown in
Fig. P8.21. Olives drop down in front of
an electronic eye. Black olives just
proceed down. When a green olive
passes the eye, air is released from a
nozzle deflecting the green olive to
another bin. It is necessary to construct a
similar system for olives, which are twice
Water
Sand
Figure P8.17
Figure P8.21 A machine for
sorting olives.
274 Fluid Mechanics
the linear size of those treated now. Find relations between the radii of the
old and new air jets, between their velocities and between the distances the
rejected green olives travel.
8.22 A glider model is 1/10 of the full-size glider. The model is connected to a
car by a wire, and at a speed of 140 km/h the force in the wire is 1,000 N.
Find the speed of the full-size glider relative to that of the model and the
power expended by an airplane pulling the glider at that speed.
8.23 Electrical wires are subject to stresses caused by winds. The effect of the
wind may look like a force distributed along the wire, pulling it in the down-
wind direction, or it may take the form of oscillations. A 1/50 model is to be
tested in a wind tunnel. Find the necessary similarity parameters.
8.24 A new hull design for a ship is to be 100 m long. A model is 3 m long. The
model is tested in a tow tank, which can accommodate towing speeds of up
to 2 m/s. Only the Froude number is conserved for similarity, because it is
impractical to conserve the Reynolds number too. A correction is later made
for the different Reynolds numbers. Find to what ship speed corresponds the
2-m/s speed of the model, and what are the ship and model Reynolds
numbers at these speeds.
8.25 The size of raindrops depends on their speed, on the trajectory they have
traveled and on the surface tension of water. Find the similarity parameters
for this size.
8.26 Some crude oil comes out of wells mixed with pebbles, gravel and sand. It
is suggested to let the oil flow through settling tanks, in which most of these
solids sink down. The experimental investigation of these tanks utilizes
water instead of oil, because water is clear and the settling process may be
viewed, and because crude oil entails continuous cleaning of
instrumentation. Find similarity parameters. Assume that satisfactory
settling has been achieved in water and give the major design parameters for
the oil in terms of those for the water.
8.27 A turbo-pump, which runs at 2000 rpm, takes in water at atmospheric pres-
sure at the rate of 0.010 m
3
/s and discharges it at 200,000 Pa above atmo-
spheric pressure.
a. If the pump rpm is increased to 3,000, find the flow rate and the dis-
charge pressure at which the new operating conditions are completely
similar to the first ones. Neglect viscous dissipation
b. Find the ratio of the powers necessary to run the pump in the two cases.
8. Similitude and Order of Magnitude 275
8.28 A cylindrical drum completely filled with water is set on a turntable. The
turntable is suddenly set into rotation, at 2,000 rpm. After 30 seconds the
water inside the drum turns with the drum in a solid body rotation.
a. Find how long it will take the water to reach solid body rotation if the
drum is suddenly set to 3,000 rpm.
b. Find how long it will take the water in a drum twice the linear size of the
old one, set to 2,000 rpm, and set to 3,000 rpm, to rotate like a solid
body.
8.29 The water in Problem 8.28 is changed to glycerin. Find the times for the old
and the new drums, with 2,000 and 3,000 rpm.
8.30 Two fluids flow into a cylindrical
mixing chamber where they are
mixed at steady state by a paddle
wheel that rotates at 50 revolu-
tions per second, Fig. P8.30. It is
suggested to scale up the process
to twice the amounts of liquids
mixed. Because the original mix-
ing is quite satisfactory, complete
similarity is favorable. One design
suggestion was to increase all
piping to twice their cross sections
and to keep the cylindrical mixing
chamber height but to increase its
diameter by
2
. Another
suggestion was to keep the old structure but to have the flowrates in the
piping increased by 2. Find if these suggestions can satisfy similarity, and if
they can, find the size and the number of revolutions of the paddle wheel.
8.31 Find the similarity parameters for the water hammer described in Example
4.10.
Fluid A
Mixture
Fluid B
Figure P8.30 Paddle-wheel mixer.
276 Fluid Mechanics
277
9. FLOWS WITH NEGLIGIBLE
ACCELERATION
The nonlinear terms on the left-hand side of the NavierStokes equations
result from the acceleration of the fluid. We already know that these terms con-
tribute much to the difficulties one encounters in the solution of the equations.
Indeed, we have noted the relative ease with which exact solution have been
obtained in Chapter 6 for fully developed flows, where those nonlinear terms
vanish. As we now look for approximations, obtained by the solutions of approx-
imate forms of the NavierStokes equations, the first idea that comes to mind is to
remove the terms, which cause the greatest difficulty, i.e., the nonlinear accelera-
tion terms.
Our subject here is fluid mechanics, and in this context we must ask if there
exist real flows in which the acceleration terms are negligible; and if they exist,
how do we identify them? The answer to this enquiry is that there are at least two
such families of flows: flows in narrow gaps and creeping flows.
Flow in Narrow Gaps
Consider the two-dimensional flow of an incompressible fluid in the narrow
gap between the two plates shown in Fig. 9.1. For simplicity we assume the plates
to be flat. The results obtained in this analysis may be extended to cases where the
plates are curved, as long as the gap is much smaller than the radius of curvature.
Another possible extension is to three-dimensional flows, where a w-component of
the velocity also exists. Here we present the simplest examples of gap flows and
limit the analysis to two-dimensional flows between flat plates.The lower plate is
set stationary along the x-axis and the upper plate may be inclined to it with the
small angle o. If the upper plate moves, its motion is restricted to its own plane We
278 Fluid Mechanics
call such a flow a flow in a narrow gap when
o
L
<<1, o <
o
L
.
(9.1)
x
L
o
y
o
A B
Figure 9.1 Flow in a narrow gap.
Let
u , the mean velocity in the x-direction, be defined,
u =
1
o
0
o
}
u dy .
Conservation of mass requires
u
A
o = u
B
(o + Lo) .
Hence
cu
cx
~
cu
cx
~
u
B
u
A
L
= u
o
o
.
(9.2)
The equation of continuity, Eq. (5.21), states
cu
cx
=
cv
cy
.
Hence with Eq. (9.2),
cv
cy
~
u o
o
.
(9.3)
The boundary conditions for v on both plates are
v = 0 at y = 0 and at y = o.
The largest value that | v | may attain, in the vicinity of the middle of the gap, is
approximately
v
max
~
o
2
cv
cy
~
u o
2
<< u . (9.4)
8. Similitude and Order of Magnitude 279
The y-component of the velocity may be neglected in comparison with the
x-component.
Furthermore, for a moving or stationary upper plate
cu
cy
~
u
o
,
(9.5)
and with Eq. (9.2),
cu
cx
cu
cy
~ o <
o
L
<< 1.
(9.6)
The x-derivative of u is negligible compared with the y-derivative.
The approximate form of the NavierStokes equations for two-dimensional
gap flows thus becomes
0 =
dp
dx
+
d
2
u
dy
2
.
(9.7)
Reynolds Lubrication Theory
An important application of flows in narrow gaps, which are flows with their
acceleration terms neglected, is in Reynolds lubrication theory. This theory yields
the forces which appear in bearings and other lubricated sliding surfaces, provided
that indeed acceleration forces may be neglected.
1 2
h
h
1
2
L
U
x x x
o
y
Figure 9.2 Flow in a bearing.
Consider the two-dimensional steady flow between the two plates shown in
Fig. 9.2. The x-wise momentum equation for this narrow gap flow is just Eq. (9.7),
which looks exactly like the one for a fully developed flow.
*
For an upper plate
*
It is the solution, Eq. (9.8), with h (x) varying with x very slowly, that justifies
2
u/x
2
= 0.
280 Fluid Mechanics
moving with the velocity U, Eq. (9.7) has the solution [see shear flow with pres-
sure gradient, Eq. (6.24)]
u =
1
2
dP
dx
|
\
|
.
| y y h(x) | |+U
y
h x
( )
.
(9.8)
Assuming w = 0 (a very wide bearing), the mass flow between the plates is con-
served
m = u dy
o
h( x)
}
=
1
2
Uh(x)
12
dP
dx
|
\
|
.
| h
3
x ( ) = const =
1
2
Uh
o
(9.9)
or
dP
dx
= 6U
h h
o
h
3
.
(9.10)
The lower plate, Fig. 9.2, is inclined to the horizontal by the angle
o ~
dh
dx
.
Now
dP
dx
=
dP
dh
dh
dx
= o
dP
dh
,
which may be substituted in Eq. (9.10) to yield
dP
dh
= 6
U
o
h h
o
h
3
.
(9.11)
Integration yields
P = 6
U
o
1
h
+
h
o
2h
2
(
(
+ B
and h
o
and B have to satisfy the two boundary conditions:
P = P
o
at h
1
and at h
2
.
This is achieved by setting 2 h
o
= 1 h
1
+1 h
2
. Hence
P P
o
= 6
U
o
h h
1
( )
h h
2
( )
h
1
+ h
2
( )
h
2
.
(9.12)
The mass flow rate is obtained from Eq. (9.9) as
m =
1
2
Uh
o
=
1
2
U
2h
1
h
2
h
1
+ h
2
h
1
> h
o
> h
2
( )
.
(9.13)
The lift force L per unit width acting on the upper plate is
8. Similitude and Order of Magnitude 281
L =
x
1
x
2
}
P P
o
( )
dx =
1
o
h
1
h
2
}
P P
o
( )
dh
= 6
U
o
2
2 h
2
h
1
( )
h
1
+ h
2
ln
h
2
h
1
(
(
(
=
6UL
2
k 1
( )
2
h
2
2
lnk
2 k 1
( )
k +1
(
(
(
,
(9.14)
where k = h
1
h
2
>1 , L = x
2
x
1
=
1
o
h
2
h
1
( ). The drag force D per unit width
acting on the upper plate is
D =
x
1
x
2
}
du
dy
|
\
|
.
|
|
y=h
dx =
h
1
h
2
}
1
2
h
dP
dh
+
U
o
1
h
(
( dh =
U
o
h
1
h
2
}
3
h h
o
h
2
+
1
h
(
( dh .
Hence,
D =
2UL
k 1 ( )h
2
2lnk
3(k 1)
k +1
(
( .
(9.15)
To maximize the lift, let dL/dk = 0. This yields
k = 2.2 ,
which corresponds to
L = 0. 4UR , D= 1. 2UR , D/ L = 3 / R ,
(9.16)
where
R =
2L
h
1
+ h
2
.
Example 9.1
Figure 9.3 shows a rotating disk with a floating magnetic reader. The linear
velocity of the disk just below the reader is 60 m/s. The width of the reader plate is
L = 0.02 m and the weight it must support is 5 g/cm in the radial direction. The
reader's arm is set to an angle o which yields the maximum lift but is otherwise
free to float, i.e., it finds its own height h
l
and h
2
. The whole system is placed in
nitrogen at 300 K. Find h
1
, h
2
and o. Find the drag force acting on the reader.
Figure 9.3 Rotating disk and magnetic reader.
U
h
1
h
2
U
L
282 Fluid Mechanics
Solution
The lift force needed to support 5 g/cm is
L =
5 10
3
9.81
10
2
= 4. 905 N / m.
For maximum lift
k = h
1
/h
2
= 2.2
h
1
h
2
L
=
h
2
k 1 ( )
L
= 1.2
h
2
L
= tano.
For nitrogen at 300
K, = 17.8410
-6
kg/m
.
s. Equation (9.14) for the lift force
now yields h
2
,
L =
1
h
2
2
6UL
2
k 1
lnk 2
k 1
k +1
(
or
4. 905 =
1
h
2
2
6 17.84 10
6
60 0. 02
2
1. 2
ln2. 2
2 1. 2
3. 2
(
(
=
8. 2310
8
h
2
2
.
Hence,
h
2
= 1.3 10
4
m = 0.13mm,
h
1
= kh
2
= 0.286 mm,
o = 0
o
26.8' = 0.0078radians.
The drag force is obtained from Eq. (9.15) as
D =
2UL
k 1 ( )h
2
2 lnk
3 k 1 ( )
k +1
(
(
(
=
2 17.84 10
6
60 0. 02
1. 2 1.3 10
4
( )
2ln2. 2
3 1. 2
3.2
(
(
= 0.124 N m.
Creeping Flows
We are still looking for approximate forms of the NavierStokes equations, in
which the nonlinear terms are neglected. Because these troublesome terms are
nonlinear in the velocities, they are forced to become very small when the veloci-
ties are smaller and smaller. Hence the name creeping flows. This idea is expressed
more exactly by the use of the Reynolds number.
8. Similitude and Order of Magnitude 283
The Reynolds number, Re = VL/v, can become small because V, the charac-
teristic velocity, is small or because L, the characteristic length, is small or because
v, the kinematic viscosity, is large. Of course, it is the combination VL/v which
determines the magnitude of Re. However, observing the time number O = L/tV
and the Euler number Eu = t /V
2
, we choose to say that Re is small because V is
small, implying that O and Eu are not necessarily small. Equation (8.14) multiplied
by Re becomes
Re O
cu
ct
+ Re u
cu
cx
+ v
cu
cy
+ w
cu
cz
(
(
(
= EuRe
cp
cx
+
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
.
(9.17)
For Re
.
O = L
2
/vt and Eu
.
Re = t
.
L/V which are not very small, the inertia terms in
the square brackets are neglected and the equation attains the form
O
cu
ct
= Eu
cp
cx
+
1
Re
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
|
\
|
.
|
|
,
(9.18)
with two similar equations for the y- and z-components of the momentum equation.
Assuming incompressible flow, the continuity equation (8.11) remains the same.
Both equations are written in their vectorial form as
V q = 0,
(9.19)
O
cq
ct
= EuVp
1
Re
VVq.
(9.20)
The flows described by Eqs. (9.19) and (9.20) are called Stokes creeping flows.
The equations are linear and possess some properties that are generally useful in
their solution.
Let the divergence of Eq. (9.20) be taken. The term on the left-hand side
becomes
V
cq
ct
|
\
|
.
| =
c
ct
Vq ( )=
c
ct
0 ( ) = 0,
and this is because V
.
q = 0, Eq. (9.19).
On the right-hand side,
V V V q ( ) = V V V q ( ) | |= 0,
and this term vanishes because the divergence of a curl of any vector is identically
zero. The only term left of Eq. (9.20) is
V
2
p = 0.
(9.21)
Thus the pressure in creeping flows is a harmonic function, i.e., it satisfies the
Laplace equation.
284 Fluid Mechanics
Equation (9.21) is not as useful as it looks because the boundary conditions in
terms of p are not always known; still this equation is used in many cases.
For time-independent flows an alternative general form can be obtained by
taking the curl of Eq. (9.20). Since ( ) 0 p V V = for any scalar p, this results in
V V V q = 0.
(9.22)
Equation (9.22) is particularly useful in flows for which the direction of the
Vq
vector is a priori known, e.g., two-dimensional and axisymmetric flows.
Two-dimensional Flows
In two-dimensional flows with the velocity vector
q = iu + jv ,
V q = i
c
cx
+ j
c
cy
|
\
|
.
|
|
iu + jv ( ) = k
cv
cx
cu
cy
|
\
|
.
|
|
.
Taking the curl of Vq results in
V V q ( ) = i
c
cy
cv
cx
cu
cy
|
\
|
.
|
|
j
c
cx
cv
cx
cu
cy
|
\
|
.
|
|
,
and the final curl,
V V V q ( )
| |
= k
c
2
cy
2
+
c
2
cx
2
|
\
|
.
|
|
cv
cx
cu
cy
|
\
|
.
|
|
.
Thus, for two-dimensional flows in rectangular coordinates Eq. (9.22) becomes
c
2
cy
2
+
c
2
cx
2
|
\
|
.
|
|
cu
cy
cv
cx
|
\
|
.
|
|
= 0.
(9.23)
With the two-dimensional stream function as defined by Eq. (5.57)
u =
c
cy
, v =
c
cx
.
(5.57)
Equation (9.23) may be written in terms of as
c
2
cx
2
+
c
2
cy
2
|
\
|
.
|
|
c
2
cx
2
+
c
2
cy
2
|
\
|
.
|
|
= 0
(9.24)
or
( )
2 2
0 V V =
(9.25)
Equation (9.25) is sometimes written as
8. Similitude and Order of Magnitude 285
V
4
=
c
2
cx
2
+
c
2
cy
2
|
\
|
.
|
|
2
= 0,
(9.26)
which means that the operator inside the parentheses has to be applied twice in
succession. The stream function is thus biharmonic.
Axisymmetric Flows
One of the most important creeping flows is that of the flow around a sphere.
To treat this case efficiently, we must extend the concept of the stream function,
which has been introduced in Chapter 5 for two-dimensional flows, to flows,
which are axisymmetric.
A
B
C
D
P ( r , z )
M( R, |)
-dz
Rd|
r
q
r
dr q
z
R
q
R
|
z
q
|
-
dR
r
Figure 9.4 Axisymmetric representative plane and streamlines.
A flow is defined axisymmetric when its description in polar coordinates (z, r,
u) shows no q
u
velocity and no dependence on the u-coordinate. Such a flow can
be described in the u = 0 plane, with the velocity vector and the streamlines lying
in this plane, Fig. 9.4. Furthermore, u = C planes, which are obtained by rotating
the u = 0 plane around the z-axis, show a flow pattern identical to that in the u = 0
plane. The u = 0 plane is therefore called the representative plane of the
axisymmetric flow. Figure 9.4 shows such a representative plane, with four
streamlines,
denoted by A,B,C,D. The streamlines in this representative plane are
also the lines of intersections of stream sheets with the representative plane, in a
fashion similar to that in two-dimensional
flows. In this case, however, these
286 Fluid Mechanics
stream sheets are really stream tubes with circular cross sections, obtained by the
rotation of the whole figure around the z-axis. Each streamline may now have a
number attached to it signifying the mass flow inside its corresponding stream
tube.
The number zero flow is naturally assigned here to the z-axis which, in
axisymmetric flows, is always a streamline. These numbers, measuring the mass
flow inside the stream tubes, are called Stokes stream functions or, where no
confusion with the two-dimensional stream function is expected, just stream
functions.
As in the treatment of the two-dimensional case, let
D
=
C
+ d. The cross
section available for q
z
in this case is 2tr dr, and that for q
r
is 2tr dz. Hence
d = 2tr dr ( ) q
z
( ) = 2tr dz ( ) q
r
( ),
from which follows
q
z
=
1
2tr
c
cr
, q
r
=
1
2tr
c
cz
.
(9.27)
The spherical coordinates (R, |, u) also appear in the representative plane, with no
dependence on u, of course. Using these with
B
=
A
+ d ,
one obtains
( )( ) ( )( )
( )( ) ( )
( )
2
2 2 sin
2 2 sin ,
R R
d rRd q R d q
r dR q R dR q
|
|
t | t | |
t t |
= =
= =
from which follows
q
R
=
1
2tR
2
sin|
c
c|
|
\
|
.
|
|
R
, q
|
=
1
2tRsin|
c
cR
|
\
|
.
|
|
.
(9.28)
As in the two-dimensional case, is ignored for incompressible flows. The
1/2t factor in Eqs. (9.27) and (9.28) is sometimes also suppressed, and one must
always check whether this has been done when considering numerical computa-
tions.
Obviously the Stokes stream function should also satisfy identically the
continuity equation. It is important to remember that the use of stream functions is
equivalent to the inclusion of the continuity equation in the considerations.
The considerations of the Stokes stream function up to this point are quite
general and are not limited to creeping flows.
We now return to axisymmetric creeping flows. Using cylindrical coordi-
nates, in which q
u
= 0,
q
z
=
1
r
c
cr
, q
r
=
1
r
c
cz
.
(9.29)
8. Similitude and Order of Magnitude 287
Substitution of these in Eq. (9.28) results in the axisymmetric equivalent of
Eq. (9.26):
c
2
cr
2
1
r
c
cr
+
c
2
cz
2
|
\
|
.
|
|
2
= 0,
(9.30)
which, however, is not the biharmonic equation.
Similarly, for axisymmetric flows using spherical coordinates
q
R
=
1
R
2
sin|
=
c
c|
, q
|
=
1
R sin|
c
cR
,
(9.31)
one obtains
c
2
cR
2
+
sin|
R
2
c
c|
1
sin|
c
c|
|
\
|
.
|
|
(
(
(
2
= 0.
(9.32)
It is noted again that Eqs. (9.30) and (9.32) are linear and methods for their
solutions are fairly well known.
Stokes Flow around a Sphere
Figure 9.5 shows a stationary sphere set in a parallel flow. The radius of the
sphere is a and its center is chosen to coincide with the origin of the coordinate
system. There exists a creeping flow field around the sphere, with the velocity far
from the sphere satisfying
q
z
= U
= const, q= kq
z
.
Figure 9.5 Creeping flow around a sphere.
288 Fluid Mechanics
Spherical coordinates are convenient to apply to this configuration, and the
Stokes stream function is related to the velocity vector by Eq. (9.31):
q
R
=
1
R
2
sin|
c
c|
, q
|
=
1
Rsin|
c
cR
.
(9.31)
The differential equation for the stream function is now Eq. (9.32):
c
2
cR
2
+
sin|
R
2
c
c|
1
sin|
c
c|
|
\
|
.
|
|
(
(
(
2
= 0.
(9.32)
This equation is of the fourth order, and there are four boundary conditions that
must be satisfied: the two velocity components, q
R
and q
|
, must vanish on the
sphere and they must become the undisturbed parallel flow velocity components
far away from the sphere. Formally,
q
R
=
1
R
2
sin|
c
c|
= 0,
c
c|
= 0 at R = a ,
q
|
=
1
R sin|
c
cR
= 0,
c
cR
= 0 at R = a,
(9.33)
and because the undisturbed parallel flow is just
q = kU
=e
R
q
R
+e
|
q
|
=U
e
R
cos| e
|
sin|
| |
,
then
q
R
= U
cos| =
1
R
2
sin|
c
c|
at R ,
q
|
= U
sin| =
1
R sin|
c
cR
at R ,
or
c
c|
= U
R
2
cos|sin| at R ,
c
cR
= U
Rsin
2
| at R .
(9.34)
Equations (9.34), which hold for very large R, may be integrated to yield
=
1
2
U
R
2
sin
2
|.
Now this result must hold for any |, and one is tempted to try
= f R ( )sin
2
| for any R.
Substitution in Eq. (9.32) yields
d
2
dR
2
2
R
2
|
\
|
.
|
|
2
f ( R) = 0,
(9.35)
8. Similitude and Order of Magnitude 289
which is homogeneous in R and dR and therefore may have a solution in the form
of a polynomial. Indeed, Eq. (9.35) is satisfied by
f R ( )=
A
R
+ BR +CR
2
+ DR
4
.
Because from the integration of Eq. (9.34), for R , f (R) =
1
2
U
R
2
,
D = 0 and C =
U
2
,
and A and B do not affect the velocity at R . The stream function thus becomes
=
A
R
+ BR +
1
2
U
R
2
|
\
|
.
| sin
2
| .
(9.36)
Equation (9.36) must be satisfied for R = a. Hence
A
a
+ Ba +
1
2
U
a
2
= 0,
A
a
2
B U
a = 0
or
A =
1
4
U
a
3
and
B =
3
4
U
a,
and finally
= U
a
2
1
4
a
R
3
4
R
a
+
1
2
R
a
|
\
|
.
|
2
(
(
(
sin
2
| ,
(9.37)
q
R
= U
1 +
1
2
a
R
|
\
|
.
3
3
2
a
R
(
(
cos|, (9.38)
q
|
= U
1
1
4
a
R
|
\
|
.
3
3
4
a
R
(
(
sin|. (9.39)
We may now proceed to calculate the pressure. Equation (9.20) is written again in
dimensional form for the present case
0 = Vp V V q,
(9.40)
and the expression for the velocity, Eqs. (9.38) - (9.39), is used to calculate
V q = e
|
3
2
U
a
R
2
sin|
|
\
|
.
|
and
290 Fluid Mechanics
( )
3
3
2cos 2sin .
2
R
U a
R
|
| |
VV = + q e e
Substitution into Eq. (9.40) in component form results in
cp
cR
= 3
U
a
R
3
cos|,
(9.41)
1
R
cp
c|
=
3
2
U
a
R
3
sin|
and integration yields
2
3
cos .
2
U a
p
R
= (9.42)
To calculate the drag on the sphere, we note that because of symmetry, the
drag force F
D
on the sphere acts in the z-direction only. Consider the forces acting
on the differential strip shown in Fig. 9.6: S is a tangential shear stress and p is the
(normal) pressure. Hence
dF
D
= 2tasin| ( )ad| S sin| pcos| | |
R=a
,
(9.43)
where p is given by Eq. (9.42) and S is found from
S =
cq
|
cR
|
\
|
.
|
R=a
= 3
A
R
4
B
R
2
|
\
|
.
R=a
sin| =
3
2
U
a
sin|. (9.44)
Thus
dF
D
= 3U
ta sin
2
| + cos
2
|
( )
sin| d| = 3U
at sin| d|
and the expression for the drag force, known as Stokes law, is obtained as
F
D
=
o
t
}
dF
D
= 6taU
.
(9.45)
Figure 9.6 Forces on the sphere.
8. Similitude and Order of Magnitude 291
The case just solved has been that of a fluid flowing steadily around a
stationary sphere. The transcription of the results to the case where the bulk of the
fluid is stationary while the sphere moves with a constant velocity (U
) is
obtained by a simple transformation of coordinates.
The drag force given by Eq. (9.45) is just the hydrodynamic contribution. In
the presence of body forces, e.g., gravity, and when the density of the sphere, ', is
different from the fluid density, , there is an additional hydrostatic force
F
B
=
4
3
ta
3
g ' ( )
(9.46)
caused by buoyancy. The net force which now acts on the sphere is the sum
F
N
= F
B
+ F
D
+
4
3
ta
3
g ' ( )+ 6taU
.
(9.47)
We now consider a small sphere dropped into a fluid at rest. The sum of the
forces acting on the sphere is at first different from zero. Newton's second law of
motion requires that the sphere accelerate. The term F
B
in Eq. (9.47) remains the
same, while F
D
, which opposes F
B
, increases in magnitude with the increase in the
velocity. The net force F
N
thus eventually becomes zero. The sphere is then said to
have reached its terminal velocity, which becomes
( )
3 2
4 2 ( )
3 6 9
t
a g a g
U
a
t
t
' '
= =
(9.48)
or
=
2
9
a
2
g
U
t
' ( ).
(9.49)
The results derived above are verified experimentally for
Re
a
=
U
t
a
< 0. 5.
(9.50)
Equation (9.49) may be used to obtain the viscosity of a fluid from experimental
measurements of the terminal velocity of a sphere dropped into it.
Example 9.2
a. The viscosity of printing ink is approximately = 3.0 kg/m
.
s. The exact
values of the viscosity must be measured, and it is suggested to drop glass
balls into the ink and measure their terminal velocities. The densities of
printing ink and glass are 1,000 and 2,000 kg/m
3
, respectively. Find the size
of the balls to be used.
b. Balls of radius a = 1.0 cm were ordered, and in a certain experiment a termi-
nal velocity of 12 cm/s was measured. Find the viscosity of the ink in that
experiment.
292 Fluid Mechanics
Solution
a. The equations obtained for Stokes creeping flow are valid for, Eq. (9.50),
Re
a
=
U
t
a
< 0. 5 ,
i.e., the velocity should not exceed
U
t
= 0. 5
=
0. 5 3
1, 000
=1. 5 10
3
m / s.
On the other hand, from Eq. (9.48),
U
t
=
2
9
a
2
g
' ( ) =
2
9
a
2
9.81
3
1,000= 726.67a
2
.
Thus,
726.67a
3
=1.510
3
,
and the maximal sphere radius is found as
a
3
=2.0610
6
, a =1.27310
2
m=1.27 cm.
b. From Eq. (9.49),
=
2
9
a
2
g
U
t
' ( ) =
2
9
10
4
9.81
0.12
1, 000 = 1.817kg/m s.
Example 9.3
Small dust particles come out of a chimney of a cement factory. The particles
are approximately spherical, with radii in the range of 10
-5
- 10
-7
m. The density of
the particles is 1,000 - 2,000 kg/m
3
. The viscosity of air is =1510
-6
kg/m
.
s and
its density is 1.2 kg/m
3
. Find how long it will take the particles to settle on the
ground.
Solution
Equation (9.48) states
U
t
=
2
9
a
2
g
'
~
2
9
a
2
9.81
1510
6
=145,333a
2
.
For a = 10
-5
m with = 2,000 kg/m
3
U
t
= 0.029 m/s = 29 mm/s .
8. Similitude and Order of Magnitude 293
For a = 10
-6
m with =1,000 kg/m
3
U
t
=1. 4510
3
m/ s = 0.145 mm / s.
For a = 10
-7
m with =1,000 kg/m
3
U
t
=1. 4510
6
m/ s =1. 4510
3
mm / s.
For a chimney 30 m high, the a = 10
-5
m particles will settle in
30
0.029
= 1034s = 35min;
the a = 10
-6
m particles will settle in
30
0.14510
3
= 2 10
5
s = 58h;
the a =10
-7
m particles will settle in
30
1. 4510
6
= 2 10
7
s = 240 days.
In reality the a < 10
-6
m particles will not settle at all because of gravity, as
winds and thermal convection air movements are quite faster than their gravity
settling speeds.
Stokes Flow around a Cylinder
Figure 9.7 shows a stationary cylinder set in a two-dimensional parallel flow.
The radius of the cylinder is a and its center is chosen to coincide with the origin
of the coordinate system. This flow seems to be the two-dimensional analog of the
flow around a sphere, and we want to find out whether indeed this is the case.
Figure 9.7 Two-dimensional creeping flow around a cylinder.
294 Fluid Mechanics
The convenient coordinate system to describe this flow is the cylindrical one.
The relations between velocity and stream function are given by Eq. (5.58):
q
r
=
1
r
c
cu
, q
u
=
c
cr
.
(5.58)
The differential equation satisfied by this stream function is Eq. (9.26), which
in cylindrical coordinates becomes
V
4
=
1
r
c
cr
r
c
cr
|
\
|
.
| +
1
r
2
c
2
cu
2
(
(
(
2
= 0.
(9.51)
The boundary conditions on the cylinder are
q
r
=
1
r
c
cu
=0, q
u
=
c
cr
= 0 at r = a .
(9.52)
The undisturbed parallel flow is
q = e
r
q
r
+ e
u
q
u
=U
e
r
cosu e
u
sinu | |
and the other two boundary conditions are adjusted such that for very large r
q iU
.
Hence
c
cu
=U
r cosu,
(9.53)
c
cr
=U
sinu at larger.
In a way similar to the treatment of the flow around the sphere, Eq. (9.53)
may be integrated to
=U
r sinu ,
and, again, one is tempted to try
= f r ( ) sinu for any r.
Substitution in Eq. (9.51) yields
1
r
d
dr
r
d
dr
|
\
|
.
|
1
r
2
(
(
(
2
f = 0,
(9.54)
which is analogous to Eq. (9.35). Application of standard methods of ordinary
differential equations yields
8. Similitude and Order of Magnitude 295
f r ( ) = Ar
3
+ Br lnr
1
2
( )
+Cr + Dr
and
3
1
ln sin
2
Ar Br r Cr Dr u
( | |
= + + +
|
(
\ .
(9.55)
One might try to proceed here along the way which proved successful for the
flow around the sphere, i.e., by applying the boundary conditions Eqs. (9.53) at
r to Eq. (9.55). This would result in A = 0, B = 0, C = U
= 0.1 m/s, find the velocities when the boundary conditions "at infinity" are
satisfied:
a. At a distance of 1 m from the cylinder's center, i.e., at r = R = 100.
b. At a distance of 2 m, i.e., at r = R = 200.
Solution
The dimensionless radius, r, is measured here in units of a (see Example 9.4),
which happens to be 1 cm. Accordingly, in part a, R = 100, and in part b, R = 200.
a. With reference to Example 9.4:
R = 100, G(R) = 36,057.3, H(R) = 36,051.7,
q
r
= 0. 5r
2
+10, 001 lnr
1
2
( )
+1 +
10, 000
2r
2
(
(
cosu
G(R)
,
q
u
= 1. 5r
2
+10, 001 lnr +
1
2
( )
+ 1
10, 000
r
2
(
(
sinu
G(R)
,
b. R = 200, G(R) = 171,939, H(R) = 171,933,
q
r
= 0. 5r
2
+ 40, 001 lnr
1
2
( )
+1+
40, 000
2r
2
(
(
cosu
G(R)
,
q
u
= 1. 5r
2
+ 40, 001 lnr +
1
2
( )
+1
40, 000
r
2
(
(
sinu
G(R)
.
The results of part a may be put as
q
r
= F
r,100
(r) cos u ,
q
u
= F
u,100
(r) cos u .
while those of part b may be put as
q
r
= F
r,200
(r) cos u ,
298 Fluid Mechanics
q
u
= F
u,200
(r) sin u ,
with F
r
and F
u
being the expressions in the square brackets divided by G(R). Some
values of F
r
and F
u
are given in Table 9.1.
r 1 2 4 8 16 32 64 100
F
r,100
0 0.088 0.254 0.439 0.627 0.809 0.908 1.000
F
u,100
0 0.296 0.514 0.711 0.897 1.057 1.122 1.000
F
r,200
0 0.074 0.213 0.369 0.528 0.687 0.839 0.926
F
u,200
0 0.248 0.431 0.598 0.759 0.914 1.048 1.100
Table 9.1. Velocity dependence on r in Stokes cylinder flow.
It is instructive to note that although R = 100 and R = 200 both seem to be
very far from the cylinder, the differences obtained between the two flows are
quite significant. Thus one is advised to choose an R, i.e., to satisfy the boundary
conditions, as close to the real physical situation as possible.
References
J. Happel and H. Brenner, "Low Reynolds Number Hydrodynamics with Special
Application to Particulate Media," Prentice-Hall, Englewood Cliffs, NJ, 1965.
W.F. Hughes, "An Introduction to Viscous Flow," Hemisphere, Washington, DC,
1979, Chapter 3.
H. Lamb, "Hydrodynamics," Dover, New York, 1945.
D. Pincus and B. Sternlicht, "Theory of Hydrodynamic Lubrication," McGraw-
Hill, New York, 1961.
Problems
9.1 A spherical balloon filled with hydrogen flies away off a child's hands. The
radius of the balloon is 0.2 m, the density of the air is 1.18 kg/m
3
and its
viscosity is 2 10
-5
kg/m
.
s. Find the terminal velocity of the balloon. Find
the Reynolds number at this terminal velocity, and decide whether this is a
creeping flow.
8. Similitude and Order of Magnitude 299
9.2 A copper ball has a radius of 5 mm and the density of 8,700 kg/m
3
. The ball
is dropped into a tank full of a fluid whose density is 833 kg/m
3
. The ball
descends at a constant velocity of 10 cm/s. Find the value of the Reynolds
number at this terminal velocity. Find the viscosity of the fluid.
9.3 Crop dusting is done from an airplane flying at the altitude of 5 m. The solid
dust particles of insecticide are approximately spherical with a diameter of d
= 10
-2
cm and a density of 2,000 kg/m
3
. Estimate the time required for the
dust particles to fall to the ground.
The nearest population center is located 1 km from the dusted field. At what
wind velocity, directed from the field toward the town, must the dusting
operation be stopped?
9.4 Dust particles have approximately the density of water. Find the diameter of
the dust particles that have a terminal velocity in air of
a. 1 m/s, b. 0.1 m/s, c. 0.01 m/s, d. 0.001 m/s.
Calculate the approximate Reynolds numbers for each case. Does the simple
Stokes formula apply in each of these cases?
9.5 Apply the boundary conditions Eq. (9.53) at r = R = 300 and then the
boundary conditions Eq. (9.52) and complete the solution for the Stokes
two-dimensional creeping flow around a cylinder. Obtain an expression for
the pressure field and for the drag force on the cylinder.
9.6 Use the results of Problem 9.5 and compute the drag force on a cylinder
when the boundary condition at large r, Eq. (9.53), is applied at r = b = 4a ,
at b = 5a, at b = 7a, at b = 10 a. Draw schematically some streamlines for
the four cases. Which do you consider a better approximation?
9.7 The linear velocity of the slide bearing in Example 9.1 is increased to
100 m/s. As a result the temperature of the nitrogen is raised to 350 K
( = 19.910
-6
kg/m
.
s). Find h
1
, h
2
and the drag force.
9.8 A wide slide bearing is constructed of a steel shaft with a diameter of 0.1 m
and a bronze sleeve with a diameter of 0.1008 m. The gap between shaft and
sleeve is filled with oil, = 880 kg/m
3
, = 2 10
-2
kg/m
.
s. The shaft turns
at 3,000 rpm. To avoid contact between the surfaces of the shaft and the
sleeve, they must be separated by at least 0.1 mm. Find the transverse load,
in N/m (per meter width), that may be applied to the shaft.
300 Fluid Mechanics
9.9 A plastic cup 0.12 m long and 0.08 m in diameter is manufactured by inject-
ing molten plastic into a flat gap inside a mold. The mold looks like an
annulus between two cylinders and may be approximated as a 0.0005-m-
wide gap between two parallel plates. The molten plastic has the apparent
viscosity = 80
kg/m
.
s.
a. Find the pressure necessary to have the liquid plastic front advance in
the mold at a rate of at least 0.1 m/s.
b. What is the flow rate of the molten plastic at these conditions?
9.10 Glycerin flows at a creeping flow
in a rectangular channel having the
sides 0.05 m by 0.02 m. The
maximum velocity in the channel is
0.2 m/s. Find the pressure gradient
along the channel.
9.11 A simple hydraulic transmission is
shown in Fig. P9.11. The trans-
mission fluid has a viscosity of
= 1.0 kg/m
.
s. The flow in the gap
is a creeping flow and is assumed
to be approximated by a shear flow
between two flat plates. The drive disk turns at 3,000 rpm.
a. Find the moment transmitted to the follower disk as a function of the
number of turns per minute of the follower disk.
b. Find the efficiency of the transmission at the various speeds.
9.12 A two-dimensional rectangular cavity
is shown in Fig. P9.12. Three sides of
the cavity are enclosed by stationary
walls, and a plane slides along the
fourth side. Write, in terms of the
stream function, the differential equa-
tion and the boundary conditions that
govern this flow.
Figure P9.11
Figure P9.12
8. Similitude and Order of Magnitude 301
9.13 A shock absorber is shown in Fig.
P9.13. It is common engineering
practice to designate shock ab-
sorbers by a constant C in the
form F = C V , where F is the
force the absorber transmits, and
V is the velocity of the piston.
Find C for this absorber.
9.14 The hydraulic oscillation damper
shown in Fig. P9.14 consists of a
bent tube section of 0.06 m in
diameter filled with oil of viscos-
ity 0.01 kg/m
.
s, in which a tube of
0.05 m in diameter moves concen-
trically at an arc of 0.4 m radius.
The flow in the annular gap is
creeping. Find the braking mo-
ment as a function of the angular
velocity.
9.15 Creeping flow is associated with small Reynolds numbers. Is a fully devel-
oped flow in a pipe different for creeping flows and for flows where the
Reynolds number is not very small?
9.16 A pipe ends with a nozzle shaped
like a cone, Fig. P9.16 . The flow in
the cone is creeping and is approx-
imated by a fully developed pipe
flow, i.e., at each cross section the
flow is like in a pipe of that cross
section. The cone is 0.1 m long, and
the pipe feeds fluid to the cone with
a maximum velocity of 1.0 m/s. The fluid viscosity is 0.1 kg/m.s. Find the
pressure drop in the cone.
9.17 The flow in Problem 9.16 is reversed, i.e., the cone feeds fluid to the pipe.
The velocity field is exactly reversed. Find the pressure drop in the cone.
9.18 Three spheres are connected by a string, as shown in Fig. P9.18. The three
are dragged through a fluid at a constant velocity, and the flow is creeping.
It is suggested to estimate the total drag force on the three spheres as three
times the drag force on a single sphere in Stokes flow. Find the distance
between any two spheres, expressed in sphere diameters, necessary to make
Figure P9.13
Figure P9.14
0.02 m
0.03 m
Figure P9.16
302 Fluid Mechanics
the approximation of the order of 1% error.
L
Figure P9.18
9.19 Can the Rayleigh problem be solved for a creeping flow? Are the results the
same as those obtained for non-creeping flows?
9.20 It can be shown that in a creeping flow the velocity field is such that the
total viscous dissipation over the whole field is minimized. Show that when
the fluid contains solid particles, which are moving freely with the flow, the
total dissipation increases. In other words, the apparent viscosity of the
seeded fluid is higher than that of the pure fluid.
303
10. HIGH REYNOLDS NUMBER
FLOWS REGIONS FAR FROM
SOLID BOUNDARIES
Negligible Shear The Euler Equations
The interpretation of the physics of flows with vanishing viscosity has already
been attempted in Chapter 5, where a simple formal substitution of v 0 in the
NavierStokes equations has led to the Euler equation (5.54). The more elaborate
formulation in Chapter 8 has transferred the role of a perturbation parameter from
the kinematic viscosity, v, to the Reynolds number, Re = UL / v, with the condition
v 0 being expressed by Re . We have also been forewarned that for
Re the second order NavierStokes equations yield the Euler equations,
which are of the first order and therefore cannot satisfy both boundary conditions
q
n
= 0
q
t
= 0
`
)
at solid boundaries.
(5.46)
Still, a legitimate question may come to mind at this point: Have the Euler
equations no physical meaning at all? Are the solutions of these equations not even
approximations to real flows? We know that near rigid boundaries, where the
boundary conditions dominate the flow, the Euler equations yield wrong solutions;
but are there no other regions, far away from the boundaries, where they correctly
predict the physics?
