River Hydraulics

GEOLOGICAL SURVEY WATER-SUPPLY PAPER 1369

This water-supply paper was published as

separate chapters A-E
UNITED STATES DEPARTMENT OF THE INTERIOR

STEWART L. UDALL, Secretary

GEOLOGICAL SURVEY

Thomas B. Nolan, Director

 CONTENTS

[The letters in parentheses designate separately published chapters]

Page
(A) Selected topics of fluid mechanics, by Carl E. Kindsvater_______ 1
(B) Energy losses associated with abrupt enlargements in pipes, with
special reference to the influence of boundary roughness, by Carl E.
Kindsvater ___________________________________ 53
(C) Surges in natural stream channels, by S. B. Rantz ___ 77
(D) Flow through openings in width constrictions, by Jacob Davidian,
P. H. Carrigan, Jr., and John Shen____________________ 91
(E) Stream gaging control structure for the Rio Grande conveyance chan-
nel near Bernardo, New Mexico, by D. D. Harris and E. V.
Richardson ___________________________________ 123

O
Selected Topics
Of Fluid Mechanics
GEOLOGICAL SURVEY WATER-SUPPLY PAPER 1369-A
Selected Topics
Of Fluid Mechanics
By CARL E. KINDSVATER

RIVER HYDRAULICS

GEOLOGICAL SURVEY WATER-SUPPLY PAPER 1369-A

A discussion of the concepts of

fluid mechanics which are the
foundations of river hydraulics

UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1958

 UNITED STATES DEPARTMENT OF THE INTERIOR
FRED A. SEATON, Secretary

GEOLOGICAL SURVEY

Thomas B. Nolan, Director

For sale by the Superintendent of Documents, U. S. Government Printing Office

Washington 25, D. C. - Price 55 cents
 CONTENTS
Page
Abstract- ___________________________________...._-___------------__.. 1
Introduction._____________________________________________________ 2
Technical symbols, terms, and basic concepts ______________________ 2
Letter symbols and units____-______________------------_------- 2
Boundary conditions.______________________________________ 2
Fluid properties.__________________________________________ 3
Motion (flow) properties__________________________________ 3
Force properties_____________-__________-__-------_-------- 4
Ratios and coefRcients________________________-_-_-_-_---__ 4
Mathematical symbols___________-________------------_-------- 5
Basic concepts of mechanics______-_____________-__-_-------_-_- 5
Basic concepts of fluid mechanics________________________________ 7
Equations of motion_______________________________________________ 10
Fundamental equation of motion________________________________ 10
External forces in fluid motion._________________________________ 10
Kinematics of fluid motion___________________________________ 11
Equations for a simple fluid motion______________________________ 12
Hydrostatics.. _______ ___ ____ 16
Piezometry___________________________________________________ 18
One-dimensional equation___________________________-_-_----___ 20
Equation of eontmuity___________-_________--___--_----_-----_- 22
Discharge equation.___________________________________________ 23
Flow net.-_______--____---_-----____----_----_------________ 24
The experimental method______________________________________ 28
Influence of shear____________________________________________ 29
Experiments and the theory of similitude.________________________ 29
Influence of a free surface___________________________-__-----___ 32
Recapitulation ________________________________________________ 33
Example.__________________________________________________ 34
The energy equation_________________ 37
Review of the Bernoulli equation________________________________ 37
General energy equation______________________________________ 40
The momentum equation______________ ____ _ 42
Uniform and gradually varied flow_______________________________ 46
Characteristics of uniform flow_______________________-__________ 46
The Chezy equation_____________________________________-_____ 47
Alternate derivation of the Chezy equation _______________________ 48
Gradually varied flow in open channels_________________________ 50

ILLUSTRATIONS
[Plates follow p. 52]
PLATE 1. Flow through a two-dimensional orifice, piezometers in pipes
and open channels, and three-dimensional orifices.
2. Typical two-dimensional flow nets, two-dimensional sluice gate,
forces acting on a fluid mass, and uniform flow in pipes and
open channels.

XXX
 RIVER HYDRAULICS

SELECTED TOPICS OF FLUID MECHANICS

By CARL E. KINDSVATER*

ABSTRACT

The fundamental equations of fluid mechanics are specific expressions of the

principles of motion which are ascribed to Isaac Newton. Thus, the equations
which form the framework of applied fluid mechanics or hydraulics are, in ad-
dition to the equation of continuity, the Newtonian equations of energy and
momentum. These basic relationships are also the foundations of river hydraulics.
The fundamental equations are developed in this report with sufficient rigor
to support critical examinations of their applicability to most problems met by
hydraulic engineers of the Water Resources Division of the United States Geo-
logical Survey. Physical concepts are emphasized, and mathematical procedures
are the simplest consistent with the specific requirements of the derivations.
In lieu of numerical examples, analogies, and alternative procedures, this treat-
ment stresses a brief methodical exposition of the essential principles.
An important objective of this report is to prepare the user to read the literature
of the science. Thus, it begins with a basic vocabulary of technical symbols,
terms, and concepts. Throughout, emphasis is placed on the language of modern
fluid mechanics as it pertains to hydraulic engineering. The basic differential
and integral equations of simple fluid motion are derived, and these equations are,
in turn, used to describe the essential characteristics of hydrostatics and piezom-
etry. The one-dimensional equations of continuity and motion are defined
and are used to derive the general discharge equation. The flow net is described
as a means of demonstrating significant characteristics of two-dimensional ir-
rotational flow patterns. A typical flow net is examined in detail.
The influence of fluid viscosity is described as an obstacle to the derivation of
general, integral equations of motion. It is observed that the part played by
viscosity is one which is usually dependent on experimental evaluation. It
follows that the dimensionless ratios known as the Euler, Froude, Reynolds,
Weber, and Cauchy numbers are defined as essential tools for interpreting and
using experimental data. The derivations of the energy and momentum equa-
tions are treated in detail. One-dimensional equations for steady nonuniform
flow are developed, and the restrictions applicable to the equations are empha-
sized.
Conditions of uniform and gradually varied flow are discussed, and the origin
of the Chezy equation is examined in relation to both the energy and the mo-
mentum equations. The inadequacy of all uniform-flow equations as a means of
describing gradually varied flow is explained. Thus, one of the definitive problems
of river hydraulics is analyzed in the light of present knowledge.
Regents-Professor of Civil Engineering, Georgia Institute of Technology, Atlanta, Oa.; consultant to tbe
U. S. Geological Surrey.
1
2 RIVER HYDRAULICS

INTRODUCTION

This report is the outgrowth of a series of short schools conducted

during the spring and summer of 1953 for engineers of the Surface
Water Branch, Water Resources Division, U. S. Geological Survey.
The topics considered are essentially the same as the topics selected
for inclusion in the schools. However, in order that they might serve
better as a guide and outline for informal study, the arrangement of
the writer's original lecture notes has been considerably altered.
The purpose of the report, like the purpose of the schools which
inspired it, is to build a simple but strong framework of the funda-
mentals of fluid mechanics. It is believed that this framework is
capable of supporting a detailed analysis of most of the practical
problems met by the engineers of the Geological Survey.
It is hoped that the least, accomplishment of this work will be to
inspire the reader with the confidence and desire to read more of the
recent and current technical literature of modern fluid mechanics.
TECHNICAL SYMBOLS, TERMS, AND BASIC CONCEPTS
The material in this section is the basic language of fluid mechanics.
The reader is urged to become thoroughly acquainted with these
symbols, terms, and fundamental concepts before reading the subse-
quent text.
LETTER SYMBOLS AND UNITS
BOUNDARY CONDITIONS

A Area; sq ft or ft2.
6, B Width of section; ft.
D Diameter of a circular section; ft.
L A length, as defined; ft.
n Direction or distance perpendicular to streamlines in
the s, n plane; as a radial distance it is measured
positively toward the center of curvature; ft.
P Wetted perimeter or perimeter of the wetted cross
section; ft.
R Hydraulic radius; ratio of the area to the wetted
perimeter; ft.
r Radius of a circular arc or radial distance; as a
distance it is measured positively from the center
of curvature; ft.
s Direction or distance along or parallel to streamlines;
ft.
S Hydraulic "slope"; in uniform-flow equations it is
the flow-distance gradient of either the total energy
head or the piezometric head.
 SELECTED TOPICS OF FLUID MECHANICS

V Volume; cu ft or ft3.
m Direction or distance perpendicular to streamlines in
the s, m plane and at right angles to the n direction;
ft.
A direction or distance; as a plane coordinate it is
usually horizontal; ft.
y A direction or distance; as a plane coordinate it is
usually vertical; in open channels it is the depth of
flow referred to the average bottom level; ft.
Elevation of a point referred to a horizontal datum;
ft.
FLUID PROPERTIES

7 (gamma) Specific or unit weight; weight per unit volume; Ib

per cu ft or lb/ft3.
P (rho) Mass density; ma'ss per unit volume; also, ratio of
specific weight to acceleration of gravity; slugs per
cu ft, slugs/ft3, or Ib-sec2/ft4.
M (mu) Viscosity; dynamic viscosity; lb-sec/ft2.
<r (sigma) Surface tension; lb/ft.
e (epsilon) Modulus of elasticity; psf or lb/ft2.
MOTION (FLOW) PROPERTIES

Acceleration; rate of change of velocity with respect

to time; a vector quantity; ft/sec2.
The acceleration due to gravitational force, usually
taken to be essentially constant in magnitude
(32.2) and fixed in direction (vertically downward);
ft/sec2.
Piezometric head; in general, equivalent to pressure
head plus elevation; for uniform flow in open
channels, it is equivalent to the elevation of the
free surface; ft.
H Total energy head; energy in ft-lb/lb of fluid flowing;
ft.
Volume rate of flow per unit width; discharge per
foot of width; cfs per ft or ft3/sec/ft.
Q Volume rate of flow; total discharge at a cross sec-
tion; cfs or ft3/sec
t,T An interval of time; sec.
v Velocity at a point; instantaneous time rate of dis-
placement of a particle or point; a vector quantity
as distinguished from speed, which is scalar; fps
or ft/sec.
4 RIVER HYDRAULICS

V Average velocity in a cross section; ratio of total dis-

charge to the normal area of a cross section; fps or
ft/sec.
FORCE PROPERTIES

F Force; a vector quantity; Ib.

M Mass; ratio of weight to the acceleration due to-
gravity; a scalar quantity; slugs.
p Pressure intensify; normal (pressure) force per unit
area; a scalar quantity; psf or lb/ft2.
W Weight; total force due to gravity; a vector quantity,
always vertical; Ib.
T (tau) Shear stress; tangential (shear) force per unit area;
psf or lb/ft2.
it (eta) Eddy viscosity; apparent viscosity due to turbu-
lence; lb-sec/ft5.
RATIOS AND COEFFICIENTS

C Any coefficient; usually dimensionless.

Ce Coefficient of contraction; ratio of area of contracted
stream to gross area of opening; dimensionless.
Cen Coefficient in the Chezy equation; has dimensions of
V#, units of ft^/sec.
Cd Coefficient of discharge; dimensionless.
CL Energy-loss coefficient; dimensionless.
CT Shear-force coefficient; dimensionless.
K A constant.
n Coefficient in the Chezy-Manning equation; has
dimensions of RH, units of ft*.
a (alpha) Velocity-head coefficient; ratio of the true average
velocity head in a cross section to the velocity
head computed on the basis of the average velocity
in the section; dimensionless.
j8 (beta) Momentum coefficient; ratio of true rate of transfer
of momentum (momentum flux) at a cross section
to the momentum ,flux computed on the basis of
the average velocity in the section; dimensionless.
E Euler number; ratio of a unit inertia! reaction to a
unit pressure force; a basic flow parameter.
F Froude number; ratio of a unit inertial reaction to a
unit weight force; a number which varies inversely
with the relative influence of weight on the flow
pattern; dimensionless.
 SELECTED TOPICS OF FLUID MECHANICS 5

R Reynolds number; ratio of a unit inertial reaction to

a unit viscous shear force; a measure of the relative
influence of viscosity; dimensionless.
W Weber number; ratio of a unit inertial reaction to a
unit surface-tension force; a measure of the relative
influence of surface tension; dimensionless.
C Cauchy number; ratio of a unit inertial reaction to a
unit elastic force; a measure of the relative in-
fluence of elasticity; dimensionless.
MATHEMATICAL SYMBOLS
A (delta) Change in value; final value minus initial value;
increment.
S (sigma) Summation.
<* Proportional to ; varies as.
** Approximately equal.
H-» Vector addition.
BASIC CONCEPTS OF MECHANICS

Newton's laws of motion. Engineering mechanics, which includes

the mechanics of solid and rigid bodies as well as fluids, is based on
Newton's hypotheses or "laws" concerning the behavior of masses
acted upon by external forces. The fundamental proposition is
stated in the second law of motion: a mass acted upon by an un-
balanced external force or force system has an acceleration which is
proportional to and in the direction of the resultant force. This
proposition is expressed algebraically in the familiar relation 2F<x.Ma,
in which both F and a are recognized as vector quantities. In a
rectilinear coordinate system, for example, *2Fx<x.Max, 2Fy<x.Mav, and

The first law is: A mass remains at rest or moves with constant
velocity in a straight line until acted upon by an unbalanced resultant
external force; or, in accordance with the second law, if S,F=0, then
a=0. The first law, as applied to fluid mechanics, describes both
hydrostatic and uniform-flow conditions.
The third law is: A mass acted upon by another mass exerts an
equal, opposite, and simultaneous reaction. If "2F is defined as the
resultant external force acting on a mass, then Ma is the inertial
reaction of the mass.
It is apparent that Newton's second law of motion provides a
quantitative expression for the relationship between force, mass, and
acceleration which is the basis for all equations of motion in engi-
neering mechanics,
6 RIVER HYDRAULICS

Dimensions. The dimension of a quantity is its classification ac-

cording to a system of independent physical categories. Because
Newton's second law is a complete description of a mass in motion,
the fundamental dimensions are the dimensions of the quantities in-
volved in a physical statement of the law. Thus, the fundamental
dimensions are mass, length, time, and force (M, L, T, and F) from
the relationship F<xMa<xML/T2. Quantities such as acceleration,
power^ and momentum are derived quantities; their dimensions are
expressed in terms of the fundamental dimensions. Because the 4
fundamental dimensions are related by means of the Newtonian equa-
tion, an adequate dimensional system need involve only 3 dimen-
sions. Commonly used dimensional systems involve F, It, and T, or
M, L, and T.
Units of measure. For each of the fundamental dimensions there
is an accepted unit of measure. Thus, L is measured in feet,, f in
pounds, T in seconds, and M in slugs. Only three units nee<| be in-
volved in a complete dimensional system, however. For example,
slugs need not be used in the F, L, T system. The units of derived
quantities such as acceleration, power, and momentum are expressed
only in terms of the units of the fundamental dimensions,
The American engineering system of units is an absolute system;
that is, in the basic relationship Foc(ML/T2)=K(ML/T*), the pro-
portionality constant, K, is taken to be unity. Thus, a force of 1 pound
produces in a mass of 1 slug an acceleration of 1 foot per second per
second. It follows that, expressed in the absolute system of units,
Newton's second law can be written 2F=Ma.
Kinematics. Kinematics deals with the description of motion
without respect to the forces which cause or influence this motion,
Kinematics involves only time and space.
Dynamics. Dynamics deals with the effect of forces in causing or
changing the motion of a mass. It is further subdivided into statics
and kinetics.
Statics. Motions described under this heading correspond to the
condition of zero acceleration in the directions considered.
Kinetics. Kinetics is that branch of dynamics which deals with the
accelerated motion of a mass under the influence of unbalanced exter-
nal forces.
Work. Work is the product of a force and the distance over which
the force acts in accomplishing motion. Work is a scalar quantity;
it does not involve the time during which the force acts. Work is
measured in foot-pounds (ft-lb).
Power. Power is the time rate of doing work. The power deliv-
ered to a moving body is proportional to the product of the force
producing the motion and the velocity of the body in the direction of
 SELECTED TOPICS OF FLUID MECHANICS 7

the motion. Power is a scalar quantity. The unit of power is foot-

pounds per second (ft-lb per sec) or horsepower. One horsepower is
equivalent to 550 ft-lb per sec.
Energy. A body upon which work is accomplished acquires thereby
the capacity to do an equal amount of work. The capacity to do
work is called energy. Thus, energy is a measure of work, and vice
versa. Energy is a scalar quantity. The unit of both work and
energy is the foot-pound (ft-lb).
Momentum. Linear momentum is the product of mass and velocity.
It is a vector quantity, and its unit is the pound-second (Ib-sec).
Impulse. Linear impulse is the product of a force and the time
during which the force acts. It is a vector quantity. From New-
ton's second law, impulse accomplishes and is equivalent to the change
in momentum experienced by the mass acted upon. The unit of im-
pulse is, therefore, the same as that of momentum, the pound-second
(Ib-sec).
BASIC CONCEPTS OF FLUID MECHANICS

Continuity and discharge. Fluid mass can be neither created no-

destroyed by virtue of its motion. Thus, in a steady fluid flow ber
tween solid boundaries or through a streamtube, the rate of passage
of fluid mass through successive cross sections is constant. If the
fluid be one of negligible compressibility (constant density), the rate
of passage of fluid volume past successive sections is also constant.
This is the simple basis for the continuity principle. The principle
is expressed, for example, in the equation <2=.AV==constant, where
Q is the volume rate of flow, or discharge, in cubic feet per second and
V is the average velocity normal to the surface represented by the
area (A). In a section of nonuniform velocity, the total discharge is
the summation (or integral) of the discharges through the elementary
areas comprising the total area of the section.
Uniform and steady motion. Velocity in a fluid motion is a vector
quantity and it is a function of both time and space. When the
velocity remains constant with respect to time, the flow is described
as steady. Most problems in engineering hydraulics involve steady
motion.
When the velocity is constant with respect to distance in the direc-
tion of motion, the flow is described as uniform. Uniform flow occurs
in long pipes. It seldom occurs in natural open channels.
Rotational and irrotational motion. The distinction between rota-
tional and irrotational motion is essentially mathematical. The most
significant characteristics of an irrotational flow are those usually
associated with the classical "ideal" or "frictionless" fluid (a fluid
without compressibility or viscosity). It is pertinent to the subse-
8 RIVER HYDRAULICS

quent use of the flow net in these notes to observe that irrotationality
is corequisite to the existence of a flow net.
Laminar and turbulent motion. Laminar fluid motion is character-
ized by a steady, translatory movement of all particles (small elements)
of the fluid. Turbulence imposed on a laminar flow is eventually
damped out by the viscous shear forces. Because viscous forces are
large relative to inertial reactions in laminar motion, this condition
corresponds to low values of the Reynolds number.
Turbulent motion is characterized by an erratic, small-scale eddying
motion which is superposed on the translatory movement of the fluid
mass. Local disturbances are rapidly dispersed throughout a turbu-
lent fluid flow because the viscous shear forces are small relative to
the inertial reactions resulting from the turbulence. Thus, turbulent
flow corresponds to high values of the Reynolds number. For most
practical purposes, motion which is turbulent is described in terms of
the 'average translatory velocity at a point or in a cross section.
Critical, tranquil, and rapid motion. It the mean total energy head
referred to the bottom of an open channel is a minimum for a given
discharge, the flow is described as critical. When critical flow occurs
in a straight, rectangular, or nearly rectangular channel in which the
velocity in the cross section is essentially constant, the average velocity
is theoretically equal to the velocity of a small gravity wave, and the
Froude number is equal to unity. The flow is described as tranquil if
the average velocity is less than the critical value. If the average
velocity is greater than the critical value, the flow is described as
rapid.
Shear and pressure. The total force on any surface within a fluid
in motion can be resolved into tangential and normal components.
The tangential component of the total force on the surface is the shear
force (FT). The limit of the ratio of the shear force to the area of the
surface as the area approaches zero is the shear stress at a point (T).
The normal component of the total force on an area is the pressure
force (Fp). The limit of the ratio of the pressure force to the area
as the area approaches zero is the pressure intensity (p) at a point.
Pressure intensity is a scalar quantity.
A fluid at rest with respect to its boundaries is in a state of zero
shear stress throughout. It follows that pressure changes imposed
at one point in a fluid mass at rest are transmitted undiminished to
all points in the fluid mass (Pascal's principle). Furthermore, because
shear stresses cannot exist in a fluid at rest, the resultant force on any
surface within the fluid is a pressure force, and it is normal to the
surface.
Viscous and turbulent shear. The viscous-shear force (FJ is a con-
sequence of internal stresses which resist fluid deformation.
 SELECTED TOPICS OP FLUID MECHANICS 9