We do not set out on this path of enquiry just as adventurers. We follow the
footsteps of such giants as Euler and Lagrange, who also wondered where it led,
and Prandtl, who walked it and found the answer.
304 Fluid Mechanics
Before embarking on the analysis of flows for which v 0, we consider an
analogous case of such perturbations, which is simpler because the equations in it
are linear. In the following example we ask under what conditions does the solu-
tion of the second order differential equation which describes the deflection of a
membrane approximate the deflection of a thin plate, which is governed by a
fourth order differential equation.
Example 10.1
The vertical deflection of a two-dimensional thin plate clamped at both ends
and subject to a uniform continuous load, Fig. 10.1, is given by the dimensionless
ordinary differential equation
B
d
4
y
dx
4
M
d
2
y
dx
2
= f ,
with the boundary conditions
y = 0 at x = 1,
dy
dx
= 0 at x = 1.
The dimensionless parameters in the equation are B the beam parameter and M the
membrane parameter.
*
For very thick plates B >> M and for membranes B << M:
a. Find the exact deflection of the plate.
b. Consider plates, which become thinner and thinner, i.e., B 0. This causes
the differential equation to reduce to the second order and then it cannot
satisfy both boundary conditions. Find the approximate deflection of the
plate: apply the regular perturbation B 0 and solve the perturbed equation.
Consistent with the perturbation, which boundary condition is ignored?
Where does the approximation hold?
*
A thick beam bends according to
B
d
4
y
dx
4
= f
with the same boundary conditions, while a membrane deflects according to
M
d
2
y
dx
2
= f
with the boundary conditions y = 0 at x = 1.
10. High Re Flows Regions far from Solid Boundaries 305
Solution
a. A particular solution of the full equation is
y
p
=
f
2M
x
2
,
and the homogeneous solution is
y
h
= C
1
+ C
2
x + C
3
sinhx + C
4
coshx ,
2
= M/B.
x =- 1
x = 0
y=0
exact
approximate
Figure 10.1 Deflection of a two-dimensional clamped plate.
Addition of the two solutions and the use of the boundary conditions yield
y =
f
M
1
2
x
2
1
( )
+
cosh coshx
sinh
(
,
which for large may be put as
y =
f
M
1
2
x
2
1 ( )+
1
1
e
|x|
e
|
\
|
.
|
(
(
and
dy
dx
=
f
M
x
e
|x|
e
(
(
(
.
b.
With B 0, the equation undergoes a regular perturbation into the mem-
brane
equation,
306 Fluid Mechanics
M
d
2
y
dx
2
= f ,
with the solution again constructed of a particular part and a homogeneous
one,
y =
f
2M
x
2
1
( )
,
dy
dx
=
f
M
x .
The kept boundary condition is that for the membrane,
y = 0 at x = 1.
Because B 0 implies , the perturbed equation yields fairly good
approximations for y; however, while the error in dy/dx is still fairly small far
from the boundary, it is very large (about 100 %) near the boundaries.
The limit B 0 is consistent with a very thin plate and in the limit with a
membrane. Now a membrane cannot transmit bending moments. Even if the
membrane had been forced to satisfy
dy/dx = 0 at x = 1,
this angular deflection could not be passed on to points not at the boundary
because it takes moments to transmit this information. Thus ignoring
dy / dx =0 is consistent with the perturbation B 0.
U
F
C
y
x
L
Figure 10.2 High-Reynolds-number flow.
Point F far from boundary. Point C close to boundary.
Fortified by Example 10.1 we return now to the implication of 0 v .
Consider a real fluid, which satisfies both boundary conditions, Eq. (5.46), at a
rigid boundary but which has a very small viscosity. The boundary condition q
t
= 0
at the wall states that the fluid layer touching the wall does not slip with respect to
it. But if the second fluid layer, adjacent to the first, may slide freely with respect
to the first, a third layer further away from the wall would have no way of learning
that the first layer did not slip. In other words, the transmission of the no-slip
boundary condition to points not touching the boundary is done through shear
10. High Re Flows Regions far from Solid Boundaries 307
stresses, and for sufficiently small v this transmission decays very quickly. Thus,
regions far from rigid boundaries in small-v large-Reynolds-number flows are not
affected by the no-slip boundary condition. In these regions, e.g., at point F in Fig.
10.2, the Euler equations with the boundary condition q
n
= 0 only yield good
approximations. At points very close to the boundaries, point C, Fig. 10.2, both
boundary conditions prevail and the Euler equations are no approximations at all.
The layer of fluid very close to the rigid boundaries is denoted the Boundary
Layer. This region is considered in the next chapter.
The region outside the boundary layer, where the Euler equations serve as
good approximations, is sometimes referred to as the region of negligible shear, in
reminiscence of the v 0 formulation. The Euler equations are solved there with
the remaining boundary condition q
n
= 0 satisfied at the rigid boundaries, Fig. 10.3.
The geographical fact that thin boundary layers do lie near the rigid boundaries is
ignored and does not affect the solutions far from the boundaries. One remembers,
however, that inside the boundary layers these negligible shear solutions are not
valid at all.
x
= U q i
= U q i
q = 0
n
Figure 10.3 Negligible shear flow over a flat plate.
The Euler equation, (5.54), is now written in detail,
O
c q
ct
+
1
2
Vq
2
q V q = EuVp +
1
Fr
2
g,
or dimensionally,
cq
ct
+
1
2
Vq
2
q V q =
1
Vp + g
(10.1)
with the boundary condition
q
n
= 0 at rigid boundaries. (10.2)
The remainder of this chapter deals with solutions of the Euler equation.
308 Fluid Mechanics
Example 10.2
A two dimensional flow is given at x << 0 by
q = iU.
At x = 0 the flow encounters a flat plate parallel to it as shown in Fig. 10.3. Assume
this a flow with negligible shear. How is the flow modified at x > 0?
Solution
Euler equation (10.1) is satisfied by
q = iU.
This velocity also satisfies q
n
= 0 on the flat plate. Therefore q = iU is the solution
everywhere and the negligible shear flow is not modified at all. Note, however,
that q = iU is a good approximation to the physical flow only outside the plate
boundary layer and certainly not for x > 0, y = 0.
Irrotational Motion
Euler's equation (10.1) for incompressible flows may be written as
*
c q
ct
q V q = V
p
+
1
2
q
2
+ gh
(
(
(
.
(10.3)
Taking the curl of this equation results in
c
c t
V q ( ) V q V q ( )= 0.
(10.4)
Define a vorticity vector = Vq and Eq. (10.4) becomes
c
ct
V q ( ) = 0.
(10.5)
One way to satisfy Eq. (10.5) is by = 0. Flows for which = 0 are called
irrotational flows.
The vorticity vector written in cartesian coordinates is
=
cw
cy
cv
cz
|
\
|
.
|
|
i +
cu
c z
cw
cx
|
\
|
.
| j +
cv
cx
c u
cy
|
\
|
.
|
|
k ;
(10.6)
*
Note that for compressible barotropic flows this form still holds; however, the term
( ) / p V
on
the right hand side changes to / dp V
}
10. High Re Flows Regions far from Solid Boundaries 309
in cylindrical coordinates it is
=
1
r
c q
z
cu
cq
u
cz
(
( e
r
+
cq
r
c z
cq
z
cr
(
( e
u
+
1
r
c rq
u
( )
cr
1
r
cq
r
cu
(
(
(
e
z
;
(10.7)
and in spherical coordinates
=
1
r sinu
c q
|
sinu
( )
cu
c q
u
c|
(
(
(
(
e
r
+
1
r sinu
c q
r
c |
c rq
|
sin u
( )
cr
(
(
(
(
e
u
+
1
r
c rq
u
( )
cr
c q
r
cu
(
(
(
e
|
.
(10.8)
The vorticity vector equals twice the angular velocity defined in Chapter 5. Hence,
the term irrotational flow, which implies flows with no angular velocity.
Returning to Eq. (10.6) we note that for a flow to be irrotational, i.e., to
satisfy Vq = 0, it must also satisfy the following conditions simultaneously:
c w
cy
=
c v
c z
,
cu
cz
=
cw
cx
,
cv
cx
=
cu
cy
,
(10.9)
and for two-dimensional flows these conditions simplify to
c v
c x
=
c u
cy
.
(10.10)
a. rotational flow b. irrotational flow
direction of flow direction of flow
Figure 10.4 Rotational vs. irrotational flow.
To gain some physical insight into the character of irrotational flows, we note
that the motion of any solid object can always be described in terms of the motion
of its center of gravity, or of some other fixed point on the object, and of its
angular motion around this point. We place a small object in the flow and expect
its motion to approximate that of the fluid if has displaced. In irrotational flow
such a small object has no angular velocity. Figure 10.4 shows a small arrow-
310 Fluid Mechanics
shaped object in rotational and in irrotational motions. In irrotational motion,
though the path of the object may be curved, there is no angular velocity and the
arrow retains its direction in space.
We now return to Eq. (10.3) and note that for flows that are steady and
irrotational the left hand side of the equation vanishes, resulting in
V
p
+
1
2
q
2
+ gh
(
(
(
= 0,
(10.11)
which implies
p
+
1
2
q
2
+ gh = const.
(10.12)
Equation (10.12) holds for steady irrotational flows in the whole domain. It is the
strong Bernoulli equation already discussed in Chapter 7.
Equation (10.5) states that all irrotational flows satisfy the Euler equation;
thus solutions based on irrotationality need not be checked again on that point.
These properties make it desirable to establish conditions under which real flows
are irrotational.
In irrotational flows = V q = 0.
Using the Stokes theorem, we define a new
concept related to fluid rotation,
S
c
}
V q ( ) ndS = q ds
C
}
= I
c
,
(10.13)
where Ic
is the circulation along the closed
curve c, S
c
is a surface supported by this
curve and n is the outer normal to this
surface, Fig. 10.5.
In what follows we prove that when
Euler's equation describes correctly the flow,
the fluid carries with it its circulation, which
is preserved. This is known as the Kelvin
theorem, which states:
In a barotropic flow that satisfies Eulers equation the circulation does not
change.
A corollary of this theorem is that a flow, which is irrotational at one
moment, is always irrotational. It starts with I
c
= 0 on any c and therefore must
remain with zero circulation.
Proof:
S c
C
d s
n
n
Figure 10.5 Curve C support-
ing surface S
c
.
10. High Re Flows Regions far from Solid Boundaries 311
Let AB be a curve consisting of a string of fluid elements that move as a part
of the motion of the whole fluid, Fig. 10.6.
surface
curve at t
curve at t + At
A
A'
B'
B
q .
At
A
q .
At
B
Figure 10.6 Surface A'B'BA generated by the curve AB moving with the fluid.
Consider
D
Dt
q d s
AB
}
,
where D/Dt denotes the rate of change while following the same fluid elements.
This concept of the material derivative has already been used in deriving the
Reynolds transport theorem in Chapter 4. Referring to Fig. 10.6,
D
Dt
AB
}
q ds = lim
At0
1
At
' A ' B
}
q t + At ( ) ds
AB
}
q t ( ) ds
{ }
= lim
At0
1
At
q
' A ' B
}
t + At
( )
ds q
' A ' B
}
t
( )
ds + q t
( )
d s
' A ' B
}
q
AB
}
t
( )
ds
{ }
,
(10.14)
lim
At 0
1
At
q(t + At) q(t ) | | ds
' A ' B
}
=
cq
ct
d s
AB
}
,
(10.15)
' A ' B
}
q d s
AB
}
q ds =
' A ' B BA ' A
}
q d s
' B B
}
q ds
A ' A
}
q ds,
(10.16)
lim
At 0
1
At
A ' A
}
q ds = lim
At0
1
At
Atq
A
2
| |
= q
A
2
.
(10.17)
312 Fluid Mechanics
Similarly
lim
At 0
1
At
B ' B
}
q ds = q
B
2
.
(10.18)
Now, by the Stokes theorem
q ds = V q ( )
S
' A ' B BA
}
n dS = n dS
} }
,
(10.19)
where n is the unit normal to the surface, Fig. 10.6.
For At0
ndS
= At and q ds, with ds taken along AB, directed from A to B.
Thus
q d s = At
AB
}
q ds = At
AB
}
q ds
}
(10.20)
and
lim
At 0
1
At
q ds =
AB
}
q d s
}
=
AB
}
(q V q) d s.
(10.21)
Collecting all terms,
D
Dt
AB
}
q d s =
AB
}
c q
ct
q V q
(
( ds + q
B
2
q
A
2
.
(10.22)
Now let AB be a closed curve, i.e., point B coincides with point A, and let this
curve be denoted c. For this case
q ds = I
c
c
}
is the circulation of q along c, and
Eq. (10.22) becomes
DI
c
Dt
=
cq
ct
q V q
(
( ds =
S
c
}
V
c q
c t
q V q
(
( ndS ,
}
(10.23)
where the Stokes theorem has been used again. However, inspection of Eq. (10.3)
reveals that the integrand in the surface integral in Eq. (10.23) has the structure of
a curl of a gradient and therefore vanishes. Thus
DI
c
Dt
= 0,
(10.24)
which completes the proof.
To describe what happens in rotational flows which satisfy Euler's equation,
we need a new concept, that of the vortex lines. Let vortex lines be defined, as the
field lines of the vectorial field V q, i.e., as curves that are everywhere tangent to
=V q. We now consider a vortex tube, Fig. 10.7, i.e., a surface that consists of
vortex lines and encloses a tube within. Let three curves be chosen on this surface,
one which is not linked with the tube, C, and two which are linked with it, A and B.
Now the Helmholtz theorem states:
10. High Re Flows Regions far from Solid Boundaries 313
The circulation along a curve lying on the surface of a vortex tube and not
linked with the tube is zero and remains so. The circulation along all curves
lying on the surface of a vortex tube and linked with it is the same and
remains constant in time.
The proof of the Helmholtz theorem is as follows: For curve C:
I
c
=
C
}
q ds =
S
c
}
n dS = 0,
(10.25)
and this is so because lies on the tube surface while n is normal to it. Because of
Kelvin's theorem, however, I
c
= 0 always.
Auxiliary lines
C
B
A
A
Surface generated by curves A,
B and auxiliary lines does not
cut through vortex lines.
B
Surface generated by curve A
cuts through vortex lines.
B
Figure 10.7 A vortex tube.
Curve A is considered together with curve B. For clarity these curves are
drawn again in Fig. 10.7 without the tube. Now let one point be cut out of curve A
and one point be cut out of curve B. Also let two auxiliary lines be drawn as
shown, and compute
q
}
ds
over the whole figure, as indicated by the arrows.
This new figure comprises a curve not linked with the vortex tube, and therefore
q
}
ds = 0
. The integration on the auxiliary lines, which coincide in the limit,
cancels out, and thus
B
}
q ds
A
}
q ds = 0,
or for all times
I
B
= I
A
,
(10.26)
which completes the proof.
314 Fluid Mechanics
Thus while the fluid particles move according to the velocity field, those
lying on the surfaces of vortex tubes remain on those surfaces, and as they move,
they take with them the tubes. Denoting the integral over the tube cross section,
}
.
n dA, the intensity of the tube, this intensity persists in time. Furthermore, a
vortex tube cannot start or end in the fluid. It must either be linked into itself in a
toroidal shape or start and end at surfaces of discontinuity, such as rigid walls or
free surfaces.
Armed with Kelvin and Helmholtz theorems we now proceed to consider
irrotational solutions to Euler's equation, and for simplicity we assume these
barotropic flows to be also incompressible.
Potential Flow
When 0 V q , i.e., when the flow is irrotational, there always exists a func-
tion | such that
q = V| ,
(10.27)
where | is called the velocity potential and the flow is known as potential flow.
Irrotational flow is thus also potential flow. The equation of continuity for incom-
pressible flows is V
.
q = 0. Hence
V
2
| = 0.
(10.28)
The velocity potential is a harmonic function and can be obtained by the solution
of the Laplace equation (10.28). A proper set of boundary conditions must
guarantee no normal flow relative to rigid surfaces. It is noted that the solution of
Eq. (10.28) is really a direct integration of the equation of continuity, while the
Euler equation is implicitly satisfied by irrotationality. Euler's equation is still used
through the Bernoulli equation to evaluate the pressure. The momentum and
continuity equations thus become uncoupled. Since the continuity equation is
linear, as is Eq. (10.28), superposition of solutions is permissible, and elaborate
flows may be constructed by superposition of simpler ones. It is noted, however,
that Euler's equation is not linear, and neither is Bernoulli's equation. Therefore the
pressure in a flow obtained by superposition of two flows is not the sum of the
pressures of the two partial flows.
We now consider two-dimensional flows. Equation (10.27) states
u =
c|
cx
, v =
c|
cy
.
Another differentiation yields
10. High Re Flows Regions far from Solid Boundaries 315
c
2
|
c xcy
=
cu
cy
=
cv
cx
,
(10.29)
which can be identified as a condition for irrotationality, Eq. (10.10). Equation
(10.29) can be combined with Eq. (5.13) for the stream function
u =
c
cy
, v =
c
cx
to become
c
2
cx
2
+
c
2
cy
2
= 0.
(10.30)
Hence, in two-dimensional irrotational flow the stream function is also harmonic.
Furthermore,
u =
c|
cx
=
c
cy
, v =
c|
c y
=
c
cx
,
(10.31)
or in polar coordinates
q
r
=
c|
cr
=
1
r
c
c u
, q
u
=
1
r
c|
cu
=
c
cr
.
(10.32)
Equations (10.31) and (10.32) are known as the CauchyRiemann conditions.
The family of | = const lines and that of = const lines intersect orthogo-
nally. This is easily shown by noting that
V| = iu + jv , V = iv + ju, V| V = 0.
(10.33)
Example 10.3 Source Flow
Consider a steady two-dimensional axisymmetric flow from a line source.
Such a flow can be approximated by a radial flow emitted through the wall of a
long slender tube made of porous material. The tube lies along the z-axis and the
fluid flows radially in the x -y plane. Let Q be the fluid flow rate per unit length of
tube. This flowrate is called the intensity of the line source. Continuity yields the
fluid velocity,
q
r
=
Q
2tr
, q
u
= 0.
(10.34)
The stream function for source flow can be calculated from Eq. (10.32),
Q
2tr
=
1
r
c
cu
,
c
c r
= 0.
316 Fluid Mechanics
Integration of
c cu
yields
=
Q
2t
|
\
|
.
u + f r ( ),
and
c cu = 0
yields f = const, which is chosen to be zero. Thus
=
Q
2t
|
\
|
.
| u .
(10.35)
Constant lines are therefore radii, as shown in Fig. 5.6. Similarly, the flow
potential is evaluated from
c |
cr
=
Q
2tr
,
1
r
c|
cu
= 0,
which results in
| =
Q
2t
|
\
|
.
| lnr .
(10.36)
The origin itself is a singular point, but everywhere else the flow is well defined.
A source with a reversed flow is a sink. Accordingly, for the flow into a sink
= Cu , | =-Clnr ,
(10.37)
where C = Q / 2t.
Example 10.4 Potential Vortex
In potential vortex flow the radial velocity component q
r
is zero and only q
u
exists. The condition of irrotationality results in = 0. Hence, Eq. (10.7) requires
1
r
c(rq
u
)
c r
= 0
(10.38)
which upon integration yields q
u
= C/r + f(u), and because nothing depends on u,
f = const = 0 is chosen. Thus
q
u
=
C
r
.
(10.39)
Substitution into Eq. (10.32) results in expressions for and |,
= Clnr , | = Cu .
(10.40)
10. High Re Flows Regions far from Solid Boundaries 317
Comparison with the
source flow, Fig. 10.8, shows
that | and lines have simply
changed their roles. Indeed,
changing the roles of | and
lines in any potential flow
yields a new flow pattern.
The constant C in Eq.
(10.40) is related to the circu-
lation around the vortex, as can
be shown using Eq. (10.25):
I =
c
}
q ds =
C
r
|
\
|
.
| e
u
e
r
dr + e
u
r du
( )
=
0
2t
}
C du = 2tC.
}
Hence
C =
I
2t
(10.41)
and
=
I
2t
lnr , | =
I
2t
u .
(10.42)
There is no contradiction between I 0 and the irrotationality of this flow.
The circulation along any closed path not linked with the origin is still zero. The
singular point at the origin contributes the I value in Eq. (10.42), and the same I is
obtained along any contour linked with the origin.
Example 10.5 Parallel Flow
Consider uniform flow with a constant velocity, U, parallel to the x-axis.
Thus,
q
x
= u = U , q
y
= v = 0.
(10.43)
Equation (10.31) yields
= Uy , | =Ux.
(10.44)
Constant |
Constant
Figure 10.8 Vortex flow.
318 Fluid Mechanics
Since in potential flows any streamline may be interpreted as a rigid boundary, the
parallel flow may also be interpreted as a flow over a flat plate (say the x-axis
between
x = a and x = a
) at zero angle of incidence; see Example 10.2.
Example 10.6 Flat Plate Stagnation Flow
.
We now describe a flow perpendicular to
a flat plate, Fig. 10.9. Far away from the plate
the flow is parallel to the x-axis. Approaching
the plate, the flow gradually turns until it be-
comes parallel to it, as required by the bound-
ary conditions. This gradual turn reminds us of
a hyperbolic arc, which we may try as a
streamline. We are also guided toward this
choice by the fact that an expression quadratic
in x and y may be amenable to satisfy the
Laplace equation. Trying
= Cxy ,
(10.45)
we find that indeed V
2
= 0 and the resulting flow field is the two-dimensional
stagnation flow on a flat plate. The velocity components are obtained as
u =
c
cy
= Cx =
c|
c x
,
v =
c
cx
= Cy =
c|
cy
.
(10.46)
The flow potential may now be computed as
| =
x
}
c|
c x
dx + f (y) =
x
}
Cx dx + f (y) =
1
2
Cx
2
+ f (y).
(10.47)
Differentiating Eq. (10.47), we obtain
c |
cy
=
df
dy
= Cy.
Hence
f =
1
2
Cy
2
(10.48)
and
| =
1
2
C x
2
y
2
( )
,
(10.49)
which is also a hyperbola.
x
y
Figure 10.9 Stagnation flow
10. High Re Flows Regions far from Solid Boundaries 319
Example 10.7 Doublet Flow
Consider the superposition of two flows:
a
= Cu
a
, which is a source located
at point a, Fig. 10.10, and
b
= Cu
b
, which is a sink located at point b. The
combined flow has the stream function
=
a
+
b
= C u
a
u
b
( ).
(10.50)
From Fig. 10.10, u = u
b
u
a
, and thus lines of constant are also lines of constant
u, i.e., circles. We may, of course, use the expressions (10.36) and also obtain
| =
Q
2t
|
\
|
.
lnr
a
lnr
b
( ).
(10.51)
u
a
u
b
r
a
r
b
u
y
x
m
a
b
Figure 10.10 Doublet flow.
Let point a coincide with the origin of the coordinates, and let point b be
located at x = m, y = 0. Thus
lnr
a
=
1
2
ln x
2
+ y
2
( )
and
lnr
b
=
1
2
ln (x m)
2
+ y
2
| |
.
Hence
320 Fluid Mechanics
| =
Q
4t
ln
x
2
+ y
2
(x m)
2
+ y
2
.
(10.52)
We note that the argument showing the streamlines to be circles does not
depend on m. It is interesting therefore to check what happens as we let m 0.
Substitution of m = 0 in Eq. (10.52) leads of course to | = 0, as the source and the
sink counteract and cancel each other. However, we may imagine a process by
which as m 0, Q
. We want Q to increase at such a rate that the limit ob-
tained for | is well defined and finite. Thus,
| =
1
4t
lim
m0
Q ln(x
2
+ y
2
) ln (x m)
2
+ y
2
| | { },
(10.53)
and we note that if Q
as M/m does when m 0, the expression becomes
| =
1
4t
lim
m0
M
m
ln(x
2
+ y
2
) ln[(x m )
2
+ y
2
]
( )
(
=
M
4t
c
cx
ln x
2
+ y
2
( ) | |
(10.54)
or
| =
M
4t
c
cx
ln(x
2
+ y
2
) =
M
2t
x
x
2
+ y
2
=
M
2t
rcosu
r
2
.
Hence
| =
M
2t
cosu
r
.
(10.55)
This new flow potential, |, describes a flow pattern known as doublet flow. Its
streamlines are still circles, and it has one singular point at the doublet itself, i.e.,
in this example the origin.
Similarly, the coalescence of a source and a sink located on the y-axis leads to
a doublet with a potential |
a
/y. The results thus obtained could be expected, at
least in the sense that |
a
/x and |
a
/y yield new flow fields. If | satisfies the
Laplace equation, so do its derivatives provided the required higher partial
derivatives exist. We may also find that
2
|
a
/x y leads to a new field describing
two coalescing doublets.
Example 10.8 Flow over Streamlined Bodies
First we consider the superposition of a parallel flow and a two-dimensional
source flow, Fig. 10.11. The source is located at the origin and has the stream
function
S
=
Q
2t
u .
(10.56)
The parallel flow is taken to proceed from left to right and thus has the stream
function
10. High Re Flows Regions far from Solid Boundaries 321
p
= Uy = Ur sinu .
(10.57)
The combined flow has the stream function
=
Q
2t
u + Ur sinu
(10.58)
and therefore the velocity components
q
r
=
1
r
c
cu
=
Q
2tr
+ U cosu ,
(10.59)
q
u
=
c
cr
= U sinu .
= Q/2
A
U
y
x
= Q/2
= Q/2
Figure 10.11 Superposition of source flow and parallel flow.
Equation (10.58) may require pointwise numerical calculation to actually give
the shape of the streamlines, i.e., lines of constant . Still some qualitative
information can be obtained by general considerations. We recognize the need for
the stagnation point A. Inspecting Eq. (10.59), we see that for u = t and r =Q/2tU
both q
r
and q
u
vanish, and a stagnation point is obtained. We could have reached
the same conclusion by the argument that the velocity along the splitting
streamline, = Q/2, as it comes from the left towards the source could not proceed
into the source, and therefore the streamline must split. At the splitting point the
velocity vector has more than one direction, and its magnitude must therefore be
zero.
We also note that the streamline = Q/2 encircling the source cannot close
again. Far to the right the flow becomes parallel again, but the splitting streamline
does not close because if it did, the output of the source would have nowhere to
go.
Can we superimpose another flow on what we already have, so as to close the
splitting streamline? We already know that in flows with negligible shear any
streamline may represent a solid boundary, simply because it satisfies the boundary
322 Fluid Mechanics
condition of no normal velocity, i.e., the only boundary condition required in such
flows. If we could close the splitting streamline, it would represent a finite solid
two-dimensional body immersed in the flow. This can be done in a number of
ways. We could place a sink of the same intensity Q somewhere to the right of the
source. This sink would swallow up all the output of the source and thus allow the
splitting streamline to close. Another possibility is to place several sinks to the
right of the source, such that the sum of their intensities equals Q. We could also
replace the source by several sources and have a series of sources and sinks,
which, as long as the sum total of their intensities is zero, would yield a closed
splitting streamline. By varying the distributions of these sources and sinks, we
could obtain different shapes.
The whole idea may be carried into three dimensions, where the stream
function does not satisfy Laplace equation. The potential still does, and superpo-
sition is still permissible.
The shapes thus obtained, in both two and three dimensions, are known as
Rankine Ovals, Fig. 10.12.
y
x
U = 10 m/s
=
Q = 10 t
m
2
/ s
3 m
Q
Figure 10.12 Rankine oval.
Example 10.9 Flow over a Circular Cylinder
A particular case of a two-dimensional Rankine oval is that of parallel flow,
one source and one sink. We let the sink coalesce with the source in a limiting
process to become a doublet. The superposition of the parallel flow and the dou-
blet, Eqs. (10.43) and (10.55), yields the potential
| =Ur cosu +
M
2t
cosu
r
.
(10.60)
Thus,
10. High Re Flows Regions far from Solid Boundaries 323
q
r
=
c|
cr
= 1
M
2tr
2
U
(
(
U cosu =
1
r
c
cu
,
q
u
=
1
r
c|
cu
= 1+
M
2tr
2
U
(
( U sinu =
c
cr
,
(10.61)
from which
= 1
M
2tr
2
U
(
( Ur sin u.
(10.62)
Setting R
2
= M/2tU gives
= 1
R
r
|
\
|
.
2
(
(
Ur sin u
(10.63)
and
q
r
= 1
R
r
|
\
|
.
|
2
(
(
(
U cosu,
(10.64)
q
u
= 1+
R
r
|
\
|
.
|
2
(
(
(
U sinu.
This is a flow around a circular cylinder. Indeed we see that for u = 0 and
u = t, = 0. Also, for r = R, = 0, and the = 0 line is as shown in Fig. 10.13a.
a. Without circulation b. With circulation
Figure 10.13 Flow around a cylinder.
Far from the cylinder , i.e., for large r R , r R
2
>>1 r , and
324 Fluid Mechanics
= lim
r
R
1
R
r
|
\
|
.
|
2
(
(
(
Ur sinu = Ur sinu = Uy ,
(10.65)
which is parallel flow.
Inspection of Eq. (10.64) reveals that on the cylinder, i.e., at r = R, q
r
= 0 as
required by the boundary condition, and
q
u
= 2U sinu.
(10.66)
We may use this q
u
in Bernoulli's equation (10.12) to obtain the pressure distribu-
tion on the cylinder. Neglecting gravitational effects and performing a Bernoulli
balance between r
and r = R, we obtain
p
+
U
2
2
=
p
+
q
u
2
2
|
\
|
.
|
r =R
= const.
(10.67)
Hence
p
R
= p
+
U
2
2
1 4sin
2
u
( )
,
(10.68)
and we find that because sin
2
u is symmetric with respect to both the x-axis and the
y-axis, the flow exerts no drag force or lift force on the cylinder.
We may try to destroy this symmetry by the superposition of yet another flow,
the potential vortex flow. In this choice we are guided by the following
observation:
The streamlines in potential vortex flow are circles. One of these circles will
coincide with the r = R circle, which is considered as the rigid cylinder. Thus,
while the flow pattern in the field is modified by the addition of the potential
vortex, that particular circular streamline r = R remains circular. Hence we still
have a flow around a cylinder.
We note, however, that for the potential vortex rotating clockwise,
V
=
I
2t
lnr ,
(10.69)
and the field in general is affected nonsymmetrically as required. Thus the com-
bined stream function becomes
= +
I
2t
lnr + 1
R
r
|
\
|
.
2
(
(
Ur sinu
(10.70)
and
10. High Re Flows Regions far from Solid Boundaries 325
q
u
=
I
2tr
1+
R
r
|
\
|
.
|
2
(
(
(
U sinu ,
(10.71)
which on the cylinder reduces to
q
u
r =R
=
I
2tR
2Usinu .
(10.72)
Substitution of this q
u
in Eq. (10.67) now yields
p
R
= p
+
U
2
2
1 4sin
2
u
| |
2
I
2tR
|
\
|
.
|
2
IU
tR
sinu
. (10.73)
This pressure distribution is symmetrical with respect to the y-axis but not so with
respect to the x-axis. The nonsymmetrical term
IU tR ( )sinu
now yields a lift
force. From Fig. 10.12b
dL = pds sinu = pR sinu du =
IU
t
sin
2
u du .
Hence
L =
IU
t
0
2t
}
sin
2
u du = IU
(10.74)
and a lift force has been obtained.
One is tempted to look for another potential flow such that when superim-
posed on the one already considered will destroy the symmetry with respect to the
y-axis and thus produce drag. No such flow exists. The only mechanism, which
can produce entropy in incompressible flows, is viscous dissipation. Viscosity
effects are barred from potential flows. Drag force, i.e., a force in the direction of
the undisturbed flow far from the cylinder, is always associated with entropy
production; we therefore must not expect drag in any potential flow. To produce
drag we must admit viscosity back into the field, at least in some thin layers near
the boundaries of solid bodies, as will be done in the next chapter.
Example 10.10 Magnus Effect
A cylinder of radius R is dragged sidewise under water and is made to rotate
at the same time. It is believed that the flow around it is potential and that the
stream function and the circumferential velocity are given by Eqs. (10.70) and
(10.71), i.e.,
326 Fluid Mechanics
=
I
2t
lnr + 1
R
r
|
\
|
.
|
2
(
(
(
Ur sinu ,
q
u
=
I
2tr
1+
R
r
|
\
|
.
2
(
(
U sinu .
Find the lift force acting on the cylinder.
Solution
This example seems the same as Example 10.8. However, the density of
water is high and we, therefore, include the corresponding term in the Bernoulli
polynomial, Eq. (10.67), which becomes for this case
p
+
U
2
2
+ 0=
p
+
q
u
2
2
+ gy
|
\
|
.
|
r =R
= const.
Because sin
r R
y K u
=
= the pressure on the cylinder is [see Eq. (10.73)]
( )
2
2
2
1 4sin sin sin
2 2 2
R
U U
p p gR
R R
I I
u u u
t t
| |
= +
|
\ .
The integral of the first three terms in this expression is still zero, as in Exam-
ple 10.8. However, the last two terms yield
L =
IU
tR
+ gR
|
\
|
.
| R
0
2t
}
sin
2
u d u = IU + gtR
2
.
Note that the addition to the lift force in this case, i.e., gtR
2
, equals the buoyancy
force one would expect to find in a hydrostatic situation.
References
L.M. MilneThompson, "Theoretical Aerodynamics," 4th ed., Macmillan, London,
1966.
A.H. Shapiro, "Shape and Flow," Doubleday, New York, 1961.
10. High Re Flows Regions far from Solid Boundaries 327
Problems
10.1 Suppose that in Fig. 10.12 we have two line sources perpendicular to the
plane of the paper.
a. Find the stream function.
b. Find the potential function.
c. Find the velocity components.
d. Are the streamlines and equipotential lines orthogonal?
e. Sketch the streamlines and equipotential lines (surfaces).
10.2 A two-dimensional source of strength Q =100 t (m
3
/s)/m is located at A (2 ,
3). A sink of the same strength is located at B (2 , 3). All lengths are mea-
sured in meters.
Add a flow parallel to the x-axis and find a streamline that may represent a
body of maximum width of 1.5 m.
10.3 A flow is given in Fig. 10.12 consisting of a superposition of a sourcesink
and uniform flow.
a. Find the stagnation point
b. Sketch to scale the streamline that represents the rigid body.
c. Determine the pressure distribution along the surface of this body.
d. Is there a drag force?
10.4 Instead of the sink in Fig. 10.12 there are two sinks at x = 0 and x = 1.5. The
intensity of the sinks is Q = 5 t m
2
/ s each. Repeat Problem 10.3 for this
case.
10.5 In two-dimensional incompressible flow the velocity vector is given as
q = iu + jv ,
where
u = 3y m/s, v = 3x m/s.
a. What is the equation of the stream function?
b. Sketch a few streamlines and calculate their values.
10.6 In two-dimensional incompressible flow the velocity vector is
q = e
r
q
r
+ e
u
q
u
,
where
q
r
=
Q
2tr
, q
u
= 0.
328 Fluid Mechanics
Derive an expression for the streamlines.
10.7 A solid wall is added to the field in Problem 10.2 along the x-axis,
Fig. P10.7.
a. Find the velocity at point C (0,1) with and without the parallel flow.
b. Find the volumetric flux between the origin (0 , 0) and point C(0 , 1) with
and without the parallel flow.
Hint: Consider the x-axis to be an axis of symmetry, or a mirror.
A(-2,3) B(2,3)
C(0,1)
Figure P10.7 Source, sink and wall.
10.8 Consider a two-dimensional flow field with the potential | and the stream
function formulated in the complex form |(z) = | + i. For example
( ) z z x iy i | | = = + = + . Draw some streamlines and interpret the flow
(e.g., flow around a circle?)
Repeat for
|(z) = z
2
= x
2
y
2
+ i2xy = | + i
. Do | and satisfy the Laplace
equation?
10.9 Repeat Problem 10.8 for |(z) = z
3
, |(z) = z
4
.
10.10 Repeat Problem 10.2 for a three-dimensional source and sink of intensity
100t m
3
/s.
10.11 Repeat Problem 10.7 for a three-dimensional source and sink for the
following cases:
a. With parallel flow at C(0, 1, 0).
b. Without the parallel flow.
10. High Re Flows Regions far from Solid Boundaries 329
Figure P10.12 Flow around a rotating cylinder.
10.12 Figure P10.12 shows the cylinder of Example 10.10, with the flow around
it. The other curves and the arrows were drawn by someone who wanted to
show the rotationality (or irrotationality) of the field, the same as in
Fig. 10.4. The curves represent streamlines. Draw the arrows as they look in
other locations in the field.
10.13 Given a potential function in spherical coordinates
| = A / R,
show that it is harmonic. Find its velocity components. Sketch this flow.
Does this flow have a stream function? If yes, find it.
10.14 Figure P10.14 shows a two-dimensional flow inside a sharp bend. A stream
function is suggested:
= Ar
t
o
sin
u t
o
|
\
|
.
| .
a. Show that indeed this is a stream function and that it satisfies the
Laplace equation.
b. Show that it is the stream function of this flow, that its velocity satisfies
the boundary conditions.
c. Find the potential of the flow.
Figure P10.14 Two-dimensional flow in a bend.
330 Fluid Mechanics
10.15 a. A three-dimensional point source of intensity Q = 5 m
3
/s is located at
z = 1. Find the stream function of the resulting flow. Sketch the flow. Are
there stagnation points?
b. Superimpose a parallel flow of the speed w = 2 m/s in the z-direction.
Repeat part a.
10.16 a. A three-dimensional point source of intensity Q = 5 m
3
/s is located at the
origin. A sink of the same strength is located at z = 1. Find the stream
function of the resulting flow. Sketch the flow. Are there stagnation
points?
b. Superimpose a parallel flow of a speed w = 2 m/s in the z-direction.
Repeat part a.
10.17 A circular rod with rounded tips has a diameter of 0.05 m and a length of
1 m from tip to tip. The rod is fixed in a parallel flow of the speed of 3 m/s,
with the rod axis in the direction of the flow. Using one source and one sink
find the "best fit" to the rod and the flow resulting from it.
A "best fit" is defined here as a closed stream surface with stagnation points
at the tips of the rod and with a girdle that does not bulge out of the cylinder
prescribed by the rod yet touches this cylinder.
10.18 A two-dimensional source of intensity Q
1
= 100 m
3
/m
.
s is located at x = 1,
y = 1. A similar source is located at x = 1, y = 1. A two-dimensional sink of
intensity Q
3
= 200 m
3
/m
.
s is located at x = 0, y = 1. The field is two-dimen-
sional and contains the two points A (0.5; 0.5) and B(0.5; 0.5).
a. Sketch the flow.
b. Find the velocity vector at point A.
c. Find the volume flux between point A and point B.
10.19 The angle between two diverging walls is
36
o
, Fig. P10.19. A two-dimensional source of
strength Q = 2 m
3
/s
.
m is located at x = 1 m.
a. Find the velocity field between the walls.
b. Find the volumetric flux between the
walls through a plane x = 15 m.
c. Find the flux through a plane at x = 16 m.
d. Find the force acting on the part of one
wall between x = 15 m and x = 16 m.
10.20 Put a two-dimensional source at the origin such that the volume flux
through a sector of 36
o
is 2 m
3
/s
.
m.
a. Is this a possible solution of the Euler equation for the flow between two
1.0
y
x
Figure P10.19
10. High Re Flows Regions far from Solid Boundaries 331
diverging walls?
b. Find for this flow all quantities sought in Problem 10.19 .
10.21 The walls in Fig. P10.21 are two-dimensional.
A three-dimensional source of strength Q = 4
m
3
/s is located at x = 4 m, y = 1, z = 0.
a. Find the velocity at x = 8 m, y = 1, z = 0.
b. Find the velocity at x = y = z = 1.
10.22 Find the total force on the wall x = 0 and on the
wall y = 0, in Fig. P10.21, under the flow condi-
tions of Problem 10.21. Use analytic or
numerical integration.
10.23 A two-dimensional wing is
schematically shown in Fig. P10.23.
The length of the underside arc of the
wing is 1 m, and the length of the
convex arc is 1.2 m. The velocity of
the air streaming toward the wing is
50 m/s. Assuming the air velocities on
the wing to be proportional to the arcs
lengths, estimate the lift force gener-
ated by the wing.
10.24 A potential flow is obtained by the superposition of a two-dimensional sink
of strength Q = 2 m
3
/ s
.
m and of a potential vortex of circulation I = 3
m
2
/s, both located at the origin.
a. Find the flow velocity at r = 1 m, at r = 0.5 m.
b. Find the pressure difference between these points and a point at r = 20 m
.
10.25 The flow field around a sharp rod sticking in front of an airplane may be
obtained by the superposition of three flows: a parallel flow at 45 m/s in the
x-direction, a three-dimensional source (Q
a
= 10 m
3
/s) located at the origin,
and a source (Q
b
= 1 m
3
/s) located on the x-axis at x = 0.2 m.
Find the stagnation point and draw the streamline representing the rod.
10.26 A parallel flow is superimposed on a two-dimensional sourcesink pair
y
x
4.0
1.0
Figure P10.21
A
B
Figure P10.23
332 Fluid Mechanics
flow, resulting with a Rankine oval flow. The Rankine oval is denoted an
obstacle. Find how far from the obstacle the modification of the parallel
flow by the obstacle is less than 1%.