relative movement of the fluid on adjacent stream surfaces in a laminar

flow results in a viscous-shear stress (TM) . In a motion of this kind, TV is
a function of the fluid viscosity and the velocity "gradient", a prop-
erty of the flow defined as the rate of change of the tangential velocity
with respect to distance perpendicular to the motion. The viscous-
shear stress at any point in the flow pattern, therefore, is TM =
fj,(dvs/dn), where ju is the dynamic viscosity of the fluid and dvjdn is the
velocity gradient at the point. It is significant that TM is independent
of fluid density and pressure.
The total shear force (FT) acting on a fluid in turbulent motion is
the sum of the viscous shear (F^ and the turbulent shear (FJ. Tur-
bulent shear is a consequence of the momentum exchange between
small masses of fluid moved transversely by the action of the turbu-
lence eddies. Unlike viscous-shear stress, turbulent-shear stress (T,)
is a function of fluid density, and it is independent of fluid viscosity.
It can be shown that the turbulent-shear force acting on a fluid
element is directly proportional to its inertial reaction. It follows
that the Reynolds number is proportional to the ratio of turbulent
shear to viscous shear, or R oc T,/TM. The influence of viscous shear is
negligible in fully developed turbulence.
Boundary layer. The boundary layer was originally conceived as a
means of correlating real fluid motion with the theoretical behavior
of the hydrodynamicist's ideal fluid. Thus, Prandtl suggested that
in flows at high Reynolds numbers the effect of fluid viscosity could
be assumed to be limited to a narrow region near the confining bound-
aries. This concept is of fundamental importance in the analysis of
flow around bodies immersed in relatively large fluid spaces. It is
also essential to an understanding of certain phenomena associated
with the establishment of uniform flow at entrances to conduits or
following local disturbances in an otherwise uniform conduit. In
regions of fully established flow, as in long pipes and open channels,
the boundary layer is virtually the entire cross section of the flow.
Streamlines and streamtubes. A streamline is a line drawn through
a flow in such a manner that at every point along its length it is tangent
to the velocity vector. It follows that there can be no flow across a
streamline; in this sense, the streamline imposes a boundary condition
on the adjacent flow.
A streamtube is a surface formed entirely of streamlines; in other
words, it is the surface defined by the streamlines passing through a
closed curve. As the boundary of a streamtube is comprised of
streamlines, there can be no velocity and, therefore, no flow through
the walls of a streamtube. Any solid boundary or separation surface
which effectively guides the flow has the properties of a streamtube
surface.
10 RIVER HYDRAULICS

Stagnation and separation. A stagnation point in fluid motion is

defined as a point of zero velocity. When boundary conditions cause
a flow to undergo an abrupt convergence, a tendency for stagnation
occurs at a point on the boundary. In a sharp, concave boundary
corner, or on the nose of a blunt obstruction, for example, the flow
along the boundary would apparently experience an instantaneous
change in direction of motion. As this implies an infinite normal
acceleration if the tangential velocity were anything but zero, stagna-
tion occurs.
When flow boundaries diverge abruptly, as in sudden enlargements
or following blunt obstructions, the boundary ceases to guide the
flow effectively and separation occurs. Thus, the apparent require-
ment that the fluid undergo an infinite acceleration at the boundary
discontinuity does not, in this example, result in stagnation. Instead,
the live stream separates from the boundary and follows a "free-
stream" surface bounded by eddying fluid. As a consequence of
viscosity, such eddy-filled separation zones also occur in corners near
regions of boundary convergence, thereby preventing stagnation.
EQUATIONS OF MOTION
FUNDAMENTAL EQUATION OF MOTION

The principal objective of fluid mechanics is a mathematical

description of fluid motion. Newton's second law is the basis for
equations of fluid motion. The sense of Newton's proposition is
contained in the algebraic expression l 2FccMa=KMa. This is
a vector relationship which states that the resultant of all external
forces acting in a particular direction on a mass is proportional to
the product of the mass and its acceleration in the direction of the
resultant force. If the quantities involved in this expression are
evaluated in accordance with an absolute system of units, then K 1.0,
or 'LF Ma. In the Cartesian space system, for example, 2Fx=Max,
and 2Fg
EXTERNAL FORCES IN FLUID MOTION

The independent forces which influence fluid motion are the con-
sequence of 1 property of state (pressure) and 4 physical properties
of the fluid (density, viscosity, surface tension, and elasticity). An
additional force, the apparent shear force due to fluid turbulence
(-F7,), is not recognized as an independent force because it is directly
proportional to the inertial characteristics of the flow.
Pressure, or pressure intensity, is a scalar quantity which describes
i Except where emphasis Is Intended, the symbols and technical terms defined In the preceding section
are used in the subsequent text without further explanation.
 SELECTED TOPICS OF FLUID MECHANICS 11

a particular kind of fluid stress. Pressure is the cause of a resultant

external force (Fp) only when it varies in the direction of motion.
Density is involved in fluid motion in two distinctly different ways.
Thus, density accounts for an accelerative force (Fg) which is the
component of the weight (W) in the direction considered. The defini-
tive equation, W=Mg=pVg, is a particular form of Newton's second
law in which p is the density and g is described as the acceleration due
to gravity. On the other hand, the total inertial reaction due to the
resultant of all the external forces, which might or might not include
a component of the weight force, also involves the density; that is,

Viscosity is a fluid property which accounts for a shear force

whenever fluids are in motion with respect to their boundaries.
Viscous shear is the result of molecular forces which oppose fluid
deformation. In turbulent fluid motion the total shear force (FT)
is the sum of F? and the turbulent shear force (F,) .
Surface tension is the name given to certain manifestations of
molecular forces which occur at the interface between different
substances; for example, between a free jet of water and the surround-
ing air. The surface-tension force (Fa) is ordinarily of appreciable
magnitude only when the other forces involved are extremely and
unusually small.
Elasticity, or forces due to elastic deformation (Ft), are involved
primarily in the unsteady flow of liquids and in the flow of gases under
high pressure gradients. Neither of these problems is pertinent to
this discussion.
KINEMATICS OF FLUID MOTION
A mathematical description of fluid motion necessarily involves the
space-time characteristics which are known as velocity and accelera-
tion.
Velocity is the time rate of displacement of a mass or a point. For
example, if As is a small displacement or distance traversed along a
streamline in a short time interval (M) , and if vs is the average tangen-
tial velocity in the interval on that part of the streamline, then
vs=As/M. It follows that the instantaneous tangential velocity at
a point on a streamline is vs =^dsjdt.
Velocity is a vector quantity. Thus, the total or resultant velocity
at a point is the vector sum of its components on the various vector
axes. From the definition of a streamline, the total velocity in the
"natural" or s, n, m system of coordinates is vt, and the normal com-
ponents, vn and vm, are zero. For many purposes, however, the
Cartesian or x, y, z system of describing motion in three-dimensional
space is preferred. Thus, in the Cartesian system v=vx-+*
12 RIVER HYDRAULICS

VV+V+^j where -f-> indicates vector addition and vx =dx/dt,

vv=dy/dt, and vt=dz/dt.
Acceleration is time rate of change of velocity. As this discussion
is concerned with steady motion only, the variation of the velocity
at a point, with respect to time, is always zero. Thus, in the general
definition a=dv/dt, the change in velocity represented by dv is the
change which occurs over the^ distance traversed during the time
interval dt. It follows that the total acceleration in steady motion
is described as convective acceleration.
In the natural system of coordinates, the total acceleration includes
both tangential and normal components. Thus, as dvs/.dt is the tangen-
tial acceleration. Similarly, an =dvn/dt and am dvm/dt are the or-
thogonal components of the normal acceleration. In the Cartesian
system the total convective acceleration is comprised of the com-
ponents ax dvx/dt, av=dvv/dt, and az dvg/dt.
For subsequent use, the tangential and normal accelerations in
two-dimensional motion (am =0) are more conveniently expressed
dvt -j-
a,=-j ds dvt
, ds
~ji Vt dvs
-j-)
at ds rig at as
or, from the calculus,

a" ==Vs ~ds == 2~ds" (1)

From any mechanics or physics text, the normal (centripetal)

acceleration at a point on a streamline is

an= dt = Gfe" Jf* Vt ~ds =~r~' (2)

in which r is the radius of curvature of the streamline at the point.

It is significant that equation 1 describes at as a measure of the rate
of change of the magnitude of vs, whereas equation 2 describes an as
a measure of the change in direction of vt. Equation 2 also explains
the anomalous situation which appears to exist when it is observed
that an is not necessarily equal to zero even though vn is always zero
on a streamline.
EQUATIONS FOB A SIMPLE FLUID MOTION
The next logical step in writing an equation of fluid motion is to
evaluate all of the forces acting on a typical accelerated fluid mass
and to write an equation of the form

(3)
 SELECTED TOPICS OF FLUID MECHANICS 13

in each of the coordinate directions. First, however, in view of the

purpose of this discussion, the problem can be simplified by omitting
further consideration of the relatively minor surface-tension and
elastic forces, F, and Ff, respectively. Furthermore, in order to demon-
strate certain basic principles without the complication of the most
troublesome of all forces those due to viscosity and turbulence
a kind of motion in which the influence of F, is negligible is initially
selected for analysis.
The flow of a fluid from a large quiescent reservoir through a small ori-
fice is shown in plate 1 (a-c). It is assumed that the orifice discharges
into another reservoir containing a fluid of the same density in order
that certain independent influences of fluid weight can also be ignored.
Sufficient, generality and considerable convenience is achieved in this
example by assuming that the flow is two-dimensional in the s, n plane.
Thus, the only accelerations involved are aa and an, and the only
forces involved are those which have components acting parallel and
perpendicular to the streamlines. The natural system of coordinates
is ideally suited to this analysis.
Plate 1 (b) shows a typical small mass of the fluid in motion. For
convenience it is shown as a prismatic element of volume (As AA).
Its axis is a streamline and it is as small as necessary in order that its
motion can be described in terms of the motion of its mass center. In
plate 1 (c), showing the same element as a free body, the pressure force
on each end of the element is represented by the product of the aver-
age pressure and the end area. The weight force in the direction of
motion (Fg) is shown as the s component of the total weight of the
element (W). It should be apparent that any errors resulting from
neglecting area, velocity, or pressure variations over the small volume
(As &A) are of a second order of magnitude and that they disappear
when the volume of the element is allowed to approach the infinitesi-
mal (ds dA).
In the kind of motion represented by plate 1 (a-c) the influence of the
total shear force (FT) is negligible. This follows from the fact that
motion in the reservoir is assumed to be nonturbulent and that the
velocity of the fluid relative to the boundaries is negligible except
immediately adjacent to the opening. Thus, neither viscous- nor
turbulent-shear forces are appreciable. Flow patterns of this kind
are essentially similar to the theoretical flow patterns described in
hydrodynamics for an ideal fluid. Thus, for the motion pattern
described,
2F.=F,-}-*Fg=Ma., (4)
in which Fp and Fg are the s components of the forces due to pressure
and weight, respectively.
14 RIVER HYDRAULICS

From plate 1 (c), the resultant pressure force is

Fp=p AA-(p+Ap)AA= Ap AA.