10.27 Use the results of Problem 10.26 to formulate the potential flow field in a
two-dimensional channel with an obstacle in it.
10.28 Repeat Problem 10.26 with a three-dimensional sourcesink pair.
10.29 A 10-m-high circular cylinder has the diameter of 1 m and rotates at
800 rpm. A wind blows across the cylinder with the velocity of 10 m/s. Find
the sideways lift force on the cylinder.
10.30 A three-dimensional source of strength 20 m
3
/ s is located on the x-axis at
x = 5 m. Another source of strength 10 m
3
/s is also located on the x-axis, at
x = 10 m. Draw the flow field at the x-y plane.
333
11. HIGH REYNOLDS NUMBER FLOWS
THE BOUNDARY LAYER
In this chapter we still consider flows in which Re
. In the previous
chapter we sought the flow field in regions, which were far away from rigid
boundaries. Now we want to complete the description and treat regions quite close
to rigid boundaries. The geometry of these boundaries is generally flat, e.g., a flat
plate, and the large Re = UL/v associated with the flow is based on the large
velocity U, which is approximately parallel to the rigid surface. The NavierStokes
equations for these flows may be approximated by the boundary layer equations,
Eqs. (8.38) - (8.42). We already know that in large Re flows the flow not near the
boundaries is approximated rather well by solutions of the Euler equation. The
deviations from this approximation become pronounced in a very thin layer close
to the boundary, in the boundary layer.
Given a boundary layer flow we start with the solution of the potential flow
equation (10.28), which is linear. Once the flow potential, |, is found, we derive u
and v and hence U by differentiation. We then use u and v in the Bernoulli equa-
tion (7.5) to obtain p. We now have p everywhere, including at y = 0, because
p/y = 0 there, as given by Eq. (8.41).
Inside the potential flow Bernoulli's equation holds:
p
+ gh +
1
2
U
2
= const.
Differentiation for cases where h is not important yields
dp
dx
= U
dU
dx
,
(11.1)
which makes Eq. (8.38) read
334 Fluid Mechanics
u
cu
cx
+ v
cu
cy
=U
dU
dx
+ v
c
2
u
cy
2
.
(11.2)
Here U is the potential flow velocity just outside the boundary layer. Note that
when h is important, Eqs. (8.37) and (8.38) should both include gdh/dx; Eq. (11.2),
however, remains unchanged.
Assuming the potential flow solved, we proceed to solve the boundary layer
equations. The solution is easier than that of the original NavierStokes equations
because there are two equations only, the pressure being known, and because these
equations are parabolic rather than elliptic. The boundary condition at y Re
1/2
is sometimes approximated at finite y = o, where o is defined as a boundary layer
thickness. We now illustrate this procedure by some examples.
Potential flow
velocity profile
Boundary layer
velocity profile
y
x
Figure 11.1 Flow over a flat plate.
Flow over a Flat Plate The Blasius Solution
Consider a flow in the x-direction parallel to a flat plate, as shown in Fig.
11.1. The potential flow for this example has already been solved, Eq. (10.43):
| = Ux , u = U = const.
(11.3)
Equation (11.3) yields UdU/dx = 0, and the remaining set of equations to be solved
is
u
cu
cx
+ v
cu
cy
= v
c
2
u
cy
2
,
(11.4)
cu
cx
+
cv
cy
= 0,
(11.5)
with the boundary conditions
11. High Re Flows The Boundary Layer 335
u = v = 0 at y = 0,
(11.6)
u = U at yRe
1
2
.
(11.7)
Physically y Re
1/2
also requires y
. At the plate u = 0 because of the
no-slip boundary condition of Eq. (11.6). Adjacent layers are also slowed down
because of shear stress. Thus the velocity profile approaches that of potential flow
only asymptotically, Fig. 11.1, and an exact definition of the boundary layer
thickness as that y for which there is no more modification of the potential flow,
i.e., as y for u = U, requires y
.
The continuity equation (11.5) can be eliminated from the set of Eqs. (11.4) -
(11.7) by the use of the stream function ,
u =
c
cy
, v =
c
cx
,
resulting in
c
cy
c
2
cxcy
c
cx
c
2
cy
2
= v
c
3
cy
3
.
(11.8)
c
cy
= U at y ,
(11.9)
c
cx
= 0,
c
cy
= 0 at y = 0.
(11.10)
To solve these equations, some additional insight is necessary. We review Chapter
8 and find the dimensionless form of the x-component of the full NavierStokes
equation, Eq. (8.12), with the similarity parameter Re. We have therefore the
dimensionless form of Eq. (11.4) as
u
cu
cx
+ v
cu
c y
=
1
Re
c
2
u
c y
2
.
The dimensionless continuity equation (8.11) becomes
cu / cx + cv/ cy = 0
(see Eq. (11.5)) and contributes no similarity parameter, and we find that the only
similarity parameter, as far as similitude is concerned, is the Reynolds number,
Re = LU /v
.
Now we read in Chapter 8 that the characteristic physical quantities used to
construct the similarity parameters can be chosen arbitrarily, provided they are well
defined and correspond to the same geometrical locations in both cases. Therefore,
to construct Re
1
= Re
2
for two boundary layer flows, we should choose U
1
and U
2
and v
1
and v
2
; but what are the characteristic lengths?
Inspection of the geometry reveals that the only well-defined length in this
flow is x itself, i.e., the distance along the plate measured from its forward tip.
336 Fluid Mechanics
Thus for two such flows to be similar,
Re
1
= Re
2
,
U
1
x
1
v
1
=
U
2
x
2
v
2
.
To emphasize the fact that this Reynolds number has x as its characteristic length,
we denote it Re
x
.
This interesting result indicates that the flow may be viewed as similar to
itself: We can cover a whole range of Re
x
by considering one x and several U
values, but also by considering one U and several x values. There seems to be
nothing special about the x value we have chosen. Could it be that the velocity
profiles are always similar, with U and v just scales of magnitude? If so, then there
must be a similarity solution. Armed with this hint, we now define a new single
independent variable which we call a similarity variable,
q = Ayx
b
,
and try a solution of the form = (q). The boundary condition at y
must of
course be satisfied by
u(x ,)=
c
cy
y
=
d
dq
q
cq
c y
= U ,
or
Ax
b
' ( ) = U.
(11.11)
Because U is constant while x
b
is not, the chosen form (q) must be modified such
that the left-hand side of Eq. (11.11) also becomes a constant. We therefore try the
form
= Bx
b
f q ( ),
(11.12)
which now satisfies the boundary condition at infinity in the form
u(x,) = ABf ' q ( )= U,
(11.13)
and by letting AB = U, we obtain
f '
( )
= 1
(11.14)
and f ' is identified as the dimensionless velocity u/U. The boundary conditions at
y = 0, Eq. (11.10) become
u (x,0) =
c
cy
y=0
= AB ' f (q)
q=0
= 0,
or
' f 0 ( ) = 0,
and
v(x,0) =
c
cx
y=0
= Bb x
b1
q ' f f ( )
q=0
= 0.
11. High Re Flows The Boundary Layer 337
Hence
f 0 ( ) = 0.
Substitution of from (11.12) into the differential equation (11.8) with the neces-
sary algebra and the selection of the free constants A, B and b such that x and y do
not appear in the equation result in
f f "+2f '"= 0
(11.15)
with the boundary conditions
f 0 ( ) = f ' 0 ( ) = 0,
(11.16)
f '
( )
= 1
(11.14)
and with
= vxU f (q),
(11.17)
q = y
U
vx
=
y
x
Re
x
1
2
, Re
x
=
Ux
v
,
(11.18)
u =
c
cy
= Uf '(q), f '(q) =
u
U
,
(11.19)
v =
c
cx
=
1
2
vU
x
q f 'f ( ).
(11.20)
For a particular x, q may be interpreted as a dimensionless distance y from the
plate. The function f '(q), Eq. (11.19), is the normalized velocity u along the plate,
while f(q), which is
f (q) =
0
q
}
f ' q ( )dq,
counts the flow rate between the plate and the considered q value, i.e., it is the
dimensionless stream function.
f" f' f
0 0.332 0 0
1 0.323 0.330 0.166
2 0.267 0.630 0.650
3 0.161 0.846 1.397
4 0.0642 0.956 2.306
5 0.0159 0.992 3.283
6 0.0024 0.999 4.280
0 1 .000
q
Table 11.1 f(q) in the boundary layer.
Equation (11.15), known as the Blasius equation, must now be solved with the
boundary conditions (11.14) and (11.16). Blasius solved this equation in 1908 by
338 Fluid Mechanics
matching a power series expansion for small q with an asymptotic form for large
q. A digital computer, however, would solve this equation in seconds. The results
of the numerical solution are summarized in Table 11.1 and in Fig. 11.2.
One important result of the numerical computation is f"(0) = 0.332, which is
related to the shear rate on the plate. From Eq. (11.19)
c u
cy
= Uf "(q)
cq
c y
= U
U
vx
f "(q).
u
U
h
0
10.0
2.0
4.0
6.0
8.0
0.2 0.4 0.6 0.8 1.0
Figure 11.2 Velocity in boundary layer.
On the plate where y = 0,
cu
cy
y=0
= U
U
vx
f "(0) = 0.332
U
2
Re
x
1
2
.
The shear stress that the fluid exerts on the plate at a given x is
t
o
=
cu
cy
y =0
=
0.332U
2
Re
x
,
(11.21)
or in its dimensionless form
C
f
=
t
o
1
2
U
2
=
0.664
Re
x
,
(11.22)
where C
f
is defined as the skin friction coefficient.
11. High Re Flows The Boundary Layer 339
The average shear stress over the region 0 x L is also found,
t
o
=
1
L 0
L
}
t
o
dx = 0.664 U
U
vL
=
0.664 U
2
Re
L
, Re
L
=
UL
v
,
(11.23)
and the average friction coefficient,
C
f
,
becomes
C
f
=
1.328
Re
L
.
(11.24)
We now consider more closely the concept of the boundary layer thickness. If
the boundary layer is defined to extend to where u = U, we already know that this
would result in o
. We therefore seek other definitions. Suppose we define o
as that y at which
u = 0.99 U,
or
f ' q ( ) = 0.99.
Table 11.1 reveals that this corresponds to
5, q ~
and a workable definition has
been obtained, which by Eq. (11.18), for y = o, becomes
o
x
=
5.0
Re
x
.
(11.25)
Of course, values different from 0.99 may be chosen to define o, resulting in
different coefficients in Eq. (11.25).
Other useful definitions for the thickness of the boundary layer are the
displacement thickness
o
d
=
1
U
0
}
(U u )dy =
0
}
1
u
U
|
\
|
.
dy
(11.26)
and the momentum thickness
2
0 0
1
( ) 1
m
u u
u U u dy dy
U U U
o
| |
= =
|
\ .
} }
(11.27)
Figure 11.3 shows the velocity distribution in the boundary layer. Region A
is the difference between the flows as assumed by the potential model and as
presented by the boundary layer model. The difference in the velocities at some y
is U - u , and the total volumetric flux deficiency is
0
} (U u)dy,
i.e., because of
the presence of the plate, the flux between y = 0 and y = is smaller by the amount
0
} (U u)dy
than it would have been had the flow really been potential. One could
cause this same difference in a purely potential flow by moving the plate parallel
to itself upward by the amount o
d
; hence
340 Fluid Mechanics
Uo
d
=
o
}
(U u)dy .
Thus o
d
, Eq. (11.26), may be interpreted as the distance by which the flat plate
must be displaced into a potential flow to cause the same mass flux deficiency as
that caused by the boundary layer. Going through an analogous argument for the
momentum, o
m
may be forced into an analogous interpretation with respect to
momentum flux deficiency. The real importance of o
d
and o
m
, however, is their
reappearance in later expressions.
Figure 11.3 Displacement thickness.
Having the Blasius solution for the flat plate at our disposal, it is interesting
to compute o
d
and o
m
just defined. Equation (11.26) can be rewritten in a dimen-
sionless form as
( )
1
2
0
Re 1 ' .
d
x
f d
x
o
q
=
}
(11.28)
Similarly, the dimensionless momentum thickness, Eq. (11.28), becomes
( )
0
1
' 1 ' .
Re
m
x
f f d
x
o
q
=
}
(11.29)
Numerical integrations, using values of f ' vs. q from Table 11.1, yield
o
d
x
=
1.73
Re
x
(11.30)
and
1.328
.
Re
m
x
x
o
=
(11.31)
Comparison of Eq. (11.31) and Eq. (11.24) suggests the interpretation that the
plate, which causes this loss of momentum, must indeed withstand a corresponding
b. Potential flow
y
a. Viscous flow
A
o
d
11. High Re Flows The Boundary Layer 341
shear stress.
Example 11.1
A thin metal plate of dimensions a = 0.2 m, b = 0.5 m is held in water
(v = 10
-6
m
2
/s, = 1000 kg/m
3
) flowing parallel to it. The water velocity is
U = 1 m/s. What is the drag force acting on the plate? Which edge of the plate
should be in the direction of the flow to obtain a lower drag force?
Solution
The drag force may be directly calculated using Eq. (11.23):
F = t
o
A = 2ab t
o
=
2 0.664abU
2
Re
L
.
The area is taken twice, because the force is exerted on both sides of the plate.
Hence,
F =
2 0.664 0.2 0.51000 1
2
Re
L
=
132.8
Re
L
.
For the case L = a = 0.2 m
Re
L
=
LU
v
=
0.2 1
10
6
= 200, 000
and
F =
132.8
200, 000
= 0.297N.
For L = b = 0.5 m
F =
132.8
500,000
= 0.188N.
Hence, holding the plate with the longer edge parallel to the flow results in a lower
drag.
Von KarmanPohlhausen Integral Method
Up to this point we have seen just one exact solution to the boundary layer
equations, that corresponding to the flat plate in parallel flow. Foreseeing situa-
tions where exact solutions to the boundary layer equations cannot be found, an
approximate method due to von Karman and Pohlhausen is now introduced. The
approximation is that the boundary layer equation is not satisfied pointwise but
rather on the average over the region.
342 Fluid Mechanics
The method consists of first transforming the differential boundary layer
equations into an integral equation. This stage is exact. We then proceed to satisfy
this integral equation by the selection of an appropriate velocity profile inside the
boundary layer. This stage introduces an approximation into the method, because
satisfying the integral relation is a necessary condition for the solution, but not a
sufficient condition.
Let h denote some constant value of y well outside the boundary layer,
h > o.
Consider again the boundary layer equation
u
cu
cx
+ v
cu
cy
=U
dU
dx
+ v
c
2
u
cy
2
.
(11.18)
Let the continuity equation be multiplied by u,
u
cu
cx
+
cv
cy
|
\
|
.
|
|
= 0,
and added to Eq. (11.18), which results in
c u
2
cx
+
c (uv )
cy
= U
dU
dx
+ v
c
2
u
cy
2
.
(11.32)
Equation (11.32) is called the conservative form of the boundary layer equation.
Integrating Eq. (11.32) between y = 0 and y = h, we obtain
0
h
}
cu
2
cx
dy +
0
h
}
c(uv )
cy
dy =
0
h
}
U
dU
dx
dy + v
0
h
}
c
2
u
cy
2
dy .
Using Leibnitz's rule on the first term and performing the integration on the second
and fourth terms result in
d
dx
u
2
dy
0
h
}
+ uv
h
= U
dU
dx
dy v
c u
cy
y=0
0
h
}
.
(11.33)
In
v
cu
cy
y=0
=
t
o
we identify t
o
, the shear stress on the plate. We also use the equation of continuity,
Eq. (11.5), to express v as
v =
0
y
}
cu
c x
dy ,
leading to
uv ( )
h
= U
o
h
}
cu
cx
dy = U
d
dx
o
h
}
udy .
11. High Re Flows The Boundary Layer 343
Rearranging Eq. (11.33) accordingly, we obtain
t
o
=
d
dx
0
h
}
(U u)u dy +
dU
dx
0
h
}
(U u)dy
(11.34)
or
t
o
=
d
dx
(U
2
o
m
)+ (Uo
d
)
dU
dx
(11.35)
with o
m
and o
d
defined by Eqs. (11.27) and (11.26), respectively.
Example 11.2
Apply the von KarmanPohlhausen integral method to the flow over a flat
plate and compare your results with the exact ones. For the velocity profile in the
boundary layer assume:
a. A second order polynomial.
b. A third order polynomial.
Solution
The boundary conditions for both cases are
0 0,
, 0 ,
at
at theedgeoftheboundarylayer.
u y
u
uU y
y
c
o
c
= =
===
(11.36)
a. The assumed profile is
u
U
= a + bq + cq
2
,
where q = y / o . The boundary conditions of Eq. (11.36) are satisfied for a = 0,
b = 2, c = 1. Hence
u = U 2q q
2
( )
Equations (11.26) and (11.27), with dy = o dq, yield
o
d
=
1
3
o , o
m
=
2
15
o ,
while
t
o
= v
cu
cy
y=0
= v
U
o
2.
Equation (11.35) now becomes
344 Fluid Mechanics
2v
U
o
=
2
15
U
2
do
dx
or
30
v
U
=
do
2
dx
.
Now, since o = 0 at x = 0, integration yields
o = 30
v
U
x =
5.48
Re
x
x,
(11.37)
This compares favorably with Eq. (11.25). The local shear stress on the plate
is
t
o
=
cu
cy
y=0
=
U
o
2= 0.365U
U
vx
= 0.365U
2
Re
x
1/ 2
and the average one is
t
o
=
1
L
0
L
}
t
o
dx = 0.73U
2
Re
L
1
2
.
(11.38)
b. The assumed profile is
u
U
= a + bq + cq
2
+ dq
3
.
Equation (11.36) yields only three boundary conditions. However, the
boundary layer equation (11.4) itself, with u = v = 0 at y = 0, yields
c
2
u
cy
2
y=0
= 0.
The boundary conditions of Eq. (11.36) and this additional condition are
satisfied by a = c = 0, b + d = 1, b + 3d = 0. Hence
u = U
3
2
q
1
2
q
3
( )
.
(11.39)
Again, Eqs. (11.26) and (11.27), with dy = o dq, yield
o
d
=
3
8
o , o
m
=
39
280
o
while
t
o
= v
cu
cy
y=0
= v
U
o
3
2
,
and Eq. (11.35) now becomes
11. High Re Flows The Boundary Layer 345
3
2
v
U
o
=
39
280
U
2
do
dx
or
280
13
v
U
=
do
2
dx
.
Since for x = 0, o = 0, integration yields
o =
280
13
v
U
x =
4.64
Re
x
x .
Compare this with Eq. (11.25). The shear stress on the plate is now computed
as
t
o
=
cu
cy
y=0
=
3
2
U
o
= 0.323U
U
vx
or
t
o
= 0.323U
2
Re
x
1
2
.
(11.40)
Similarly
t
o
=
1
L
0
L
}
t
o
dx = 0.646U
2
Re
L
1
2
(11.41)
which compare rather favorably with the exact results, Eqs. (11.21) and
(11.23).
It is noted that applications of higher order polynomials, with additional conditions
of the form
c
2
u
cy
2
y=o
= 0,
c
3
u
c y
3
y=o
= 0,
i.e., "smoother" transition to the potential flow, yield worse and worse results. We
do not know that these higher derivatives even exist and the assumption of a very
large number of derivatives vanishing at y = o completely destroys the profile of
the boundary layer.
The Two-dimensional Laminar Jet
Another example of an exact solution of the boundary layer equations is the
two-dimensional jet, Fig. 11.4. It has a mass flux Q = bu
o
and a momentum flux
M = bu
o
2
.
We assume the width of the slit through which the jet emerges, b, to
approach zero,
while the momentum M is kept constant. This would make
u
o
there, but because M is constant, Q = M/u
o
0. Thus for small b it is
reasonable to consider a two-dimensional momentum source, which is a very weak
mass source.
346 Fluid Mechanics
b
u
u
o
Figure 11.4 Two-dimensional jet.
For large y, u = U = 0, and since the fluid extends to y = , the original
momentum flux must be conserved:
M
x
= M
o
=
}
u
2
dy .
(11.42)
The system to be solved is
u
cu
cx
+ v
cu
cy
= v
c
2
u
cy
2
,
(11.43)
v = 0 ,
cu
cy
= 0 at y = 0,
(11.44)
u = 0 at y .
(11.45)
To completely specify the problem, the momentum of the jet, Eq. (11.42), has to be
given.
We note that this flow also has no characteristic length except for x, and
hence we follow Bickley (1939) and Schlichting (1933) in seeking a similarity
transformation:
=
Mvx
|
\
|
.
|
|
1/3
f (q), q =
M
v
2
x
2
|
\
|
.
|
|
1/3
y ,
(11.46)
whence
u =
c
cy
=
M
2
2
vx
|
\
|
.
|
|
1/ 3
f ' (q), v =
c
cx
=
1
3
Mv
x
2
|
\
|
.
|
|
1/ 3
2qf ' f ( ),
(11.47)
which transforms Eq. (11.43) to
3f' ' ' +f'
2
+ff ' ' =0 (11.48)
and the boundary conditions to
f 0 ( ) = f ' ' 0 ( ) =0,
(11.49)
f ' ( ) =0.
(11.50)
Integration of Eq. (11.48) yields
11. High Re Flows The Boundary Layer 347
3f ' ' +f f '= 0
with the constant of integration A = 0 because of Eq. (11.49). A second integration
yields
6f '+f
2
= B
2
or
d q = 6
df
B
2
f
2
=
3
B
df
B + f
+
df
B f
(
(
(
,
which is integrated to
B
3
q = ln
B + f
B f
= ln
1+ f / B
1 f / B
or
f = 2B tanh
Bq
3 2
.
(11.51)
Substitution in u, Eq. (11.47), and then in (11.42) yields
B
3
=
9
4 2
,
resulting in
f =
9
2
|
\
|
.
1/3
tanh
q
48
1/ 3
(
.
Finally
=
9Mxv
2
|
\
|
.
|
1/3
tanh
q
48
1/3
|
\
|
.
and
u =
3M
2
32
2
vx
|
\
|
.
|
|
1/3
1
cosh
2
q 48
1/3
( )
,
v =
Mv
36x
2
|
\
|
.
|
|
1/3
q
cosh
2
q 48
1/ 3
( )
(
(
(
(
tanh
q
48
1/ 3
|
\
|
.
|
.
References
W. Bickley, "The Plane Jet," Phil. Mag., Ser. 7, 23, 727 (1939).
N. Curle, "The Laminar Boundary Layer Equations," Oxford University Press,
London, 1938.
348 Fluid Mechanics
S.I. Pai, "Viscous Flow Theory," Van Nostrand, Princeton, 1957.
H. Schlichting, "Laminare Strahlenausbreitung," ZAMM, 13, 260 (1933).
H. Schlichting, "Boundary-Layer Theory," 7th ed., McGraw-Hill, New York, 1979.
Problems
11.1 Assume a velocity distribution in the boundary layer in the form
u =Usin ty 2o ( )
and apply the von KarmanPohlhausen method to find the
shear stress on a flat plate.
11.2 A flat plate 1 m by 0.5 m is set in a stream of air flowing at 1 m/s parallel to
the plate. What is the minimal drag force on the plate?
11.3 A thin tube, as shown in Fig. P11.3, is put in a parallel flow. The fluid veloc-
ity is 0.1 m/s, the kinematic viscosity is 1.510
-5
m
2
/s and its density is
1.3 kg/m
3
. Find the force required to hold the tube in place if its length is
a. 1 m, b. 3 m.
Figure P11.3
11.4 Is the information that the boundary layer flow admits a similarity solution
used in the von KarmanPohlhausen integral method? If yes, where?
11.5 A flat plate of dimensions 0.1 m by 0.05 m is inserted into a flow such that a
boundary layer is believed to develop along its 0.1-m side. The shear force
on the plate is 0.1 N. Find the force expected to act on a similar plate of
dimensions 0.2 m by 0.1 m.
11. High Re Flows The Boundary Layer 349
11.6 The stream function in a potential flow
into a 90
o
corner shown in Fig. P11.6
is
= xyU.
Estimate the shear stress on the L
long part of the structure along the
x-axis.
11.7 A two-dimensional flow field is given
approximately by
=x
2
y
2
.
A small metal flag with the dimensions
a and b is inserted into the flow field,
where it arranges itself such that the flat
plane lies (approximately) tangent to a
stream sheet, as shown in Fig. P11.7.
The force with which the flag must be
held in place is then measured. The
viscosity of the fluid is = 10 kg/m
.
s,
and its density is = 1000 kg/m
3
.
Neglect the change of the flow field
velocity along the flag, assume laminar flow and compute the force acting
on the flag when its leading edge is located at
a. (1 ; 2), b. (2 ; 3), c. (10 ; 1).
Hint: The flow along the flag is a boundary layer flow, with representing
the potential flow outside the boundary layer.
11.8 Consider the flow along the flag in Problem 11.7. Assume the velocity to
change linearly along the length of the flag. Obtain the differential equation
for the boundary layer thickness resulting from the integral approach.
11.9 Two wide, parallel flat plates are
shown in Fig. P11.9. The flow
streaming toward the plates is at
U = 0.25 m/s, and the flow is assumed
two-dimensional. The density of the
fluid is = 1,000 kg/m
3
, and its vis-
cosity is = 10
3
kg/m
.
s. The two
plates are part of a structure, and
Figure P11.6 Flow into a 90
o
corner.
Figure P11.7 Small flag in a
flow field.
Figure P11.9
350 Fluid Mechanics
a = 1 m, b = 0.5 m. Find the force transferred from the plates to the structure.
11.10 What is the smallest dimension b in Problem 11.9, which still permits a
boundary layer approximation for this configuration.
11.11 Figure P11.9 now represents a cross section of a thin-walled cylinder. Is the
smallest diameter that still admits a boundary layer approximation smaller
or greater than the value of b obtained in Problem 11.10. Explain.
11.12 Consider the boundary layer flow
shown in Fig. P11.12. Use the von
KarmanPohlhausen integral method,
with a third order polynomial.
a. Obtain the velocity profile and the
boundary layer thickness at
x = 1 m and x = 3 m.
b. Calculate the mean velocity inside
the boundary layer, at x = 1 m. Now use this mean velocity as the free
stream velocity and obtain the velocity profile and the boundary layer
thickness for a flat plate at x = 2 m. Compare these results with those
obtained in part a for x = 3 m.
11.13 Oil with the kinematic viscosity of 0.05 m
2
/s enters a 0.1-m-diameter pipe.
Using boundary layer approximation, find how far downstream will the
boundary layers meet.
11.14 Repeat Problem 11.13 for a flow
in an annulus with the diameters
0.2 m and 0.1 m.
11.15 Calculate the total shear force on
the entrance section of the pipe in
Problem 11.13 between the
entrance and up to the point
where the boundary layers meet,
Fig. P11.15. Compare the result
with the shear force on the same pipe section for a fully developed laminar
flow of the same mass flux.
11.16 Let the oil in Problem 11.15 enter the pipe, as shown in Fig. P11.15. The en-
trance section, from point A to point B, approximates the boundary layer
flow, with B the location where the boundary layers meet. The section from
U
x = 1 m x = 3 m
Figure P11.2
Figure P11.15
11. High Re Flows The Boundary Layer 351
B to C, which is of the same length as that from A to B, represents the fully
developed flow. Three manometers read the pressures at points A, B and C.
The pipe is horizontal. Express the readings of the manometers in terms of
the shear forces calculated in Problem 11.15 and of the pipe geometry.
11.17 Express the friction head losses between points A and B and B and C in
Problem 11.16.
11.18 Consider the boundary layer flow over a porous flat plate, with constant
suction applied along the plate. The boundary conditions in this case are
u =U at y and u = 0, v = S = const at y = 0. Try u/x = 0, and obtain
an exact solution of the boundary layer equations.
11.19 An airplane flies at 1,000 km/h. An air scoop is designed in the belly of the
airplane to take in air to cool electronic equipment. Air is to be taken from
the potential flow, avoiding the boundary layer region. Compute b,
Fig. P11.19.
Figure P11.19
11.20 An old way to determine the speed of a ship was to drop into the water a log
of wood tied with a rope. The length of the rope that the log dragged off the
ship per some predetermined time was then measured, e.g., by counting
knots made in the rope. The force needed to drag a rope off a ship is 20 N.
The log is a wooden cylinder 1.2 m long and 0.12 m in diameter. Neglect
the contribution of the pressure field, and compute the speed with which the
log must be dragged to begin measuring, i.e., the log drift correction.
11.21 An estimate of the forces of wind on a tall rectangular building is obtained
as follows: The front of the building is assumed to be a 20 m wide by 50 m
high flat plate perpendicular to the flow, subjected to the stagnation pressure
of the wind. The backside is at the static pressure. The sides of the building
are flat plates 50 m high by 60 m long in the direction of the wind, which
suffer shear forces as in a boundary layer flow. The top is a flat plate 20 m
352 Fluid Mechanics
wide by 60 m long, also subjected to shear forces. Compute the total force
on the building for winds of 10, 20 and 40 km/h.
11.22 A glass sinker has the teardrop shape shown in Fig.
P11.22, which is drawn to scale. The density of glass is
2,000 kg/m
3
. The drag force on the sinker is caused by
viscous friction only. Measure the figure and compute
the terminal speed of the sinker in water.
11.23 An arrow is drawn in Fig. P11.23, to scale. Find the dis-
tribution of the drag force along the arrow. Find the
correcting moment for small deviation angles, i.e., that
moment which keeps the arrow tip pointing forward, for
arrow speeds of 10 and 30 m/s.
100 mm
600 mm
2 mm
14 mm
6 mm
Figure P11.23
11.24 A sleeve of cloth is towed behind a ship. The sleeve is 2 m long and 0.5 m
in diameter. The ship sails at 3 m/s. Find the drag force on the sleeve.
11.25 A bullet fired into a pond enters the water at the speed of 300 m/s. The
bullet has a diameter of 0.009 m and is 0.015 m long. Its mass is 0.008 kg.
Assume just viscous friction as in a boundary layer flow and estimate the
bullet's deceleration in the water.
11.26 An amphibian car floats on the water with its rear wheels spinning. The
outer radius of the tires is 0.8 m, and they are 0.3 m wide; the wheels are just
half submerged. The speedometer shows that on land the speed would be 80
km/h. Estimate the force driving the car forward in the water.
11.27 A javelin has an average diameter of 0.03 m and is 2 m long, with an appar-
ent density of 800 kg/m
3
. A man throws that javelin at an elevation angle of
45
o
and reaches a distance of 50 m. Find the distance he could have reached
in a vacuum.
Figure P11.22
11. High Re Flows The Boundary Layer 353
11.28 A streamlined boat has most of its drag derived from viscous friction on its
wet skin. It is suggested to use results known for an old well-designed boat
to construct a new double-hulled boat that has the same total carrying
capacity as the old one. Using boundary layer considerations, estimate by
how much must the power driving the new boat be increased relative to that
driving the old boat.
11.29 A rope, which has a diameter of
0.05 m and is 20 m long dangles in
the water behind a boat sailing at
5 m/s. Find the drag caused by the
rope.
11.30 A 0.014-m-diameter pipe 3 m long
is set in the side of a 10-m-high wa-
ter tank, Fig. 11.30. Find the mass
flowrate of the water, including
boundary layer effects in the pipe.
Figure 11.30
354 Fluid Mechanics
355
12. TURBULENT FLOW
Reynolds Experiment
Consider the experimental apparatus shown in Fig. 12.1. It consists of a small
tube inserted inside a large transparent pipe. The inner tube has a small diameter
and its opening is located far from the bend, i.e., L
B
is sufficiently long to
guarantee fully developed flow. The fluid in the inner tube is the same as that in
the outer pipe, and they have the same and . However, the inner fluid is dyed
while the outer one is clear. In all experiments the velocities of the two fluids are
adjusted so as to have as little shear strain between them as possible.
L
B
L
B
Laminar flow Turbulent flow
Figure 12.1 Reynolds experiment.
As the experiment proceeds, a dyed stream cylinder is seen to form down-
stream from the opening of the inner tube. The stream tube, which makes the
envelope of this stream cylinder, is well defined, and the dye is restricted to the
inside of this stream tube. The dye begins to spread out by molecular diffusion
only far downstream.
The experiment is repeated for various fluids and for various velocities, and
as discussed in Chapter 8, all these can be summarized using the Reynolds number
as the single parameter involved. It is found that the description given above fits
356 Fluid Mechanics
flows whose Reynolds number, calculated using the pipe diameter D as the
characteristic length, is below a certain numerical value, namely,
Re
D
=
VD
v
< 2,000,
(12.1)
where V is the average velocity in the pipe, i.e. the flowrate per unit cross-sectional
area. Above this Reynolds number the envelope of the dyed stream tube is not
clearly defined. The dye is seen to mix with the outer fluid right from the opening
of the inner tube. The laminar character of the flow is lost and the flow becomes
turbulent.
At this point it is necessary to reconsider the definition of the velocity vector.
Consider a body of fluid, with a system S within, and with the point A inside the
system, Fig. 1.1. Let the momentum of the system be M and let its mass be m. An
average velocity for the system may be defined as
q =
M
m
and the velocity at point A may be defined as
q = lim
mm
c
M
m
.
(12.2)
The definition of m
c
(or rather of the volume V
c
that contains the mass m
c
) as well
as the limitations on this definition are presented in the sections on the continuum
and on local properties in a continuum, in Chapter 1.
Returning to the description of Reynolds' experiment, we are faced with one
of two uncomfortable choices: Either for Re > 2,000 the limit in Eq. (12.2) does
not exist, q then is not defined and our continuity and momentum equations cannot
be used; or, the momentum and continuity equations, with boundary conditions
that do not depend on time and which have had time-independent solutions for
Re< 2,000, do not have such time-independent solutions for Re > 2,000. Although
the boundary conditions remain time independent, the resulting flows persist to be
time dependent, nonperiodic and non-decaying.
While one should always bear in mind the first possibility, the second choice
is the correct one. To explain the appearance of the turbulent flow, use must be
made of the theory of hydrodynamic stability, which is outside the scope of this
book. Still it is important to note that the explanation relies on the amplification of
random perturbations, and therefore a certain degree of randomness in the ensuing
turbulent flow is to be expected.
Consider the turbulent flow in a pipe. If one were to plot the velocity of the
fluid at a given point in the pipe, a pattern similar to that of Fig. 12.2 would result.
As seen, there are small rapid velocity fluctuations about some average value. This
average value may or may not depend on time. The velocity then is a superposition
12. Turbulent Flow 357
of an average velocity
u
and a velocity fluctuation u',
u = u + u.
(12.3)
u
t
steady
unsteady steady
u
-
Figure 12.2 Velocity in turbulent flow.
Similar relations may be written for the other velocity components as well as
for other properties, such as the pressure, the temperature, etc.:
v = v + ' v , p = p + ' p , T = T + ' T .
(12.4)
The average quantities are defined according to the form
q =
1
At
t
t +At
}
q dt =
1
At
t
t +At
}
q + ' q ( )dt
(12.5)
and
t
t +At
}
' q dt = 0,
(12.6)
i.e., the time average of the fluctuations vanishes.
The Nature of Turbulence
The concept of the coefficient of viscosity has already been used to obtain the
Navier-Stokes equations. It was treated as a coefficient, i.e., as a multiplier which
serves to adjust units and to formally yield good approximations to physical
phenomena. Some insight into the mechanism of viscosity is helpful to explain its
modification by turbulence.
Consider the differential layer shown in Fig. 12.3 and note that any point in a
three-dimensional flow can be locally described by that scheme. In addition to the
continuum velocity u, there exists the microscopic random thermal motion of the
358 Fluid Mechanics
molecules. This thermal motion has an average zero mass flux, and a molecule A
traveling a distance +dy is compensated for by a molecule B traveling dy. The A
molecules have an average x-wise velocity of u, while the B ones have u + du
velocities.
u + du
u
y
y + dy
B
A
y
x
L
Figure 12.3 Exchange of momentum on the molecular scale.
As seen directly from Fig. 12.3, the shear stress between the layers is
proportional to both du/dy and the speed of the random motion of the molecules. In
turbulent flow, in addition to the random motion of the individual molecules,
whole chunks of fluid move at random. For a given du/dy the transfer of
momentum is thus enhanced by turbulence.
The increase in shear stresses in turbulent flows is sometimes said to result
from eddy viscosity, which is just another statement describing the action of the
random turbulent fluctuations.
The Equations of Motion in Turbulent Flow
Consider the x-component of the momentum equation for incompressible
flow. Let the equation of continuity be multiplied by u,
u
cu
cx
+
cv
cy
+
cw
c z
|
\
|
.
|
|
= 0,
and added to the x-component of the momentum equation, to yield:
cu
c t
+
c(u
2
)
cx
+
c(uv)
cy
+
c(uw)
c z
|
\
|
.
|
|
=
cp
cx
+ g
x
+
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
(
(
(
.
(12.7)
This form of the equation of motion is sometimes referred to as the conservative
form.
In turbulent flow, for each term in Eq. (12.7) we substitute the sum of its
12. Turbulent Flow 359
mean value and the corresponding fluctuations. Thus
q = q + ' q , p = p + ' p ,
(12.8)
c (u
2
)
c x
=
c u
2
( )
cx
+ 2
c u ' u ( )
cx
+
c ' u
2
( )
cx
,
(12.9)
c (uv )
cy
=
c(u v )
cy
+
c ( ' u v )
cy
+
c (u ' v )
cy
+
c ( ' u ' v )
cy
.
(12.10)
We now time-average the resulting equation by integration over the time
interval At. Thus each term in the equation is replaced by its mean value. We note
that
u ' =
1
At
t
t +At
}
udt = 0
and
v' = w' = p' = 0.
Also
c u ' u
cx
=
c
cx
t
t +At
}
u ' u dt = 0
and
c ' u v
cy
=
cu ' v
cy
= 0.
However
c ( ' u
2
)
cx
=
c
cx
1
At
t
t +At
}
' u
2
( )
dt
(
(
doesnot vanish.
Using these expressions, Eq. (12.9) yields
c (u
2
)
cx
=
c(u
2
)
cx
+
c( ' u
2
)
cx
.
(12.11)
Time averaging eliminates all the terms in which the fluctuations are linear.
However, the terms nonlinear in the fluctuations remain in the time-averaged
equations. Time smoothing of Eq. (12.7) results in
cu
ct
+
c(u
2
)
cx
+
c(uv )
cy
+
c (uw )
cz
|
\
|
.
|
=
cp
cx
+ g
x
+
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
(
(
c ( ' u
2
)
cx
+
c ( ' u ' v )
cy
+
c( ' u ' w )
cz
|
\
|
.
|
|
.
(12.12)
360 Fluid Mechanics
Comparison of Eqs. (12.7) and (12.12) reveals a new group of terms in Eq. (12.12)
which do not have corresponding ones in Eq. (12.7). These terms account for the
turbulent momentum transfer due to the random velocity fluctuations. Their role is
similar to that of momentum transfer by molecular viscosity resulting in the
laminar shear stresses. These terms are known as the turbulent Reynolds stresses
and may be written as
t
xx
t
= ' u
2
,
(12.13)
t
xy
t
= ' u ' v ,
(12.14)
t
xz
t
= ' u ' w .
(12.15)
Repeating the procedure for the y- and z-components of the time-averaged Navier
Stokes equation, three additional terms of the turbulent stress tensor t t
t
are
obtained:
t
yy
t
= ' v
2
,
(12.16)
t
yz
t
= ' v ' w ,
(12.17)
t
zz
t
= ' w
2
.
(12.18)
Let Eq. (12.7) be divided by and rewritten with a slight modification, i.e.,
instead of the kinematic viscosity, v = / , we substitute v
t
+ v ,
cu
ct
+
c(u
2
)
cx
+
c(uv)
cy
+
c(uw)
cz
=
1
cp
cx
+ g
x
+ v
t
+ v
( )
c
2
u
cx
2
+
c
2
u
cy
2
+
c
2
u
cz
2
(
(
.
(12.19)
Equation (12.19) is now compared with Eq. (12.12), and we see that Eq. (12.19)
has the extra terms resulting from v
t
, while Eq. (12.12) has the extra terms of the
Reynolds stresses. May Eq. (12.19) be used instead of Eq. (12.12)? In many prac-
tical problems Eq. (12.19) yields good approximations to measured results, which
justifies its use. The term v
t
is called turbulent viscosity or eddy viscosity.
Prandtl Mixing Length, Generalized Profiles
In early work on turbulent flows close to rigid boundaries Prandtl (1925)
attempted to obtain the Reynolds stresses from an a priori estimate of the
' u ' v
quantities. He defined the macroscopic analog of the microscopic mean free path,
i.e., that distance traveled by a lump of fluid laterally to the mean velocity direc-
tion, before it randomly changed direction. This distance is called the Prandtl
Mixing Length. This mixing length is not a property of the fluid, as is the mean
free path, but rather a property of the flow. Certainly the distance traveled by a
12. Turbulent Flow 361
chunk of fluid cannot exceed the distance from some nearby wall. Prandtl chose
his mixing length as proportional to that distance, i.e.,
L = k
L
y , (12.20)
with k
L
the proportionality factor. We look again at Fig. 12.3 and consider it now
as corresponding to turbulent flows. Instead of
L
, the molecular mean free path,
we substitute L, the Prandtl mixing length, i.e., that length traveled by a chunk of
fluid laterally to the direction of the average flow. We note from the figure that
' u ~ ' v ~L
d u
dy
|
\
|
.
|
|
.
(12.21)
Hence the turbulent stresses are
t
t
= L
du
dy
|
\
|
.
|
|
2
.
Close to a solid surface we may take
t
t
~ t
w
= const
so that
L
d u
dy
|
\
|
.
2
= t
w
or
d u
dy
=
1
L
t
w
,
which together with (12.20) gives
d u
dy
=
1
k
L
y
t
w
or
d u =
dy
y
1
k
L
t
w
.
Integration yields
u
t
w
/
=
A
k
L
lny + B.
(12.22)
A general velocity profile near a rigid wall has been obtained. Though this
treatment seems rather qualitative, the resulting profile is confirmed
experimentally for many classes of turbulent flows, at least in the vicinity of the
wall.
The constants of Eq. (12.22) must be found experimentally for each flow. The
equation recast in dimensionless form for pipe flow becomes
362 Fluid Mechanics
u
+
= 2.5lny
+
+ 5.5,
(12.23)
where u
+
=u/u* is the dimensionless velocity with
u
*
= t
w
/
known as the shear
velocity. The dimensionless distance from the wall is defined in the form of a
Reynolds number as
y
+
=
yu
*
v
=
y
v
t
w
.
(12.24)
At the wall itself, because of the viscous boundary conditions, all velocities
must damp out and the flow cannot remain turbulent. Equation (12.23) does not
hold at the wall, as can be seen from the ln y
+
term.
A more elaborate treatment divides the flow in a pipe into a laminar sublayer
near the wall, a turbulent core and a buffer layer in between. The velocity profile
obtained is given in the form of three equations, one for each region. It is known as
the universal profile and it holds for the whole cross section of the pipe.
For the laminar sublayer, which extends from the wall to y
+
= 5, the profile is
linear:
u
+
= y
+
, 0s y
+
< 5.
(12.25)
For the buffer layer the profile is given by
u
+
= 5.0lny
+
3.05, 5s y
+
< 30,
(12.26)
while for the turbulent core Prandtl's simple expression still applies,
u
+
= 2.5lny
+
+ 5.5, y
+
> 30.
(12.23)
The universal profile, given by Eqs. (12.23), (12.25) and (12.26), applies to
turbulent flows in smooth pipes. It is particularly useful in the analysis of heat and
mass transfer to walls in turbulent conduit flows, where the processes depend
strongly on the flow field details near the wall. The use of this profile, ignoring the
laminar sublayer and the buffer zone, yields very good results in many cases.
The Moody Diagram
Inspection of the time-averaged momentum equation for turbulent flow, Eq.
(12.12), reveals that for the same number of equations as in laminar flow we now
have three additional dependent variables, u', v', w'. Attempts to construct a
sufficient number of additional equations were not successful, and solutions of
turbulent flows rely largely on experiments.
12. Turbulent Flow 363
While Eq. (12.12) cannot be solved, it still implies that the only similarity
parameter in the flow is the Reynolds number, in addition, of course, to geometri-
cal similarity.
All round pipes are geometrically similar. The fact that the pipe walls may
not be smooth leads to the use of the concept of wall roughness. Wall roughness is
defined as the average height, c, of the asperities on the wall surface, made di-
mensionless with respect to the pipe diameter. We thus expect the dimensionless
pressure drop along the pipe to depend on the relative roughness, c / D, and on the
Reynolds number.
A convenient way to formulate the problem is to define a friction factor f as
f =
AP
L / D
1
1
2
V
2
, f = f Re,c / D ( ),
(12.27)
where V = Q / A is the average velocity in the pipe. Using Eq. (6.49) one may write
for a round pipe
AP
L
=
1
r
c
cr
r t
rz
( )
with the boundary conditions at the centerline (r = 0) and at the pipe wall (r = R),
respectively,
t
rz
= 0 at r = 0,
t
rz
= t
w
at r = R .
Integration yields
t
w
=
R
2
AP
L
or
t
w
=
D
4
AP
L
.
(12.28)
This result, combined with Eq. (12.27), defines the friction factor in terms of the
wall shear stress, t
w
,
t
w
=
f
8
V
2
, f =
8t
w
V
2
(12.29)
Comparison of Eq. (12.29) with Eq. (11.22) shows that the definition of the
friction factor is similar to that of the skin friction coefficient, C
f
, for a flat plate.
Equation (12.27) is now rewritten in terms of the friction head loss, h
f
,
h
f
= f
L
D
V
2
2g
|
\
|
.
|
|
.
(12.30)
364 Fluid Mechanics
This result may be used in the modified Bernoulli Equation, Eq. (7.23).
Generally, the friction factor has to be evaluated experimentally, except in
rare cases where an analytical solution exists.
For the special case of laminar flow Eq. (12.27) may be combined with the
Poiseuille Equation, Eq. (6.58),
V =
APD
2
32 L
,
to yield
f = 64
DV
|
\
|
.
|
or
f =
64
Re
.
(12.31)
The friction factor, f, is presented for the general case as a function of the
Reynolds number and the relative roughness, c/D. This was done graphically by
Moody in 1944 in what is called the Moody diagram, Fig. 12.4. The diagram
contains three distinct zones:
a. Laminar flow, for which an analytical solution was given by Eq. (12.31).
b. Transition zone, where head losses depend both on the molecular viscosity
(linear with the velocity) and on Reynolds stresses (proportional to the square
of the velocity).
c. Turbulent zone, where Reynolds stresses prevail and f hardly changes with
Re. It is really the existence of this region that justifies the choice of V
2
in the
definition of f in Eq. (12.27).
Example 12.1
The universal velocity profile for turbulent flow in a smooth pipe is given by
Eqs. (12.25), (12.26) and (12.23). The profile is written in terms of
u
+
= u / t
w
/ and y
+
= y t
w
/
( )
v.
To perform calculations it is convenient to have the profile put in terms of the
parameters used in the Moody diagram. Rewrite the universal profile in terms of
the friction factor, f, and the Reynolds number, Re.
365
R
e
y
n
o
l
d
s
n
u
m
b
e
r
F
i
g
u
r
e
1
2
.
4
M
o
o
d
y
d
i
a
g
r
a
m
o
f
f
r
i
c
t
i
o
n
l
o
s
s
c
h
a
r
a
c
t
e
r
i
s
t
i
c
s
f
o
r
l
a
m
i
n
a
r
a
n
d
t
u
r
b
u
l
e
n
t
f
l
o
w
i
n
p
i
p
e
s
.
(
L
.
F
.
M
o
o
d
y
,
T
r
a
n
s
.
A
S
M
E
6
6
(
8
)
,
p
.
6
7
1
,
1
9
4
4
.
)
R
e
=
D
V
v
366 Fluid Mechanics
Solution
The velocity profile is given by
u
+
= f y
+
( )
,
where
u
+
=
u
t
w
/
, y
+
=
y t
w
/
v
.
Wesubstitute t
w
/ fromEq. 12.29 ( ),
t
w
= V
f
8
,
and obtain
u
+
=
u
V f / 8
.
Also,
y
+
=
yV
v
f
8
=
y
D
Re
f
8
,
and for the various regions the universal velocity profile becomes:
a. For the laminar sublayer, 0 y
+
< 5, from Eq. (12.25),
u
V f / 8
=
y
D
Re f / 8
or
u
V
=
y
D
Re
f
8
|
\
|
.
| at 0s
y
D
Re
f
8
< 5.
b. For the buffer layer, 5 y
+
< 30, Eq. (12.26) yields
u
V f / 8
= 5.0ln
y
D
Re
f
8
(
(
(
3.05 at 5s
y
D
Re
f
8
< 30.
c. For the turbulent core, y
+
30, from Eq. (12.23)
u
V f / 8
= 2.5ln
y
D
Re
f
8
(
(
(
+ 5.5 at
y
D
Re
f
8
> 30.
12. Turbulent Flow 367
Example 12.2
Water (density = 1,000 kg / m
3
, kinematic viscosity v = 10
-6
m
2
/ s ) flows
with the average velocity of V = 10 m / s in a smooth pipe of diameter D = 10 cm.
a. Find the thickness of the laminar sublayer.
b. Where does the turbulent core begin?
c. Write the equations of the velocity distribution in the pipe.
d. What are the velocities at the edge of the laminar sublayer, at the beginning of
the turbulent core, and at the pipe centerline?
Solution
We first establish the flow regime by computing the Reynolds number
Re =
VD
v
=
10 0.1
10
6
= 10
6
.
Hence, the flow is turbulent. From the Moody diagram, Fig. 12.4, we obtain
f = 0.0117.
The variable y
+
is
y
+
=
y
D
Re
f
8
= y
10
6
0.0117/ 8
0.1
= 382, 400y
or y = 2.615 10
-6
y
+
.
a. The thickness of the laminar sublayer, y
+
= 5, is
y = 2.61510
6
5.
b. The turbulent core begins at y
+
= 30, i.e.,
y = 0.0000784 m = 0.078 mm.
Hence in this pipe the buffer zone extends only up to 0.16% of the radius; and
the turbulent core covers 99.8% of the radius.
c. The velocity profile is obtained from the results of Example 12.1. Hence, for
the laminar sublayer,
u =
V
D
Re
f
8
|
\
|
.
| y =
10 10
6
0.0117
0.1 8
y,
u = 146, 000y.
For the buffer zone
368 Fluid Mechanics
u = 5V
f
8
ln
y
D
Re
f
8
(
(
(
3.05V
f
8
,
where f / 8 = 0.0117/ 8= 0.03824.
Thus,
u = 5 10 0.03824 ln
10
6
0.03824y
0.1
(
(
(
3.05 10 0.03824,
u = 1.912 ln (382,400 y ) 1.166.
For the turbulent core
u = 2.5V
f
8
ln
y
D
Re
f
8
(
(
(
+ 5.5V
f
8
,
u = 2.5 10 0.03824 ln (382,400 y ) + 5.5 10 0.03824 ,
u = 0.956 ln (382,400y ) + 2.103 .
d. The velocity at the edge of the laminar sublayer, where y = 1.30710
-5
m, is
u = 146000 1.307 10
-5
= 1.91 m/s.
The velocity at the beginning of the turbulent core, where y = 0.0000784 m, is
u = 0.956 ln (382,400 7.84 10
-5
) + 2.103 = 5.35 m / s.
At the centerline y = 0.05 m, and the velocity is
u = 0.956 ln (328,400 0.05 ) + 2.103 = 11.38 m/s.
It is worthwhile to note that although the velocity at the centerline seems
reasonable, the velocity gradient is not. Differentiation shows that it does not
vanish at the centerline, as it should.
Example 12.3
It has been found experimentally that turbulent flow in rough pipes is not
affected by the roughness elements as long as the latter do not protrude outside the
laminar sublayer; i.e., the pipe walls are said to be hydraulically smooth for
c
+
=
cu
*
v
s 5,
12. Turbulent Flow 369
where c is the height of the roughness elements.
For turbulent flow in a pipe of 5 cm diameter and Re = 50,000, find the
maximum height of roughness elements allowing flow under hydraulically smooth
conditions.
Solution
We are essentially looking for the thickness of the laminar sublayer for a flow
with Re = 50,000. We use the expression for y derived in Example 12.2
y
+
=
y
D
Re
f
8
, y =
y
+
D
Re f / 8
,
and let y
+
= 5. For Re = 50,000 we read, from Fig. 12.4, f = 0.021. Thus,
c
max
= y =
5 5
50,000 0.021/ 8
= 9.75 10
3
cm~ 0.1mm.
Hence, for roughness elements smaller than c
max
the pipe walls can be considered
hydraulically smooth.
One should also note how thin the laminar sublayer is. It is not surprising that
Prandtl's formula, Eq. (12.23), which neglects it altogether, is nevertheless quite
accurate for hydrodynamic calculations.
Flow in Pipes
The problems encountered in the use of the Moody diagram to compute pipe
flow are of two kinds:
1. The direct problem: given the mass flow in a pipe, what is the head loss?
2. The indirect problem: given the pressure drop, what is the flowrate?
The indirect problem is nonlinear. Most problems in water distribution belong
to the indirect class, where solutions are obtained by iterations and the use of a
computer.
Example 12.4
A cast iron pipe 15 cm in diameter was installed in a water system 15 years
ago. Experience indicated that this type of pipe will increase its wall roughness
with time according to the relation
370 Fluid Mechanics
c = c
o
1+ 0.25t ( ),
where c
o
= 2510
-5
m is the wall roughness of a new pipe and c is the roughness
after t years. A new pump is to be installed in the system, which is capable of
delivering a maximum discharge of 0.03 m
3
/s of water at 27
o
C. Estimate the
pressure drop per meter of horizontal pipe at the maximum discharge.
Solution
After 15 years of service,
c = 2510
5
1+ 0.25 15 ( ) = 0.00119m.
The relative roughness is then given by
c
D
=
0.00119
0.15
= 0.0079.
The average velocity of the water flow in the pipe is
V =
Q
A
=
0.03
0.15 ( )
2
t 4 ( )
= 1.698m/ s.
Now, at 27
o
C, v = 0.008610
-4
m
2
/s and = 1,000 kg/m
3
. Thus, the Reynolds
number is
Re =
VD
v
=
1.698 0.15
0.0086 10
4
= 2.96 10
5
.
For the values of c/D and Re calculated above the Moody diagram, Fig. 12.4,
yields f = 0.036. The friction loss is then given by
h
f
= f
L
D
V
2
2g
= 0.036
1
0.15
|
\
|
.
|
1.698
2
2 9.8
= 0.035m.
Example 12.5
A pipe carries water from a reservoir and discharges it as a free jet, Fig. 12.5.
Find the flowrate to be expected through a 20-cm-inner-diameter commercial steel
pipe for which c = 4.610
-5
m.
12. Turbulent Flow 371
1
2
h = 35 m
h = 0
V
L = 135 m
Figure 12.5
Solution
Points 1 and 2 are both at atmospheric pressure. Thus, Bernoulli's equation,
Eq. (7.23), which applies between these two points is simplified to
h
1
=
V
2
2
2g
+ h
f
,
where the velocity at which the reservoir surface recedes, V
1
, has been neglected.
Now,
h
f
= f
L
D
V
2
2
2g
.
Combining this with the Bernoulli equation leads to
h
1
=
V
2
2
2g
1+ f
L
D
|
\
|
.
| .
Substituting the data results in
35=
V
2
2
2 9.8
1+ f
135
0.2
|
\
|
.
or
V
2
=
686
1+ 675f
.
The problem cannot be solved directly in the case of turbulent flow, since f as a
function of V is not known, and so V cannot be isolated algebraically. Instead, an
iterative process involving the Moody diagram must be used.
We use the Moody diagram and as a first guess estimate the friction factor for
the roughness of c/D = 4.610
-5
/ 0.2 = 0.00023 to be f = 0.014. We then solve for
V
2
:
V
2
=
686
1+ 675 0.014
= 8.1m/ s.
The Reynolds number is then
Re =
V
2
D
v
=
8.1 0.2
0.0086 10
4
= 1.8810
6
.
372 Fluid Mechanics
For this number, the Moody diagram gives a friction factor of 0.0155. We now
repeat the above computations using this new friction factor until there is no
appreciable change in the factor. We find
V
2
= 7.736 m / s,
f = 0.0155,
the friction factor remaining essentially unchanged. The volume flow rate is then
Q = V
2
A = 7.736
t
4
|
\
|
.
| 0.2
2
= 0.243m
3
/ s,
and the pressure drop is
Ap = f
L
D
V
2
2
2
= 0.0155
135
0.2
1
2
7.736
2
1,000= 3.13 10
5
N/ m
2
.
Example 12.6
Figure 12.6 shows a line diagram of a piping system: Pipe AB is 500 m long
and has an inner diameter of 0.1 m. Pipe BC is 200 m long and has an inner
diameter of 0.05 m. Pipe BD is 300 m long and has an inner diameter of 0.025 m.
The pressures were measured at points A, D, C and found to be p
A
= 600 kPa, p
C
=
p
D
=
100 kPa. Find the flowrates in the pipes for a fluid of v = 10
-6
m
2
/s, = 1000
kg/m
3
and c / D = 0.002 for all pipes.
A B
C
D
Figure 12.6 Piping system.
Solution
All flows are assumed turbulent. From Eq. (12.27)
Ap = f
L
D
1
2
V
2
.
Therefore
Ap
AB
= 2.5 10
6
f V
AB
2
,
Ap
BC
= 210
6
f V
BC
2
,
Ap
BD
= 610
6
f V
BD
2
.
12. Turbulent Flow 373
Also Ap
BC
= Ap
BD
, and because of conservation of mass
V
AB
= 0.5
2
V
BC
+ 0.25
2
V
BD
.
The solution is iterative:
a. Assume: V
BD
= 5 m/s Re
BD
= 125,000, f = 0.0185,
Ap
BD
= 2,775 kPa > 500 kPa.
b. Assume: V
BD
= 1 m/s Re
BD
= 25,000, f = 0.03,
Ap
BD
= 180 kPa.
Assume V
BC
= 2 m/s Re
BC
= 100,000, f = 0.025,
Ap
BC
= 200 kPa > 180 kPa.
Assume V
BC
= 1.9 m/s Re
BC
= 95,000, f = 0.025,
Ap
BC
= 180.5 kPa ~ 180 kPa
V
AB
= 0.5
2
1.9 + 0.25
2
1 = 0.5375 m/s,
Re
AB
= 53,750, f = 0.025, Ap
AB
= 1.95 kPa,
Ap
AB
+ Ap
BC
= 182 kPa < 500 kPa.
c. Assume: V
BD
= 1.67 m/s Re
BD
= 41,665, f = 0.027,
Ap
BD
= 450 kPa.
Assume V
BC
= 3.17 m/s Re
BC
= 158500, f = 0.0245,
Ap
BC
= 492 kPa > 450 kPa.
Assume V
BC
= 3.03 m/s Re
BC
= 151500, f = 0.0245
Ap
BC
= 450 kPa.
V
AB
= 0.5
2
3.03 + 0.25
2
1.67 = 0.86 m/s
Re
AB
= 86,000, f = 0.025 ; Ap
AB
= 46 kPa.
Ap
AB
+ Ap
BC
= 496 kPa ~ 500 kPa.
Results: V
AB
= 0.86 m/s ; V
BC
= 3.03 m/s ; V
BD
= 1.67 m/s.
Flow through Pipe Fittings
Pipe fittings are such parts as connections, flow dividers, bends and elbows,
reducers, valves and many more, all of which have fluids flowing through them.
Pipes used to transfer fluids invariably involve pipe fittings, and the combination
of pipes and fittings is sometimes referred to as a piping network. The municipal
water supply in a city is an example of a piping network. In the previous section
we have considered friction losses in pipes. We now extend our treatment to the
374 Fluid Mechanics
contribution of the various fittings to the friction losses in the piping network.
Suppose we had only one type of fitting, say a 90
o
short bend, which we use
when we want to connect two pipes perpendicular to one another. We could look
for similarity conditions and find that in addition to similarity in geometry there is
one similarity parameter, the Reynolds number. Closer scrutiny shows that surface
roughness is also important and that a dimensionless roughness parameter may be
defined in exactly the same manner as in pipe flow. Armed with this information
we could perform experiments and sum up our results in a chart somewhat like the
Moody diagram. We could denote this diagram as the friction loss chart for a 90
o
short bend.
We could now repeat the procedure for a 90
o
long bend, for a tee-shaped flow
divider, etc. Indeed under certain circumstances this elaborate treatment is
adopted. In general, however, one tries to avoid such a detailed investigation for
each and every fitting; there are very many fittings, and the amount of the
experimental work involved can become considerable.
Suppose we had obtained the friction loss diagram for a 90
o
short bend. The
friction coefficient is denoted by F, and by analogy to Eqs. (12.30) and (12.27), the
friction head loss may be expressed as
h
f
= F
V
2
2g
(12.32)
with
F = F Re,c / D ( ).
(12.33)
Simplification of treatment now depends on the behavior of F.
Some experimental data published in the literature show this F to be a
constant for each particular fitting, i.e., F = k,
h
f
= k
V
2
2g
(12.34)
with k being referred to as a loss coefficient. A partial explanation for this
somewhat surprising simplification is that many fittings are designed for particular
forms of use, and the range of the Reynolds numbers at which they are used is not
large. In other words, the function F does not vary much over the range in which
its argument, Re, varies. Another possible explanation is that the losses result from
turbulent eddies which are activated by deceleration and draw their energy at a rate
proportional to the kinetic energy of the flow. In such cases the loss coefficient k
in Eq. (12.34) is simply the proportionality coefficient, and the losses may be
interpreted as a certain percentage of the kinetic energy. Some k values are
presented in Table 12.1.
12. Turbulent Flow 375
Fitting
Loss coefficient,
Eq. (12.34)
k = h
f
V
2
/ 2g ( )
Equivalent length,
Eq. (12.36)
L
e
=
L
D
= h
f
f V
2
/ 2g ( )
Elbow, 90
o 0.9 30 - 40
Elbow, 45
o 0.4 15
Tee, flow turns 90
o 1.8 60
Tee, flow straight through
0.9 30
Return bend
2.2
Gate valve, fully open
0.2 10
Globe valve, fully open
10 100 - 300
Sharp-edged pipe entrance
1.0
Rounded pipe entrance
0.02 - 0.1
Inward projecting pipe entrance
0.8
Sharp-edged pipe exit
0.5
Square expansion from d to D
d/D = 0.2
0.92
d/D = 0.4 0.72
d/D = 0.8 0.16
Square contraction from D to d
d/D = 0.2
0.49
d/D = 0.4 0.42
d/D = 0.8 0.18
Table 12.1 Friction losses in pipe fittings.
The total friction losses of a fluid flowing through a pipe with fittings is
found by combining Eq. (12.30) with Eq. (12.34),
h
f
= f
L
D
+ k
i
i
|
\
|
.
|
|
V
2
2g
.
(12.35)
Another approach to an approximation relies on the speculation that perhaps
the F function, Eq. (12.33), may behave like the Moody friction factor f, Eq.
(12.27), in the form
F Re, e / D ( )= L
e
f Re, e / D ( ) ,
with L
e
= L / D, in which L
e
is a dimensionless equivalent length, expressed as an
376 Fluid Mechanics
equivalent number of pipe diameters, which is a constant. Experimental results
confirm this speculation for the 90
o
short bend and for many other pipe fittings,
with a constant L
e
value for each fitting. To compute friction losses under these
circumstances, the F function is obtained for a particular fitting by reading its L
e
value from an appropriate table and multiplying it by the friction factor f read from
the Moody diagram for a pipe of the corresponding diameter. Thus the friction
head loss due to a fitting may be computed as
h
f
= F
V
2
2g
= L
e
f
V
2
2g
,
(12.36)
and comparison with Eq. (12.30) suggests that L
e
may be indeed expressed as L
e
=
L / D, in which L is a dimensionless equivalent length, i.e., the losses associated
with the fitting are the same as for a pipe of the same diameter and an equivalent
length expressed in a number of pipe diameters. Table 12.1 also lists the equivalent
lengths for some fittings. It should be noted that for some flow patterns the use of
the loss coefficient k yields better approximations, while for other patterns the L
e
approximation is better. When one approach is definitely better, the table does not
list the other parameter at all.
Example 12.7
A piping network is shown in Fig. 12.7. The pipe from A to E and from F to
G has a diameter of 0.05 m and a total length of 60 m. The pipe section EF is
0.025 m in diameter and 20 m long. All the pipes have a relative roughness
c/D = 0.001. The mean velocity needed at G, when the globe valve D is fully open,
is 2 m/s. Find the water level h needed to sustain such a flow of water.
V
h
A B
C D E
F
S
G
Figure 12.7 Piping system.
12. Turbulent Flow 377
Solution
The Bernoulli equation between points S and G yields
h = h
f
+
V
G
2
2g
.
To find the friction factor, we first calculate the Reynolds number for the pipes
with D = 0.05 m,
Re
G
=
VD
v
=
2 0.05
10
6
= 100,000.
The Moody diagram yields
f = 0.0220.
Similarly, for the pipe with the 0.025 m diameter,
f = 0.0215.
The friction losses due to piping alone are
h
f
p
= f
L
D
V
2
2g
|
\
|
.
|
|
= 0.022
60
0.05
2
2
2g
+ 0.0215
20
0.025
8
2
2g
= 61.5.
From Table 12.1, the losses due to the fittings are given in terms of k-values as:
A (sharp pipe entrance) 1.0; B (90
o
elbow) 0.9; C (tee, 90
o
) 1.8; D (globe valve) 10;
E (contraction,d/D = 0.5) 0.36; G (sharp pipe exit) 0.5;F (expansion,d /D =0.5) 0.58.
Hence,
h
f
F
= k
i
i
|
\
|
.
|
V
2
2g
= 1+ 0.9+ 1.8+ 10+ 0.36+ 0.5 ( )
2
2
2g
+ 0.58
8
2
2g
= 4.86m
.
The total friction losses are
h
f
= h
f
p
+ h
f
F
= 61.5+ 4.86= 66.4 m,
and the water level h required is
h = h
f
+
V
G
2
2g
= 66.4+
2
2
2g
= 66.6 m.
Noncircular Pipes
A concept, which emerges from the application of the idea of the universal
velocity distribution to the similarity of turbulent flows, is that of the hydraulic
diameter.
Suppose we seek the head loss in a turbulent flow in an elliptical pipe. In the
main part of the pipe cross section the velocity profile is practically flat, Fig.12.8.
378 Fluid Mechanics
Near the walls, in a narrow layer, the velocity is given by the universal profile,
which must support the forces corresponding to the head loss. This situation is
common to the elliptical pipe and to a circular pipe, and the mechanisms acting in
both seem to be the same, provided both have the same ratio of cross-sectional
area to circumference, i.e., the same shear loading of the universal velocity profile.
Figure 12.8 Turbulent velocity profile in elliptical pipe.
For a circular pipe the ratio of cross section to the "wetted" circumference is
A / C =
t
4
D
2
tD
=
D
4
and
D = 4
A
C
.
We therefore define the hydraulic diameter as
D
H
= 4
A
C
.
(12.37)
Thus for an ellipse A = tab, C ~t (a + b) and
D
H
= 4
A
C
= 4
ab
a + b
.
The Moody diagram may now be used for the elliptical pipe. The same treatment
may be used for rectangular pipes and for pipes of other cross sections.
Coefficient of Drag
In turbulent flows around bluff bodies the use of the friction coefficient is not
convenient, and a better concept is that of the coefficient of drag, C
D
, defined by
F
D
= C
D
A
U
2
2
,
(12.38)
where F
D
is the drag force acting on the body, or the force per unit span for infinite
span bodies and A is the cross-sectional area normal to the flow. The drag
coefficient, C
D
, is usually found experimentally. Some representative values of C
D
are given in Fig. 12.9 and in Table 12.2.
12. Turbulent Flow 379
Figure 12.9 Drag coefficient of a cylinder and a sphere.
Shape and orientation to flow Reynolds number range
C
D
Flat strip, normal to flow Greater than 10
3
1.95
Disk, normal to flow Greater than 10
3
1.10
Hemispherical shell:
Hollow upstream Greater than 10
3
1.33
Hollow downstream Greater than 10
3
0.34
Cylinder, normal to flow 10
3
210
5
1.10
Greater than 5 10
5
~ 0.35
Sphere 10
3
2 10
5
0.45
Greater than 3 10
5
~ 0.20
Model of airship Greater than 2 10
5
0.04
Airfoil at zero angle of attack Boundary layer transition
0.006
Table 12.2 Coefficients of drag of various shapes.
[Compiled from data in Hunsaker and Rightmire (1947) and Schlichting (1979).]
Example 12.8
A spherical particle has the density of 3,000 kg/m
3
and the diameter of
0.004 m. The kinematic viscosity of water is v =10
-6
m
2
/s. Find the terminal
sinking velocity of this sphere in water.
380 Fluid Mechanics
Solution
The terminal velocity is achieved when the drag force is balanced exactly by
the buoyancy force. Thus
F
D
=
4
3
t 0.002
3
w
( )g = 0.657 10
3
.
Since Re is not known, assume C
D
= 0.45 (Table 12.2). From Eq. (12.33),
U = 2F
D
/ C
D
A
w
= 0.482 m/ s,
Re = UD/v = 1,928,
which confirms the assumed C
D
,
Table 12.2. Thus
U = 0.482 m/s.
Flow-induced Lift Forces
The previous section, which dealt with drag forces, treated flows around
closed configurations such that far from the configuration the velocity field could
be assumed undisturbed, i.e., it was the same as if the closed configuration did not
exist. The undisturbed velocity vector was essentially constant. The total force that
such a flow applied to the configuration had a component in the undisturbed flow
direction; this component was denoted the drag force. The total force acting on the
configuration may also have a component in the direction perpendicular to that of
the undisturbed flow. This perpendicular component is now denoted the lift force.
The lift force is expected to appear in all flows not symmetrical with respect to the
configuration, even when the configuration itself is symmetrical. In some cases the
lift force may not be as large as the drag force, but this does not necessarily reflect
on its importance. The experienced engineer is always aware of the lift force,
which sometimes he or she may choose to ignore.
Some attempts to calculate lift forces for nonsymmetrical flows around
cylinders have been done in the chapter on potential flows. There are many
examples in which lift forces may be derived analytically or numerically. The lift
forces may also be obtained experimentally for cases that cannot be otherwise
resolved and also as experimental verifications of theoretical results. We note that
in our examples of computations of lift forces we have used potential flow con-
siderations. Indeed this is the more frequent case in the design of lifting configu-
rations. For such cases we conclude that the viscosity, and with it the Reynolds
number, have only a second order effect on the lift forces. The exception to this
conclusion is that the lift forces do depend on whether the flow is laminar or
12. Turbulent Flow 381
turbulent and for configurations not designed for lift, the influence of the Reynolds
number need be checked.
Considerations of similarity still lead us to the formulation
F
L
= C
L
A
1
2
V
2
(12.39)
in which F
L
is the lift force, A is a characteristic area, is the fluid density and V is
the undisturbed flow velocity. However, unlike the drag coefficient, the lift
coefficient C
L
does not depend on the Reynolds number, at least for well-designed
lifting configurations. In many cases airplane wings may be treated as two
dimensional. Then A, the characteristic area, may be replaced by L, the
characteristic length, which is usually the cord of the wing's profile. The lift force
thus obtained is, of course, per unit width of the wing.
As long as the flow may be considered incompressible, there is only one
similarity parameter, the Reynolds number. In the previous section C
D
, the drag
force coefficient, has been presented as a function of the Reynolds number. Since
here C
L
seems not to depend on the Reynolds number, it comes out to be constant
for a particular configuration. The angle between the cord of a wing and the
direction of the undisturbed flow is called the angle of attack, and during flight the
same wing may operate with various angles of attack. Each angle of attack
corresponds to a different flow field and therefore has a different lift coefficient. It
is common practice to present the lift coefficient as a function of the angle of
attack, as shown in Fig. 12.10. The drag coefficient of the same wing is also shown
in Fig. 12.10 as a function of the angle of attack, and with its dependence on the
Reynolds number deleted.
Example 12.9
An airplane is designed for a nominal cruising speed V
c
= 800 km/h = 222 m/s
and take-off speed V
o
= 100 km/h = 28 m/s. The weight of the airplane is G = 30,000
N. The airplane wing is shown in Fig. 12.10. Compute the dimensions of the wing
and the power expanded to propel just the wing forward.
Solution
We assume that the lift forces arise mainly from the wings. The highest lift
per wing area is needed at takeoff because the air speed is then the lowest. At the
same time we want to reduce drag at this stage of operation. Inspecting Fig. 12.10,
we find that a lift coefficient of 1.0 is still below the point where the drag begins to
become large. We therefore choose for takeoff C
L
= 1.0. Correspondingly, from
Fig. 12.10, C
D
= 0.08.
382 Fluid Mechanics
o
U
-5 0 5 10 15 20
-0.50
-0.25
0
0.25
0.50
0.75
1.00
0
0.05
0.10
0.15
0.20
o (
)
C
L
C
D
C
L
C
D
Figure 12.10 Lift and drag coefficients for a wing.
The density is assumed as = 1.1 kg / m
3
. Then
F
L
= AC
L
1
2
V
2
( )
, A =
F
L
1
2
V
2
C
L
.
At takeoff
A
o
=
F
L
1
2
V
o
2
C
L
=
30, 000
1
2
1.1 28
2
1.0
= 69.6m
2
.
Choosing a wing length of 10 m on each side, with L being the wing cord,
L =
69.6
2 10
= 3.48m.
At cruising speed,
12. Turbulent Flow 383
C
L
=
F
L
1
2
V
2
A
=
30,000
1
2
1.1 222
2
69.6
= 0.016.
Accordingly, from Fig. 12.10, C
D
= 0.02. The force required to overcome the drag
at takeoff is
F
D
= AC
D
1
2
V
2
( )
= 69.6 0.08
1
2
1.1 28
2
( )
= 2, 400N
and the power at takeoff is
P = F
D
V
o
= 2400 28= 67,200W=67.2kW.
At cruising
F
D
= AC
D
1
2
V
2
( )= 69.6 0.02
1
2
1.1 222
2
( )= 37,700N
and the power is
P = F
D
V
c
= 37,700 222 10
3
= 8,380kW.
Flowrate Measurement
Two devices used to measure flow velocities, the Venturi meter and the Pitot
tube, have already been considered in Chapter 7. Some more devices are shown in
Fig. 12.11.
b. Short flow nozzle
q
a. Venturi meter
d. Thin-plate orifice
q
c. Sharp-edge orifice
q
q
Figure 12.11 Flow measuring devices.
384 Fluid Mechanics
The measurements in all these devices rely on relations between velocities
and differences in pressure, which are derived from the Bernoulli equation. We
note that with the exception of the Pitot tube, the other devices presented may be
arranged in a descending order of elaboration, the Venturi meter being the most
elaborate, and then the measuring flow nozzle, the sharp edge orifice and the thin
plate orifice.
The expression for the velocity of an incompressible fluid, measured by a
Pitot tube, was derived in Example 7.4 as
v =
2
p
1
p
2
( ),
(12.40)
where
p
1
p
2
is the difference between the total and the static pressure. For a
manometer attached to both legs of the Pitot tube and reading Ah
m
the expression
for the velocity is
v = 2g Ah
m
m
.
(12.41)
The relations for the flow of an incompressible fluid through a Venturi meter
was derived in Example 7.2 as
v =
2
p
1
p
2
( )
1 d D
( )
4
(12.42)
and
Q =
tD
2
4
V =
td
2
4
v =
td
2
4
2
p
1
p
2
( )
1 d D
( )
4
.
(12.43)
where v is the velocity at the throat of the Venturi meter, V is the velocity in the
pipe, Q [m
3
/s] is the volumetric flowrate, A = td
2
/4 is the cross-sectional area at
the throat and d/D is the ratio of the diameters of the throat and the pipe,
respectively.
An assumption inherent in the derivation of Eq. (12.43) is that the flow is
incompressible and that friction head losses and changes of stream tube cross
sections due to accelerations, sometimes denoted vena contracta, are negligible.
Relations for compressible flows may be formulated using equations derived
in Chapter 13. Alternatively, the velocities and flowrates can still be obtained by
equations like Eqs. (12.40) and (12.43),respectively, provided a correction for
compressibility is included. For example, the velocity of a compressible fluid,
measured by a Pitot tube, can be expressed as
12. Turbulent Flow 385
v = C
c
2
p
1
p
2
( ) .
(12.44)
Corrections in the form of a compressibility factor C
c
, are presented in Table 12.3.
Ratio of pressures Ratio of throat diameter to pipe diameter d/D
P
2
P
1
P
1
P
2
( ) P
1
0.25 0.50 0.75
For Venturi meters and flow nozzles
0.98 0.02 0.989 0.988 0.981
0.96 0.04 0.978 0.976 0.964
0.92 0.08 0.956 0.953 0.932
0.88 0.12 0.933 0.928 0.898
0.84
0.80
0.16
0.20
0.909
0.885
0.903
0.878
0.865
0.834
For square-edged orifices
0.98
0.96
0.92
0.88
0.84
0.80
0.02
0.04
0.08
0.12
0.16
0.20
0.994
0.986
0.974
0.964
0.952
0.941
0.994
0.986
0.974
0.963
0.950
0.938
0.993
0.985
0.970
0.956
0.940
0.925
Table 12.3 Approximate values of the compressibility factor C
c
for flow of air or
gases with k = 1.4 through a Venturi, flow nozzle or orifice.
(For gases with k = 1.3 the error is less than 1.0%.)
Another correction coefficient,
C
d
, the discharge coefficient, may be
used to include friction head losses and
effects of strong curvature of the
streamline near the device, i.e., of vena
contracta. Some values for discharge
coefficients are shown in Table 12.4
and in Fig. 12.13. Equation (12.43),
thus corrected, provides the flowrate
through the device as
1
.
5
D
D
1.75D
A
0.6D
1
.
5
D
D B
Figure 12.12 Types of flow nozzles,
A long, B short.
386 Fluid Mechanics
Q = C
c
C
d
td
2
4
2
p
1
p
2
( )
1 d D ( )
4
.
(12.45)
Orifice Re
Diameter ratio, d/D
D
i
s
c
h
a
r
g
e
c
o
e
f
f
i
c
i
e
n
t
,
C
0.2 0.4 0.6 0.8
0.590
0.600
0.610
0.620
R
e
1
0
6
=
R
e
=
1
0
4
R
e
=
1
0
3
d
Figure 12.13 Discharge coefficients for thin plate orifice.
The compressibility correction, C
c
, depends on the ratio of the diameters of
the device and also, slightly, on the kind of device used. The discharge coefficient,
C
d
, depends also on the ratio of the diameters of the device and the kind of device
used. The more elaborate devices present smaller resistance to the flow and have
higher discharge coefficients. For a well-designed Venturi meter the discharge
coefficient approaches 1. In principle, C
d
also depends on the Reynolds number.
Reynolds number Diameter ratio, d/D
at throat
0.20 - 0.40 0.60 0.80
50,000
100,000
200,000
500,000
1,000,000
5,000,000
0.966
0.978
0.985
0.991
0.993
0.994
0.965
0.976
0.984
0.989
0.991
0.993
0.958
0.970
0.978
0.985
0.986
0.988
Table 12.4 Discharge coefficients for long nozzles.
(2 10-in. pipe; pipe-wall taps one diameter upstream
and half a diameter downstream from inlet face.)
12. Turbulent Flow 387
In many practical cases it is helpful to consider the discharge coefficient C
d
as a multiplier obtained from similarity considerations, not unlike the drag coeffi-
cient. In some cases the measuring device, e.g, an orifice, is manufactured to fit
some particular configuration and is then calibrated. It is common then to express
the calibration information in the form of a calibrated discharge coefficient C
d
.
Example 12.10
A Pitot tube is used to measure air flow velocity. The air pressure is 10
5
Pa,
its temperature is 300 K and the measured pressure difference of the air is 5,000 Pa.
Find the velocity.
Note that this example is quite similar to Example 7.4, except for the need to
correct for compressibility.
Solution
The density of the air is
=
p
RT
=
10
5
8,314/ 28.9 ( ) 300
= 1.15kg/ m
3
.
Using Eq. (12.44),
C
c
= 0.98
, we obtain
v = 0.98
2
1.15
5,000 ( ) = 91.4m/ s.
Example 12.11
Air flows in a 6-in. pipe in which a flowrate measuring device with a diame-
ter ratio of 2 has been installed. The air pressure is 10
5
Pa and its temperature is
300 K. The measured pressure drop of the air across the device is 6,000 Pa. Find
the flowrate in each of the following measuring devices:
a. A Venturi meter.
b. A flow nozzle.
c. A thin plate orifice.
388 Fluid Mechanics
Solution
The flowrate is calculated by means of Eq. (12.45). The density of the air is
=
p
RT
=
10
5
8,314/ 28.9 ( ) 300
= 1.15kg/ m
3
.
The cross-sectional area at the throat of the device is
A =
t
4
d
2
=
t
4
3 0.0254 ( )
2
= 4.56 10
3
m
2
.
Hence, Eq. (12.45) yields
Q = C
c
C
d
td
2
4
2
p
1
p
2
( )
1 d D
( )
4
= C
c
C
d
4.56 10
3
2
1.15
6,000
1 1/ 2 ( )
4
= 0.466C
c
C
d
.
a. For a Venturi meter:
From Table 12.3, the compressibility correction is C
c
= 0.965. The discharge
coefficient for a well-designed Venturi meter is 1 and Eq. (12.45) yields
Q = 0.466C
c
C
d
= 0.466 0.9651= 0.450m
3
s.
b. For a flow nozzle:
From Table 12.3, the compressibility correction is C
c
= 0.965. The discharge
coefficient for a nozzle, Table 12.4, depends on the Reynolds number at the
throat. We find Re approximately,
Re =
dv
v
=
4Q
t dv
=
4 0.450
t 3 0.0254 ( ) 1.57 10
5
= 479,000
and the discharge coefficient is C
d
= 0.990. Equation (12.45) now yields
Q = 0.466C
c
C
d
= 0.466 0.965 0.990= 0.445m
3
s.
c. For a thin plate orifice:
From Table 12.3, the compressibility correction is C
c
= 0.980. For the
Reynolds number of part b, Fig. 12.13 leads to a discharge coefficient,
C
d
= 0.611. Equation (12.45) yields
Q = 0.466C
c
C
d
= 0.466 0.980 0.611= 0.279m
3
s.
This leads to a lower Re. Still the discharge coefficient is roughly the same.