If it is assumed that the pressure intensity increases at a uniform rate

(dp/ds) in the direction of motion, then Ap= (dp/ds) As is the total
change in pressure over the distance As. Substituting and simplifying,

F^-^CAsAA). (5)

The weight of the element is y(AsAA), The s component of the

weight is, from plate 1 (c),
dz
Fg= -y(As AA) sin 6= -y(As AA) ^- (6)
CLS

The mass of the element is p(As AA), and the inertial reaction due
to Fp and Fg is *
Mas=p(As AA)as. (7)

Combining equations 5, 6, and 7 with equation 4,

-^ (As AA)-y ^ (As AA)=pas(As AA). (8)

as as

'Reducing this to a force-per-unit-volume relationship by dividing by

the volume (As AA) of the element,

- dp dz

This equation shows that acceleration in the direction of motion re-

sults only when there is a gradient (a change with respect to distance)
of the sum p-\-yz in that direction. It shows also that when dz/ds Q,
which is true of any motion in a horizontal plane, the pressure gradient
( dp/ds) is the sole accelerative force per unit volume, and it is
exactly equal to the inertial reaction per unit volume (pas). It in-
dicates furthermore that when dp/ds 0, then y(dz/ds)=pas, or
dz/ds=as/g, equations which, for example, describe flow along any
free (constant-pressure) surface; thus, when ds= dz, as in free fall,
a,s=g, or the total .acceleration is equal to the acceleration due to
gravity. Finally, it indicates that when as=0, the pressure gradient
is dp/ds= y(dz/ds), or, in the distance ds, dp= ydz; this describes
the condition of hydrostatic pressure distribution.
 SELECTED TOPICS OF FLUID MECHANICS 15
* T 2

If as=7j f- from equation 1 is substituted in equation 9 and the

a ds
terms rearranged,

From this equation, the s rate of change of the quantity within paren-
theses, at any point on the streamline, is zero.
If equation 10 is multiplied by the differential distance (ds),

Q, (11)

which shows that the differential change in the parenthetical quantity

over the differential distance (ds) is also zero. It follows that the
quantity within the parentheses is a constant along the streamline.
Thus,
iH-rz+^f-W, (12)

which is the integral of equation 11, is a relationship between pressure,

elevation, and velocity in which K is the constant of integration for a
particular streamline. At successive points along this line, therefore,
av* pv£
^+A+rBlSSI__J+ft+7gl| etc. (13)

Equation 13 can be written in an alternate "head" form, known

more familiarly to civil-hydraulic engineers as the Bernoulli equation.
Dividing by the constant specific weight of the liquid (7) and noting
that yp=

The algebraic operations which led from the Newtonian equation

to equation 13 involved dividing a force by a volume and multiplying
by a distance. Thus, the units of every term in the equation are
foot-pounds per cubic foot. It follows that the units of every term
in equation 14 are foot-pounds per pound or, simply, feet.
Equations 13 and 14 are both properly described as Bernoulli
equations. They are relationships between integral values of pres-
sure, elevation, and velocity at successive points on a single stream-
line. In order to develop similar relationships between points on
adjacent streamlines it is necessary that the foregoing analysis be
repeated with the axis of the definitive fluid element oriented in the
16 RIVER HYDRAULICS

n direction. As a result, a differential equation is obtained which is

the counterpart of equation 9:

However, when the equivalent expression for an in terms of the

velocity (vs) is substituted, the similarity in the two developments
ceases. Thus, an=vs2/r from equation 2. When this is substituted
in equation 15, it appears to yield a nonintegrable equation. Re-
turning to equation 2, however, let an=vs (dvnfds). Adding and sub-
tracting the quantity vs (dvs/dn) ,
dvn . dvs dvg dvs . /dvn dv

From hydrodynamics, the quantity within the parentheses in the

last term is a measure of rotationality. Irrotational flow corresponds
to the condition that (dvjds) (dvs/dri)=Q. Whereas this distinction
between rotational and irrotational flow is essentially mathematical,
it is significant that hydrodynamics also shows that the absence of
viscous shear is corequisite to irrotationality. Thus, because shear
forces are negligible here, it can be assumed that the flow is irrota-
tional, and, therefore,

an=vs dvs
-7-=o1 dvs2
-7 ,(16).
dn 2 dn ^
When this is substituted in equation 15, the modified equation
leads to differential relationships which are the counterparts of equa-
tions 10 and 11 and integral relationships which are identical with the
Bernoulli equations (12, 13, and 14). Thus, when the shear forces are
negligible, the same relationships between velocity, pressure, and
elevation will apply between adjacent streamlines as apply along
single streamlines. In other words, the constant of integration in
equation* 12 is the same for all streamlines and the Bernoulli equation
is applicable to all points in an irrotational or frictionless flow pattern.
Several important applications of the equations of motion, including
hydrostatics, piezometry, the one-dimensional equation, and the dis-
charge equation, are discussed in the following sections.
HYDROSTATICS
When only pressure and weight forces are effective in a fluid motion,
as in the preceding example, the condition described as hydrostatic
pressure distribution occurs along any line on which the acceleration
at all points is zero. Thus, with as =0 in equation 9, Irydrostatie
 SELECTED TOPICS OP FLUID MECHANICS 17

pressure distribution in the s direction implies that the pressure in an

incompressible fluid (7 is constant) varies inversely with the eleva-
tion along a streamline. Similarly, from equation 15, pressure dis-
tribution is hydrostatic along lines normal to the streamlines when
an=0. This also means, of course, that pressure is not hydrostatically
distributed in the n direction if the streamlines are curved. It follows
that hydrostatic pressure distribution occurs throughout a fluid in
motion only if shear forces are negligible and the flow is truly uniform;
that is, if a,=0 and a»=0.
However, when the shear force (FT] cannot be ignored, it should be
apparent from plate 1 (c) that the pressure must vary with the influence
of this force as well as with elevation along the streamline. Thus, in
general, pressures are hydrostatically distributed in the s direction of
a uniform motion only when the relative velocity of fluid and boundary
is so small that the shear force is negligible. On the other hand,
pressures are hydrostatically distributed in the n direction as long as
an=0. This follows from the fact that, from the definition of a
streamline, the average normal velocity (w») is zero and, as a conse-
quence, the net shear force in the n direction is always zero.
Hydrostatics, as the term is generally used in technical literature,
refers to the behavior of fluids, particularly liquids, at rest. The
equations derived for a simple fluid motion are also applicable to hydro-
statics. For this application, as there is no flow and consequently
there are no streamlines, the s direction is arbitrarily defined as any
direction. From equation 9, when the fluid is at rest the acceleration
in all directions is zero, and the equation

defines the condition of hydrostatic pressure distribution throughout

the fluid. Thus, along any line, in a distance ds,
dp= ydz. (18)
In horizontal planes, dz=Q, whence, from equation 18, dp=0.
Therefore, one of the basic principles of hydrostatics is that the
pressure is the same at all points on horizontal planes in a fluid at rest.
When 7 is constant, which is sensibly true for liquids, integration of
equation 18 yields
Za)' (19)
Equation 19 is apparently a particular form of the Bernoulli equation
(13). It is also the basic gage-pressure equation if conditions at point 1
are defined as the conditions on a liquid surface open to the atmos-
18 RIVER HYDRAULICS

phere. Thus, if pi=0 (gage) and if 21 22=2/2 is the vertical depth of

point 2 below the free surface, then #2 =7j/2- It follows that the gage
pressure at any point in a continuous body of liquid at rest is

p=yy, (20)
in which y is the vertical depth of the point below the real or imaginary
free surface of the liquid.
PIEZOMETBY
A piezometer is a small hole located in a surface which comprises a
fluid boundary. The purpose of the piezometer is to provide a means
of measuring the fluid pressure at the point where the hole is located.
In its definitive application to a liquid, as shown in plate 1 (d), the pie-
zometer is connected to an open tube called a manometer. When the
pressure in the liquid at the piezometer is exactly equal to the pres-
sure due to the weight of the liquid in the manometer column, the
liquid in the manometer is at rest. Thus, the pressure at the point
of attachment can be computed from the gage-pressure relationship
(eq 20). When a piezometer is used to measure the pressure on the
boundary of a flowing fluid, the piezometer orifice must be small and
very carefully made in order that flow along the boundary is not
disturbed by the presence of the piezometer.
The level of the liquid in a simple open-tube manometer is appropri-
ately called the piezometric level. As its height above the level of
the piezometer (y) is equal to pfy from equation 20, this distance is
described as the pressure head at the point of attachment. The
elevation of the point with respect to a horizontal datum is called its
elevation head. Thus, the vertical distance between the piezometric
level in the manometer and the elevation datum is described as the
piezometric head at the point where the piezometer is located. From
this definition, if the symbol h is used to denote the piezometric
head at any point,
A=£+z. (21)

Thus, from equation 15 (divided by 7),

~j~ r n
dn\y / dn 7 g
This equation shows that the piezometric-head distribution in
a section normal to the flow is proportional to the normal acceleration.
The hydrostatic condition in the n direction occurs when an=0. For
this condition the gradient of the piezometric head is zero; that is,
the piezometric head is constant. Thus, when flow past a piezometer
 SELECTED TOPICS OF FLUID MECHANICS 19

in a flow boundary is truly uniform, the piezometer indicates not only

the piezometric head at the point of attachment but also the piezo-
metric head at all points in the adjacent uniform flow. Under these
circumstances, the piezometric level is independent of the elevation
of the piezometer. These conclusions are illustrated in plate 1 (d).
When a liquid flows in an open channel, as in plate 1 (f), the upper
boundary of the flow is both a stream surface and a piezometric sur-
face. The gage pressure over the upper stream surface, regardless of
its configuration or velocity, is zero. Thus, a piezometer located on
the boundary of a uniform flow in an open channel indicates not only
the piezometric level but also the level of the flowing liquid at the
point of attachment. It follows that, if the piezometric datum is
defined as the bottom of the channel, a piezometer located any-
where in the cross section indicates the depth of the liquid in the
channel at, that section.
When a piezometer is located on the upstream end of a small object
placed in a fluid flow, stagnation occurs over the piezometer. As
viscosity has little influence on this occurrence, the Bernoulli equation
is applicable between the stagnation point and any point on the same
streamline in the undisturbed flow upstream from the obstruction.
Two such points (on streamline AB) in a uniform pipe flow are desig-
nated points 1 and 2 in plate 1 (e). If 02=0, and h= (p/y) -\-z is sub-
stituted in equation 13, the piezometric head at the stagnation point
is
A2=Ai+^- (23)

In plate 1 (e), the piezometer at point 2 is connected to an open

manometer tube. The stagnation pressure on the nose of the tube
causes liquid in the tube to rise to a height equal to p2/y above the
level of the piezometer or h2 above the elevation datum. If the flow
in the pipe is uniform at this section, a second piezometer located
at any point on the boundary indicates the piezometric head in the
undisturbed flow at point 1. Thus, from equation 23, the velocity
head at point 1 is indicated as the difference in the piezometric levels
in the two manometers. It follows that the velocity in the undis-
turbed flow on streamline AB is

). (24)

A tube such as that in plate 1 (e) which indicates the stagnation-

pressure head is called a stagnation or pitot tube. It is apparent that
the pitot tube provides an accurate means of measuring velocities in
fluid motion. Its application to open-channel flows is illustrated in
plate 1 (g).
20 RIVER HYDRAULICS