12. Turbulent Flow 389
Turbulent Boundary Layers
In the introduction to turbulent flows in this chapter no violation of the
NavierStokes equations has been implied. The equations are still valid; though for
turbulent flows the task of finding solutions becomes even more formidable,
because these are time dependent. The boundary layer equations were derived from
the NavierStokes equations by singular perturbation and therefore are also valid
for turbulent flows.
When turbulence exists inside the boundary layer, we speak of turbulent
boundary layers. Turbulence in the boundary layer is determined by the local
Reynolds number, Re
x
= ux/v, where x is the distance from the leading edge. Thus
for sufficiently small x the flow in the boundary layer is laminar. For larger
distances from the leading edge, i.e., for Reynolds numbers exceeding 500,000, the
flow in the boundary layer becomes turbulent, Fig. 12.14. It is also noted that on
rigid boundaries q = 0 and therefore q' must also subside; hence the flow must
always become laminar at the wall itself, at least within some thin sublayer. This
means that velocity profiles believed to exist inside a turbulent boundary layer
cannot be simply extended to the wall, and some additional information may be
necessary to affect a solution.
x
Laminar Turbulent
x
crit
dx
Figure 12.14 Laminar and turbulent boundary layer on a flat plate. The critical
Reynolds number for transition from laminar to turbulent flow is
Re
crit
= 500,000.
Another property typical to turbulent flow is the possibility of the appearance
of boundary layers in fully developed or almost fully developed flows. This is a
direct result of the ability of the turbulent profile to sustain high shear stress by
Reynolds stresses. The velocity profile is thus rather flat in the main flow,
Fig.12.8. However, approaching the walls, the intensity of the turbulence
390 Fluid Mechanics
decreases. Here the size of the eddies cannot exceed the distance from the wall,
and the Prandtl mixing length must decrease. Since the shear forces must
eventually be transferred to the wall and the available mechanism for shear stress
is now mostly viscous stress, the velocity profile must exhibit a sharp gradient in
the vicinity of the wall, i.e., a boundary layer.
One practical result of this behavior is the possibility to use measurements
obtained for pipe flow in boundary layer calculations, pipe flows being consider-
ably more amenable to experiments than flat plate flows.
Integral Solutions for Boundary Layers
We now seek an integral solution method, similar to that obtained for the
laminar boundary layer in Chapter 11. We want to apply it to flows where our only
justification for the use of the term boundary layer is that the velocity profile
changes rapidly within a narrow region. We therefore do not integrate the
boundary layer equations, as in Chapter 11, but rather start from basic concepts.
Consider the layer shown in Fig. 12.14. The mass flowrate into the dx strip in
the boundary layer is just
o
o
}
u dy ,
and that out of it is
o
o
}
u dy + dx
d
dx
o
o
}
u dy .
The mass flowrate into the strip from the main flow, i.e., into the strip from above,
is therefore
dx
d
dx
o
o
}
u dy
with the corresponding flux of momentum
dx U
d
dx
o
o
}
u dy .
The excess of momentum entering the strip at x over that leaving at x +dx is
dx
d
dx
o
o
}
u
2
dy
and the total change in momentum is compensated by the pressure gradient o
(dP/dx) and the shear at the wall t
o
dx. Thus
12. Turbulent Flow 391
U
d
dx
o
o
}
u dy
d
dx
o
o
}
u
2
dy = o
dP
dx
+ t
o
or
t
o
=
d
dx
o
o
}
U u ( )udy
(
(
dU
dx
p
o
}
u dy o
dP
dx
,
(12.46)
and because in the main flow we still assume
dP
dx
= U
dU
dx
,
(12.47)
Eq. (12.34) becomes
t
o
=
d
dx
o
o
}
U u ( )u dy
(
( +
dU
dx
o
o
}
U u ( )dy ,
(12.48)
which is exactly the same as Eq. (11.34). Equation (12.48) is now used to obtain an
approximate solution for the flat plate turbulent boundary layer. The velocity
profile is assumed, after von Karman, as the seventh-root law
u
U
=
y
o
|
\
|
.
1/ 7
= q
1/7
(12.49)
while the shear stress is taken from a correlation determined experimentally by
Blasius,
t
o
= 0.046
v
Uo
|
\
|
.
|
1/4
1
2
U
2
.
(12.50)
Substitution in Eq. (12.48) and integration yield
o
x
= 0.37Re
x
0. 2
.
(12.51)
Eliminating o between Eqs. (12.51) and (12.50) leads to
t
o
= 0.030U
2
v
Ux
|
\
|
.
|
1/ 5
.
(12.52)
It should be noted that t
o
was given here, rather than computed from
(u/y)|
y=0
, as for the laminar boundary layer; the turbulent profile, Eq. (12.49),
does not hold down to y = 0, where a laminar sublayer exists. Where no informa-
tion on t
o
, such as Eq. (12.50), is given, this laminar sublayer must be considered
with its velocity distribution, which makes the solution more involved.
392 Fluid Mechanics
Example 12.12
A thin metal plate of dimensions a = 0.2 m, b = 0.5 m is held in water
( =1000 kg/m
3
, v = 10
-6
m
2
/s), which flows parallel to it, with a velocity U = 4 m /s.
Find the drag force exerted on the plate for the following cases:
a. The flow being parallel to a.
b. The flow being parallel to b.
Compare your finding to those of Example 11.1.
Solution
The force exerted by the fluid on a plate of length L and width w is
F = w
o
L
}
t
o
dx .
For the case of Re
L
being higher than the critical Reynolds number for transition
from laminar to turbulent flow, namely, higher than Re
c
= 500,000, the integration
has to be split into two parts. In the laminar region the wall shear stress t
o
integrated from x = 0 to x = x
c
is given by Eq. (11.23). In the turbulent region the
integration is carried out from x = x
c
to x = L using Eq. (12.49). Thus
F = w
o
x
c
}
t
o
( )
lam
dx + w
x
c
L
}
(t
o
)
turb
dx ,
which upon substitution and integration yields
F = wU
2
0.664x
c
Re
c
1/2
+ 0.037 L Re
L
1/ 5
x
c
Re
c
1/ 5
( ) | |
In the present case x
c
is found from
6
Re 10 500,000
Re , 0.125 m.
4
c c
c c
x U
x
U
v
v
= = = =
Hence, the force on each side of the plate is given by
( )
2 1/2 1/5 1/5
1,000 4 0.664 0.125 500,000 0.037 Re 0.125 500,000
L
F w L
(
= +
or
( )
1/5
1.878 592 Re 0.00906
L
F w L
(
= +
Case a.
L = a = 0.2 m,
w = b = 0.5 m,
12. Turbulent Flow 393
Re
L
=
UL
v
=
4 0.2
10
6
= 800,000.
The force on each side is
F = 0.5 [1.878 + 592 ( 0.2800,000
-1/5
0.00906 )] = 2.163 N
and on both sides
F = 4.326 N .
Case b.
L = b = 0.5 m,
w = a = 0.2 m,
Re
L
=
UL
v
=
4 0.5
10
6
= 2,000,000.
The force on each side is
F = 0.2 [1.878 + 592 (0.52,000,000
-1/5
0.00906)] = 2.555 N
and on both sides
F = 5.110 N .
As seen, the plate with the longer edge parallel to the flow yields here a
higher drag force. This is contrary to what was found in Example 11.1 for the
laminar case. In the case of the longer edge parallel to the flow a larger part of the
plate is in turbulent flow, which leads to a higher overall drag force.
Example 12.13
Repeat Example 12.12 for a plate with the sides 2 m by 5 m, i.e., ten times as
large as in Example 12.12.
Solution
The formulation of Example 12.12 is repeated, resulting in
( )
( ) ( )
1/5
0.8 0.2 0.8
1.878 592 Re 0.00906
592 4,000,000 3.4855 28,308 3.486
L
F w L
w L L w
(
= +
= =
Case a. L = a = 2 m,
w = b = 5 m,
F = 229.0 N, on one side.
394 Fluid Mechanics
Case b. L = a = 5 m,
w = b = 2 m,
F = 198.2 N, on one side.
The plate with the longer side parallel to the flow now yields the lower drag force,
i.e., the same behavior as in laminar flow.
References
S.K. Friedlander and L. Topper, "Turbulence, Classic Papers on Statistical Turbu-
lence," Interscience, New York, 1961.
J.O. Hinze, "Turbulence," McGraw-Hill, New York, 1959.
J.C. Hunsaker and B.G. Rightmire, "Engineering Applications of Fluid Me-
chanics," McGraw-Hill, New York, 1947.
L.F. Moody, "Friction Factors for Pipe Flow," Trans. ASME, p. 671, November
1944.
H. Schlichting, "Boundary Layer Theory," 7th ed., McGraw-Hill, New York,
1979.
H. Tennekes and J.L. Lumley, "A First Course in Turbulence," MIT Press, Cam-
bridge, MA, 1972.
Problems
12.1 Use the universal velocity profile to construct a plot of the velocity as a
function of radial position when air with a density of 1.2 kg/m
3
flows at an
average velocity of 20 m/s through a smooth 0.6-m-diameter pipe. The vis-
cosity of air is 0.02 cp. What is the thickness of the laminar sublayer for this
case.
12.2 Use the universal velocity distribution to calculate the local velocity at a point
situated midway between the pipe axis and the pipe wall in a 0.5-m-
diameter smooth pipe carrying air at an average velocity of 30 m/s and a
pressure of 200 kPa gauge. Viscosity of air is 0.02 cp.
12.3 A major pipeline carrying crude oil (density 860 kg/m
3
and viscosity 5 cp)
drops 600 m in elevation over the last 5 miles, as shown schematically in
12. Turbulent Flow 395
Fig. P12.3. The main part of the line is 0.75 m in diameter in which the
average velocity of the oil is 2 m/s. The oil arrives at point A at a pressure
of 300 kPa gauge and flows downhill to point B where the pressure is
desired to be no more than 100 kPa gauge. What diameter of line would you
recommend for the section AB? Assume the pipe to be hydrodynamically
smooth.
600 m
5 mi
B
A
Main line
Figure P12.3 Oil pipe.
12.4 Repeat Problem 12.3 for p
A
= 1,500 kPa.
12.5 Oil (20
o
C, viscosity 3 cp, specific gravity 0.95) is to flow by gravity through a
1-mile-long pipeline from a large reservoir to a lower station, the average
difference in levels being 14 m. The required discharge rate is 1.6 m
3
/min.
Calculate the diameter of steel pipe necessary for these conditions. Compare
your solution and answer to those of Problem 7.10.
12.6 A pump is connected to the end of the pipe, point 2 in Fig. 12.5, and is used
to recharge the reservoir. The water level in the reservoir is h = 35 m. The
pressure at point 2, just outside the pump is 500 kPa. What is the water flow
into the reservoir?
12.7 Find the water flux as in Problem 12.6 when the circular pipe is replaced by a
triangular pipe with all sides equal, having the same cross-sectional area as
the circular pipe.
12.8 A car travels at 80 km/h. The driver puts his hand out of the window to cool
off. Assuming the hand resembles a flat strip perpendicular to the flow, 0.1
m by 0.5 m, find the additional power expanded by the car.
12.9 Repeat Problem 12.8 when the "strip" is parallel to the flow.
12.10 Figure P12.10 is a scheme of a pipe connected to a water tower open to the
396 Fluid Mechanics
atmosphere. The atmospheric pressure is 100 kPa. The bend in the pipe
presents a resistance to the flow equivalent to 2 m of pipe. The pipe diame-
ter is 0.1 m, and it is a smooth pipe. The water density is 998 kg/m
3
,
and its
viscosity is = 0.001 Pa
.
s. The pressure at point A, shown in the figure, is
measured to be P
1
= 160 kPa, absolute pressure.
a. Find the volumetric flowrate in the pipe.
b. Find the dimensions of a square pipe that gives the same volumetric
flowrate as the circular pipe.
Figure P12.10
12.11 To simplify maintenance, it was suggested to use two 0.1-m-diameter pipes
instead of one 0.1414-m-diameter pipe. For the same volume flux of water,
find the ratios of the pressure drops:
a. When the flows are laminar.
b. When the flows are turbulent.
12.12 Measurements show that the flow between the two plates in Problem 6.7 is
turbulent. The pressure between the plates is assumed to depend on the
radial coordinate only; the pressure drop is approximated by using relations
for pipe flow with the same hydraulic diameter as the radius-dependent
circular cross section available to the flow between the plates. The plates are
very smooth. With reference to Fig. P6.7, F = 5 N, h = 0.002 m, L = 0.1 m,
find the air supply pressure and rate.
12.13 Water flows in a 100-m-long pipe with a diameter of 0.025 m. The average
velocity in the pipe is 5 m/s and the pipe discharges to the atmosphere.
Compute the Reynolds number.
a. Is the flow turbulent?
b. Calculate the pressure at the entrance to the pipe.
12.14 It is suggested to change the pipe described in Problem 12.13 and use two
smaller pipes such that the average velocity of the flow remains the same,
L = 20 m
Bend
h
1
= 5 m
h
2
=10 m
A
p
a
= 100 kPa
12. Turbulent Flow 397
i.e., 5 m/s. The flowrate remains that of Problem 12.13.
a. Calculate the required pressure drop in these pipes.
b. For a pressure, which is that used in Problem 12.13, find the average
velocity and the flow rate in these two pipes.
12.15 The same amount of fluid as in Problem 12.13 must now be supplied using
a 0.0125-m-diameter pipe.
a. Calculate the pressure drop in this pipe and compare your result with
that of Problem 12.13.
b. Note that for a constant flow rate, the pressure drop is proportional to the
diameter of the pipe raised to the power n. Find n.
c. Also note that for a constant pressure drop the flow rate is proportional
to the diameter of the pipe raised to the power s. Find s.
12.16 A fluid flows in a pipe that has the diameter D. The same fluid flows in the
gap between two parallel flat plates. The size of the gap is also D. The same
pressure gradient exists in both systems. Calculate the ratio between the
flow rate of the fluid in the pipe and the flow rate between the plates per
width of D.
12.17 The pipe and the plates of Problem 12.16 are now used such that the flow
rate in the pipe is the same as that between the plates per width D. Calculate
the ratio between the power needed to pump the fluid through the pipe and
that needed to pump the fluid between the plates.
12.18 Water flows through a rectangular square duct with the sides D = 0.1 m. The
average velocity is 8 m/s. The duct is 100 m long. Calculate the power
needed to pump the water. Compare your results with those obtained for the
same flow rate through a circular pipe.
12.19 Water flows through an annulus with an inner radius of 0.10 m and an outer
radius of 0.12 m at an average velocity of 6 m/s. The length of the annulus
is 100 m. Find the pressure drop along the annulus.
12.20 Repeat Problem 12.13 for a pipe of the same cross section, which is:
a. Square.
b. Rectangular, with sides ratio of 1 : 2.
c. Elliptic, with sides ratio of 1 : 2.
12.21 A municipal water main has the diameter of 0.1 m. The mean velocity in the
main is 10 m/s. The line has a tee connection every 50 m and a bend every
100 m. Calculate the pressure drop expected per 1000 m of pipe.
398 Fluid Mechanics
12.22 A truck is designed to travel at a cruising speed of 110 km/h. A preliminary
design is shown in Fig. P12.22. The width of the truck is 4 m, and to esti-
mate the air drag force on the truck, one
may use the frontal projection, i.e., the
same drag as for a flat plate 4 m 4 m
perpendicular to the flow. It has been
suggested to streamline the geometry
such that the drag coefficient becomes
0.6. Estimate the power saved at
cruising speeds.
12.23 A car has the mass of 700 kg. In normal driving the car tires are not ex-
pected to skid as long as the normal force between a tire and the road is at
least 700 N. The maximum velocity technically safe for the car is 140 km/h.
The projected area of the car is 6 m
2
. Find the maximum allowable lift coef-
ficient of the car geometry. Note that this coefficient is defined for the
projected area. Find the maximum allowable driving speeds with wind
velocities of 80 km/h and 30 km/h.
12.24 A plumber had to connect a 0.025-m-diameter pipe to a hose. Not having
the necessary parts he inserted a connection of 0.013 m diameter in the line
and then continued with the 0.025 m diameter. For a mean flow velocity of
5 m/s, estimate the extra pressure drop caused by the plumber's improvisa-
tion.
12.25 An air turbine used to measure wind
velocities is shown in Fig. P12.25. The
turbine is connected to an electric
generator, and the energy generated is
measured and wasted. For wind velocity
of 30 m/s the turbine turns at 12 rpm.
Estimate the energy wasted.
12.26 The airplane considered in Example in
12.9 is now using takeoff strips that
permit ground speeds of 200 km/h. Find
the dimensions of the airplane wings
necessary under these new conditions.
12.27 The mean velocity of a fluid having properties like water is between 1 m/s
and 6 m/s. The fluid flows in a pipe with a diameter of 0.1 m. Estimate the
0.3 m
0.08 m
Hemispherical
shell
Wind
Figure P12.25
4
m
Figure P12.22 A truck.
12. Turbulent Flow 399
reading of the following flowrate measuring devices at the extreme veloci-
ties:
a. Venturi meter.
b. Flow nozzle.
c. Thin plate orifice.
12.28 Repeat Problem 12.27 for an air like fluid, with the limit mean velocities
10 m/s to 60 m/s.
12.29 Consider water flowing in a smooth pipe of 0.3 m diameter at the mean
velocity of 10 m/s. Calculate the pressure drop along 100 m of pipe. Now
assume the flow to be a boundary layer flow along the pipe walls and
calculate the pressure drop expected from the entrance to the pipe up to a
point of a 100 m downstream.
12.30 Consider the turbulent velocity profile in a pipe to be approximated by a
mid region of constant velocity and a boundary region, near the wall, where
the velocity change is strong. Use the integral approach to the boundary
layer solution and find the thickness of this region of strong velocity change
such that the correct pressure drop is obtained.
12.31 Consider Fig. 12.6 and Example 12.6. We are
now told that point A, where the pressure P
A
=
600 kPa has been measured, is located inside a
water tank, and the exit to the pipe is a perpen-
dicular square entrance, for which Table 12.2
yields k = 0.50. Point B is followed by a tee
connection, for which Table 12.2 suggests L
e
=
60, for the direction BC and for the direction
BD. Between B and C there is an additional
bend with L
e
= 20 and another bend of the
same kind between B and D. Find the flow
rates now.
12.32 Water is being discharged from a large tank
open to the atmosphere through a vertical tube,
as shown in Fig. P12.32. The tube is 10 m
long, 1 cm in diameter, has a roughness of
510
-5
m, and its inlet is sharp edged and
located 1 m below the level of the water in the
tank.
10 m
h = 0
h = 11 m
V
1
m
1
2
Figure P12.32 Water dis-
charge from a tank.
400 Fluid Mechanics
a. Find the velocity and the volumetric flowrate in the pipe.
b. Compare your answers to those of Example 7.5.
12.33 Consider Problem 12.32. A turbine is connected at the tube outlet, point 2.
Find the maximum power that can be obtained from the turbine.
401
13. COMPRESSIBLE FLOW
Speed of Infinitesimal Disturbances Sonic Speed
In all our considerations up to this point the density of the fluid was assumed
constant. Suppose we have a fluid continuum at rest and cause a small pressure
disturbance at one point in it. We may now ask how fast this disturbance travels in
the continuum. To answer this question, we may use measurements or try to
calculate the speed of the disturbance. Consistent with the previous assumption
that
= const , d = 0,
we find this speed of propagation to be infinite, i.e., the disturbance is felt at once
everywhere [see Eq. (13.5)]. This means, of course, that strict incompressibility
does not occur, though it may be a useful approximation. This also explains why
this question was not asked before, while treating incompressible flows. Once we
admit compressibility, the speed of propagation of disturbances becomes finite.
Then it also becomes an important dependent variable of the field of flow, which
we want now to calculate.
Consider the propagation of a disturbance in a continuum, as shown in
Fig. 13.1. The undisturbed part of the medium is separated from the disturbed
region by a surface of discontinuity, called the front of the disturbance. We
concentrate on a small but finite area of this front, which has the forward normal n
and the local velocity of propagation
u = nu.
Consider an observer sitting in a coordinate system, which moves with u.
From the observer's point of view the undisturbed medium is streaming into the
considered area of the disturbance front with the velocity ( u). As the material
passes the front, its properties change:
402 Fluid Mechanics
u into u + Au,
p into p + Ap,
into + A,
T into T + AT,
etc.
u
n
y
x
Observer's
coordinates
Control volume
Undisturbed medium
p , , T
Disturbance front
Disturbed
medium
+ A
T + A T
p + A p
u
Figure 13.1 Propagation of disturbance.
Now consider a small control volume in the shape of a stream tube cutting
through the disturbance front. Mass balance for the control volume requires
u = + A ( ) u + Au ( )
(13.1)
and the momentum theorem for the control volume states
( ) ( ) ( )( ) ( ). p u u p p u u u u + = +A + +A +A +A (
(13.2)
Substitution of Eq. (13.1) into Eq. (13.2) and rearrangement yield
u
2
=
Ap
A
u Au.
(13.3)
An infinitesimal disturbance is now defined as that which has the limits
Au du, Ap dp, A d, AT dT , ... .
The speed of propagation of this disturbance is denoted by c, and Eq. (13.3)
becomes
c =
dp
d
.
(13.4)
The process of passing through the disturbance is taken to be adiabatic, and
the undisturbed region is taken to be in thermodynamic equilibrium. Moreover, the
13. Compressible Flow 403
disturbance, which is infinitesimal, causes all thermodynamic properties in the
disturbed region to deviate from those in the undisturbed one only infinitesimally,
and therefore the disturbed region is also assumed to be in thermodynamic equi-
librium. The process of passing through the infinitesimal disturbance is thus taken
to be reversible, hence isentropic. To clearly show this, Eq. (13.4) is rewritten as
c =
cp
c
|
\
|
.
|
|
s
,
(13.5)
where c is also called the sonic speed, and measurements show it to be a very good
approximation to the speed of sound, i.e., to the speed with which sound waves
propagate. Equation (13.5) is quite general and applies to solids, liquids and gases,
whether ideal or not.
In what follows the considered fluid is air, assumed to obey the ideal gas law,
i.e.,
p = RT ,
(13.6)
with R = 287 J/kg K for air. In an isentropic process
p /
k
= const,
(13.7)
which together with Eq. (13.6) yields
c p
c
|
\
|
.
|
|
s
= k
p
= kRT ,
and for an ideal gas Eq. (13.5) becomes
c = k
p
= k RT .
(13.8)
Example 13.1
A large pressure vessel contains air (k = 1.4, R = 287 J/kg K) at 2 bars, 300 K.
A small nozzle is fitted into the wall of the vessel through which air expands isen-
tropically to the environment pressure of 1 bar. Find the speed of the air as it leaves
the nozzle. Is this speed above or below the speed of sound, i.e., is the flow there
subsonic or supersonic?
Solution
The first part of this problem is very similar to part b of Example 7.5. Since
the process is isentropic,
404 Fluid Mechanics
T
2
= T
1
p
2
p
1
|
\
|
.
|
k 1
k
= 300
1
2
|
\
|
.
|
0.4
1.4
= 246 K.
Denoting the enthalpy i and the specific heat of air c
p
= Rk /(k 1) = 1,005 J / kg,
we obtain
i
1
i
2
= c
p
T
1
T
2
( )
= 1,005 300 246
( )
= 54,270J/kg =
1
2
u
2
2
u
1
2
( )
,
u
1
= 0, u
2
= 329m/s.
Thus the speed of the air at the outlet of the nozzle is 329 m/s. Is this subsonic or
supersonic speed?
The speed of sound of the air in the vessel
is
c
1
= kRT
1
= 1.4 287 300 = 347m/s.
However, the speed of sound of the air as it emerges from the nozzle is
c
2
= kRT
2
= 1.4 287 246 = 314m/s .
Hence,
u
2
>
c
2
and the speed is supersonic.
Propagation of Finite Disturbances
The sonic speed, obtained for infinitesimal disturbances, has been uniquely
determined by Eq. (13.5). Is there such a general constraint on the speed of finite
disturbances moving in a compressible medium? An example of a finite distur-
bance is a solid body moving through the medium: a meteor, an airplane, a bullet, a
car, a speck of dust, a feather. Such a moving body influences certain regions in
the fluid by means of propagating disturbance fronts.
Consider a small body (e.g., a meteorite) moving in the x-direction at a
constant velocity v, Fig. 13.2. As the body moves, it disturbs the stationary envi-
ronment by pushing the air ahead of it, and the disturbance advances outward with
its front shaped as a spherical shell of radius
r = c t t
o
( ),
where t
o
is the time of emission and r is measured from the location of the body at
the moment of emission. Let the Mach number M be defined by
M =
v
c
.
13. Compressible Flow 405
c
|
v
M >
1
v > c
M =
1
v
=
c
v
c
M <
1
v
< c
c
v
Figure 13.2 Subsonic, sonic and supersonic motions.
In subsonic motion M < 1, i.e., v < c, and the front of the disturbance, which
advances with the speed c, precedes the moving body itself, Fig. 13.2. The front of
the disturbance may be detected by some pressure-sensitive instrument, e.g., the
human ear, and someone hearing that noise may look up and await the approach of
the moving body. The front of the disturbance is called a Mach surface.
Now let M = 1. The body moves in the x-direction just as fast as the distur-
bance front. The front does not precede the body in the x-direction, and the region
to the right of the body, Fig. 13.2, is known as the zone of silence. The moving
body is not heard in the zone of silence, and no pressure rise precedes it. Once
heard, the listener is surprised by the appearance of the body itself.
In supersonic motion M > 1, i.e., v > c, and the front of the moving body
moves faster than the disturbance. It pulls the front of the disturbance forward in
the direction of its own motion and the disturbance front attains the form of a cone
of half apex angle
| = arcsin
1
M
|
\
|
.
(13.9)
which moves forward with the speed v along its axis of symmetry. The cone itself
is called the Mach cone, and it represents a conical front, which propagates with
the speed c normal to itself. The region in front of the cone is a zone of silence, and
once the moving body is heard, the listener must look rather forward, in the
direction of its motion, to see the retreating body somewhat from behind.
If the moving body is a long thin cylinder perpendicular to the plane of
406 Fluid Mechanics
Fig. 13.2, the phenomenon becomes two dimensional, with the Mach cone
changing to a Mach wedge, having the same half apex angle of Eq. (13.9).
Example 13.2
A jet airplane flies with the speed of 400 m/s in air whose temperature is
280 K. What is the half apex angle of its Mach cone? The airplane flies at the
height of h = 1 km. A man on the ground hears the airplane. How far ahead of the
man is the airplane at that moment?
Solution
c = kRT = 1.4 287 280 = 335.4m/ s,
M =
v
c
=
400
335.4
= 1.19, | = arcsin 1/ M ( )= 57
o
= 1rad.
Assuming the plane flies parallel to the ground,
Ax = h tan | = 1000 tan 1 = 1550 m ahead.
The Piston Analogy
The description of the motion of supersonic disturbances, as just presented,
still leaves the details of what happens just in front of the body for M 1 unclear.
c
c
v
c
c
1
3
4
2
Expansion front Compression front
Figure 13.3 Piston and disturbances.
To clarify this point let us consider a simpler, one-dimensional case. Let the
moving body be an insulated frictionless piston moving in an infinitely long insu-
lated cylinder filled with an ideal gas, Fig. 13.3. The gas is initially at rest, and at
the time t
o
= 0 the piston is suddenly set in motion at the constant speed v. The gas
13. Compressible Flow 407
in the direction of the piston motion is compressed, with the disturbance running to
the right as a continuous series of successive compression fronts. The gas on the
reverse side of the piston expands, with the disturbances moving to the left as
successive expansion fronts. All these fronts represent Mach surfaces.
Consider two compression fronts, 1 and 2, where the second front started
later than the first one. The first moves at the speed
c
1
= kRT
and leaves behind it the temperature , T dT + and because the compression is
assumed isentropic,
dT > 0.
The second front moves into the already disturbed gas and therefore has the speed
c
2
= kR T + dT ( ) > c
1
.
The second front overtakes the first front and coalesces with it. No matter how
many fronts we have on the compression side, those closer to the piston run faster
and overtake those ahead of them, which eventually overtake the first one. The
disturbances are concentrated into one front of finite magnitude, called a Shock
Wave. The speed of the shock wave is no longer given by Eq. (13.5), but rather by
Eq. (13.3). From the point of view of Fig. 13.1, Au is negative, and u Au > 0.
Furthermore, the process is no longer isentropic, but is still adiabatic, and therefore
requires A s > 0, which for compression results in
Ap
A
>
cp
c
|
\
|
.
|
|
s
.
Equation (13.3) then yields u > c. The shock wave moves at speeds higher than the
sonic speed.
Now consider the expansion fronts on the left side of the piston, Fig. 13.3.
The first one moves at the speed
c
3
= kRT ,
leaving behind it the temperature T + dT, and because the expansion is assumed
isentropic
dT < 0.
The second front moves at the speed
c
4
= kR T + dT ( ) < c
3
,
and it therefore lags behind the first front. No matter how many expansion fronts
we have, no two of them coalesce, and no amplification or concentration of
408 Fluid Mechanics
disturbances occur. A finite disturbance of the expansion type simply cannot be
created.
Going now back to the case M > 1 in Fig. 13.2, we see that the moving body
carries in front of it a shock wave, which also moves supersonically at the speed v.
Quasi-one-dimensional Flow, Stagnation Properties
A flow that satisfies the following two requirements is called quasi-one-
dimensional:
1. The divergence and convergence of all stream tubes in the flow are small, in
the sense that the velocity component along the tube, u, and the other velocity
components, v and w, which are perpendicular to u, satisfy
v
u
<<1,
w
u
<< 1.
2. The ratio of the radius of curvature of the stream tube to the linear dimension
of the tube cross section, b, measured along the radius of curvature, is much
larger than 1.
A quasi-one-dimensional-flow is shown in Fig. 13.4. In quasi-one-dimen-
sional flows the x-axis points in the local direction of the flow. Conservation of
mass now becomes
const uAM = =
(13.10)
with A the cross-sectional area normal to u. Differentiation of Eq. (13.10) yields
d
+
du
u
+
dA
A
= 0,
1
d
dx
+
1
u
du
dx
+
1
A
dA
dx
= 0.
(13.11)
We consider here fast flows, with rather large Reynolds numbers, and we
assume the conditions leading to the Euler equation (10.1) to be satisfied. Thus the
momentum equation has the Euler form, Eq. (10.1), which for negligible body
forces and for quasi-one-dimensional flows becomes
u
du
dx
=
dp
dx
or u du = dp.
(13.12)
Because the flow is fast, it can be assumed adiabatic. Heat transfer may still
be considered, say into the wall of a nozzle. However, for the fast moving gas in
that same nozzle the process is still adiabatic. The apparent contradiction here is
resolved by the observation that a negligible heat interaction per unit mass of
flowing gas may amount to a large heat interaction when multiplied by a very large
mass flux. Thus, except for some singular locations in the field, the flow may be
13. Compressible Flow 409
taken as adiabatic. Moreover, because the flow in the considered region has also
been assumed frictionless, it is now both adiabatic and frictionless, i.e., isentropic.
y
x
u
b
Radius of
curvature, R
b
R
<< 1
Figure 13.4 A stream tube in quasi-one-dimensional flow.
For an isentropic process the momentum equation can be integrated to yield
an energy equation, and indeed Eq. (13.12) can be integrated to
1
2
u
2
+
dp
p
o
p
}
= const
which is just a particular form of the Bernoulli equation, Eq. (7.7). Now, for a
thermodynamic system undergoing a frictionless process the first law requires
oQ = di
1
dp,
where Q is the heat per unit mass and i is the specific enthalpy. For an adiabatic
process this becomes di = 1/ dp, and with Eq. (13.12) it takes the form
i +
1
2
u
2
= i
o
= const,
(13.13)
where i
o
is denoted the stagnation enthalpy. It is the value of the enthalpy, which
may be obtained by stopping the flow adiabatically. The more useful form of the
energy equation is that of Eq. (13.13).
We now summarize the system of equations governing quasi-one-dimen-
sional frictionless flows:
The energy equation
i +
1
2
u
2
= i
o
= const,
(13.13)
the equation of state
410 Fluid Mechanics
p = RT ,
(13.6)
i + c
p
T = const
(13.14)
and the equation for an isentropic process
p
k
= const.
(13.7)
We note that the last three equations hold for an ideal gas only, while Eq. (13.13) is
general.
The stagnation properties are those that can be obtained for a point in the
flow by an isentropic process that ends in stagnation, i.e., in u = 0. The process
need not be physically performed and the stagnation properties may be simply
calculated. These stagnation properties "belong" to the point at which the process
starts. However, as long as the flow is isentropic and contains no singular points,
the same stagnation properties apply to all points in the flow.
It is convenient to render most relations for the flow dimensionless, and a
natural characteristic velocity here is the sonic speed. Let Eq. (13.13) be written
again using Eq. (13.14),
c
p
T +
1
2
u
2
= c
p
T
o
.
(13.15)
Using Eq. (13.8) and noting
c
p
= kR k 1 ( ), Eq. (13.15) becomes
T
o
T
= 1+
1
2c
p
u
2
T
= 1+
k 1
2
u
2
kRT
or
T
o
T
= 1+
k 1
2
M
2
.
(13.16)
Using Eq. (13. 7) and noting that T
2
T
1
= p
2
p
1
( )
k1
k , Eq. (13.16) becomes
p
o
p
= 1+
k 1
2
M
2
(
(
k
k 1
,
(13.17)
and
=
p
o
p
|
\
|
.
|
|
1
k
= 1+
k 1
2
M
2
(
(
1
k1
.
(13.18)
Obviously, for an isentropic process
s
o
s
= 1.
13. Compressible Flow 411
Example 13.3
Point B is in a compressible flow field of air (ideal gas, R = 287 J/kg
.
K,
k=1.4). It is also known that
p
B
= 210
5
Pa, T
B
= 350 K, U
B
= 350 m/s.
Find the stagnation properties p
Bo
, T
Bo
,
Bo
, s
Bo
of point B.
The streamline through B eventually passes through point D, but between B
and D it passes through a shock wave (an irreversible adiabatic process). Which of
the stagnation properties of B changes and in which direction?
Solution
B
=
p
B
RT
B
=
2 10
5
287 350
= 2.0kg/ m
3
,
c
B
= kRT
B
= 375m/ s, M
B
=
u
B
c
B
=
350
375
= 0.933.
For an isentropic process, s
Bo
= s
B
.
From Eqs. (13.17), (13.18) and (13.16), respectively,
p
Bo
= p
B
1+
k 1
2
M
B
2
(
(
k
k 1
= 210
5
1.754 = 3.5110
5
Pa,
Bo
=
B
1+
k 1
2
M
B
2
(
(
l
k1
2.0 1.494 = 2.988kg/ m
3
,
T
Bo
= T
B
1+
k 1
2
M
B
2
(
(
= 3501.174= 411K.
After the shock wave:
i
Do
= i
Bo
,
T
Do
= T
Bo
,
s
Do
> s
Bo
= s
B
,
p
Do
< p
Bo
(because T
Do
= T
Bo
and s
Do
> s
Bo
),
Do
<
Bo
(because T
Do
= T
Bo
and s
Do
> s
Bo
).
412 Fluid Mechanics
Nozzle Flow
Consider a stream tube in quasi-one-dimensional flow, Fig. 13.4. Let du from
Eq. (13.12) be substituted in Eq. (13.11) and this equation rearranged to
A
u
2
dp
dx
u
2
dp / d
1
(
(
+
dA
dx
= 0
or
dA
dx
=
dp
dx
M
2
1
( )
A
u
2
=
du
dx
M
2
1
( )
A
u
.
(13.19)
Thus for subsonic flow, with M < 1, a converging stream tube, with dA/dx < 0,
corresponds to expansion, with
dp / dx < 0
, and to increase in velocity, with
du / dx > 0
. For supersonic flow, however, the behavior of dp/dx and du/dx is the
opposite, and the various combinations are listed in Table 13.1.
M dA/dx dp/dx du/dx
_______________________________________________
M < 1 > 0 > 0 < 0
M < 1 < 0 < 0 > 0
_______________________________________________
M > 1 > 0 < 0 > 0
M < 1 < 0 > 0 < 0
_______________________________________________
M = 1 0 > 0 > 0
< 0 < 0
Table 13.1 Convergingdiverging stream tube.
It is noted that
dA / dx = 0
is a necessary condition for M = 1. Sonic speed can
appear only where the tube has a local minimum, or a so-called throat. However
dA / dx = du / dx = 0
may also occur at that location, i.e.,
dA / dx = 0
is not a
sufficient condition for M = 1.
At this stage we consider a flow in a nozzle. A nozzle is a conduit of rigid
walls satisfying the conditions for quasi-one-dimensional flow. This nozzle is
connected to a vessel where a gas at the stagnation pressure p
o
and the stagnation
temperature T
o
is assumed to be supplied indefinitely. The nozzle's exit is also
assumed to be kept at a constant pressure p
e
. The nozzle is assumed to have a
throat, as shown in Fig. 13.5. The rigid walls of the nozzle are considered to be the
stream tube envelope, and we also assume no separation of the flow from the
nozzle wall. We want to obtain some relations for the design and performance of
such nozzles.
13. Compressible Flow 413
exit
o c e
stagnation
p
o
T
o
e
e
p T
critical
Figure 13.5 A diagram of a convergentdivergent nozzle.
Example 13.4
Show that the sonic speed can appear where the nozzle has a local minimum,
i.e., a "throat," and cannot appear where the nozzle has a local maximum.
Solution
We refer to Table 13.1: For a local minimum the flow toward the location
where dA/dx = 0 is in a convergent part of the nozzle. Thus if the approaching flow
is subsonic it accelerates and may become sonic. If the approaching flow is
supersonic, it decelerates and may become sonic.
On the other hand, where the nozzle has a local maximum, the approach
stream is divergent. Thus a subsonic stream would slow down, and a supersonic
stream would speed up, and neither can reach sonic speed.
Mass Flux through the Nozzle
Suppose the sonic speed is attained in the nozzle. This can happen only at the
throat c. We then call this section "critical" and denote by asterisks the properties
corresponding to it. If the sonic speed is not realized at the throat, section c, then
this section is not called critical. Substitution of M = 1 in Eqs. (13.16), (13.17) and
(13.18) yields
T
*
T
o
=
1
1+ k 1 ( ) 2
=
2
k +1
= 0.83333,
(13.20)
414 Fluid Mechanics
p
*
p
o
=
2
k +1
|
\
|
.
|
k
k 1
= 0.52828,
(13.21)
o
=
2
k + 1
|
\
|
.
|
1
k 1
= 0.63394,
(13.22)
where the computed numerical values are for k = 1.4. We also note that because
Eqs. (13.16) - (13.18) have been obtained for isentropic processes, such processes
are also implied for Eqs. (13.20) - (13.22).
The mass flow through the nozzle is
m = uA =
*
u
*
A
*
=
*
c
*
A =
o
o
c
o
T
*
T
o
|
\
|
.
|
1
2
A
*
.
And for
o
= p
o
/ RT
o
, c
o
= kRT
o
,
o
c
o
= p
o
k
RT
o
,
this becomes
m = p
o
A
*
k
RT
o
2
k + 1
|
\
|
.
|
1
k 1
+
1
2
= p
o
A
*
k
RT
o
2
k + 1
|
\
|
.
|
k+1
2 k 2
= 0.6847
p
o
A
*
RT
o
.
(13.23)
The mass flux is completely determined by the stagnation pressure and
temperature and by the critical cross section A
*
. Since the same mass flows
through all sections
uA =
*
u
*
A
*
=
*
c
*
A
*
and
A
A
*
=
*
c
*
u
=
*
c
*
c
c
u
=
*
T
*
T
|
\
|
.
|
1
2
1
M
.
(13.24)
Using Eqs. (13.16) - (13.18) and (13.20) - (13.22), we obtain
p
*
p
=
p
*
p
o
p
o
p
=
2
k + 1
|
\
|
.
| 1+
k 1
2
M
2
|
\
|
.
|
(
(
(
k
k1
,
(13.25)
=
*
=
2
k +1
|
\
|
.
|
1+
k 1
2
M
2
|
\
|
.
|
(
(
(
1
k 1
,
(13.26)
13. Compressible Flow 415
T
*
T
=
T
*
T
o
T
o
T
=
2
k +1
|
\
|
.
| 1+
k 1
2
M
2
|
\
|
.
| .
(13.27)
Equations (13.26) and (13.27) are substituted into Eq. (13.24), which becomes
A
A
*
=
1
M
2
k +1
|
\
|
.
|
1+
k 1
2
M
2
|
\
|
.
|
(
(
k+1
2(k1)
.
(13.28)
Using the Mach number M as a parameter, p, , T and A are determined every-
where in terms of the critical properties p
*
,
*
, T
*
and A
*
.
Example 13.5
Show that the mass flux per unit cross-sectional area is the highest at the
critical section.
Solution
Because the same mass flux passes through all sections,
m = uA =
*
u
*
A
*
= const.
Thus
u = m / A,
and u has a maximum where A has a minimum. Now a
minimum for A can be only where A has a local minimum. If the nozzle has only
one throat and sonic speed is attained, this throat is both the critical one and the
only one with a local minimum for its area; the proof is then complete.
Consider now a nozzle, which has several throats and achieves sonic speed.
The mass flux per unit cross-sectional area, u, still has a maximum where A is the
smallest. Hence, of all the throats the smallest one achieves the highest value for
u. Let the critical section, i.e., the one where sonic conditions appear and which
must be one of the throats, be A
*
. Let any other section, throat or not throat, be A.
Then Eq. (13.28) gives
A
A
*
= f M ( ) =
1
M
2
k +1
|
\
|
.
| 1+
k 1
2
M
2
|
\
|
.
|
(
(
(
k +1
2( k1)
or
lnf = lnM +
k +1
2(k 1)
ln 1+
k 1
2
M
2
|
\
|
.
+ ln
2
k + 1
|
\
|
.
(
(
.
Differentiation with respect to M yields
416 Fluid Mechanics
1
f
df
dM
= 0=
1
M
+
k +1
2(k 1)
k 1 ( )M
1+ k 1
( )
2
| |
M
2
=
1
M
+
(k + 1)M
2+ (k 1)M
2
,
with the solution M = 1. To ensure that the function f(M) = A/A
*
has a true mini-
mum at f (1) = 1, i.e., that the critical section is the smallest one, and the proof is
complete, the reader may take the second derivative and find it positive.
The Design and Performance of a Nozzle
We return now to the nozzle in Fig. 13.5. Suppose its critical section and the
mass flux are already known. Still conditions at its exit section e must be consid-
ered. Two different engineering situations may arise:
a. A design problem: Given p
e
, find A
e
.
Using p
*
/p
e
in Eq. (13.25), M
e
is obtained and substituted in Eq. (13.28) to
yield A
e
/A
*
.
b. A performance problem: Given A
e
, find the corresponding p
e
.
Using A
e
/A
*
in Eq. (13.28), two M
e
values are obtained, and with them two p
e
values.
This two-valued situation may be seen at once from the structure of
Eq. (13.28) with its minimum (see Example 13.5). It could also, however, be
inferred from Table 13.1.
The flow at c, Fig. 13.5, is assumed to be at M = 1. Suppose just downstream
from c the flow becomes slightly subsonic, i.e., M < 1. Since the nozzle diverges,
the flow would slow down further, ending with M
e
<1. Conversely, if the flow just
downstream from c becomes even slightly supersonic, it would continue to accel-
erate because the nozzle diverges.
These two M
e
values, when used in Eq. (13.25), yield two p
e
values. Thus the
same nozzle corresponds to two p
e
values, and the flow in its diverging part,
whether subsonic or supersonic, is determined by the pressure p
e
actually imposed
at the exit e.
The entrance section, o, which is assumed at stagnation, evades similar treat-
ment. Because of M
o
~ 0, Eq. (13.28) yields A
o
, and because of u ~ 0,
Eq. (13.19) yields dA
o
/dx . The nozzle should start, therefore, with its walls
tangent to the stagnation vessel, as shown in Fig. 13.5.
Let a scheme of the convergentdivergent nozzle be drawn again, with the
pressure ratio p / p
o
also shown, Fig. 13.6. The two exit pressures corresponding to
sonic speed at the throat are shown as point C for the subsonic exit and point D for
the supersonic one. Point A corresponds to no flow at all, while point B represents
a flow which is subsonic everywhere.
13. Compressible Flow 417
We find that all exit pressures between points A and C correspond to
subsonic flows, while point D corresponds to a supersonic flow in the diverging
part of the nozzle. Yet the nozzle can be made to discharge into a vessel where the
pressure is controlled. What then happens for exit pressures below point D? Or
between C and D?
A
B
C
S
1.0
0.52828
D
p
p
o
x
Figure 13.6 Pressure along a nozzle.
Below p
D
the flow inside the nozzle is the same as for p
D
The information
that the pressure in the discharge vessel has dropped below p
D
is expressed in the
form of an expansion wave and therefore cannot propagate against the supersonic
flow into the nozzle. An expansion wave, as already shown in this chapter,
propagates at the sonic speed. Hence the exit is a "zone of silence" with respect to
the diverging part of the nozzle.
Coming out of the nozzle the gas encounters a zone of lower pressure and it
expands further. However, this takes place outside the nozzle and the flow is free
to choose its envelope for the streamlines. The phenomenon then becomes non-
quasi-one-dimensional.
Between p
C
and p
D
the situation is more complex, because, assuming a flow
down to p
D
to have been established, the information that the pressure is above p
D
is in the form of a compression wave. It can build up to a shock wave, which
418 Fluid Mechanics
moves at a supersonic speed, and thus can propagate upstream into the nozzle.
Indeed this is what happens, and some additional information on the behavior of
shock waves is now needed in order to explain what really takes place.
Example 13.6
A large pressure vessel contains air at the stagnation pressure p
o
= 410
5
Pa
and the stagnation temperature T
o
= 350 K. The atmospheric pressure is
p
a
= 10
5
Pa. Design a convergentdivergent nozzle that discharges 1.5 kg/s air to
the atmosphere at atmospheric pressure. Find the speed of the discharged air.
Solution
From Eq. (13.23),
m =1.5=
0.6847 410
5
A
*
287 350
.
The critical section is
A
*
= 1.736 10
-3
m
2
,
D
*
=
4
t
A
*
= 4.70 10
2
m= 47mm.
From Eq. (13.17)
p
o
p
e
= 4= 1+
k 1
2
M
e
2
(
(
k
k 1
.
Hence M
e
= 1.56, and from Eq. (13.28)
A
e
= A
*
1
M
e
2
k + 1
|
\
|
.
| 1+
k 1
2
M
e
2
|
\
|
.
|
(
(
(
k +1
2(k1)
= 1.73610
3
1
1.56
2
2.4
|
\
|
.
| 1+ 0.2 1.56
2
( )
(
(
3
= 2.116 10
3
m
2
,
D
e
=
4A
e
t
= 5.19 10
2
m= 51.9mm.
From Eq. (13.16)
13. Compressible Flow 419
T
e
=
T
o
1+
k 1
2
M
e
2
(
(
= 235K,
c
e
= kRT
e
= 1.4 287 235 = 307.3m/ s,
u
e
= c
e
M
e
= 480 m/s.
Half the cross section of the designed nozzle is shown in Fig. 13.7. The radius
of curvature of the converging part has been chosen equal to D
*
, and half the
spread angle of the diverging part is chosen to be 8
o
to guarantee no flow
separation. The reason for this is beyond the scope of this presentation.
o c
e
Figure 13.7 Supersonic convergentdivergent nozzle.
Example 13.7
The pressure vessel of Example 13.6 is connected to another vessel at
p
B
= 310
5
Pa, and a nozzle is used to transfer 1.5 kg / s m= of air from the first
vessel to the second one. Design this nozzle.
Solution
We may still utilize the concept of the critical section although in this case
this section is not built but rather serves as an auxiliary computational step. Thus
Eq. (13.23) yields (see Example 13.6)
A
*
= 1.736 10
3
m
2
.
From Eq. (13.17) we obtain
p
o
p
e
=
4
3
= 1+
k 1
2
M
e
2
(
(
k
k 1
.
Hence M
e
= 0.654. Equation (13.28) now gives
420 Fluid Mechanics
A
e
=
A
*
M
e
2
k +1
1+
k 1
2
M
e
2
|
\
|
.
|
(
(
(
k +1
2( k1)
=
1.736 10
3
0.654
2
2.4
1+ 0.2 0.654
2
( )
(
(
3
= 1.97 10
3
m
2
,
D
e
=
4A
e
t
= 5 10
2
m = 50.0mm.
Note that this nozzle can be obtained from that designed in Example 13.6, which is
shown in Fig. 13.7, by cutting it off at D = 50.0 mm, before the critical section.
If the nozzle of Example 13.6 is cut at D = 50.0 mm but after the critical
section, it still satisfies the conditions of Example 13.7 .
Tables of One-dimensional Compressible Flow
Equations (13.16), (13.17), (13.18), (13.23) and (13.28) may be used for vari-
ous flow conditions, and the results tabulated with the Mach Number used as a
parameter. Such tables may be used to obtain a quick estimate of the principal
dimensions of a nozzle. Examples of such tables are Table D-1, Table D -2 and
Table D -3 for ideal gases with k = 1.4, k = 1.667 and k = 1.3, respectively.
Example 13.8
Repeat Examples 13.6 and 13.7 for helium (R = 2079.7 J/kg
.
K, k = 1.667)
using the compressible flow tables.
Solution
For Example 13.6, from Table D-2,
m =
0.7262 4 10
5
( )
A
*
2079.7 350
or
1.5 = 340.5 A
*
, A
*
= 4.40610
-3
m
2
and
D
*
= 7.4910
-2
m = 74.9 mm,
13. Compressible Flow 421
p
e
p
o
=
1
4
= 0.25, hence from Table D-2 , M
e
= 1.49,
o
= 0.4357,
T
e
T
o
= 0.5747,
A
e
A
*
= 1.1430.
Hence
T
e
= T
o
T
e
T
o
= 350 0.5747 = 201K,
c
e
= kRT
e
= 1.667 2079.7 201 = 835m/ s,
u
e
= M
e
c
e
= 1244m/ s,
A
e
= A
*
A
e
A
*
= 5.0410
3
m
2
, D
e
= 8.01 10
2
m= 80.1mm.
For Example 13.7
p
e
p
o
=
3
4
= 0.75.
Hence from Table D-2, M
e
= 0.605 and
A
e
A
*
= 1.1706, A
e
= 1.1706 4.406 10
3
( )
= 5.1610
3
m
2
,
D
e
= 8.1010
-2
m = 81.0 mm.
Nonisentropic One-dimensional Flows
To gain some insight into the physics of shock waves as well as to extend the
treatment to include additional important and frequently realized flows, we widen
our scope and consider some nonisentropic flows. We do not want, however, to
extend our treatment too much. Therefore, having relaxed one condition, that of
the flows being isentropic, we impose one condition: In this treatment the flows
are not just quasi-one-dimensional, but must be really one dimensional, i.e., the
continuity equations (13.10), (13.11) now become
const m u = = (13.29)
d
+
du
u
= 0.
(13.30)
We know of two phenomena, friction and heat transfer, which make a process
nonisentropic.
422 Fluid Mechanics
Pipe Flow with Friction
The first family of nonisentropic flows we consider is adiabatic flows with
friction. The details of the mechanism of friction are ignored, and we admit its
effect by not using the momentum equation (13.12) in our considerations. Because
the flow is adiabatic, the energy equation in its form (13.13) or (13.14) still holds,
and so do the ideal gas relations.
Thus, from Eq. (13.15)
c
p
dT + u du =
k
k 1
R dT + u
2
du
u
= 0
or with Eq. (13.30)
R
k 1
dT
T
u
2
kRT
R
d
= c
v
dT
T
M
2
R
d
= 0.
(13.31)
From basic thermodynamics for an ideal gas
ds = c
v
dT
T
R
d
,
which together with Eq. (13.31) yields the forms
ds = M
2
1
( )
R
d
= 1 M
2
( )
R
du
u
= 1
1
M
2
|
\
|
.
| c
v
dT
T
.
(13.32)
Since in an adiabatic process ds 0, so must be (1 M
2
)du/u c 0. Thus for super-
sonic flows, M
2
> 1, du < 0, and the effect of friction is to slow the flow, while for
subsonic flows, M
2
< 1, du > 0, and the flow must accelerate.
Repeating this same argument for the sign of
1
1
M
2
|
\
|
.
| c
v
dT
T
,
we find that for supersonic flows the gas temperature goes up along the conduit,
while for subsonic flows the effect of friction is to reduce the gas temperature.
Let the x direction coincide with the direction of the flow in the pipe. The
results just obtained are summarized in Table 13.2, which is somewhat analogous
to Table 13.1. It is also noted that since friction, and hence any real adiabatic
process, requires ds > 0, no time-independent flow with M = 1 may exist in a pipe,
Eq. (13.32). Even an assumption of negligible friction still does not allow M = 1
inside the conduit, because M = 1 corresponds to ds = 0 and not to ds ~ 0.
13. Compressible Flow 423
M
du
dx
dT
dx
d
dx
M< 1 > 0 < 0 < 0
M> 1 < 0 > 0 > 0
M= 1
no time-independent flows
Table 13.2 Adiabatic flow with friction in a pipe.
Equation (13.32) defines a curve on the T - s diagram, having a subsonic
branch and a supersonic one, Fig. 13.8, separated by the point M = 1 which corre-
sponds to ds = 0. This curve is called the Fanno Line .
M = 1
M > 1
Fanno line
Rayleigh line
T
s
M = 1
M < 1
A
B
M < 1
M > 1
Figure 13.8 Fanno and Rayleigh lines.
A flow in a conduit of constant cross section corresponds to a particular
Fanno line and a point in the flow corresponds to a point on that line. As a fluid
particle proceeds on its way in the conduit, its point on the diagram moves along
its Fanno line and always toward larger s. At M = 1 no further motion along the
Fanno line toward larger s is possible. Hence M = 1 cannot appear in the conduit,
but only at its end.
Suppose M = 1 does appear at the exit of a conduit. What is the velocity at
that point or what is the sonic speed there? We note that Eq. (13.16) has been
obtained from the energy equation (13.15) and the expression (13.8) for the sonic
speed, both of which hold for adiabatic flows. Thus putting M = 1 in Eq. (13.16)
we obtain
424 Fluid Mechanics
T
T
o
=
2
k +1
,
which is the same as Eq. (13.20) for isentropic flow! The temperature, the velocity
and the sonic speed obtained at M = 1 by adiabatic expansion from stagnation do
not depend on whether the expansion is isentropic or adiabatic with friction. The
pressure and the density and, of course, the entropy come out to be different there.
Still this means that in designs of convergentdivergent nozzles, where sonic
speeds do appear, rough designs of the convergent part yield surprisingly good
results.
Let an initial point in the conduit be chosen, e.g., at its entrance. Let the gas
properties there be
i
, u
i
, T
i
, p
i
, s
i
and note: u
i
there is not zero. Let the stagnation
properties corresponding to the initial point be
o
, T
o
, p
o
, s
o
, with s
o
= s
i
. Now,
again, from basic thermodynamics
ds = c
v
dT
T
R
d
= c
v
dT
T
+ R
du
u
.
Equation (13.15), which holds for adiabatic processes, states
c
p
T
o
= c
p
T +
1
2
u
2
, u
2
= 2c
p
(T
o
T ).
Hence
c
p
dT + udu = 0,
du
u
= c
p
dT
u
2
=
dT
2(T
o
T )
and
ds = R
1
k 1
1
T
1
2(T
o
T )
(
(
(
dT .
(13.33)
The coefficient of friction in pipe flow, as defined in Eq. (12.29), yields the head
loss
dh
f
= f
u
2
2g
|
\
|
.
|
dL
D
,
which affects the change of entropy
ds = g
dh
f
T
= f
u
2
2T
dL
D
= f
c
p
D
T
o
T
T
(
( dL .
(13.34)
Equation (13.33) combined with Eq. (13.34) yield
f k
dL
D
=
k +1
2
1
1 q
k 1
2
1
(1 q )
2
(
(
(
dq,
13. Compressible Flow 425
where
q =
T
T
o
=
1
1+
k 1
2
M
2
(
.
Integration yields the Fanno number Fn = f L/D,
Fn = f
L
D
=
k + 1
2k
ln
1 q
1 q
i
k 1
2k
1
1 q
1
1 q
i
|
\
|
.
|
|
,
(13.35)
which describes how q = T/ T
o
changes along the conduit. Of course, once T is
known, and u may also be computed. Substitution of q = T*/ T
o
= 2/(k+1) in Eq.
(13.35) yields the maximum length the conduit may have, i.e., that length after
which sonic speed appears. Equation (13.35) then takes the form
L
D
|
\
|
.
max
=
1
2f
k +1
2k
ln
k 1
(k +1)(1 q
i
)
k 1
2k
k + 1
k 1
1
1 q
i
|
\
|
.
|
(
(
with
q
i
= 1+
k 1
2
M
i
2
(
(
1
.
Some numerical values are presented in Tables D-4 , D-5 and D-6 in Appendix D
arranged such that Fn = f L/D
is the dimensionless length still remaining before M
= 1 is reached.
Example 13.9
A certain cooking gas, which may be considered an ideal gas with k = 1.3, is a
mixture of CO, C
2
H
6
, CH
4
and N
2
. It is introduced into a pipe of 0.05 m diameter
and flows at this point with M = 0.01. The Moody diagram, Fig. 12.4, yields for
this pipe the friction factor f = 0.02, which is assumed constant. The pipe is to be
15 km long.
Find the maximum length this pipe may have before the flow is choked. Find
the Mach number at the 15-km point and the values of T
2
/ T
01
, p
2
/ p
01
and
2
/
01
there.
Solution
At the point of introduction the gas has
M
1
= 0.01, T
1
/ T
01
= 1.0, p
1
/ p
01
= 1.000,
1
/
01
= 1.000,
426 Fluid Mechanics
where the various values have been read from Table D-1, which holds for isen-
tropic relations.
Table D-6 yields, for M = 0.01,
f
L
max
1
D
|
\
|
.
|
|
= 7683.
Hence L
max
1
= 7683 D f =19182m.
At L
2
= 15,000 m: L
max
2
= 19,182 L
2
= 4,182m and f
L
max
2
D
= 1,673.
Table D-4 yields, by interpolation, M
2
~ 0.022, and certainly 0.02< M
2
<0.03.
The table also yields T
2
/T
o
= 0.9999. The sonic speed at point 2 is the same as at
point 1, and therefore
u
2
u
1
=
M
2
M
1
= 2.2=
1
2
,
2
=
1
2.2
,
2
01
=
1
01
|
\
|
.
|
|
1
2.2
= 0.4545,
p
2
p
01
=
p
1
p
01
|
\
|
.
|
|
2
T
2
1
T
1
= 0.9999 0.454511.0 = 0.4545.
Frictionless Pipe Flow with Heat Transfer
The second family of nonisentropic flows we consider is frictionless flows
with heat transfer. Using the same strategy as for adiabatic flows with friction, we
keep the one-dimensional continuity equations (13.29) and (13.30) and the
momentum equation (13.12) and ignore the energy equation.
Substitution of du = u d from Eq. (13.30) into Eq. (13.12) yields
u
2
d = dp = R dT + T d ( ).
Hence,
u
2
RT
d
=
dT
T
+
d
,
and because RT = c
2
/k, this becomes
( )
2
1 .
d dT
kM
T
= (13.36)
13. Compressible Flow 427
From general thermodynamics
ds = c
v
dT
T
R
d
,
and from Eq. (13.36),
ds =
M
2
1
kM
2
1
c
p
dT
T
.
(13.37)
Again, a relation between T and s has been obtained, which can be described
as a line on the T - s diagram, the Rayleigh Line, Fig. 13.8.
The frictionless motion of a fluid particle along the conduit, with heat trans-
fer, is represented by points on this line. Since heat transfer can be into and out of
the moving fluid, the points on the line in the diagram move both to the left and to
the right.
The Rayleigh line also breaks into subsonic and supersonic branches, with
M = 1 at the point of maximum s. It has a point of maximum T at M
2
= 1/k, and for
k < M
2
< 1
the temperature goes down with the addition of heat and increases with
heat extraction. The general behavior is presented in Table 13.3, which is analo-
gous to Table 13.2 for the Fanno line.
m dT du
M
2
< > 0 > 0
M
2
> 1 > 0 < 0
M dT du
< 0
< 0
> 0
M
2
> 1 < 0 > 0
M
2
<
1
k
<
2
< 1
1
k
1
k
M
With heat addition
With heat extraction
Table 13.3 Frictionless flow with heat transfer in a pipe.
428 Fluid Mechanics
We now endeavor to explain shock waves using the concepts just developed,
i.e., those represented by the Fanno line and by the Rayleigh line.
Let a shock wave be defined as a local singularity, i.e., a local nonisentropic
occurrence. Then:
- It may be considered locally one dimensional.
- It must permit local heat transfer and internal friction.
Let the gas just before the shock wave be represented by a point on the T-s
diagram, say, point A in Fig. 13.8. Both Fanno and Rayleigh lines may be drawn
through this point. The shock is a one-dimensional phenomenon and the state after
the shock is restricted to its Fanno and Rayleigh lines; hence the shock must be
represented by a jump from point A to point B or from B to A. However, s
B
> s
A
.
Hence a shock can occur only at supersonic speeds.
Shock Wave Relations
Consider an observer moving with a normal
shock, Fig. 13.9. The observer sees the shock
stationary, with the flow going into and out of the
shock. A momentum balance through a stream
tube yields
1
u
1
2
+ p
1
=
2
u
2
2
+ p
2
=
1
M
1
2
c
1
2
+ p
1
= M
1
2
1
k
p
1
1
+ p
1
or
( ) ( )
2 2
1 1 2 2
1 1 . p kM p kM + = + (13.38)
Equation (13.16) applies both before and after the shock, and because i
o
and with it
T
o
are conserved through the shock,
T
1
1+
k 1
2
M
1
2
(
( = T
2
1+
k 1
2
M
2
2
(
( .
(13.39)
Continuity requires
1
u
1
=
2
u
2
=
1
c
1
M
1
=
1
kR T
1
M
1
=
2
kR T
2
M
2
,
or
1
M
1
T
1
1
2
=
2
M
2
T
2
1
2
.
(13.40)
Shock
2. Disturbed 1. Undisturbed
Figure 13.9 Normal shock.
13. Compressible Flow 429
Now p/T = R, and p, T and may be eliminated between Eqs. (13.38), (13.39)
and (13.40). This done, the resulting relation between M
1
and M
2
is written as
M
2
2
=
(k 1)M
1
2
+ 2
2kM
1
2
(k 1)
.
(13.41)
It is noted that algebraically M
1
> 1 implies M
2
< 1 and M
1
< 1 implies M
2
> 1. Thus,
unless
M
1
= M
2
,
*
the flow must change from supersonic to subsonic. A change
from subsonic to supersonic flow requires decrease of entropy, which cannot be in
adiabatic flows. Hence a normal shock may occur only in supersonic flows and the
flow after the shock is subsonic.
The pressure ratio is given by
p
2
p
1
=
1+ kM
1
2
1+ kM
2
2
= 1+
2k
k +1
M
1
2
1
( )
.
(13.42)
Once M
2
is obtained, Eqs. (13.38), (13.39) and (13.40) yield the temperature
ratio
T
2
T
1
=
1+
k 1
2
M
1
2
1+
k 1
2
M
2
2
= 1+
2(k 1)
(k + 1)
2
kM
1
2
+ 1
M
1
2
M
1
2
1 ( )
(13.43)
and the density ratio
1
=
u
1
u
2
=
M
1
M
2
T
1
T
2
=
k + 1 ( )M
1
2
(k 1)M
1
2
+ 2
.
(13.44)
The relations of Eqs. (13.41) - (13.44) are also incorporated into Appendix D.
Using Eqs. (13.16), (13.17) and (13.18), the stagnation properties before and after
the shock can be found. Using Eqs. (13.25), (13.26) and (13.27), the critical prop-
erties can be found.
The shock is a compression wave, and p
2
is always greater than p
1
. A Shock
Strength may thus be defined as (p
2
p
1
)/p
1
= Ap
1
/p
1
. In general the increase of
entropy may be found using properties before and after the shock. However, for
shocks of small strength, i.e., weak shocks, a simpler approach is helpful. Let
m = M
1
2
1. Then
s
2
s
1
= c
v
ln
p
2
p
1
c
p
ln
1
= R ln 1+
2km
k + 1
|
\
|
.
|
1
k 1
m + 1 ( )
k
k1
k 1
k + 1
m + 1
|
\
|
.
|
k
k1
(
(
(
(
.
(13.45)
Expanding for m << 1
*
Division by M
1
M
2
has been affected. Hence, this possibility exists, which means no shock.
430 Fluid Mechanics
s
2
s
1
R
=
2k
k +1 ( )
2
m
3
3
+ 0(m
4
) ~
2k
(k + 1)
2
M
1
2
1
( )
3
3
or
s
2
s
1
R
=
k + 1
12k
2
Ap
1
p
1
|
\
|
.
|
|
3
.
(13.46)
For weak shocks the increase of entropy is proportional to the third power of
the shock strength.
At this stage we can answer partially a question raised when considering
nozzle flow: Exit pressures between points C and E, Fig. 13.6, are accommodated
by some supersonic flow in the diverging part of the nozzle; then a normal shock
occurs and the flow continues subsonically to the exit pressure. Point C corre-
sponds to zero strength shock, at the throat, and point E to a shock at the exit.
Example 13.10
The nozzle designed in Example 13.6 is made to discharge into another
vessel, where the pressure is
a. p
e
= 3.210
5
Pa ; b. 310
5
Pa ; c. 210
5
Pa.
The stagnation pressure where the flow originates is still p
o
= 410
5
Pa. Use values
from Table D-1, without interpolation, and describe the resulting flows.
Solution
a.
p
e
p
o
=
3.210
5
4 10
5
= 0.8> 0.5283.
Could the flow be all subsonic?
A
e
A
*
=
2.12 10
3
1.73610
3
= 1.221.
From Table D-1: M
e
= 0.57, p
e
/ p
o
= 0.8.
Indeed the flow is all subsonic!
b.
p
e
p
o
=
3 10
5
410
5
= 0.75> 0.52,
A
e
/A
*
= 1.221.
From Table D-1: M
e
= 0.57, p
e
/p
o
= 0.8 > 0.75.
13. Compressible Flow 431
The flow cannot be all subsonic. Thus supersonic speed must appear and the
part of the flow between the entrance and the critical section is exactly the
same as in Example 13.6, with sonic conditions at A
*
. Now
p
*
= (0.5283)(410
5
) = 211320 Pa.
The air must expand further, the flow becomes supersonic and since
p
e
= 310
5
Pa
is higher than 10
5
Pa, the design exit pressure in Example 13.6,
we expect a shock wave; the flow then becomes subsonic, and the compres-
sion in the divergent part of the nozzle raises the pressure up to 310
5
Pa.
The location of the shock wave must be found by trial and error, and we know
that 1 < M
1
< 1.56; see Example 13.6.
Try M
1
= 1.2. From Table D-1:
M
2
= 0.8442,
p
1
= 0.4124 p
o
= 1.6510
5
Pa,
p
2
= 1.5133 p
1
= 2.510
5
Pa,
A
1
= A
*
(A
1
/A
*
) = 1.73610
-3
1.0304 = 1.789 10
3
m
2
.
We now look again at Table D-1, at the line of M = 0.8442, and from
p
2
= 2.510
5
deduce p
o2
= 2.510
5
/0.63 = 3.9710
5
Pa. The stagnation pressure
is reduced because the flow went through an irreversible process, a shock
wave. We also find there A/A
*
= 1.0237 and deduce
A
2
*
= A
1
/1.0237= 1.747 10
3
m
2
.
The critical cross section has been increased, again because of the shock
wave. Note that this critical cross section is not realized physically but is
rather an answer to the question of what critical cross section is required to
accommodate the flow after the shock wave. Now at the exit
A
e
A
2
*
=
2.12 10
3
1.74710
3
= 1.20,
and from Table D-1 we read
M
e
= 0.59, p
e
/p
o2
= 0.7901 and
p
e
= 0.790 3.9710
5
= 3.1410
5
Pa > 310
5
Pa.
Try M
1
= 1.5 and repeat:
M
2
= 0.7011, p
1
= (0.2724)p
o
= 1.0910
5
Pa,
p
2
= (2.4583) p
1
= 2.710
5
Pa,
432 Fluid Mechanics
A
1
= 1.736 10
3
1.1762= 2.04210
3
m
2
,
p
o2
=
2.710
5
0.7209
= 3.74510
5
Pa,
A
2
*
=
2.042 10
3
1.0944
= 1.866 10
3
m
2
,
A
e
A
2
*
=
2.12 10
3
1.866 10
3
= 1.14, M
e
= 0.65, p
e
/ p
o2
= 0.7528
and
p
e
= 0.75283.745 10
5
= 2.82 10
5
Pa < 3 10
5
Pa.
Try
M
1
= 1.35 and repeat:
M
e
= 0.6, p
e
= 3.0310
5
Pa.
Try M
1
= 1.34.
c. We are led in this case by the calculations in case b, and try right away
M
1
= 1,56, M
2=
= 0.6809,
p
1
= 0.2496p
o
= 0.998410
5
Pa,
p
2
= 2.6725p
1
= 2.668 10
5
Pa,
and no solution can be obtained. The lowest exit pressure that may still
correspond to a normal shock wave is p
e
= 2.66810
5
Pa.
The flows obtained are:
a. Complete subsonic flow.
b. Subsonic flow up to the critical section followed by supersonic flow up to
A = 1.9 10
-3
m
2
. Then a shock wave and isentropic compression to p
e
. Solu-
tions of the type in part b are possible for 2.66810
5
Pa < p
e
< 3.210
5
Pa.
c. Subsonic flow up to the critical section followed by supersonic flow up to the
exit. Whatever adjusts the exit pressure to the imposed one must take place
after the exit.
Oblique Shocks and PrandtlMeyer Expansion
Consider a flow w
1
encountering an oblique shock, Fig. 13.10, i.e., the veloc-
ity component normal to the shock, u
1
, undergoes a normal shock, while the
component parallel to the shock, v
1
, is not altered at all. For this to happen u
1
itself
must be supersonic, resulting in subsonic u
2
. As a result of this, the flow turns by
the angle u. A way to force an oblique shock is therefore to make a supersonic
flow turn by the introduction of a wedge into the flow. Two oblique shock waves
appear, Fig. 13.11.
13. Compressible Flow 433
Shock
v
1
|
u
u
1 w
2
1
w
2
v
1
2
u
Figure 13.10 Oblique shock. Figure 13.11 Shocks at a wedge.
For large enough v
2
= v
1
, w
2
may still be supersonic and Ap
1
/p
1
may indeed
be quite small. The flow then approaches an isentropic one. Furthermore, it can be
shown that for small turning angles u [see Eq. (13.46)],
Ap
1
p
1
u , As u
3
.
(13.47)
u
w
1
shock
u
w
1
shock
u
w
1
shock
a. One step b. Two steps c. Many steps
Figure 13.12 Compressive turning of supersonic flow.
Consider the flow w
1
, which must turn by the angle u, Fig. 13.12. Case a is a
turn as already discussed. In case b the turn is reached in two steps, say, each of
u/2. The change of entropy in this case, by Eq. (13.47), is
As = 2
u
2
|
\
|
.
|
3
=
1
4
u
3
,
i.e., one quarter of case a. We note, however, that each step generates its own
shock wave and that the two waves meet. In case c, the change of entropy is
434 Fluid Mechanics
3
3
2
lim lim 0
n n
s n
n n
u u
| |
A = = =
|
\ .
and the flow is isentropic. Still, each differential step generates its differentially
weak shock, i.e., its Mach line, and these do meet. In the close vicinity of the
rounded bend, however, these Mach lines hardly meet and the flow is isentropic,
i.e., reversible.
Now consider a supersonic flow along a wall, which turns away from the
flow, Fig. 13.13a. Since every point in a supersonic flow has a right to a Mach line,
we may draw some of these, and note that they diverge. We may repeat the
argument leading to case c in Fig. 13.12 to show that now we have isentropic
expansion. Furthermore, because the Mach lines diverge, case b in Fig. 13.13 is
also an isentropic expansion.
w
1 w
1
a. Gradual turn b. Corner
Figure 13.13 Supersonic expansion by turning.
Such a supersonic expansion yields M as a function of the turn angle, called
the PrandtlMeyer angle. There is no compressive counterpart to this function,
because in compression the Mach lines converge to generate partial shocks and
these meet to form stronger shocks, in a way similar to the piston flow, Fig. 13.3.
PrandtlMeyer angle v is also given in Table D-1.
We may now complete the description of the nozzle flow, Fig. 13.6: For exit
pressures between points E and D, oblique shocks occur outside the nozzle exit.
For pressures below D there is a PrandtlMeyer expansion.
Example 13.11
Complete case c in Example 13.10. Also find what flows result for the exit
pressures
13. Compressible Flow 435
d. p
e
= 10
5
Pa,
e. p
e
= 0.510
5
Pa.
Solution
c. There are oblique shocks outside the exit, and the phenomenon becomes three
dimensional.
d. The flow in the divergent part is fully supersonic. The exit pressure is the
design pressure.
e. The flow in the nozzle is at the design conditions. At the exit the flow is
supersonic with M = 1.56. There is a PrandtlMeyer outward expansion with
the angle Av = 0.4652 0.2387 = 0.2265 rad, up to M = 2.01. This is followed
by three-dimensional phenomena.
w
Shock
1
Expansion
Expansion
Shock
Shock
Figure 13.14 Diamond-shaped wing in supersonic flow.
Lift and Drag
Finally we note that because of the increase in entropy through shock waves,
the equations for supersonic flow yield both lift and drag, as seen in Fig. 13.14.
This result is different from that obtained from the equations for subsonic flows
with negligible shear, which could predict only lift.
436 Fluid Mechanics
References
A.B. Cambel and B.H. Jennings, "Gas Dynamics," McGraw-Hill, New York, 1958.
A. Chapman and W.F. Walker, "Introductory Gas Dynamics," Holt, Rinehard and
Winston, New York, 1971.
H.W. Liepmann and A. Roshko, "Elements of Gas Dynamics," Wiley, New York,
1957.
R. von Mises, "Mathematical Theory of Compressible Fluid Flow," Academic
Press, New York, 1958.
J.A. Owczarek, "Fundamentals of Gas Dynamics," McGraw-Hill, New York, 1971.
A.H. Shapiro, "The Dynamics and Thermodynamics of Compressible Fluid Flow,"
2 vols., Ronald Press, New York, 1953.
Problems
13.1a. What is the speed of sound in air at 300
K, 1 bar? In helium at the same
state? In hydrogen? In a 1 : 1 molar mixture of helium and hydrogen?
b. Equation (13.4) was obtained by Newton, who first assumed the process to
be isothermal rather than isentropic. Use
dp
d
=
cp
c
|
\
|
.
|
T
in Eq. (13.4) and recalculate the speed of sound c. What are the errors?
13.2a. The isentropic compressibility of water at 1 bar, 90
o
C, is
1
c
c p
|
\
|
.
|
|
s
= 4.410
10
m
2
/ N.
Find the speed of sound in water.
b. The isentropic compressibility of steel-like solids is
1
c
cp
|
\
|
.
|
s
=
31 2v ( )
E
,
where v is the Poisson ratio, and E is the Young modulus. Find the speed of
sound in steel.
13. Compressible Flow 437
13.3 Helium at stagnation, at 210
5
Pa, 300 K, expands isentropically:
a. to p
1
= 10
5
Pa, and then continues;
b. to p
2
= 0.510
5
Pa, and then continues;
c. to p
3
~ 0.
Find Mach numbers, speeds, sonic speeds and stagnation pressures for a, b, c
and for the pressure at which the sonic speed appears. Assume ideal gas
behavior.
13.4 Assume one-dimensional adiabatic frictionless flow through the nozzle shown
in Fig. P13.4. In section 1: u = 20 m/s, p = 810
6
Pa, T = 300 K. Calculate
the velocities in sections 2 and 3 for the fluids:
a. water, b. molten lead, c. air, d. hydrogen.
20 mm
35 mm 50 mm
Figure P13.4 Nozzle.
13.5 Design, i.e., find the dimensions at the throat and at the exit, and draw a
scheme of a convergentdivergent nozzle which transfers 1 kg/s air from
vessel A, where p
o
= 10
6
Pa, T = 300 K, to vessel B, where p
e
= 1.510
5
Pa.
13.6 In Problem 13.5 the pressure at the exit, p
e
, is changed to
a. 10
5
Pa; b. 9.510
5
Pa.
Find the new mass flowrate,
m ,
and the pressure at the throat for the same
nozzle.
13.7. The throat cross-sectional area in Problem 13.5 is A
*
, and that of the exit is A
e
.
Let A
1
= (A
*
+A
e
)/2. With the same stagnation conditions as in Problem 13.5,
the sonic speed is reached at A
*
; but then there is a shock wave at A
1
. Find
the Mach number and the pressure at the throat, at A
1
just before the shock
and at the exit.
13.8 A compressor takes in air at the atmospheric pressure p = 10
5
Pa and dis-
438 Fluid Mechanics
charges the compressed air into a high pressure settling tank. An engineer
who has to measure the performance curve of the compressor, i.e., the mass
flux of the compressed air as a function of the compression pressure,
connects to the settling tank a thermometer and a convergent nozzle with a
minimal cross-sectional area of 20 10
-4
m
2
.
In all the measurements the
temperatures were between 350 K and 355 K. The measured gauge pressures
were:
1.110
5
Pa, 1.610
5
Pa, 2.510
5
Pa, 4.010
5
Pa and 6.010
5
Pa.
Find the corresponding mass flowrates.
13.9 A second engineer, who saw what the one in Problem 13.8 did, suggested that
a convergentdivergent nozzle be used, such that only one equation had to
be used to compute the mass fluxes in the five measurements taken there.
Design such a nozzle, which has the same minimal cross section as in
Problem 13.8.
13.10 A Pitot tube is used to measure the speed of an airplane. Assuming atmo-
spheric pressure of 10
5
Pa and isentropic processes except through shock
waves, find the pressures read on the Pitot differential manometer and the
speeds of the airplane, corresponding to Mach numbers of 0.1, 0.3, 0.7,
1.0, 1.5, 2.0.
13.11 Helium at p
o
= 310
5
Pa, T
o
= 400 K, enters a pipe of an inner diameter
0.015 m. The friction coefficient is f = 0.023. Find the pipe length at which
sonic speed is reached.
13.12 Design a nozzle which discharges 1 kg/s air from a vessel at stagnation
conditions, p
o
= 8.010
5
Pa, T
o
= 400 K, to the atmosphere at p =10
5
Pa.
Find A
*
, A
e
and the total thrust force which acts on the vessel. For reasons of
material availability the nozzle designed above has been manufactured with
its divergent part opening only up to the cross-sectional area of A
1
= (A
*
+ A
e
)/2. This truncated nozzle is used instead of the designed one. Find the total
thrust force applied to the vessel now.
13.13 Both nozzles, the designed and the truncated one, of Problem 13.12 are
tested under the conditions of Problem 13.12. However, instead of using air,
hydrogen is being used. Find the thrust forces acting on the vessel now.
13.14 The four tests of Problems 13.12 and 13.13 are repeated in space. Find the
four new thrust forces.
13. Compressible Flow 439
13.15 A normal shock wave moves through quiescent air at p = 10
5
Pa, T = 300 K.
The speed of the shock is u
s
= 694 m/s. Find the pressure left immediately
behind the shock. Is the air immediately behind the shock quiescent? If not,
what is its velocity and in which direction?
13.16 The shock wave of Problem 13.15 hits a wall in a direct frontal collision. It
is then reflected back. Find the highest pressure suffered by the wall. Find
the speed of the reflected wave.
13.17 A large pressure vessel contains gas at the stagnation properties:
p
o
= 400 kPa, T
o
= 420 K. The gas is approximately ideal, with
R = 287 J/kg
.
K, and k=1.4. The outside atmospheric pressure is p
a
= 100kPa.
A convergentdivergent nozzle is designed to pass a mass flux of 1 kg/s
from the vessel to the outside. Find:
a. The speed of the gas at the exit from the nozzle, V
e
.
b. The exit Mach number, M
e
.
c. The critical cross-sectional area, A
*
.
d. The exit cross-sectional area, A
e
.
13.18 The nozzle of Problem 13.17 is cut after the critical section at point 1,
where A
1
=(A + A
e
)/2. It still connects the pressure vessel to the outside.
Find:
e. The pressure, p
1
, at section A
1
.
f. The speed of the air, V
1
, at A
1
.
g. The Mach number, M
1
, at A
1
.
h. The mass flux,
1
m at A
1
.
13.19 The nozzle of Problem 13.18 is further cut, before the critical section at
point 2, where A
2
= A
1
. It still connects the pressure vessel to the outside.
Find:
e. The pressure, p
2
, at section A
2
.
f. The speed of the air, V
2
, at A
2
.
g. The Mach number, M
2
, at A
2
.
h. The mass flux,
2
m at A
2
.
13.20 The pressure vessel of Problem 13.17 is also connected to the outside by a
pipe of a constant cross-sectional area of A
3
= A
2
. At the exit of the pipe the
speed of the gas is V
3
, and the mass flux through the pipe is . m
i. Is V
3
smaller, equal to or greater than V
2
?
j. Is
3
m smaller, equal to or greater than
2
m ?
440 Fluid Mechanics
13.21 A rocket, shown in Fig. P13.21, has a cylindrical exhaust pipe of the diame-
ter d = 0.025 m instead of a nozzle. The rocket moves with a constant speed
and exhausts gas at the rate of 0.5 kg/s. The gas is approximately ideal, with
k = 1.4, and R = 287 J/kg
.
K. The stagnation properties of the gas in the
rocket are p
o
= 400 kPa, T
o
= 3,000 K. The outside pressure is 100 kPa. Find
the force on the rocket.
U
d = 0.025 m
Figure P13.21 Rocket.
13.22 A jet-propelled airplane flies at 900 km/h. The air intake into the engine is
designed as a diffuser, i.e., as part of a nozzle that takes in the outside air
and brings it to the speed of the airplane while increasing its pressure. The
outside air is at 10
5
Pa, 300 K. Find the maximum attainable pressure and
the ratio between the area of the diffuser at the intake and that where the
higher pressure is obtained.
13.23 The airplane of Problem 13.22, with the diffuser designed in that problem,
flies now at a higher altitude, where the outside air is at 510
4
Pa, T = 280 K.
Find the highest pressure which can be obtained now at the exit of the
diffuser. Assuming the airplane operates at that new highest pressure, find
the ratio between the mass rate of air intake under the conditions of this
problem and that of Problem 13.22.