The velocity in a normal section through a uniform fluid flow

between fixed boundaries varies with distance from the boundary.
This is incidentally a consequence of the shear stresses which result
from the relative motion of a viscous fluid and a solid boundary.
When a pitot tube is used to traverse such a section, the piezometric
level indicated by the tube varies from one streamline to the next.
However, the piezometric level indicated by all wall piezometers is
the same as long as an 0 at the section. Thus, only the pitot-tube
manometer indications are affected by the velocity variations in
the section. These conclusions are illustrated for an open channel in
plate 1 (h).
The equations of motion have been used to demonstrate some of the
basic principles of piezometry in uniform fluid motion. These equa-
tions are equally valuable as aids in the interpretation of piezometric
measurements in nonuniform flows. For example, piezometers are
often located at various points on bridge piers or abutments as a
means of measuring the average water-surface level hi the bridge
opening. If a piezometer is located downstream from the disturbed
flow near the entrance to the constricted waterway, and if the flow
across the entire section normal to the piezometer station is essen-
tially parallel to the face of the pier or abutment on which the piezom-
eter is located, the measurement will be quite satisfactory as an
indication of the average water-surface level at this section.
Piezometers located near the upstream end of the structure, how-
ever, are generally unsatisfactory. Thus, piezometers located on
the upstream face of a pier are ordinarily in the vicinity of a stag-
nation zone, and measurements are influenced to an indeterminate
degree by separation upstream from the pier nose. The so-called
superelevation of the water curving around the nose of a pier is
evidence of nonhydrostatic pressure distribution due to normal
acceleration in the horizontal plane. Similarly, vertical accelerations
in this zone account for nonhydrostatic pressure distributions in
vertical planes. Therefore, piezometers located on the walls of a
pier just downstream from the nose indicate neither the mean water-
surface level in the cross section nor the level adjacent to the pier.
It should be apparent that short piers, circular piers, piers with
irregular boundaries, webbed pile bents, or piers alined obliquely
with the flow are unsatisfactory as locations for stage-recorder gage
intakes unless they are calibrated in the field or hi the laboratory.
ONE-DIMENSIONAL EQUATION
The piezometric head has been shown to be constant in any normal
section over which an is everywhere zero (eq 22). From equation
16, which applies only when the shear forces are negligible, the normal
 SELECTED TOPICS OF FLUID MECHANICS 21

acceleration is proportional to the n gradient of the tangential velocity.

Thus, if a«=0, it follows that dv,,fdn~Q, or vs v is a constant. It may
be concluded, therefore, that at all points over a normal section both
the tangential velocity and the piezometric head are constant when
FT and an are negligible. Two sections which satisfy these conditions
are shown in plate 1 (i) .
Plate 1 (i), like plate 1 (a), represents the flow of a fluid from a large
reservoir through a small orifice into another reservoir. Section 1
in the figure is a hemispherical surface which is concentric with the
center of the orifice. Streamlines passing through this surface are
radial lines ; therefore they satisfy the requirement that the stream-
lines be without curvature if an be zero. As the influence of the
shear forces is negligible in this kind of flow pattern, it is concluded
that the piezometric head (hi) and the tangential velocity (PI) are
constant at all points on section 1. Section 2 is a section through the
parallel boundaries of the jet, downstream from the nonuniform flow
in the immediate vicinity of the orifice. As the streamlines in this
section are straight, an=0 and the velocity and piezometric head are
constant at all points. It follows that every term in an equation
of the Bernoulli form

applies to any and all streamlines between sections 1 and 2 in plate 1 (i).
For the circumstances illustrated by plate 1 (i), it has been deter-
mined that the velocity at all points in section 1 is equal to v\. This
velocity, therefore, can be represented by the symbol V\, for V has
been defined as the average velocity in a total cross section. Similarly,
here, 02=V2. When these substitutions are made in equation 25, the
resulting equation,
(26)

can be described as an equation of motion for a typical streamline or

an equation of motion for a streamtube which envelops the total
flow.
Plate 1 (j) shows an orifice flow pattern which differs in some very
important respects from the flow pattern shown in plate 1 (i). In
plate 1 (j) a diaphragm orifice is located in a pipe. Section 2 is again
located in the parallel jet. Thus, as in the previous example, the
velocity and the piezometric head at all points in section 2 are es-
sentially constant. Section 1 is normal to the axis of the pipe and
far enough upstream from the orifice to be removed from the zone
of streamline curvature. Thus, as the streamlines are all straight
22 RIVER HYDRAULICS

and perpendicular to section 1, the normal acceleration is everywhere

zero, and the piezometric head is constant. However, the influence
of the shear forces on the velocity distribution in section 1 can no
longer be ignored. As the velocity of a viscous fluid relative to its
fixed boundaries becomes appreciable, shear forces cause the fluid
near the boundary to be retarded. As illustrated in plate 1 (j), there-
fore, the velocities in section 1 vary with distance from the boundary.
Another effect due to the influence of shear is shown in plate 1 (j).
As there can be no slippage at the boundaries, the velocity of the
contact fluid is always zero. On the other hand, flow approaching
the upstream corners must decelerate. The apparent conflict be-
tween the condition of zero velocity at the boundary and the tendency
for deceleration along the boundary is resolved by separation and the
formation of an eddy zone in the corner region. The result is a
modification of the flow pattern and a depreciation of flow energy.
These matters obviously cannot be considered in detail at this time
because the simplified equations of motion were derived on the con-
dition that the influence of shear forces was negligible.
For many practical purposes, it is permissible to ignore certain
effects of viscous and turbulent shear. For a pattern such as that in
plate 1 (j), for example, it is generally sufficient to ignore the nonuni-
formity of tangential velocities across the normal sections and to
acknowledge that equation 26 is a good approximation. Under such
circumstances, because it ignores the influence of transverse velocity
variations, equation 26 is called a one-dimensional equation of motion.
EQUATION OP CONTINUITY
Plate 1 (k) shows a typical streamtube in a steady fluid motion. The
average tangential velocity at any section is vg, and the area of a
section normal to the axis of the tube at any station is AAn. The
product vs &An has already been defined as discharge, or volume rate
of flow, a quantity which has no counterpart in rigid-body mechanics.
The dimensions of discharge indicate that it is a measure of the time
rate of passage of fluid volume through the streamtube. The product
of discharge and density, therefore, is a measure of the rate of passage
of fluid mass. By definition, the flow normal to the streamtube sur-
faces is everywhere zero. Thus, because fluid mass can be neither
created nor destroyed, in a steady fluid motion the mass rate of flow
(pvt &An) is constant at successive sections, or, (pvs AAn)i=(pvs A^.^.
For a liquid flow, furthermore, the density is essentially constant.
Thus, the volume rate of flow (A<2) at successive sections is also
constant, or
AQ=(». *An\=(v, A4.),. (27)
 SELECTED TOPICS OF FLUID MECHANICS 23

The discharge (Q) through a total cross section between fixed

boundaries is the summation of the discharges through the component
subsections, or
(28)

in which V is the average velocity and A is the total area of the section.
Thus, because the fixed boundaries comprise limiting stream surfaces,
from equations 27 and 28, the total discharge is constant at successive
sections, and
(29)

Equation 29 is the one-dimensional equation of continuity.

DISCHARGE EQUATION

Equation 29 can be applied, for example, to sections 1 and 2 hi

plate 1 (j) . It can also be solved simultaneously with equation 26,
which is an approximate equation of motion for the same flow pattern.
The result,

(30)

is a discharge equation hi terms of area and piezometric head. As is

true of equation 26, equation 30 is applicable only when the influence
of the shear forces is small.
When the discharge equation is applied to a flow pattern such as
that shown in plate 1 (j), the area at section 2 is most conveniently
expressed hi terms of the area (A) of the orifice and the coefficient of
contraction (<7e). By definition, and as illustrated in plate 1 (1),
A2=CcA. (31)

Thus, from equations 30 and 31, a common form of the discharge equa-
tion for a convergent boundary is

Q.= . V2gft.-k) (32)

or
^), (33)
24 RIVER HYDRAULICS

in which C& is the coefficient of discharge. From equations 32 and 33,

PLOW NET

The flow net is a pattern of streamlines and "potential" lines which

indicates by its configuration the velocity distribution in a fluid flow.
Flow nets can be constructed only in accordance with the following
restrictions: The flow to be represented is a steady flow; the flow
pattern is independent of the influence of fluid viscosity, surface ten-
sion, and elasticity; the flow pattern is independent of the influence
of fluid weight (this does not preclude flow nets for completely enclosed
flows in vertical planes nor certain kinds of free-surface flows).
Within the limits of the restrictions above, flow nets can be constructed
for flows involving free-stream surfaces as well as those which are
effectively guided by rigid boundaries. A flow net is unique for the
given boundary conditions, regardless of scale (size) or any assump-
tions regarding fluid density, velocity, elevation, or pressure.
From hydrodynamics it is known that the equations of the lines
which comprise the flow net can be determined for many two-dimen-
sional flows and certain axisymmetric three-dimensional flows. From
hydrodynamics it is also known that the requirement of negligible
fluid viscosity is corequisite to the mathematical restriction of irrota-
tionality.
Truly irrotational flow, of course, is impossible; however, as already
observed, there are some kinds of fluid motion in which the influence
of the complicating fluid properties is negligible. For such flows,
usually involving rapidly accelerating motion and a completely
enclosed or submerged flow pattern, the flow net affords a powerful
means of analysis.
Whereas certain flow nets can be plotted as families of curves whose
equations are developed by mathematical techniques, many others
can be determined only by graphic or arithmetic successive-approxi-
mation procedures. Electric analogies, laminar-flow analogies, and
other laboratory techniques facilitate the approximation procedures.
Flow nets involving free-stream surfaces are more difficult to obtain
than nets for totally enclosed flows. However, measurement of the
free-stream profile under carefully controlled laboratory conditions is
an expedient.
It is apparent that, in general, flow nets are not easily obtained.
It will be made equally apparent, however, that a flow net once drawn
 SELECTED TOPICS OF FLUID MECHANICS 25

for a particular boundary shape affords a universal solution for many

fundamental flow characteristics. Some typical two-dimensional
flow nets are shown on plate 2 (a-h) .
One of the definitive characteristics of a two-dimensional flow net
is that the streamlines are so spaced that the discharge between all
adjacent streamlines in the net is a constant. Furthermore, from the
equation of continuity, the discharge past successive sections in each
of the^two-dimensional streamtubes is also constant. In plate 2 (d), v
is the average velocity and Aft is the corresponding normal distance
between any pan1 of streamlines in a typical flow net. Because
Ag=0 An, and because Ag<,=A<fr=Ag2=A<fe in plate 2 (d), it follows that
v0 kn0=Vi Afti=02 Aft2, etc., throughout the flow net. Thus, the relative
velocity at any point in the flow pattern is indicated by the distance
between streamlines at that point. For example, if v0 and An0 are
described as reference characteristics in a particular flow net, the
velocity at any other point is v=&q/&n=(v0 Aft0)/Aft, or

. I .
_o
.
t
) Bill/.
V0 Aft v0 Aft! v0 Aft
In regions of flow where acceleration is appreciable, finite values of
the normal distance between streamlines provide only an approximate
measure of the mean velocity between the streamlines. Further
subdivision of the net, however, affords a means of improving the
accuracy of the measure. The limit of the subdivision procedure
occurs when &n=dn. Then Aq dq=v dn is the constant infinitesimal
discharge in each of an infinite number of streamtubes which make up
the flow net. Actually, the decision regarding the number of stream-
lines to be drawn that is, the practical limit of subdivision is
always based on a consideration of the labor involved, the accuracy
attainable, and the accuracy required.
The potential lines on a flow net are a family of lines which are
orthogonal (perpendicular) to the streamlines, including the effective
boundaries. The potential lines are customarily drawn so that they
are everywhere equ^al in spacing (As) to the spacing of the adjacent
streamlines (An). Together, the conditions of perpendicularity and
equivalent spacing of streamlines and potential lines govern the
graphical, successive-approximations method of constructing a flow
net.
In regions where the flow is uniform the meshes formed by pairs
of streamlines and potential lines are squares. In regions of non-
uniform flow, the meshes are imperfect squares, although individually
their medians are equal and their corners form 90° angles. If the net
is further subdivided, however, all of the component meshes approach
perfect squares as Aft and As approach dn and ds, respectively, just as
26 RIVER HYDRAULICS

all curved lines approach straight lines as the length of line considered
approaches zero.
Because in any local zone As is equal to Aw, it follows that As is a
measure of the mean velocity between adjacent potential lines for the
same reason that Aw is a measure of the mean velocity between ad-
jacent streamlines. Thus, in a uniform-flow zone, An=As, and
v=-7
An0 As0 etc.
=-T-^ . /OQv
(36)
V0 An As
In nonuniform zones, either An or As might be preferable as an approxi-
mate measure of the mean velocity, depending on the characteristics
of the flow pattern and the purpose of the analysis. For example,
along any streamline, including the boundary, the maximum and
minimum velocities are best indicated by the potential-line spacing.
Typical velocity-distribution curves, determined from the flow net
for a sluice gate, are shown on plate 2 (g). Here, velocities along the
floor of the channel and along the face of the sluice gate are repre-
sented as a multiple of the downstream uniform velocity (F2). Thus,
the velocity-distribution curves shown on plate 2 (g) are plots of th«
relationship

in which v is the velocity at any point, As. is the distance between poten-
tial lines at that point, and As2 is the distance between potential lines
in the downstream, uniform-flow zone.
It has been established that the velocity variation throughout an
irrotational-flow pattern is related to the piezometric-head variation.
It follows that the flow net provides a means of analyzing the piezo-
metric-head distribution throughout any flow pattern for which a net
can be drawn. From the Bernoulli equation (14), when it is applied
between a reference point and a general point in any irrotational-flow
pattern,
, , Vg2 V2

in which h=(p/y)-\-z is the piezometric head. By dividing both sides

of this equation by the velocity head at the reference point and
observing that
v_Ang_As0
v0 An As '
then
' (37)
v '
 SELECTED TOPICS OF FLUID MECHANICS 27.

The left-hand member of equation 37, called the piezometric-head

ratio, is dimensionless. It contains only one dependent variable (h),
in addition to the reference values (h0 and v0). The terms on the right-
hand side of the equation indicate that the piezometric-head ratio is a
function of the flow-net configuration alone. It has been established
from hydrodynamics that, if one can be drawn, a flow net is unique for
any given boundary conditions; in other words, there is only one flow
net for a given boundary form. It follows that the piezometric-head
ratio is determined for all points in a given flow pattern by the shape
of its boundaries. The full significance of this conclusion is best illus-
trated by the observation that both the velocity ratio (eq 36) and the
piezometric-head ratio are independent of the size of the boundary as
well as of the fluid density, discharge, elevation, and pressure.
Typical piezometric-head distribution curves, corresponding to the
velocity curves in plate 2 (g), are shown for the sluice gate in plate 2 (h).
From equation 37, when it is applied between a general point and a
point in the uniform flow downstream from the gate,

in which As2, V22/2g, and hz are constants which, for any scale of
drawing, fix the scale of the piezometric-head graph. Thus, the only
variable in this equation is the distance As, which is the scaled dis-
tance between potential lines at the point where the piezometric head is
A. For convenience, in plate 2 (h) the piezometric head is represented as
a multiple of the gate opening (6); that is, h=Cb, where C is a function
of a single variable (As).
Because the elevation of any point on the flow pattern shown in
plate 2 (h) is readily scaled off the drawing, it follows that the pressure
head at any point along the gate face or the floor is represented as the
difference between the plotted piezometric head and the elevation. In
this manner the flow net can be used to evaluate the total fluid-pressure
force acting on any portion of the boundary.
It is emphasized that the sluice gate shown in plate 2 (f-h) represents
a fluid motion in which the influence of viscosity is almost negligible.
Also, it is observed that, in this example, the influence of fluid weight
is confined to its accounting for a variation of pressure with elevation.
In other words, this is the kind of flow pattern for which a flow net
yields useful, practical information. Many such examples could be
given.2

' Rouse, Hunter (ed.) 1960, Engineering hydraulics: New York, John Wiley <& Sons, Inc. See p. 22,
33, 47-52, 534-642.
,28 RIVER HTDRAULICS

The coefficient of discharge for convergent-boundary flow patterns

has been defined by the relationships
Q
C A.
/
The prerequisites for the applicability of this equation are essentially
the same as those which determine whether the flow pattern can be
represented by a flow net. In fact, a property of all flow nets that
the streamline configuration is uniquely determinedly the form of
the fixed boundaries leads to the conclusion that the second member
of equation 34 involves only dimensionless characteristics of the flow
boundary. In other words, Ce and, therefore, Cd, are functions of the
boundary geometry alone. This conclusion is substantiated by the
observation that the right-hand member in equation 34 is a particular
form'of the piezometric-head ratio; that is,
s.
C*= / / . .= ,~> (38)

V ft'l" /»2
~ET

where V is the average velocity through the gross orifice area and
hi hz is the difference in piezometric heads between the two uniform-
flow regions designated sections 1 and 2. Thus, Cd has all the char-
acteristics of any piezometric-head ratio, including those of being
dependent on the form of the fixed boundaries and independent of
boundary scale, fluid, density, discharge, elevation, and pressure.
THE EXPERIMENTAL METHOD
A simple kind of motion which involved mainly the forces due to
pressure gradient and fluid weight was examined in. the jpreceding
section. The discharge of an incompressible fluid through a plate
orifice between adjacent large reservoirs was described as a practical
example of this kind of motion. In general, of course, additional
forces must be considered. Thus, when a fluid flows between fixed
boundaries, viscous and turbulent shear stresses are involved. On
the other hand, when accelerated motion with a free surface occurs,
certain independent effects due to fluid weight must be considered.
Unfortunately, flow patterns which depend on the combined effects
of several fluid-property forces are seldom subject to complete mathe-
matical description. Thus, a full complement of expanded equations
of motion, counterparts of the equations for a simple fluid motion
 SELECTED TOPICS OF FLUID MECHANICS 29

derived in the preceding section, is not forthcoming. In this section

a method of analysis which provides an indirect evaluation of the
relative influence of the fluid-property forces will be considered.
INFLUENCE OF SHEAR
The effects of viscous and turbulent shear are prominent in most
fluid-flow phenomena. They are manifested, usually, by their influ-
ence on the energy of flow and th.3 external aspects of the flow pattern.
Thus, as a consequence of viscous shear, heat is generated within a
moving fluid and the energy of the flow system is thereby degraded.
On the other hand, she'ar stresses are primarily responsible for the
occurrence of boundary drag forces, separation zones, eddies, and
various phenomena associated with the boundary layer.
In the absence of comprehensive equations of motion, practical
solutions for many problems in engineering hydraulics involve em-
pirical adjustments of the simple equations of motion derived in the
preceding section. Thus, an understanding of the complex part
played by the shear forces depends, in general, on an understanding
of the techniques and limitations of dimensional analysis and experi-
ment.
EXPERIMENTS AND THE THEORY OF SIMILITUDE
The principal purpose of experiment is to determine the relative
influence of each of the variables which govern the occurrence of a
phenomenon. For occurrences lacking a general analytical solution,
however, the identification of the independent variables is often as
obscure as the evaluation of their influence. Frequently, therefore,
an auxiliary purpose of experimental observations is to isolate the
effective variables. Many practical problems have been satisfac-
torily solved by semi-empirical methods based on experimental data.
However, it is emphasized that empirical solutions lack generality by
their failure to recognize the exact functional relationship between
the primary variables.
The variables involved in fluid motion are contained in one of
following categories: boundary conditions; kinematic or flow charac-
teristics, including acceleration and velocity (or discharge); and
dynamic characteristics, or the forces which cause motion. One of
the most powerful tools of modern fluid mechanics is that which utilizes
dimensionless ratios to describe boundary geometry, flow character-
istics, and force characteristics. The dimensionless-ratio technique
provides a means of correlating similar phenomena without restriction
as to scale, flow, or fluid properties. It is fundamental to the planning
and execution of experimental programs, to the analysis of experi-
mental data, and to the application of solutions which are based on
experimentation.
30 RIVER HYDRAULICS

The general Newtonian equation of motion for fluids has already

been expressed in the symbolic form (eq 3), FP+*Fg+*FT-\-*F,-\-*Fe=
Ma.
Plate 2 (i-1) illustrates the significance of such an equation when ap-
plied to the motion of a small element of fluid wherein the effects of
Fff and Ft are negligible. Thus, in plate 2 (j),
F,+*Fg+>Fr =Ma. (39)
The size of the polygon which represents this force system depends
on the scale selected for the drawing and the relative magnitude of
the forces involved. It follows that the force picture at a correspond-
ing point in a completely similar flow pattern might be represented
to the same scale by plate 2 (k). However, if every term in equation 39
were divided by the resultant inertial reaction (Ma),
F F F
++ 1' (4e>
the force systems shown by the similar polygons in plate 2 (j, k)
might both be represented by the unit polygon shown in plate 2 (1).
At corresponding points in all truly similar flow patterns such dimens-
ionless force polygons are identical.
The magnitude of each of the ratios in equation 40 is a relative
measure of the part played by a particular accelerative force. Each
of the ratios, therefore, has obvious significance as a criterion of
similarity. For example, the first ratio is a measure of the influence
of the pressure force relative to the total inertial reaction. It is
evident that when this and each of the other ratios are fixed, the force
pattern and, therefore, the motion pattern, are determined. It
follows that two or more flow patterns are completely similar when
their boundaries are geometrically similar and the corresponding force
ratios are identical. Thus, dynamic similarity is independent of the;
size of the boundaries, the magnitudes of the velocities, or the physical
properties of the fluids involved.
Equation 40 and plate 2 (1) have been used to demonstrate the signifi-
cance of dimensionless force ratios as a means of describing a particular
fluid motion. Equivalent ratios, which are proportional to the recip-
rocals of the square roots of the ratios in equation 40, are widely
accepted as general criteria of similarity. The first of these is the
Euler-number (E),

» £= (41>
 SELECTED TOPICS OF FLUID MECHANICS 31

in which V is a typical average velocity and Ap is a significant pres-

sure difference. The Euler number is derived from the pressure-
force ratio, the first term in equation 40. Thus, the Euler number
is a measure of the relative influence of the pressure force.
Similarly, a ratio which is a measure of the influence of the gravity
force and which is derived from the second term in equation 40 is
called the Froude number (F). For free-surfaee flows, the accepted
form of this ratio is

(42)

in which L is a significant length.