13.24 At takeoff the airplane of Problem 13.22 must have the same mass rate of
air supply as in Problem 13.22. Now, however, the airplane is at zero speed.
To achieve this, the engine compressor must set a lower pressure just at the
exit from the diffuser. Find this pressure.
13.25 Relating to the airplane in Problem 13.22, the air coming out of the diffuser
passes through a compressor where it is compressed by a compression ratio
of 1 : 20. The air then flows to a combustion chamber where its temperature
is raised by 1,000 K. Then the air expands in a turbine where energy is
extracted just enough to run the compressor, and then the air expands in a
13. Compressible Flow 441
nozzle. Assume isentropic compression and expansion. For 1 kg/s air,
design the nozzle for operating under the conditions of Problem 13.22, i.e.,
find the critical diameter and the exit diameter. Find the thrust of the engine.
13.26 The turbine in Problem 13.25 is changed to a cheaper one, with an adiabatic
efficiency of 70%. This means that the expansion in the turbine is still
adiabatic but that the power extracted from the gas is only 70% of what
could be extracted by isentropic expansion. To obtain the power necessary
to run the compressor, the gas must exit from the turbine at a lower
pressure. Find the critical diameter and the exit diameter of the jet nozzle
now. Find the thrust of the engine.
13.27 The compressor in Problem 13.25 is changed to a cheaper one, with an
adiabatic efficiency of 70%. This means that the compression in the com-
pressor is still adiabatic but that the power needed to compress the gas is
1.0/0.7 times that needed in an isentropic compression. The turbine, which
remains the same as in Problem 13.25, must now supply the increased
power necessary to run the compressor, and the gas exits from the turbine at
a lower pressure. Find the critical diameter and the exit diameter of the jet
nozzle now. Find the thrust of the engine.
13.28 A continuous stream of air at
10
5
Pa, 300 K, flows with the
speed of 450 m/s perpendicular
to a wall. It hits the wall and a
normal shock wave is reflected
from the wall and propagates
against the stream of the incom-
ing air, Fig. P13.28. Now the
stream of air passes through the
reflected shock, and behind the
shock, on the wall side, the air is stationary with respect to the wall because
it must satisfy this boundary condition. Find the pressure near the wall and
the speed of the reflected shock.
13.29 A diamond-shaped supersonic wing moves in air at 10
5
Pa, 300 K, at
500 m/s, Fig. P13.29.
Find the lift and the drag induced by the shocks and the PrandtlMeyer
expansion for cases a and b.
Incoming air
at 450 m/s
Quiescent
air
Reflected
shock
Wall
Figure P13.28
442 Fluid Mechanics
Shock
P-M
Expansion
Shock
Shock
P-M
Expansion
10
o
a.
Shock
P-M
Expansion
Shock
Shock
P-M
Expansion
10
o
b.
Shock
Figure P13.29 Diamond-shaped wings in supersonic flow.
13.30 A stream of air at 10
5
Pa, 300 K, moving
at 450 m/s meets with another stream at
1.310
5
Pa, 300 K, moving at 450 m/s.
The meeting takes place at an angle of
20
o
, Fig. P13.30. The resulting flow
must have the same pressure and the
same direction everywhere. Find the
new pressure and direction.
o
20
o
Shock
Shock
Figure P13.30
443
14. NON-NEWTONIAN FLUIDS
A Newtonian fluid has been defined in Chapter 5 as one that satisfies a linear
relationship between its stress and its rate of strain. For a simple unidirectional
flow, such as the flow between two parallel plates, this relationship is expressed as
t =
du
dy
|
\
|
.
.
(14.1)
Equation (14.1) is known as Newton's law of viscosity. This equation is a
mathematical statement, and there is no reason to believe that all real fluids should
obey it. Indeed, there are more fluids that do not behave according to Eq. (14.1)
than those that do. However, the three fluids most abundant in nature, air, water
and petroleum, follow Eq. (14.1) quite accurately. Fluids whose behavior cannot
be described by Newton's law of viscosity are called non-Newtonian fluids.
Typical non-Newtonian fluids are paints, slurries, pastes, jellies and similar
food products, foams, polymer melts, blood, etc. Some of these fluids simply
exhibit nonlinear viscous effects, while others show effects of "memory" which are
related to their viscoelastic behavior. A typical example of the latter is the dough
climbing up the beater of a food mixer. Only nonlinear viscous effects will be dealt
with in this chapter. Elastic effects, which strongly influence time-dependent flows
and stretching flows, are outside the scope of this book.
Figure 14.1 shows curves of stress versus rate of strain for several classes of
fluids. The Newtonian fluid of Eq. (14.1) is represented by the straight line of
constant slope, , passing through the origin. The constant viscosity of a Newto-
nian fluid is defined as the ratio between a given shear stress and the resultant rate
of strain. Similarly, we may define an apparent viscosity of a non-Newtonian fluid
by
t =
app
du
dy
|
\
|
.
.
(14.2)
444 Fluid Mechanics
Here, however, the apparent viscosity is not a constant but depends on the rate of
strain. The relationship between the shear rate and the rate of strain is found
experimentally by a series of viscometric measurements at different shear rates.
Plotting the results as in Fig. 14.1, the curve obtained is called a flow curve. This
curve is fitted by means of an equation called the constitutive equation of the fluid.
One of the simplest constitutive equations describing the flow behavior of a non-
Newtonian fluid is that of the power-law model:
t = K
du
dy
|
\
|
.
n
.
(14.3)
For cases of n < 1, the fluid is referred to as the pseudoplastic fluid. For n > 1, the
fluid is known as dilatant. The Newtonian fluid is, of course, a special case of the
power-law fluid with n = 1.
Bingham Plastic
Pseudoplastic (n < 1)
Newtonian Fluid
Dilatant (n > 1)
t
t
o
du
dy
Figure 14.1 Stress vs. rate-of-strain relationships for different fluids.
Comparing Eqs. (14.2) and (14.3) it is noted that the apparent viscosity of a
power-law fluid may be expressed as
app
= K
du
dy
n 1
. (14.4)
14. Non-Newtonian Fluids 445
The absolute value sign is introduced to insure that the apparent viscosity is the
same for shear rates equal in magnitude but acting in opposite directions. Equation
(14.4) may be combined with (14.2) to yield
t = K
du
dy
n 1
(
(
(
du
dy
|
\
|
.
|
|
.
(14.5)
The power-law model as expressed by Eq. (14.5) is to be preferred over that
of (14.3), because in Eq. (14.5) the term in the square brackets, i.e., the apparent
viscosity, is always positive, while the sign of the shear stress is determined by the
sign of du/dy. In many cases, the power-law model describes the behavior of the
fluid only over a limited range of shear rates. In particular, it breaks down at very
low shear rates. Inspection of Eq. (14.4) shows that as the shear rate approaches
zero,
app
goes to infinity. On the other hand, experience shows that most non-
Newtonian fluids approach Newtonian behavior at low shear rates. A slightly more
complex model that takes this into account is the Ellis model:
( )
1
.
s u
a b
y
c
t t
c
= +
(14.6)
This model includes as special cases the power-law fluid when a = 0 and the
Newtonian fluid for b = 0.
Another fluid shown in Fig. 14.1 is the Bingham plastic. This fluid does not
flow below a certain yield stress t
o
. Above t
o
the increase in t is proportional to
the shear rate. Thus, it can be described by
t = t
o
+
du
dy
|
\
|
.
|
|
for | t |> t
o
,
du
dy
= 0 for 0 s t s t
o
.
(14.7)
This model may also be derived as a special case of the Ellis model.
The constitutive equations given above were written in one-dimensional form
and thus could be used to solve problems of unidirectional flows only. In order to
deal with higher dimensional flows, one has to extend the constitutive relationships
to two- and three-dimensional cases. To do this, we start again with the Newtonian
fluid and then extend our results to non-Newtonian fluids.
The stress versus rate-of-strain relationship for an incompressible Newtonian
fluid is
t
ij
=
cu
i
c x
j
+
c u
j
c x
i
|
\
|
.
|
(14.8)
446 Fluid Mechanics
or by Eq. (5.37)
t
ij
= 2c
ij
.
Similarly, one may write for a non-Newtonian fluid
t
ij
=
app
cu
i
c x
j
+
cu
j
c x
i
|
\
|
.
|
|
(14.9)
and
t
ij
= 2
app
c
ij
.
(14.10)
Again,
app
is not constant but some function of the rate of deformation:
( )
.
app ij
f c = (14.11)
Although we do not know a priori the expression for
app
, we may note the
following facts: Inspection of Eq. (14.10) reveals that the stress on the left-hand
side stands for a second order tensor. The shear rate c
ij
on the right-hand side
stands also for a second order tensor. In order to preserve the tensorial rank of the
equation, it requires that
app
be a scalar, i.e., a tensor of rank zero. The only way
for a function of a tensorial quantity to be a scalar is by being a function of the
scalar invariants of the tensor. The scalar invariants of a tensor are those combi-
nations of the components that do not vary under rotation of the coordinate system.
The three scalar invariants of the rate-of-strain tensor are
I
1
=c
ii
= c
11
+c
22
+c
33
,
I
2
= c
ij
c
ij
=c
11
2
+c
22
2
+ c
33
2
+2c
12
2
+2c
23
2
+2c
13
2
,
I
3
= det| c
ij
|,
where the third invariant is the determinant of the matrix of c
ij
. The first invariant,
I
1
, equals simply V
.
q, which vanishes for incompressible fluids. Hence
app
= f I
2
, I
3
( ).
For two-dimensional flows I
3
= 0. Hence for this case
app
= f I
2
( ).
(14.12)
Now whatever form this f(I
2
) attains, once the flow is unidirectional, it must reduce
to a simple one-dimensional form. For example, the apparent viscosity of the
power-law model for a two-dimensional flow may be put in the following form
that satisfies this condition:
app
= K 2c
ij
c
ij
n1
.
(14.13)
14. Non-Newtonian Fluids 447
Hence, the stress versus rate-of-strain relationship becomes
t
ij
= K 2
cu
cx
|
\
|
.
|
2
+ 2
cv
cy
|
\
|
.
|
|
2
+
cu
cy
+
cv
cx
|
\
|
.
|
|
2
n 1
(
(
(
(
(
cu
i
cx
j
+
cu
j
cx
i
|
\
|
.
|
|
.
(14.14)
The scalar expression in the square brackets is, of course, the apparent viscosity.
For unidirectional flows where u = u(y), Eq. (14.14) reduces to
T
xy
= K
cu
cy
n 1
(
(
(
(
cu
cy
,
(14.15)
which corresponds to the one-dimensional form of Eq. (14.5).
The constitutive equation for a power-law fluid in polar coordinates is
derived similarly. It reads
t
ru
= K 2
cv
r
cr
|
\
|
.
|
2
+ 2
1
r
cv
u
cu
+
v
r
r
|
\
|
.
|
2
+ r
c
cr
v
u
r
|
\
|
.
|
+
1
r
cv
r
cu
|
\
|
.
|
2
n1
2
(
(
(
(
(
r
c
cr
v
u
r
|
\
|
.
|
+
1
r
cv
r
cu
|
\
|
.
| ,
(14.16)
which for a flow with v
u
= v
u
(r) simplifies to
t
ru
= K r
c
cr
v
u
r
|
\
|
.
|
n1
(
(
(
r
c
cr
v
u
r
|
\
|
.
|
|
\
|
.
| .
(14.17)
Similarly for axisymmetric flows
t
rz
= K 2
cv
r
cr
|
\
|
.
|
2
+ 2
cv
z
cz
|
\
|
.
|
2
+
cv
z
cr
+
cv
r
cz
|
\
|
.
|
2
n 1
2
(
(
(
(
(
cv
z
cr
+
cv
r
cz
|
\
|
.
| ,
(14.18)
and for unidirectional flows, such as the flow in a circular pipe, with v
z
= v
z
( r ),
t
rz
= K
cv
z
cr
n1
(
(
(
cv
z
cr
.
(14.19)
The stress versus rate-of-strain relationships given by the constitutive equa-
tions may now be substituted into the momentum equations, Eqs. (5.38) - (5.40) or
Eq. (5.41), which can be solved for particular cases. The following examples
illustrate this point.
448 Fluid Mechanics
Flow of a Power-law Fluid in a Circular Tube
This problem, solved in Chapter 6 for a Newtonian fluid, is considered again
using a power-law model. The momentum equations for steady flow in a circular
pipe, Eqs. (5.48) - (5.50), simplify to
0 =
c P
cr
,
(6.45)
0 =
1
r
c P
c u
,
(6.46)
0 =
c P
c z
+
1
r
c
c r
(r t
rz
),
(14.20)
where P = p+ gh is the modified pressure. From Eq. (6.47), we conclude that P is
a function of z only and P/z in Eq. (14.20) may be replaced by AP/Az. Substitu-
tion of t
rz
from (14.19) results in
c
c r
Kr
c v
z
c r
n 1
c v
z
c r
(
(
(
=
AP
Az
r
(14.21)
with the boundary conditions
v
z
R ( ) = 0
(14.22)
and
c v
z
0
( )
c r
= 0.
(14.23)
Equation (14.21) is rewritten in a slightly more convenient form by making
use of a sign coefficient c:
c =
cu cy
cu cy
.
(14.24)
The value of c is 1 or +1, depending on the sign of v
z
/r. Thus
cK
c
c r
r
c v
z
c r
n
(
(
(
=
AP
Az
r.
Integration and the use of the boundary condition Eq. (14.23) result in
c v
z
c r
n
=
r
2Kc
AP
Az
.
The absolute value sign may now be dropped because v
z
/r is always negative in
tube flow, i.e., c = 1; thus
14. Non-Newtonian Fluids 449
cv
z
c r
=
1
2K
AP
Az
|
\
|
.
|
(
(
(
1/ n
r
1/ n
.
A second integration yields
v
z
=
n
n +1
R
n+1
2K
AP
Az
|
\
|
.
|
(
(
(
1/n
1
r
R
|
\
|
.
|
n+1
(
(
(
. (14.25)
The volumetric flow rate is obtained as
Q =
0
R
}
2tv
z
r dr = t
n
3n +1
1
2K
AP
Az
|
\
|
.
|
(
(
(
1/n
R
3n +1
n
. (14.26)
The centerline velocity, v
o
, is obtained from Eq. (14.25) by setting r = 0:
v
o
=
n
n +1
R
n+1
2K
AP
Az
|
\
|
.
|
(
(
(
1/ n
, (14.27)
which combined with (14.25) yields
v
z
= v
o
1
r
R
|
\
|
.
|
n +1
(
(
(
.
(14.28)
The volumetric flow rate is given in terms of the centerline velocity as
Q=
n +1
3n +1
tR
2
v
o
,
(14.29)
and the average velocity is
v =
Q
A
=
n +1
3n +1
v
o
.
(14.30)
The appropriate expressions in Chapter 6 for Newtonian flow in a tube may be
obtained by setting n = 1.
Example 14.1
Find the change in the volumetric flowrate of a fluid flowing in a pipe when
the pressure drop along the pipe is doubled.
a. For a Newtonian fluid.
b. For a power-law fluid with n = 0.2.
450 Fluid Mechanics
Solution
The volumetric flowrate given by Eq. (14.27) for a power-law fluid also
holds for a Newtonian fluid with n = 1. Thus:
a. For a Newtonian fluid
Q
2
Q
1
=
AP
2
AP
1
= 2.
b. For a power-law fluid
Q
2
Q
1
=
AP
2
AP
1
|
\
|
.
|
|
1/n
= 2
1
0.2
= 2
5
= 32.
Hence, doubling the pressure drop will increase the flowrate 32-fold.
Figure 14.2 Flow of a Bingham plastic between two parallel plates.
Flow of a Bingham Plastic Material between Parallel Plates
Consider the flow of a Bingham plastic material between two stationary
parallel plates separated by a distance 2d, Fig. 14.2. The momentum equation for
this case is derived similarly to Eq. (6.21):
0 =
AP
Ax
+
c t
xy
c y
(14.31)
with the boundary conditions
u d ( )=0,
(14.32)
cu 0
( )
c y
=0, or t
xy
0 ( ) =0.
(14.33)
y
1
y y
u t
2d
14. Non-Newtonian Fluids 451
Integration of Eq. (14.31) with the boundary condition Eq. (14.33) results in
t
xy
=
AP
Ax
|
\
|
.
| y.
(14.34)
The shear stress given in Eq. (14.34) is a linear function of y as shown in Fig. 14.2.
It has a maximum at y = d, goes through zero at the centerline and has a mini-
mum at y = + d. The constitutive relationship for a Bingham plastic is given by
Eq. (14.7). Substitution of t
o
for t
xy
in Eq. (14.34) yields the distance y
1
, within
which du/dy = 0, i.e.,
u = u
1
= const, y
1
s y s y
1
,
(14.35)
with
t
o
=
AP
Ax
|
\
|
.
| y
1
, y
1
=t
o
AP
Ax
|
\
|
.
| .
(14.36)
Equation (14.7) is combined with (14.34) resulting in
AP
Ax
y = t
o
+
du
dy
,
which upon substitution of t
o
from Eq. (14.36) yields
AP
Ax
y y
1
( )
=
du
dy
.
(14.37)
Integration between the limits y
1
and y gives
u u
1
=
1
2
AP
Ax
y y
1
( )
2
.
(14.38)
Equation (14.38) holds for the regions
y
1
< y < d
and
d < y < y
1. The constant
velocity u
1
may now be evaluated from the boundary condition at
y = d,
Eq. (14.32):
u
1
=
1
2
AP
Ax
|
\
|
.
| d y
1
( )
2
, y
1
s y s +y
1
(14.39)
and finally
( ) ( )
2 2
1 1 1
1
,
2
P
u d y y y y y
x
A
| |
(
= >
|
A
\ .
(14.40)
Equation (14.40) reduces to the Newtonian profile for the special case of y
1
= 0.
452 Fluid Mechanics
Example 14.2
Many paints behave like a Bingham plastic. For such paint, derive an
expression for the maximum film thickness, which just will not flow down a verti-
cal wall.
Solution
Assume that creeping flow conditions apply in this
case. With the coordinates of Fig. 14.3, Eqs. (5.42) and
(5.45) yield
1
0 ,
0 .
yx
p
g
x y
p
y
ct
c
c c
c
c
= + +
=
From the second equation, p = p( y ), and therefore
( ).
p dp
f y
x dx
c
c
= =
This equation holds in the bulk and as on the airliquid interface, where dp/dx = 0.
Hence dp/dx = 0 holds everywhere. The momentum equation for this film flow
simplifies to
0 = g +
1
c t
yx
cy
.
With the boundary conditions of zero shear at the interface and zero velocity on
the solid wall, respectively,
t
yx
y=0
= 0 and u o ( )= 0.
Integration of this momentum equation with the first boundary condition yields
t
yx
=go
and combining this with the constitutive equation for a Bingham plastic fluid,
Eq. (14.7), yields
go = t
o
+
du
dy
|
\
|
.
|
|
for t > t
o
and
du
dy
=0 for 0s t s t
o
.
The maximum thickness, o
max
, is obtained for t = t
o
and du/dy = 0. Hence
go
max
=t
o
o
14. Non-Newtonian Fluids 453
and
o
max
=
t
o
g
.
A film thicker than o
max
results in the sagging of the paint.
The Boundary Layer Equations for a Power-law Fluid
Consider the flow of a power-law fluid in a two-dimensional boundary layer.
The momentum and continuity equations for the two-dimensional flow are
u
cu
c x
+v
cu
c y
|
\
|
.
|
|
=
c P
c x
+
ct
xx
c x
+
ct
xy
c y
,
(14.41)
u
cv
c x
+ v
cv
c y
|
\
|
.
|
|
=
c P
c y
+
ct
xy
c x
+
ct
yy
c y
,
(14.42)
cu
c x
+
cv
c y
=0,
(14.43)
where the components of the stress tensor are given by
t
ij
= K 2
c u
c x
|
\
|
.
|
|
2
+ 2
c v
c y
|
\
|
.
|
|
2
+
c u
c y
+
c v
c x
|
\
|
.
|
|
2
n1
(
(
(
(
(
c u
i
c x
j
+
c u
j
c x
i
|
\
|
.
|
|
.
(14.14)
The considerations of orders of magnitudes that led to the boundary layer
equations in Chapter 8 may be repeated here. The order of magnitude analysis
simplifies the expression for the apparent viscosity given in the square brackets in
Eq. (14.14), and Eq. (14.14) is modified to
t
ij
~ K
c u
c y
n1
(
(
(
c u
i
c x
j
+
c u
j
cx
i
|
\
|
.
|
|
.
(14.44)
This expression may now be substituted in Eqs. (14.41) and (14.42):
u
c u
c x
+ v
c u
c y
|
\
|
.
|
|
=
c P
c x
+
c
c x
cK
c u
c y
n
2
c u
c x
c u
c y
(
(
(
(
(
(
+
c
c y
cK
c u
c y
n
1 +
c v
c x
c u
c y
|
\
|
.
|
|
|
|
|
(
(
(
(
(
(
,
(14.45)
454 Fluid Mechanics
u
c v
c x
+ v
c v
c y
|
\
|
.
|
|
=
c P
c y
+
c
c x
cK
c u
c y
n
1 +
c v
c x
c u
c y
|
\
|
.
|
|
|
|
|
(
(
(
(
(
(
+
c
c y
cK
c u
c y
n
2
c v
c x
c u
c y
(
(
(
(
(
(
,
(14.46)
where again
c =
cu cy
cu cy
.
Repeated application of order of magnitude analysis reduces Eq. (14.46) to
0 =
c P
c y
,
(14.47)
leading to
cP / cx = dP / dx
in Eq. (14.45); and Eq. (14.45) reduces to
u
c u
c x
+ v
c u
c y
|
\
|
.
|
|
=
dP
dx
+ K
c
c y
c u
c y
n
. (14.48)
Equation (14.48) together with the continuity equation, Eq. (14.43), constitute
the boundary layer equations for a power-law fluid.
Equation (14.48) may be put in dimensionless form to yield the definition of a
Reynolds number for a power-law fluid,
Re =
L
n
U
2n
r
K
. (14.49)
Again, for n = 1, this reduces to the Reynolds number for a Newtonian fluid.
References
R. B. Bird, R.C. Armstrong and O. Hassager, "Dynamics of Polymeric Liquids,"
Vol. 1, "Fluid Mechanics," Wiley, New York, 1977.
J. A. Brydson, "Flow Properties of Polymer Melts," Van Nostrand, London, 1970.
B.D. Coleman, H. Markowitz and W. Noll, "Viscometric Flows of Non-Newtonian
Fluids," Springer-Verlag, New York, 1966.
S. Middleman, "The Flow of High Polymers," Wiley-Interscience, New York,
1968.
M. Reiner, "Deformation, Strain and Flow," Wiley-Interscience, New York, 1960.
A.V. Tobolsky, "Properties and Structure of Polymers," Wiley, New York, 1960.
14. Non-Newtonian Fluids 455
Problems
14.1 Show that for the laminar flow of a Bingham plastic fluid the velocity dis-
tribution in a circular pipe is given by
u =
1
AP
4L
R
2
r
2
( )
t
o
R r ( )
(
( , r
o
s r < R,
and
u =
1
APR
2
4L
+
t
o
2
AP / L
t
o
R
(
(
(
= const, 0 s r < r
o
,
where r
o
is the pipe radius corresponding to t
o
,
t
o
>0. Further, show that this
may be integrated to give
Q =
tR
4
AP
8L
1
8
3
Lt
o
AP R
+
1
3
2Lt
o
PR
|
\
|
.
|
4
(
(
(
.
14.2 Consider film flow of a Bingham plastic material on the vertical wall de-
scribed in Example 14.2. Derive expressions for the velocity profile, aver-
age velocity and volumetric flowrate for this case.
14.3 The following data relate the flow rate Q to the pressure drop AP for the flow
of a non-Newtonian fluid through a capillary tube of diameter 2 mm and of
length 25 cm.
Use the data to predict the pressure drop required to cause the same fluid to
flow at u
ave
= 0.1
m/s through a pipe with a diameter of 3 cm and a
length of
100 m.
14.4
It is conventional to represent the momentum associated with fluid passing
through a given cross section of a tube as , mV
where
m
is the mass
flowrate and V is
the average fluid velocity. In reality a velocity distribution
exists, and a factor | should be introduced to take this into account,
A P ( N/cm ) 12 16 20 24 27 29 30
Q (cm /s ) 0.02 0.055 0.107 0.193 0.306 0.490 0.70
3
2
456 Fluid Mechanics
i.e., . M mV | = Determine the numerical value of | for the laminar flow of
a power-law fluid in a circular pipe.
14.5 A power-law fluid flows in a pipe, which has the diameter D. The same
fluid flows in the gap between two parallel flat plates. The size of the gap is
also D. The same pressure gradient exists in both systems. Calculate the
ratio between the flowrate of the fluid in the pipe and the flowrate between
the plates per width of D. Perform your calculations for n = 0.2, 0.4, 0.8.
Compare your results with those of Problem 6.15.
14.6 The pipe and the plates of Problem 14.5 are now used such that the flowrate
in the pipe is the same as that between the plates per width D. Calculate the
ratio between the power needed to pump the fluid through the pipe and that
needed to pump between the plates.
14.7 A power-law fluid with n = 0.5 and K = 0.001 (in SI units) flows at the rate of
0.05 m
3
/s through the annular gap with an inner radius of 0.10 m and an
outer radius of 0.12 m. The length of the annulus is 100 m. Assume the flow
laminar and fully developed.
a. Find the pressure drop along the annulus.
b. Find the magnitude and location of the maximum velocity in the annular
gap.
c. Compare your results with those of Problem 6.25.
14.8 Measurements of pressure drop along a fully developed flow in a pipe of
diameter 0.05 m gave the following readings:
Q (m
3
/s) 0.001 0.002
AP/L (Pa/m) 10
4
1.5 10
4
Assuming a power-law fluid, find K and n.
14.9 For the power-law fluid of Problem 14.8, find the volumetric flowrates in a
pipe of 0.1 m diameter for pressure drops of 0.510
4
Pa/m and of
2.010
4
Pa/m.
14.10 In a plane shear flow the upper plate is 0.1 m above the lower plate and
moves at 1.0 m/s. The fluid is that of Problem 14.8. Find the shear stress on
the plates.
14. Non-Newtonian Fluids 457
14.11 The layer of a power-law fluid (K = 10, n = 1.2) in a plane shear flow
between two parallel plates is 0.01 m thick. The shear stress on the plates is
100 N/m
2
. Find how fast the upper plate moves.
14.12 The lower plate in the shear flow system of Problem 14.11 is cooled while
the upper plate is insulated. As a result, the power-law fluid changes its n
value from 0.8 at the upper plate to 1.0 at the lower plate, linearly.
a. Find the velocity profile between the plates.
b. Find the speed of the upper plate.
458 Fluid Mechanics
459
APPENDIXES
APPENDIX A: UNIT CONVERSION FACTORS
Each of the following conversion factors is dimensionless and equals unity.
Hence, any expression may be multiplied or divided by any of these factors
without change in. its physical magnitude.
Length
2.540 cm/in. 3.281 ft/m 1.609 km/mi 12 in./ft
30.480 cm/ft 5,280 ft/mi 10
10
/m 3 ft/yd
Area
6.452 cm
2
/in.
2
10.764 ft
2
/m
2
144 in.
2
/ft
2
929.0 cm
2
/ft
2
Volume
16.387 cm
3
/in.
3
28.317 L/ft
3
7.48 gal/ft
3
3.7845 L/gal 1,728 in.
3
/ft
3
4 qt/gal
Mass
453.59 g/lbm 14.594 kg/slug 2,000 lbm/ton
2.2046 lbm/kg 32.174 lbm/slug 28.349 g/oz
Force
9.807 N/kgf 7.233 poundal/N 10
5
dyn/N
4.448 N/lbf 32.174 poundal/lbf 2.205 lbf/kgf
460 Fluid Mechanics
Density
62.428
lbm/ ft
3
g/ cm
3
1,000
kg/ m
3
g/ cm
3
32.174
lbm/ ft
3
slug/ ft
3
1.9403
slug/ ft
3
g/ cm
3
16.018
kg/ m
3
lbm/ ft
3
8.345
lbm/ gal
g/ cm
3
Viscosity
100
cp
g/ cm s ( )
1,000
cp
kg/ (m s)
0.6723
lbm/ ft s
( )
kg/ (m s)
2.42
lbm/ ft h
( )
cp
0.06723
lbm/ ft s
( )
g/ cm s ( )
Thermal Conductivity
1.730278
W/m
o
C
Btu/h ft
o
F
241.9
Btu/h ft
o
F
cal/s cm
o
C
Pressure
1.01325
bars
atm
1.0332
kgf / cm
2
atm
144
lbf / ft
2
psi
27.71
in.H
2
O
psi
10
5
N/ m
2
bar
14.696
psi
atm
0.1
dyn/ cm
2
Pa
760
mmHg
atm
100
kPa
bar
2,116.2
lbf / ft
2
atm
2.036
in.Hg
psi
6.8949
kPa
psi
Energy
1.8
Btu/ lbm
kcal / kg
550.0
ft lbf / s
hp
737.56
ft lbf / s
kW
4.18676 kJ / kcal 3,412.8Btu/ kWh 1.35582J / ft lbf ( )
2.6552 10
6
ft lbf / kWh 3,600.0kJ / kWh 0.25199kcal / Btu
1.3410hp/ kW 778.16 ft lbf / Btu 860.0kcal / kWh
0.252kcal / Btu 2,544. 46Btu/ (hp h) 101.92 kgf m/ kJ
1.05505kJ / Btu 1.9810
6
ft lbf / (hp h) 0.746 kW/ hp
Appendixes 461
APPENDIX B: PHYSICAL PROPERTIES OF WATER AND AIR
TABLE B-1: PROPERTIES OF WATER
_____________________________________________________________________
Temperature Specific Density Viscosity Kinematic Surface
weight viscosity tension
T v o
[
o
C ] [ N/m
3
] [ kg/m
3
] [ kg/m
.
s] [ m
2
/s] [ N/m ]
_____________________________________________________________________
0 9,805 999.9 1.79210
-3
1.79210
-6
7.6210
-2
5 9,806 1000.0 1.519 1.519 7.54
10 9,803 999.7 1.308 1.308 7.48
15 9,798 999.1 1.140 1.141 7.41
20 9,789 998.2 1.005 1.007 7.36
25 9,779 997.1 0.894 0.897 7.26
30 9,767 995.7 0.801 0.804 7.18
35 9,752 994.1 0.723 0.727 7.10
40 9,737 992.2 0.656 0.660 7.01
45 9,720 990.2 0.599 0.605 6.92
50 9,697 988.1 0.549 0.556 6.82
60 9,658 983.2 0.469 0.477 6.68
70 9,600 977.8 0.406 0.415 6.50
80 9,557 971.8 0.357 0.367 6.30
90 9,499 965.3 0.317 0.328 6.12
100 9,438 958.4 0.284 0.296 5.94
_____________________________________________________________________
462 Fluid Mechanics
TABLE B-2: PROPERTIES OF AIR AT ATMOSPHERIC PRESSURE
Temperature
Density
Viscosity
Kinematic
T
viscosity
v
[
o
C]
[K]
[kg/m
3
]
[kg/m
.
s]
[m
2
/s]
-50 223 1.582 1.46 10
-5
0.921 10
-5
-40 233 1.514 1.51 0.998
-30 243 1.452 1.56 1.08
-20 253 1.394 1.61 1.16
-10 263 1.342 1.67 1.24
0 273 1.292 1.72 1.33
10 283 1.247 1.76 1.42
20 293 1.204 1.81 1.51
30 303 1.164 1.86 1.60
40 313 1.127 1.91 1.69
50 323 1.092 1.95 1.79
60 333 1.060 2.00 1.89
70 343 1.030 2.05 1.99
80 353 1.000 2.09 2.09
90 363 0.973 2.13 2.19
100 373 0.946 2.17 2.30
150 423 0.834 2.38 2.85
200 473 0.746 2.57 3.45
250 523 0.675 2.75 4.08
300 573 0.616 2.93 4.75
Appendixes 463
APPENDIX C: PROPERTIES OF SOME COMMON FLUIDS
(At 300 K and atmospheric pressure)
Fluid
Density
[kg/m
3
]
Viscosity
[kg/m
.
s]
Kinematic
viscosity
v
[m
2
/s]
Thermal
conductivity
k
[W/m
.
K]
Thermal
diffusivity
o
[m
2
/s]
Prandtl
number
Pr
Air 1.1774
1.846
10
-5
1.568 10
-5
0.0262 0.222 10
-4
0.708
Hydrogen 0.0819 8.963
10
-6
1.095 10
-4
0.182 1.554 10
-4
0.706
Carbon
dioxide
1.797 1.496 10
-5
8.321 10
-6
0.0166 0.106 10
-4
0.770
Water 995.7 8.60 10
-4
8.64 10
-7
0.614 1.475 10
-4
5.85
Mercury 13,580 1.55 10
-3
1.14 10
-7
8.690 4.606 10
-6
0.025
Glycerin 1,264 1.4915 1.18 10
-3
0.286 0.947 10
-7
21.5
Engine oil
*
840 1.752 10
-2
2.03 10
-5
0.137 0.738 10
-4 276
*
At 373 K.