In the third term of equation 40,

TT irr JfTT' WH-Constant,

FT Fu . Fn fll,r\ 4. 4.

Ma Ma Ma Ma
in which FJMa is a constant because turbulent shear forces are pro
portional to the total inertia! reaction. It follows that

. (43)
Ma
The third term in equation 43 represents the relative influence of the
independent viscous shear force. The accepted form of the ratio
which is derived from this term is called the Reynolds number (R) ,

(44)

Similar ratios which have been formulated to describe the surface-

tension and elastic forces are called the Weber and Cauchy (or Mach)
numbers, respectively.
Inasmuch as the Euler, Froude, and Reynolds numbers are propor-
tional to tjie dimensionless ratios in equation 43, it follows that com-
plete similarity of two or more flow patterns will prevail when their
boundaries are geometrically similar and their Euler, Froude, and
Reynolds numbers are equal.
Of the ratios in equation 43, the Euler number (E) has special
significance as a general flow parameter. Because it involves no fluid
property other than density, E is the sole criterion of similarity for
motion which, from hydrodynamics, is described as ideal. Thus, for
an enclosed flow in a horizontal plane with negligible influence due to
shear, inertial reactions are a function of pressure forces alone. For
any given boundary conditions, therefore, E, representing a ratio of
32 RIVER HYDRAULICS

a typical inertial reaction to a typical pressure force, is a constant. In

other words, for ideal-fluid motion, E is a function of boundary condi-
tions alone. On the other hand, if forces other than pressure are
partly responsible for the total inertial reaction, the magnitude of E
reflects the relative influence of the other forces. As all the other
ratios are related to the fluid properties, the Euler number is appro-
priately selected as a dependent ratio. This leads to the statement
of a general principle of similitude: Values of E for two or more flow
patterns are identical, and the patterns, therefore, are exactly similar
if their boundary conditions are similar and if the corresponding
values of the fluid-property parameters, such as F and R, are equal.
A significant example of the Euler number is the coefficient of
discharge described on pages 24 and 28. Other examples are the
uniform-flow resistance coefficients, various minor-loss coefficients,
drag-force coefficients, etc. From equation 38,

£
A V ==E. (45)
/2
i Pa)

To all those conclusions concerning Cd which were derived previously

from the piezometric-head ratio analogy and the flow-net character-
istics, there are now added the limitations imposed by the influence
of the fluid-property forces. Thus C&, obviously an important flow
parameter, is now defined as a function of not only the boundary
geometry but also the several fluid-property parameters derived in
this section.3
INFLUENCE OF A FREE SURFACE

Certain kinds of free-surface flow patterns, such as the sluice gate

(pi. 2, f-h) and the simple measuring weir, are free of independent
weight-force effects. In other words, for those flow patterns which
are uniquely determined by the fixed boundaries, the only influence
of fluid weight is to account for a variation of pressure with elevation.
Under other conditions, as illustrated by the brink at the end of a
sloping channel or the orifice at the end of a pipe, the weight force
can have an independent influence on the flow pattern. For example,
when a fluid discharges from an'orifice into a space occupied by
another fluid of different density, the trajectory of the jet and many
related characteristics of the flow pattern depend on the velocity of
approach, the specific weight of the fluid flowing, and the specific
weight of the surrounding fluid, as well as the boundary geometry.
* Rouse, Hunter, 1946, Elementary mechanics of fluids: New York, John WUey & Sons, Inc. The reader
Is urged to study the sequence of Ct curves represented by figures 31, 62,86, and 129*
 SELECTED TOPICS OF FLUID MECHANICS 33

The effect of the weight force in this example is to deflect the jet
upward or downward, depending on its effective weight or buoyancy
relative to the. surrounding fluid. The relative magnitude of this
effect depends on the ratio of the inertia of the jet to the weight force.
Such a ratio, of course, is the Froude number. However, for an
application of this kind, F must be written in the more general form,

(46)

in which AY is the difference in specific weights of the two fluids in-

volved and p is the density of the fluid flowing.
Practical problems related to free-surface flows are seldom subject
to complete analytical solution. By means of the Froude criterion,
however, along with the other dimensionless force ratios described
previously, satisfactory solutions based on experiment can be
achieved.
RECAPITULATION

A preceding section was principally concerned with a simple kind

of fluid motion in which the only effective forces were those due to-
pressure difference and weight. The main objective of that section
was to derive equations which constitute a mathematical description
of the motion. The most important equations derived were the
Bernoulli equation, which defines the relationship between velocities,
pressures, and elevations throughout the flow pattern; the one-
dimensional equation of motion, which describes the variation in the
average flow characteristics between sections of essentially uniform
motion; the one-dimensional equation of continuity; and the dis-
charge equation. In addition, the flow net was introduced as a
valuable analytical tool and a means of demonstrating the significance
of the equations of motion.
It is now acknowledged that neither the simple equations of motion
nor the flow net are theoretically compatible with the occurrence of
shear forces. It is also acknowledged that, in general, it is not pos-
sible to derive expanded equations of motion which adequately define
the influence of shear and certain free-surface flow characteristics due
to fluid weight. Thus, for many problems involving appreciable ef-
fects due to viscosity, weight, or any other fluid property, solutions
based on field or laboratory experiments must be sought. For this
purpose the dimensionless-ratio technique of specifying boundary
characteristics (a length ratio, usually), flow characteristics (a coeffi-
cient of discharge, piezometric-head ratio, efficiency, energy-loss co-
efficient, or some other form of the Euler number), and fluid-property
34 RIVER HYDRAULICS

force characteristics (the Froude, Reynolds, Weber, and Cauchy

numbers) has been described. Flow-pattern descriptions based on
dimensionless ratios are completely independent of restrictions re-
garding scale, discharge, pressure, or fluid properties. Thus, dimen-
sionless ratios are the basis for the general principle of similitude
which governs all experimental investigations.
EXAMPLE
The manner in which the principle of similitude can be used to guide
the planning and execution of an experimental program is demon-
strated by the following example. It is proposed to make laboratory
tests on a diaphragm orifice. The purpose of the tests is to determine
the discharge characteristics of an 8-inch thick-plate metering orifice
in a 16-inch pipeline carrying crude oil. The laboratory tests are to
be made in a 4-inch pipe carrying water. The principal objective of
the investigation is to define the coefficient of discharge (Ca) for a full
range of oil-pipeline (prototype) flows. This objective is incidental,
of course, to the requirement that the flow pattern in the pipeline be
exactly simulated by the flow pattern created in the experimental
setup.
The scale ratio of lengths in this example is determined by the ratio
of pipe diameters, 1:4. In order to achieve geometric similarity,
therefore, the laboratory orifice must be one-fourth as large as the
pipeline orifice, or 2 inches hi diameter. Similarly, all other length
characteristics, such as plate thickness, rounding radius, and piez-
ometer locations, must be reproduced in accordance with the 1:4
length-scale ratio.
In order to obtain completely similar flow patterns in the vicinity
of the geometrically similar orifices, the relative influence of each of
the effective independent forces must be the same. In this example,
viscous shear forces are involved in the occurrence of the boundary
layer upstream from the orifice and in all the consequent effects de-
scribed on page 22. Because the flow pattern is entirely enclosed,
however, the weight force does not have an independent influence.
Surface-tensiQn forces are absent because there are no fluid interfaces,
and elastic forces are negligible because both of the fluids involved are
liquids.
It follows that, for similar boundary conditions, pressure differences
are functions of viscous shear forces and accelerations. Therefore,
ratios of the Euler variety, including the coefficient of discharge, are
functions of the Reynolds number alone. In other words, complete
similarity of the flow pattern (including turbulence) will be achieved
and the discharge coefficients for the two orifices will be equal when
their Reynolds numbers are equal. It is this condition which de-
 SELECTED TOPICS OF FLUID MECHANICS 35

termines the range of laboratory test conditions required to define C&

for a full range of pipeline discharges.
Thus, for similar flow patterns and equal coefficients of discharge,
R«> ( water) = (oU)j

(VLP\ }
\ M /w \ M /o

in which V is usually defined as the average velocity based on the

gross area of the orifice, and L is the diameter of the orifice. From
this relationship, the ratio of velocities for similar flow conditions is
vw /LO\/PO\/HW\ ( 7.
(47)
and, because discharge is proportional to the product of a velocity
and an area (or the square of a length) ,

'&-~Ajr
(giB_ A-w V0w -\j:J
_ /jLuA p.