464 Fluid Mechanics
APPENDIX D: COMPRESSIBLE FLOW DATA FOR IDEAL GASES
TABLE D-1: COMPRESSIBLE NOZZLE FLOW, k = c
p
/c
v
= 1.4000
Mass Flux = 0.6847p
o
A* RT
o
p* / p
o
= 0.5283 T* / T
o
= 0.8333 * /
o
= 0.6339
SUBSONIC FLOW, k = c
p
/c
v
= 1.4000
___________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* M T/T
o
p/p
o
/
o
A/A*
___________________________________________________________________________
0.010 1.000 1.000 1.000 57.874 0.020 1.000 1.000 1.000 28.942
0.030 1.000 0.999 1.000 19.301 0.040 1.000 0.999 0.999 14.481
0.050 1.000 0.998 0.999 11.591 0.060 0.999 0.997 0.998 9.666
0.070 0.999 0.997 0.998 8.292 0.080 0.999 0.996 0.997 7.262
0.090 0.998 0.994 0.996 6.461 0.100 0.998 0.993 0.995 5.822
0.110 0.998 0.992 0.994 5.299 0.120 0.997 0.990 0.993 4.864
0.130 0.997 0.988 0.992 4.497 0.140 0.996 0.986 0.990 4.182
0.150 0.996 0.984 0.989 3.910 0.160 0.995 0.982 0.987 3.673
0.170 0.994 0.980 0.986 3.464 0.180 0.994 0.978 0.984 3.278
0.190 0.993 0.975 0.982 3.112 0.200 0.992 0.972 0.980 2.964
0.210 0.991 0.970 0.978 2.829 0.220 0.990 0.967 0.976 2.708
0.230 0.990 0.964 0.974 2.597 0.240 0.989 0.961 0.972 2.496
0.250 0.988 0.957 0.969 2.403 0.260 0.987 0.954 0.967 2.317
0.270 0.986 0.951 0.964 2.238 0.280 0.985 0.947 0.962 2.166
0.290 0.983 0.943 0.959 2.098 0.300 0.982 0.939 0.956 2.035
0.310 0.981 0.936 0.954 1.977 0.320 0.980 0.932 0.951 1.922
0.330 0.979 0.927 0.948 1.871 0.340 0.977 0.923 0.944 1.823
0.350 0.976 0.919 0.941 1.778 0.360 0.975 0.914 0.938 1.736
0.370 0.973 0.910 0.935 1.696 0.380 0.972 0.905 0.931 1.659
0.390 0.970 0.900 0.928 1.623 0.400 0.969 0.896 0.924 1.590
0.410 0.967 0.891 0.921 1.559 0.420 0.966 0.886 0.917 1.529
0.430 0.964 0.881 0.913 1.501 0.440 0.963 0.875 0.909 1.474
0.450 0.961 0.870 0.906 1.449 0.460 0.959 0.865 0.902 1.425
0.470 0.958 0.860 0.898 1.402 0.480 0.956 0.854 0.893 1.380
0.490 0.954 0.849 0.889 1.359 0.500 0.952 0.843 0.885 1.340
0.510 0.951 0.837 0.881 1.321 0.520 0.949 0.832 0.877 1.303
0.530 0.947 0.826 0.872 1.286 0.540 0.945 0.820 0.868 1.270
0.550 0.943 0.814 0.863 1.255 0.560 0.941 0.808 0.859 1.240
0.570 0.939 0.802 0.854 1.226 0.580 0.937 0.796 0.850 1.213
0.590 0.935 0.790 0.845 1.200 0.600 0.933 0.784 0.840 1.188
0.610 0.931 0.778 0.836 1.177 0.620 0.929 0.772 0.831 1.166
0.630 0.926 0.765 0.826 1.155 0.640 0.924 0.759 0.821 1.145
0.650 0.922 0.753 0.816 1.136 0.660 0.920 0.746 0.812 1.127
Appendixes 465
SUBSONIC FLOW, k = c
p
/c
v
= 1.4000 (cont. )
___________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* M T/T
o
p/p
o
/
o
A/A*
___________________________________________________________________________
0.670 0.918 0.740 0.807 1.118 0.680 0.915 0.734 0.802 1.110
0.690 0.913 0.727 0.797 1.102 0.700 0.911 0.721 0.792 1.094
0.710 0.908 0.714 0.787 1.087 0.720 0.906 0.708 0.781 1.081
0.730 0.904 0.702 0.776 1.074 0.740 0.901 0.695 0.771 1.068
0.750 0.899 0.689 0.766 1.062 0.760 0.896 0.682 0.761 1.057
0.770 0.894 0.676 0.756 1.052 0.780 0.892 0.669 0.750 1.047
0.790 0.889 0.663 0.745 1.043 0.800 0.887 0.656 0.740 1.038
0.810 0.884 0.650 0.735 1.034 0.820 0.881 0.643 0.729 1.030
0.830 0.879 0.636 0.724 1.027 0.840 0.876 0.630 0.719 1.024
0.850 0.874 0.624 0.714 1.021 0.860 0.871 0.617 0.708 1.018
0.870 0.869 0.611 0.703 1.015 0.880 0.866 0.604 0.698 1.013
0.890 0.863 0.598 0.692 1.011 0.900 0.861 0.591 0.687 1.009
0.910 0.858 0.585 0.682 1.007 0.920 0.855 0.578 0.676 1.006
0.930 0.853 0.572 0.671 1.004 0.940 0.850 0.566 0.666 1.003
0.950 0.847 0.559 0.660 1.002 0.960 0.844 0.553 0.655 1.001
0.970 0.842 0.547 0.650 1.001 0.980 0.839 0.541 0.645 1.000
0.990 0.836 0.534 0.639 1.000 1.000 0.833 0.528 0.634 1.000
466 Fluid Mechanics
SUPERSONIC FLOW, k = c
p
/c
v
= 1.4000
______________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* v M
2
p
2
/p
1
T
2
/T
1
2
/
1
______________________________________________________________________________
1.10 0.8052 0.4684 0.5817 1.0079 0.02 0.9118 1.2450 1.0649 1.1691
1.20 0.7764 0.4124 0.5311 1.0304 0.06 0.8422 1.5133 1.1280 1.3416
1.30 0.7474 0.3609 0.4829 1.0663 0.11 0.7860 1.8050 1.1909 1.5157
1.40 0.7184 0.3142 0.4374 1.1149 0.16 0.7397 2.1200 1.2547 1.6897
1.50 0.6897 0.2724 0.3950 1.1762 0.21 0.7011 2.4583 1.3202 1.8621
1.60 0.6614 0.2353 0.3557 1.2502 0.26 0.6684 2.8200 1.3880 2.0317
1.70 0.6337 0.2026 0.3197 1.3376 0.31 0.6405 3.2050 1.4583 2.1977
1.80 0.6068 0.1740 0.2868 1.4390 0.36 0.6165 3.6133 1.5316 2.3592
1.90 0.5807 0.1492 0.2570 1.5553 0.41 0.5956 4.0450 1.6079 2.5157
2.00 0.5556 0.1278 0.2300 1.6875 0.46 0.5774 4.5000 1.6875 2.6667
2.10 0.5313 0.1094 0.2058 1.8369 0.51 0.5613 4.9783 1.7704 2.8119
2.20 0.5081 0.0935 0.1841 2.0050 0.55 0.5471 5.4800 1.8569 2.9512
2.30 0.4859 0.0800 0.1646 2.1931 0.60 0.5344 6.0050 1.9468 3.0845
2.40 0.4647 0.0684 0.1472 2.4031 0.64 0.5231 6.5533 2.0403 3.2119
2.50 0.4444 0.0585 0.1317 2.6367 0.68 0.5130 7.1250 2.1375 3.3333
2.60 0.4252 0.0501 0.1179 2.8960 0.72 0.5039 7.7200 2.2383 3.4490
2.70 0.4068 0.0430 0.1056 3.1830 0.76 0.4956 8.3383 2.3429 3.5590
2.80 0.3894 0.0368 0.0946 3.5001 0.80 0.4882 8.9800 2.4512 3.6636
2.90 0.3729 0.0317 0.0849 3.8498 0.83 0.4814 9.6450 2.5632 3.7629
3.00 0.3571 0.0272 0.0762 4.2346 0.87 0.4752 10.3333 2.6790 3.8571
3.10 0.3422 0.0234 0.0685 4.6573 0.90 0.4695 11.0450 2.7986 3.9466
3.20 0.3281 0.0202 0.0617 5.1209 0.93 0.4643 11.7800 2.9220 4.0315
3.30 0.3147 0.0175 0.0555 5.6286 0.96 0.4596 12.5383 3.0492 4.1120
3.40 0.3019 0.0151 0.0501 6.1837 0.99 0.4552 13.3200 3.1802 4.1884
3.50 0.2899 0.0131 0.0452 6.7896 1.02 0.4512 14.1250 3.3150 4.2609
3.60 0.2784 0.0114 0.0409 7.4501 1.05 0.4474 14.9533 3.4537 4.3296
3.70 0.2675 0.0099 0.0370 8.1691 1.08 0.4439 15.8050 3.5962 4.3949
3.80 0.2572 0.0086 0.0335 8.9506 1.10 0.4407 16.6800 3.7426 4.4568
3.90 0.2474 0.0075 0.0304 9.7990 1.12 0.4377 17.5783 3.8928 4.5156
4.00 0.2381 0.0066 0.0277 10.7187 1.15 0.4350 18.5000 4.0469 4.5714
4.10 0.2293 0.0058 0.0252 11.7146 1.17 0.4324 19.4450 4.2048 4.6245
4.20 0.2208 0.0051 0.0229 12.7916 1.19 0.4299 20.4133 4.3666 4.6749
4.30 0.2129 0.0044 0.0209 13.9549 1.21 0.4277 21.4050 4.5322 4.7229
4.40 0.2053 0.0039 0.0191 15.2098 1.23 0.4255 22.4200 4.7017 4.7685
4.50 0.1980 0.0035 0.0174 16.5622 1.25 0.4236 23.4583 4.8751 4.8119
4.60 0.1911 0.0031 0.0160 18.0178 1.27 0.4217 24.5200 5.0523 4.8532
4.70 0.1846 0.0027 0.0146 19.5828 1.29 0.4199 25.6050 5.2334 4.8926
4.80 0.1783 0.0024 0.0134 21.2637 1.31 0.4183 26.7133 5.4184 4.9301
4.90 0.1724 0.0021 0.0123 23.0671 1.33 0.4167 27.8450 5.6073 4.9659
5.00 0.1667 0.0019 0.0113 24.9999 1.34 0.4152 29.0000 5.8000 5.0000
5.10 0.1612 0.0017 0.0104 27.0695 1.36 0.4138 30.1783 5.9966 5.0326
Appendixes 467
TABLE D-2: COMPRESSIBLE NOZZLE FLOW, k = c
p
/c
v
= 1.6667
Mass Flux = 0.7262 p
o
A* RT
o
p* / p
o
= 0.4871 T* / T
o
= 0.7500 * /
o
= 0.6495
SUBSONIC FLOW, k = c
p
/c
v
= 1.6667
______________________________________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* M T/T
o
p/p
o
/
o
A/A*
______________________________________________________________________________
0.010 1.000 1.000 1.000 56.254 0.020 1.000 1.000 1.000 28.132
0.030 1.000 0.999 1.000 18.761 0.040 0.999 0.999 0.999 14.077
0.050 0.999 0.998 0.999 11.269 0.060 0.999 0.997 0.998 9.397
0.070 0.998 0.996 0.998 8.062 0.080 0.998 0.995 0.997 7.061
0.090 0.997 0.993 0.996 6.284 0.100 0.997 0.992 0.995 5.663
0.110 0.996 0.990 0.994 5.155 0.120 0.995 0.988 0.993 4.733
0.130 0.994 0.986 0.992 4.376 0.140 0.994 0.984 0.990 4.071
0.150 0.993 0.981 0.989 3.806 0.160 0.992 0.979 0.987 3.576
0.170 0.990 0.976 0.986 3.373 0.180 0.989 0.974 0.984 3.193
0.190 0.988 0.971 0.982 3.032 0.200 0.987 0.967 0.980 2.888
0.210 0.986 0.964 0.978 2.758 0.220 0.984 0.961 0.976 2.640
0.230 0.983 0.957 0.974 2.533 0.240 0.981 0.954 0.972 2.435
0.250 0.980 0.950 0.970 2.345 0.260 0.978 0.946 0.967 2.262
0.270 0.976 0.942 0.965 2.186 0.280 0.975 0.938 0.962 2.115
0.290 0.973 0.933 0.959 2.050 0.300 0.971 0.929 0.957 1.989
0.310 0.969 0.924 0.954 1.933 0.320 0.967 0.920 0.951 1.880
0.330 0.965 0.915 0.948 1.831 0.340 0.963 0.910 0.945 1.784
0.350 0.961 0.905 0.942 1.741 0.360 0.959 0.900 0.939 1.700
0.370 0.956 0.894 0.935 1.662 0.380 0.954 0.889 0.932 1.626
0.390 0.952 0.884 0.929 1.592 0.400 0.949 0.878 0.925 1.560
0.410 0.947 0.873 0.921 1.530 0.420 0.944 0.867 0.918 1.501
0.430 0.942 0.861 0.914 1.474 0.440 0.939 0.855 0.910 1.449
0.450 0.937 0.849 0.907 1.424 0.460 0.934 0.843 0.903 1.401
0.470 0.931 0.837 0.899 1.380 0.480 0.929 0.831 0.895 1.359
0.490 0.926 0.825 0.891 1.339 0.500 0.923 0.819 0.887 1.320
0.510 0.920 0.812 0.883 1.302 0.520 0.917 0.806 0.879 1.286
0.530 0.914 0.800 0.874 1.269 0.540 0.911 0.793 0.870 1.254
0.550 0.908 0.786 0.866 1.239 0.560 0.905 0.780 0.861 1.225
0.570 0.902 0.773 0.857 1.212 0.580 0.899 0.767 0.853 1.200
0.590 0.896 0.760 0.848 1.187 0.600 0.893 0.753 0.844 1.176
0.610 0.890 0.747 0.839 1.165 0.620 0.886 0.740 0.835 1.155
0.630 0.883 0.733 0.830 1.145 0.640 0.880 0.726 0.825 1.135
0.650 0.877 0.719 0.821 1.126 0.660 0.873 0.713 0.816 1.118
0.670 0.870 0.706 0.811 1.110 0.680 0.866 0.699 0.807 1.102
0.690 0.863 0.692 0.802 1.094 0.700 0.860 0.685 0.797 1.088
0.710 0.856 0.678 0.792 1.081 0.720 0.853 0.671 0.787 1.075
0.730 0.849 0.664 0.783 1.069 0.740 0.846 0.658 0.778 1.063
0.750 0.842 0.651 0.773 1.058 0.760 0.839 0.644 0.768 1.053
468 Fluid Mechanics
0.770 0.835 0.637 0.763 1.048 0.780 0.831 0.630 0.758 1.043
0.790 0.828 0.623 0.753 1.039 0.800 0.824 0.617 0.748 1.035
SUBSONIC FLOW, k = c
p
/c
v
= 1.6667 (cont. )
_____________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* M T/T
o
p/p
o
/
o
A/A*
______________________________________________________________________________
0.810 0.821 0.610 0.743 1.031 0.820 0.817 0.603 0.738 1.028
0.830 0.813 0.596 0.733 1.025 0.840 0.810 0.590 0.728 1.022
0.850 0.806 0.583 0.723 1.019 0.860 0.802 0.576 0.719 1.016
0.870 0.799 0.570 0.714 1.014 0.880 0.795 0.563 0.709 1.012
0.890 0.791 0.557 0.704 1.010 0.900 0.787 0.550 0.699 1.008
0.910 0.784 0.544 0.694 1.006 0.920 0.780 0.537 0.689 1.005
0.930 0.776 0.531 0.684 1.004 0.940 0.772 0.524 0.679 1.003
0.950 0.769 0.518 0.674 1.002 0.960 0.765 0.512 0.669 1.001
0.970 0.761 0.506 0.664 1.001 0.980 0.757 0.499 0.659 1.000
0.990 0.754 0.493 0.654 1.000 1.000 0.750 0.487 0.650 1.000
Appendixes 469
SUPERSONIC FLOW, k = c
p
/c
v
= 1.6667
______________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* v M
2
p
2
/p
1
T
2
/T
1
2
/
1
______________________________________________________________________________
1.10 0.7126 0.4286 0.6015 1.0071 0.02 0.9131 1.2625 1.0982 1.1496
1.20 0.6757 0.3753 0.5554 1.0267 0.05 0.8462 1.5500 1.1948 1.2973
1.30 0.6396 0.3272 0.5116 1.0575 0.09 0.7934 1.8625 1.2922 1.4414
1.40 0.6048 0.2845 0.4704 1.0983 0.14 0.7508 2.2000 1.3919 1.5806
1.50 0.5714 0.2468 0.4320 1.1484 0.18 0.7157 2.5625 1.4948 1.7143
1.60 0.5396 0.2139 0.3963 1.2076 0.22 0.6864 2.9500 1.6018 1.8417
1.70 0.5093 0.1851 0.3635 1.2754 0.26 0.6618 3.3625 1.7133 1.9626
1.80 0.4808 0.1603 0.3334 1.3520 0.30 0.6407 3.8000 1.8297 2.0769
1.90 0.4538 0.1388 0.3058 1.4372 0.34 0.6226 4.2625 1.9512 2.1845
2.00 0.4286 0.1202 0.2806 1.5312 0.38 0.6070 4.7500 2.0782 2.2857
2.10 0.4048 0.1043 0.2576 1.6341 0.42 0.5933 5.2625 2.2107 2.3805
2.20 0.3826 0.0906 0.2367 1.7462 0.45 0.5813 5.8000 2.3488 2.4693
2.30 0.3619 0.0788 0.2177 1.8675 0.48 0.5707 6.3625 2.4927 2.5524
2.40 0.3425 0.0686 0.2004 1.9983 0.52 0.5613 6.9500 2.6425 2.6301
2.50 0.3243 0.0599 0.1847 2.1390 0.55 0.5530 7.5625 2.7982 2.7026
2.60 0.3074 0.0524 0.1704 2.2898 0.58 0.5455 8.2000 2.9598 2.7704
2.70 0.2915 0.0459 0.1574 2.4510 0.60 0.5388 8.8626 3.1275 2.8338
2.80 0.2767 0.0403 0.1456 2.6228 0.63 0.5327 9.5501 3.3012 2.8929
2.90 0.2629 0.0354 0.1348 2.8057 0.66 0.5272 10.2626 3.4809 2.9482
3.00 0.2500 0.0313 0.1250 2.9999 0.68 0.5222 11.0001 3.6668 2.9999
3.10 0.2379 0.0276 0.1160 3.2058 0.70 0.5177 11.7626 3.8587 3.0483
3.20 0.2266 0.0244 0.1079 3.4236 0.72 0.5136 12.5501 4.0568 3.0936
3.30 0.2160 0.0217 0.1004 3.6539 0.75 0.5098 13.3626 4.2610 3.1360
3.40 0.2060 0.0193 0.0935 3.8968 0.77 0.5063 14.2001 4.4714 3.1757
3.50 0.1967 0.0172 0.0873 4.1527 0.79 0.5031 15.0626 4.6880 3.2130
3.60 0.1880 0.0153 0.0815 4.4220 0.80 0.5002 15.9501 4.9107 3.2480
3.70 0.1797 0.0137 0.0762 4.7051 0.82 0.4974 16.8626 5.1396 3.2809
3.80 0.1720 0.0123 0.0713 5.0023 0.84 0.4949 17.8001 5.3747 3.3118
3.90 0.1647 0.0110 0.0669 5.3139 0.85 0.4926 18.7626 5.6160 3.3409
4.00 0.1579 0.0099 0.0627 5.6403 0.87 0.4904 19.7501 5.8635 3.3683
4.10 0.1514 0.0089 0.0589 5.9819 0.89 0.4884 20.7626 6.1172 3.3941
4.20 0.1453 0.0081 0.0554 6.3390 0.90 0.4865 21.8001 6.3771 3.4185
4.30 0.1396 0.0073 0.0522 6.7121 0.91 0.4848 22.8626 6.6432 3.4415
4.40 0.1342 0.0066 0.0491 7.1014 0.93 0.4831 23.9501 6.9156 3.4632
4.50 0.1290 0.0060 0.0464 7.5073 0.94 0.4816 25.0626 7.1941 3.4838
4.60 0.1242 0.0054 0.0438 7.9302 0.95 0.4801 26.2002 7.4789 3.5032
4.70 0.1196 0.0049 0.0413 8.3705 0.96 0.4788 27.3627 7.7699 3.5216
4.80 0.1152 0.0045 0.0391 8.8285 0.98 0.4775 28.5502 8.0672 3.5391
4.90 0.1111 0.0041 0.0370 9.3046 0.99 0.4763 29.7627 8.3706 3.5556
5.00 0.1071 0.0038 0.0351 9.7992 1.00 0.4752 31.0002 8.6803 3.5713
470 Fluid Mechanics
5.10 0.1034 0.0034 0.0333 10.3126 1.01 0.4741 32.2627 8.9963 3.5862
TABLE D-3: COMPRESSIBLE NOZZLE FLOW, k = c
p
/c
v
= 1.3000
Mass Flux = 0.6673p
o
A * RT
o
p* / p
o
= 0.5457 T* / T
o
= 0.8696 * /
o
= 0.6276
SUBSONIC FLOW, k = c
p
/c
v
= 1.3000
_____________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* M T/T
o
p/p
o
/
o
A/A*
______________________________________________________________________________
0.010 1.000 1.000 1.000 58.526 0.020 1.000 1.000 1.000 29.268
0.030 1.000 0.999 1.000 19.518 0.040 1.000 0.999 0.999 14.644
0.050 1.000 0.998 0.999 11.721 0.060 0.999 0.998 0.998 9.774
0.070 0.999 0.997 0.998 8.384 0.080 0.999 0.996 0.997 7.342
0.090 0.999 0.995 0.996 6.533 0.100 0.999 0.994 0.995 5.886
0.110 0.998 0.992 0.994 5.357 0.120 0.998 0.991 0.993 4.917
0.130 0.997 0.989 0.992 4.546 0.140 0.997 0.987 0.990 4.228
0.150 0.997 0.986 0.989 3.952 0.160 0.996 0.984 0.987 3.712
0.170 0.996 0.981 0.986 3.500 0.180 0.995 0.979 0.984 3.312
0.190 0.995 0.977 0.982 3.145 0.200 0.994 0.974 0.980 2.994
0.210 0.993 0.972 0.978 2.858 0.220 0.993 0.969 0.976 2.735
0.230 0.992 0.966 0.974 2.623 0.240 0.991 0.963 0.972 2.520
0.250 0.991 0.960 0.969 2.426 0.260 0.990 0.957 0.967 2.340
0.270 0.989 0.954 0.964 2.260 0.280 0.988 0.951 0.962 2.186
0.290 0.988 0.947 0.959 2.117 0.300 0.987 0.944 0.956 2.054
0.310 0.986 0.940 0.953 1.994 0.320 0.985 0.936 0.950 1.939
0.330 0.984 0.932 0.947 1.887 0.340 0.983 0.928 0.944 1.839
0.350 0.982 0.924 0.941 1.793 0.360 0.981 0.920 0.938 1.750
0.370 0.980 0.916 0.934 1.710 0.380 0.979 0.911 0.931 1.672
0.390 0.978 0.907 0.928 1.636 0.400 0.977 0.902 0.924 1.602
0.410 0.975 0.898 0.920 1.570 0.420 0.974 0.893 0.917 1.540
0.430 0.973 0.888 0.913 1.511 0.440 0.972 0.883 0.909 1.484
0.450 0.971 0.878 0.905 1.459 0.460 0.969 0.873 0.901 1.434
0.470 0.968 0.868 0.897 1.411 0.480 0.967 0.863 0.893 1.389
0.490 0.965 0.858 0.889 1.368 0.500 0.964 0.853 0.885 1.348
0.510 0.962 0.847 0.880 1.329 0.520 0.961 0.842 0.876 1.311
0.530 0.960 0.836 0.871 1.293 0.540 0.958 0.831 0.867 1.277
0.550 0.957 0.825 0.863 1.261 0.560 0.955 0.819 0.858 1.246
0.570 0.954 0.814 0.853 1.232 0.580 0.952 0.808 0.849 1.219
0.590 0.950 0.802 0.844 1.206 0.600 0.949 0.796 0.839 1.193
0.610 0.947 0.790 0.834 1.181 0.620 0.945 0.784 0.830 1.170
0.630 0.944 0.778 0.825 1.159 0.640 0.942 0.772 0.820 1.149
0.650 0.940 0.766 0.815 1.139 0.660 0.939 0.760 0.810 1.130
0.670 0.937 0.754 0.805 1.121 0.680 0.935 0.748 0.800 1.113
0.690 0.933 0.742 0.795 1.105 0.700 0.932 0.735 0.789 1.097
Appendixes 471
0.710 0.930 0.729 0.784 1.090 0.720 0.928 0.723 0.779 1.083
SUBSONIC FLOW, k = c
p
/c
v
= 1.3000 (cont.)
_____________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* M T/T
o
p/p
o
/
o
A/A*
______________________________________________________________________________
0.730 0.926 0.717 0.774 1.077 0.740 0.924 0.710 0.769 1.070
0.750 0.922 0.704 0.763 1.064 0.760 0.920 0.698 0.758 1.059
0.770 0.918 0.691 0.753 1.054 0.780 0.916 0.685 0.747 1.049
0.790 0.914 0.679 0.742 1.044 0.800 0.912 0.672 0.737 1.040
0.810 0.910 0.666 0.731 1.035 0.820 0.908 0.659 0.726 1.032
0.830 0.906 0.653 0.721 1.028 0.840 0.904 0.647 0.715 1.025
0.850 0.902 0.640 0.710 1.021 0.860 0.900 0.634 0.704 1.019
0.870 0.898 0.628 0.699 1.016 0.880 0.896 0.621 0.693 1.013
0.890 0.894 0.615 0.688 1.011 0.900 0.892 0.608 0.682 1.009
0.910 0.890 0.602 0.677 1.007 0.920 0.887 0.596 0.671 1.006
0.930 0.885 0.589 0.666 1.004 0.940 0.883 0.583 0.660 1.003
0.950 0.881 0.577 0.655 1.002 0.960 0.879 0.571 0.649 1.001
0.970 0.876 0.564 0.644 1.001 0.980 0.874 0.558 0.639 1.000
0.990 0.872 0.552 0.633 1.000 1.000 0.870 0.546 0.628 1.000
472 Fluid Mechanics
SUPERSONIC FLOW, k = c
p
/c
v
= 1.3000
______________________________________________________________________________
M T/T
o
p/p
o
/
o
A/A* v M
2
p
2
/p
1
T
2
/T
1
2
/
1
______________________________________________________________________________
1.10 0.8464 0.4854 0.5735 1.0083 0.02 0.9112 1.2374 1.0506 1.1777
1.20 0.8224 0.4285 0.5211 1.0321 0.07 0.8403 1.4974 1.0995 1.3618
1.30 0.7978 0.3757 0.4709 1.0703 0.11 0.7825 1.7800 1.1480 1.5505
1.40 0.7728 0.3273 0.4235 1.1227 0.17 0.7346 2.0852 1.1971 1.7419
1.50 0.7477 0.2836 0.3793 1.1895 0.22 0.6942 2.4130 1.2473 1.9346
1.60 0.7225 0.2446 0.3385 1.2712 0.28 0.6599 2.7635 1.2991 2.1272
1.70 0.6976 0.2100 0.3011 1.3690 0.33 0.6304 3.1365 1.3529 2.3185
1.80 0.6729 0.1797 0.2671 1.4841 0.39 0.6048 3.5322 1.4087 2.5074
1.90 0.6487 0.1533 0.2363 1.6182 0.45 0.5825 3.9504 1.4668 2.6932
2.00 0.6250 0.1305 0.2087 1.7732 0.50 0.5629 4.3913 1.5274 2.8750
2.10 0.6019 0.1108 0.1841 1.9514 0.55 0.5455 4.8548 1.5905 3.0524
2.20 0.5794 0.0939 0.1621 2.1556 0.61 0.5301 5.3409 1.6562 3.2248
2.30 0.5576 0.0795 0.1427 2.3885 0.66 0.5163 5.8496 1.7245 3.3920
2.40 0.5365 0.0673 0.1255 2.6535 0.71 0.5040 6.3809 1.7956 3.5536
2.50 0.5161 0.0569 0.1103 2.9545 0.75 0.4929 6.9348 1.8694 3.7097
2.60 0.4965 0.0481 0.0969 3.2954 0.80 0.4829 7.5113 1.9459 3.8600
2.70 0.4777 0.0407 0.0852 3.6811 0.85 0.4738 8.1104 2.0253 4.0045
2.80 0.4596 0.0344 0.0749 4.1165 0.89 0.4655 8.7322 2.1075 4.1434
2.90 0.4422 0.0291 0.0659 4.6073 0.93 0.4580 9.3765 2.1925 4.2766
3.00 0.4255 0.0247 0.0580 5.1598 0.97 0.4511 10.0435 2.2804 4.4043
3.10 0.4096 0.0209 0.0510 5.7807 1.01 0.4448 10.7330 2.3711 4.5265
3.20 0.3943 0.0177 0.0450 6.4776 1.05 0.4389 11.4452 2.4648 4.6435
3.30 0.3797 0.0151 0.0396 7.2586 1.09 0.4336 12.1800 2.5613 4.7555
3.40 0.3658 0.0128 0.0350 8.1328 1.12 0.4287 12.9374 2.6607 4.8625
3.50 0.3524 0.0109 0.0309 9.1098 1.16 0.4241 13.7174 2.7630 4.9648
3.60 0.3397 0.0093 0.0273 10.2004 1.19 0.4199 14.5200 2.8681 5.0625
3.70 0.3275 0.0079 0.0242 11.4160 1.22 0.4160 15.3452 2.9762 5.1559
3.80 0.3159 0.0068 0.0215 12.7693 1.25 0.4123 16.1930 3.0873 5.2451
3.90 0.3047 0.0058 0.0190 14.2737 1.28 0.4089 17.0635 3.2012 5.3303
4.00 0.2941 0.0050 0.0169 15.9441 1.31 0.4058 17.9565 3.3180 5.4118
4.10 0.2840 0.0043 0.0151 17.7963 1.34 0.4028 18.8722 3.4378 5.4896
4.20 0.2743 0.0037 0.0134 19.8475 1.37 0.4000 19.8104 3.5605 5.5639
4.30 0.2650 0.0032 0.0120 22.1161 1.39 0.3975 20.7713 3.6861 5.6350
4.40 0.2561 0.0027 0.0107 24.6221 1.42 0.3950 21.7548 3.8147 5.7029
4.50 0.2477 0.0024 0.0095 27.3869 1.44 0.3927 22.7608 3.9462 5.7678
4.60 0.2396 0.0020 0.0085 30.4334 1.47 0.3906 23.7895 4.0806 5.8299
4.70 0.2318 0.0018 0.0077 33.7863 1.49 0.3886 24.8408 4.2180 5.8893
4.80 0.2244 0.0015 0.0069 37.4719 1.51 0.3867 25.9147 4.3582 5.9461
4.90 0.2173 0.0013 0.0062 41.5185 1.53 0.3849 27.0113 4.5015 6.0005
5.00 0.2105 0.0012 0.0056 45.9564 1.56 0.3832 28.1304 4.6476 6.0526
5.10 0.2040 0.0010 0.0050 50.8177 1.58 0.3816 29.2721 4.7967 6.1025
Appendixes 473
474
TABLE D-4: FANNO LINE, COMPRESSIBLE PIPE FLOW
k = c
p
/c
v
= 1.400
f(L/D) = Dimensionless distance to point of M = 1
_________________________________________________
M T/T
o
p/p
o
R/R
o
f(L/D)
_________________________________________________
0.01 1.000 1.000 1.000 7134.405
0.02 1.000 1.000 1.000 1778.450
0.04 1.000 1.000 0.999 440.352
0.05 1.000 0.999 0.999 280.020
0.10 0.998 0.997 0.995 66.922
0.15 0.996 0.994 0.989 27.932
0.20 0.992 0.989 0.980 14.533
0.25 0.988 0.983 0.969 8.483
0.30 0.982 0.975 0.956 5.299
0.35 0.976 0.967 0.941 3.452
0.40 0.969 0.957 0.924 2.308
0.45 0.961 0.946 0.906 1.566
0.50 0.952 0.934 0.885 1.069
0.55 0.943 0.921 0.863 0.728
0.60 0.933 0.907 0.840 0.491
0.65 0.922 0.893 0.816 0.325
0.70 0.911 0.877 0.792 0.208
0.75 0.899 0.861 0.766 0.127
0.80 0.887 0.845 0.740 0.072
0.85 0.874 0.828 0.714 0.036
0.90 0.861 0.810 0.687 0.015
0.95 0.847 0.793 0.660 0.003
1.00 0.833 0.775 0.634 0.000
1.25 0.762 0.683 0.507 0.049
1.50 0.690 0.594 0.395 0.136
1.75 0.620 0.512 0.303 0.225
2.00 0.556 0.439 0.230 0.305
2.25 0.497 0.376 0.174 0.374
2.50 0.444 0.321 0.132 0.432
2.75 0.398 0.275 0.100 0.481
3.00 0.357 0.237 0.076 0.522
3.25 0.321 0.204 0.059 0.557
3.50 0.290 0.177 0.045 0.586
3.75 0.262 0.154 0.035 0.612
4.00 0.238 0.134 0.028 0.633
4.25 0.217 0.118 0.022 0.652
4.50 0.198 0.104 0.017 0.668
4.75 0.181 0.092 0.014 0.682
5.00 0.167 0.081 0.011 0.694
5.25 0.154 0.073 0.009 0.705
5.50 0.142 0.065 0.008 0.714
5.75 0.131 0.058 0.006 0.722
6.00 0.122 0.053 0.005 0.730
6.25 0.113 0.048 0.004 0.737
6.50 0.106 0.043 0.004 0.743
Appendixes 475
7.00 0.093 0.036 0.003 0.753
TABLE D-5: FANNO LINE, COMPRESSIBLE PIPE FLOW
k = c
p
/c
v
= 1.667
f(L/D) = Dimensionless distance to point of M = 1
_________________________________________________________________________________________________
M T/T
o
p/p
o
R/R
o
f(L/D)
____________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
0.01 1.000 1.000 1.000 5992.262
0.02 1.000 1.000 1.000 1493.371
0.03 1.000 1.000 1.000 660.686
0.04 0.999 0.999 0.999 369.480
0.05 0.999 0.999 0.999 234.836
0.10 0.997 0.994 0.995 55.943
0.15 0.993 0.988 0.989 23.255
0.20 0.987 0.978 0.980 12.044
0.25 0.980 0.966 0.970 6.996
0.30 0.971 0.952 0.957 4.347
0.35 0.961 0.935 0.942 2.816
0.40 0.949 0.917 0.925 1.873
0.45 0.937 0.897 0.907 1.263
0.50 0.923 0.875 0.887 0.857
0.55 0.908 0.852 0.866 0.580
0.60 0.893 0.828 0.844 0.389
0.65 0.877 0.803 0.821 0.256
0.70 0.860 0.777 0.797 0.163
0.75 0.842 0.751 0.773 0.099
0.80 0.824 0.724 0.748 0.056
0.85 0.806 0.698 0.723 0.028
0.90 0.787 0.671 0.699 0.011
0.95 0.769 0.645 0.674 0.002
1.00 0.750 0.619 0.650 0.000
1.25 0.658 0.497 0.533 0.036
1.50 0.571 0.393 0.432 0.098
1.75 0.495 0.310 0.348 0.159
2.00 0.429 0.244 0.281 0.211
2.25 0.372 0.192 0.227 0.255
2.50 0.324 0.153 0.185 0.291
2.75 0.284 0.123 0.151 0.321
3.00 0.250 0.099 0.125 0.346
3.25 0.221 0.081 0.104 0.366
3.50 0.197 0.067 0.087 0.383
3.75 0.176 0.055 0.074 0.397
4.00 0.158 0.046 0.063 0.409
4.25 0.142 0.039 0.054 0.419
4.50 0.129 0.033 0.046 0.428
4.75 0.117 0.028 0.040 0.436
5.00 0.107 0.024 0.035 0.442
5.25 0.098 0.021 0.031 0.448
5.50 0.090 0.018 0.027 0.453
5.75 0.083 0.016 0.024 0.458
6.00 0.077 0.014 0.021 0.462
6.25 0.071 0.012 0.019 0.465
476 Fluid Mechanics
6.50 0.066 0.011 0.017 0.468
6.75 0.062 0.010 0.015 0.471
7.00 0.058 0.009 0.014 0.474
TABLE D-6: FANNO LINE, COMPRESSIBLE PIPE FLOW
k = c
p
/c
v
= 1.300
f(L/D) = Dimensionless distance to point of M = 1
_____________________________________________________________________________________________________
M T/T
o
p/p
o
R/R
o
f(L/D)
_____________________________________________________________________________________________________
0.01 1.000 1.000 1.000 7683.515
0.02 1.000 1.000 1.000 1915.510
0.03 1.000 1.000 1.000 847.851
0.04 1.000 1.000 0.999 474.429
0.05 1.000 1.000 0.999 301.746
0.10 0.999 0.998 0.995 72.202
0.15 0.997 0.996 0.989 30.183
0.20 0.994 0.992 0.980 15.732
0.25 0.991 0.988 0.969 9.201
0.30 0.987 0.983 0.956 5.759
0.35 0.982 0.977 0.941 3.760
0.40 0.977 0.970 0.924 2.520
0.45 0.971 0.962 0.905 1.714
0.50 0.964 0.953 0.885 1.172
0.55 0.957 0.944 0.863 0.800
0.60 0.949 0.934 0.839 0.541
0.65 0.940 0.923 0.815 0.359
0.70 0.932 0.912 0.789 0.230
0.75 0.922 0.900 0.763 0.141
0.80 0.912 0.888 0.737 0.080
0.85 0.902 0.875 0.710 0.041
0.90 0.892 0.862 0.682 0.016
0.95 0.881 0.848 0.655 0.004
1.00 0.870 0.834 0.628 0.000
1.25 0.810 0.761 0.496 0.055
1.50 0.748 0.685 0.379 0.156
1.75 0.685 0.612 0.284 0.261
2.00 0.625 0.543 0.209 0.357
2.25 0.568 0.480 0.152 0.441
2.50 0.516 0.423 0.110 0.514
2.75 0.469 0.373 0.080 0.575
3.25 0.387 0.291 0.042 0.673
3.50 0.352 0.258 0.031 0.711
3.75 0.322 0.229 0.023 0.744
4.00 0.294 0.204 0.017 0.773
4.25 0.270 0.182 0.013 0.797
4.50 0.248 0.163 0.010 0.819
4.75 0.228 0.146 0.007 0.838
5.00 0.211 0.132 0.006 0.854
5.25 0.195 0.119 0.004 0.869
5.50 0.181 0.108 0.003 0.882
5.75 0.168 0.098 0.003 0.893
6.00 0.156 0.090 0.002 0.904
6.25 0.146 0.082 0.002 0.913
6.50 0.136 0.075 0.001 0.921
Appendixes 477
6.75 0.128 0.069 0.001 0.929
7.00 0.120 0.063 0.001 0.935
478
INDEX
479
Absolute pressure 55, 181
Absolute stability 81
Absolute velocity 125, 132
Accelerating coordinate system 59
Acceleration 19, 23, 53, 54, 106,
116, 152
angular 24
Adiabatic flow
with friction 422-424, 426
Adiabatic process 409, 411
irreversible 411
Airplane wing 216
Angle of attack 379, 381
Angular acceleration 24
Angular momentum 124
theorem 123, 124
Angular velocity 126
Annular flow 195
Apparent viscosity 443, 444, 445, 446,
447, 453
Archimedes' law 74, 78
Atmospheric pressure 55, 56, 57
Average velocity 104, 128, 129,
130, 186
Barometer 57
Barotropic flow 222, 314
Bearing 279
Bernoulli equation 121, 213, 215-216,
231, 314, 333, 371, 377, 409
compressible flow, extension to
221-223
strong, 215, 216, 310
weak 215
Bernoulli polynomial 214, 219, 326
Bernoulli surface 215
Bingham plastic 451, 445, 450-452
Blasius equation 337
Blasius solution 334, 340
Body force 4, 5, 20, 107, 113
Boundary layer 306, 307, 333-345, 453
displacement thickness 339
transition 379, 389, 392
turbulent 389-391
Boundary layer equations 262-266,
333-341, 453-454
conservative form of 342
integral form 342-345
Buckingham's t theorem 267-268
Buffer layer 362, 366
Bulk modulus of elasticity 12-14
isothermal 13
isentropic 13
Buoyancy 291
center of 71, 81
force 71, 72, 76, 326, 380
480 Fluid Mechanics
CauchyRiemann conditions 315
Center of
buoyancy 71, 81
gravity 65, 72, 80, 81
pressure 63, 65
Centroid of area 63
of a triangle 67
Circulation 310-317
Compressibility 12, 13, 401
Compressibility factor 385
Compressible flow 14, 221, 401-420
in nozzles 412-420, 430, 434
Concentric cylinder viscometer 207
Conservation of mass 98, 99, 128, 142,
147, 164
Conservative form of
boundary layer equations 342
equations of motion 358
Constitutive equation 444, 445, 447,
452
Continuity, equation of 148-150, 163,
164, 165, 166, 169, 171, 263
Continuum 1-10, 53, 153
Control surface 102
Control volume 97-128, 144, 152
momentum theorem 109
Convergentdivergent nozzle 413,
418, 419
Convergingdiverging stream tube 412
Correction factor for
compressibility 384
momentum flux 128
Couette Flow 33, 184, 186, 198
Couette viscometer 198
Creeping flow 277, 282-293
Critical cross section 414
Critical Reynolds number 389, 392
D'Alambert force 201
Deformation 154, 155, 156
Density 3, 10, 72
Differential Manometer, 57
Discharge coefficient 385, 387
Displacement thickness 339
Divergence theorem 147
Doublet flow 319, 320
Drag 435
Drag coefficient 378, 379-387
Drag force 281, 325, 341, 380
Drag on the sphere 290
Dynamic stability 80
Dynamic viscosity 11
Eddy viscosity 360
Elasticity, bulk modulus of 13, 14
Elliptical pipe 378
Ellis model 445
Enthalpy 222
Equation of
continuity 147-149, 164, 166
motion, conservative form 358
Error function 189
Euler equation 162, 172, 174, 263,
303-307, 310, 333
Euler number 247
Extensive property 98, 100
Fanno line 423-428
Fanno number 425
First law of thermodynamics 222, 231
Float valve 78
Floating bodies 69, 73
Flow around
cylinder 293-295, 324
sphere 287
Flow over flat plate 334
Blasius solution 334-340
Flow field 142, 144, 171
Flow nozzle 384-388
Flows in narrow gaps 277
Fluid, definition of 6, 46
Fluid properties 10-14
Friction
Index 481
coefficient 424
factor 363-372
Friction head 227
Froude number 247
Fully developed flow 180, 182
Gage pressure 55
Gauss' theorem 71
Generalized velocity profile 360-362
Harmonic function 314
Helmholtz theorem 312, 313, 314
High-Reynolds-number flow 307
Hydraulic diameter 377-378
Hydraulic efficiency 227
Hydrostatic pressure 63, 67
Hydrostatic Stability 79
Hydrostatics 53
Ideal gas 403
Inclined manometer 57, 84
Incompressible flow 103, 164, 166
Incompressible fluid 111
Index notation 22
integral momentum equation 108
Integral momentum theorem 108
Irrotational flow 308-315
Isentropic compressibility 436
Isentropic flow 223
Isentropic process 224, 403, 409, 410
Jet engine 118
Kelvin theorem 310
Kelvin's theorem 313
Kinematic viscosity 12
Laminar sublayer 362, 366, 367, 391
Laplace equation 314, 318, 320, 322
Leibnitz's rule 221, 342
Lift 435
Lift coefficient 381
Lift force 216, 280, 380-381
Loss coefficient 374, 375
Mach cone 405
Mach number 404-425
Mach surface 405
Mach wedge 406
Magnetic reader 281
Magnus Effect 325
Manometer 55-58
differential 57
inclined 57, 84
multifluid 57, 83
Manometric fluid 57, 58
Mass flowrate 118
Mass flux 104, 163
Mass velocity 104
Material derivative 99, 142, 144, 145-
146, 311
of a vector 146
Mean free path 361
Membrane equation 305
Membrane parameter. 304
Metacentric height 81
Modified pressure 181-182
Moment of inertia 63, 64, 65
Momentum equations 19-23, 53, 59,
107, 108, 163, 171
Momentum flux correction factor 128
Momentum flux deficiency 340
Momentum theorem for a control
volume 109
Momentum thickness 339, 340
Moody diagram 362-371, 425
NavierStokes equations 160-162, 163,
171, 172, 174, 179-198
Newton's law of viscosity 153, 443
Newton's second law 99, 107, 108,
142, 150-152, 291
Newton's third law 20, 109
Newtonian fluid 153, 154, 157, 443
482 Fluid Mechanics
Non-Newtonian fluids 153, 443-454
Noncircular pipes 377
Nonisentropic flow 421
Normal stress 5, 21, 109
Nozzle 122, 123, 126, 132, 133
Nozzle flow 412-420, 430, 434
Oblique shock 432-435
One-dimensional SISO device 110
Order of magnitude 257-258
Parallel flow 317
Pascal's law 31, 38, 46, 53, 59
Pathline 142, 143
Perfect gas 13
Perturbation 261, 262
regular 262, 304
singular 262, 263
Pipe fittings 373-376
Pipe flow 190-193, 361-373, 448-450
Pitot tube 220, 383-387
Poiseuille
equation 192, 364
flow 35, 184, 193, 194, 227
Potential flow 314, 325, 334, 335, 339
Potential vortex 316
Power-law fluid 444, 445, 448, 449,
450, 453, 454
Prandtl mixing length 360
PrandtlMeyer expansion 432-434
Pressure
absolute 55, 181
gage 55
measurement 58
Pressure function 221, 222
Pressure head 215
Principal directions 38-40, 41, 42,
44, 45
Principal stresses 40, 41, 44, 46
Pseudoplastic fluid 444
Quasi-one-dimensional flow 408
Ram-jet 133
Rankine oval 322
Rate of deformation 153, 156, 157
tensor 157
Rate of strain 11
Rayleigh line 423-428
Rayleigh problem 187-189
Regular perturbation 262, 304
Relative roughness 363, 364, 370
Relative velocity 122, 124
Reynolds experiment 355
Reynolds lubrication theory 279
Reynolds number 247, 255, 269, 336,
355-356, 454
Reynolds stress 360
Reynolds transport theorem 97, 98-
101, 107, 152, 311
Rigid body
motion 59
rotation 175
Rocket 103, 112, 116, 133
Rocket nozzle 112
Roll angle 81
Rotation 155
Rotational flow 312
Roughness parameter, dimensionless
374
Roughness, relative 363, 364, 370
Seventh-root law 391
Shear flow 32, 34, 175
Shear stress 5, 6, 11, 21, 26, 53, 59,
109, 153
Shear velocity 362
Shock wave relations 428
Similarity 249-252, 255, 256, 258
parameter 249, 251-252, 266-269
solution 189
transformation 187
variable 336
Similitude 248-249, 254
Index 483
Single-inputsingle-output (SISO)
device 110
Singular perturbation 262, 263
Sink 316
Sink flow 175
Sonic speed 403
Source flow 168, 170, 171, 175, 315
Specific gravity 11
Specific property 100
Specific weight 11, 54
Speed of sound 403
Sprinkler 126, 127, 133, 138
Stable boat 81
Stagnation enthalpy 409
Stagnation flow 318
Stagnation point 168, 321
Stagnation pressure 414
Stagnation properties 408-411,
424, 429
Static equilibrium 80
Static head 215
Steady state 103, 126
Steiner's theorem 63
Stokes paradox 295
Stokes theorem 310, 312
Streakline 142, 143
Stream function 163, 164, 165, 166,
168, 169, 171, 315, 337
Stream sheet 144, 163
Stream tube 144
Streamline 142, 143, 163, 214
Stress,
normal 5, 6
tangential 5
Stress at a point 1, 4-5
Stress matrix 23, 28, 29, 30-36
Stress tensor 36-44, 46, 153
Strong Bernoulli equation 215,
216, 310
Submarine 80, 81
Submerged
boundaries 70
surface 61, 69, 75
Surface force 4, 6, 20, 107
Surface forces 4,5
Terminal velocity 291, 380
Thermodynamic system 98, 99, 102,
107, 124, 144
Time-independent flow 184
Total enthalpy 222
Turbulence 357-358
Turbulent boundary layer 389-391
Turbulent core 362, 366, 367
Turbulent flow 355-393
Turbulent viscosity 360
Two-dimensional
air bearing 203
flow 163, 165, 166
jet 345-346
source 165
Universal velocity profile 362,
364, 366
Vapor pressure 57
Velocity
absolute 125, 132
relative 122
Velocity field 99, 144, 166, 175
Velocity head 215
Velocity potential 314
Velocity vector 163, 169, 170, 171
Vena contracta 237, 384, 385
Venturi meter 383-388
Venturi tube 216
Viscometer, concentric cylinder 207
Viscosity 11-12
apparent 443, 444, 445, 446, 447,
453
dynamic 11
kinematic 12
Von KarmanPohlhausen integral
method 341, 343
484 Fluid Mechanics
Vortex
line 312
tube 314
Vortex flow 317
Vorticity vector 308-309
Wall roughness 363
Water turbine 124
Water-hammer 119
Work head 227
Zone of silence 405