Therefore, the discharge of water in the laboratory setup required to

simulate a given rate of flow in the pipeline can be expressed in terms
of the length ratio, the density ratio, and the viscosity ratio, or,

« {©(£)©> (48)
In this example, LWIL0=Y^ Furthermore, the density and viscosity
ratios are virtually fixed because, for given fluids, these properties
can be altered only slightly (by changes in temperature). Thus,
equation 48 is of the form Quj=K Q0, and K is essentially constant.
The example would have been considerably complicated if the
metering orifice had been located at the end of the pipeline. Because
the orifice would then discharge freely into the atmosphere, fluid
weight would be added to the independent variables already con-
sidered. Thus, the Froude number would be an additional criterion
governing the performance of the laboratory investigation. Under
certain conditions this added requirement makes complete similarity
impossible to achieve.
The difficulty of satisfying more than one fluid-property criterion
is easily demonstrated. From the requirement that the Froude
numbers be equal for similar flow patterns,
LA /JLA
36 RIVER HYDRAULICS

or,

whence,

~o- (so)

For similarity with respect to viscous forces as well as the weight

force, equation 50 must be equivalent to equation 48. Solving the-
two equations simultaneously,

or,

(51)

Equation 51 places a definite restriction on the physical properties

of the fluids involved in the investigation. In fact, assuming that
the length ratio and the experimental fluid are fixed within narrow
limits by practical considerations, equation 51 imposes an impossible
condition on the investigation. Thus, because of the small range in
magnitude represented by the viscosities and densities of ordinary
fluids, it is a conclusion of general significance that it is impossible
to satisfy simultaneously more than one of the fluid-property criteria.
Fortunately, the apparent difficulty in attempting to satisfy all
of the similitude criteria is not a hopeless barrier to experimental
investigations. It is recalled that the Reynolds number, for example,.
is proportional to the ratio of a typical inertial reaction to a typical
viscous-shear force. The relative influence of F^ , therefore, diminishes
as R increases. Experiments show that the effect of viscosity on
CA for most sharp-edged orifices is negligible at values of R above a
critical value of 100,000 * (not an unusually high value, as it cor-
responds, for example, to the discharge of water at 70° F at 6 feet
per second through a 2-inch orifice). Similarly, the influence of the-
weight force diminishes as the Froude number increases, becoming
negligible at a comparatively low critical value of F.6
In the range wherein variations in the magnitude of a given fluid-
property criterion produces negligible changes in the flow pattern,
* Bouse, Hunter, 1946, Elementary mechanics of fluids: New York, John Wiley & Sons, Inc. See fig.
129, p. 256.
« Ibid, fig. 52, p. 105.
 SELECTED TOPICS OF FLUID MECHANICS 37

that criterion can be ignored as a condition for similitude. Thus, it

is possible that neither viscous forces nor weight forces have an appre-
ciable effect on the coefficient of discharge within the full range of the
pipeline flows. If this were true, the coefficient of discharge would be
a function of boundary conditions alone, and the laboratory program
to determine Cd for the metering orifice might actually consist of a
single test. However, that test could not be made with assurance
until a series of tests had established the lower critical values for both
F and R. It is apparent, therefore, that the fluid-property parameters
are useful not only as similitude criteria but also as means of defining
the effective range of influence of the various fluid-property forces.
The practical theory of model testing makes use of the fact that a
single force can have a dominant, although not an exclusive, influence
on the flow pattern.
THE ENERGY EQUATION

REVIEW OF THE BERNOULLI EQUATION

One of the fundamental concepts of fluid mechanics and the basis

for many of the practical formulas of engineering hydraulics is the
energy equation. To bring a detailed consideration of the energy
concept into this discussion requires first a review of certain steps
taken in the derivation of the equations for a simple fluid motion.
Plate 1 (c) shows the small element of fluid which, on page 13, was
considered to be undergoing accelerated motion because of a com-
bination of pressure and weight forces. It is pertinent now to discuss
certain characteristics of this element which make it possible to
'associate with it some of the familiar concepts of particle mechanics.
The elementary volume of fluid shown in plate 1 (c) was assvmed to be
large enough to be represented by steady-motion characteristics hi a
turbulent flow. On the other hand, it was assumed to be small
enough that pressure forces could be accurately expressed in terms of
a mean pressure intensity, pressure gradients could be assumed con-
stant over the length of the element, and velocity characteristics could
be described in terms of the velocity at the mass center of the element.
Thus, as the volume of the element (As &A) was,allowed to approach
the differential quantity ds dA, the motion characteristics of the
element were described as the steady-motion characteristics at a point.
They might also be described as the characteristics of a fluid particle.
Because fluid flow is a continuous process, it is often convenient to
consider the fluid particle as the instantaneous image of the moving
fluid contained in a particular space in the flow pattern. Thus,
equations of motion for a fluid particle can be related to the rate of
38 RIVER HYDRAULICS

flow through the space designated as its location. The volume of the
element, for example, is
dY =ds dA=v dt dA=dQ dt (52)
in which v is the mean velocity in the differential length (ds) of the
infinitesimal streamtube on which the element is located. It follows
that the weight of a fluid particle (dW) can be expressed as
dW=ydY=ydQdt. (53)

An equation for the simple fluid motion represented by plate 1 (a-c)

is equation 8,

Y(As AA)=pa,(As AA). (8)

Collecting terms in the left member of this equation, substituting the

equivalent expression for a» from equation I,6 and replacing As AA
with the differential volume (ds dA),

dA. (54)

Reviewing the steps taken in obtaining equations 8 and 54 from

Newton's fundamental equation, it appears that they both represent,
on the left, the resultant of the external forces acting on the fluid
element in the s direction, and, on the right, the total inertial reaction
in the same direction.
Continuing with the particle analogy, if dW is substituted for
7 ds dA in equation 54, and terms are rearranged,

Equation 55 states simply that the s gradient (rate of change with

respect to s distance) of the quantity in parentheses is zero. If the
gradient is zero, it follows that the differential of the quantity in
parentheses is zero over a small distance (ds), or,

(S6)
The process of determining the differential of the quantity in paren-
theses consisted of multiplying the gradient by the distance (ds). It
* The subscript on v. in the definition of a, will henceforth be omitted. By definition, vn is equal to zero,
and v, is therefore equal to tbe total velocity, c.
 SELECTED TOPICS OF FLUID MECHANICS 39

is this operation which converts the vector force-inertia relationships

in the preceding equations into a scalar work-energy relationship
(eq 56). The conversion might have been more apparent had the
operation been performed while the differential equation was in the
form shown by equation 54. In that procedure, multiplying the
left-hand term, a force, by ds converts it to a measure of the work
performed by the external forces during the time interval (dt) repre-
sented by the average velocity (v) and the distance (ds). Concur-
rently, the right-hand term, [d(pv*l2)] ds dA~d[dW(i^l2g)], is a meas-
ure of the change in kinetic energy of the particle whose weight is dW.
Equation 56, therefore, is an expression of the familiar work-energy
principle of mechanics. A convenient interpretation of the concept
of work-energy equivalence and an examination of the units involved
in equation 56 lead to the practice of referring to the parenthetical
quantity as the total energy per pound of the fluid flowing. Thus,
equation 56 is interpreted to mean that the total energy of the particle
of weight (dW) remains unchanged over the distance interval (ds).
It follows that the quantity in parentheses is a constant during the
interval, or,
(57)

Equation 57 is simply the integral (along the streamline) of equa-

tion 56. When it is divided by dW it becomes the familiar Bernoulli
equation,
/» » n \
-H, (58)

in which H, the constant of integration, is the total energy per pound

of fluid flowing.
In review, the Bernoulli equation is a description of a simple fluid
motion along a streamline. In its customary form (eq 58), it states
that the sum of three quantities, collectively defined as the total
energy per pound of fluid (H), remains constant during the motion.
Individually, the quantities which constitute H may be either a
measure of work performed on the fluid or a change in its mechanical
energy. Thus, the Bernoulli equation states that the total energy or
"work potential" of 1 pound of fluid in motion along a streamline
remains constant regardless of transformations due to changes in
velocity, pressure, or elevation.
It has already been established (p. 16) that the Bernoulli equation
could be assumed to apply to an entire flow pattern as long as fluid
viscosity is ignored. However, the practical objective is a general
equation of motion which includes terms to account for the influence
of viscous and turbulent shear.
40 RIVEH HYDRAULICS

GENERAL ENERGY EQUATION

As observed previously the influence of viscous and turbulent shear

is manifested by its effect on the energy of flow and by its effect on
the external aspects of the flow pattern. Unfortunately, general
equations of motion which describe these effects are not yet available.
Thus, a practicable general energy equation is based on a simple
modification of the procedure leading to the Bernoulli equation.
The equation of motion for a single streamline (eq 56), modified to
acknowledge the loss of energy, takes the form

(59)

in which dH is the differential loss in total energy head in the distance

ds. From equation 53 the term dW in equation 59 is equivalent to
y dQ dt, in which dQ is the differential flow through the elementary
streamtube which contains and defines the particle represented by
dW. Because dQ is constant in a steady flow, equation 59 is readily
integrated with respect to distance along the streamline. Thus, be-
tween any two points designated 1 and 2 on a streamline,

(60)

In order to obtain an equation which describes the total flow

between enclosing boundaries, it is necessary to integrate equation
60 over the total cross-sectional areas which contain points 1 and 2.
The difficulty of performing this operation is made apparent by
recalling that dW y ds dA yv dt dA. It is evident that integration
with respect to area must take into account the fact that the velocity,
in general, is variable over the cross section. Therefore, unless the
velocity is essentially constant or is analytically defined as a function
of area, the-indicated integration is impossible. The assumption of
constant velocity across a section between solid boundaries would be
inconsistent with the known effects of viscosity. On the other hand,
explicit analytical relationships between velocity and area are im-
possible to formulate except for a few classic examples of uniform
motion.
Not only does variable velocity hinder integration over total cross
sections, but, as a result of boundary nonuniformities, a nonhydro-
static pressure variation may present a similar obstacle. However,
if it is assumed that the integration can be performed over a section
of parallel flow, which is usually possible, the pressure distribution
can be assumed to be hydrostatic that is, the sum of the quantities
pjy and z is constant and this obstacle is removed. Thus, if y dQ dt is
 SELECTED TOPICS OF FLUID MECHANICS 41

:substituted for dW in equation 60, and if both sides of the equation

.are divided by dt, integration with respect to the discharge (Q) in the
total area gives

Incidentally, the foregoing division by dt converts the work-energy

relationship into a power relationship. Thus, each of the expressions
in brackets on the left-hand side of equation 61 is a measure of the
total power of the flowing fluid. The expression on the right-hand
tside, therefore, is a measure of the power "lost" between the designated
cross sections.
If equation 61 is divided by the quantity <2y, it again becomes an
-energy equation, and the units of every term in the equation are
ioot-pounds per pound (or feet) of the fluid flowing. Thus,

Defining the right-hand member of this equation as the , average

«energy (head) loss (HL) between sections 1 and 2, and substituting
the piezometric-head symbol (h) for (p/y)-\-z,

-* -*« (62)
JEquation 62 is a general energy equation which is restricted only by
the assumption of hydrostatic pressure distribution at the designated
cross sections. The head-loss term (HL) ordinarily requires evalua-
tion on the basis of experiment.
In many practical applications the integration involved in the
velocity-head terms in equation 62 can be approximated by a numeri-
cal summation. Thus,

Ifrom this, the average kinetic energy or velocity head in a total cross
^section can be defined as the discharge-weighted mean of the velocity
Jieads in its constituent subsections. Each subsection is taken to
represent a zone of uniform velocity. This process is conveniently
adapted to the practice of determining discharge by means of current-
aneter or pitot-tube velocity traverses.
42 RIVER HYDRAUUCS

The minor importance of the kinetic-energy terms- in certain appli-

cations of the general energy equation does not justify involved
methods of accounting for the velocity variation in a cross section.
For those applications it is customary to approximate the true average
kinetic energy in terms of the average velocity in the section (V).
The result is a modification of equation 62 which is another form of
the one-dimensional equation of motion (eq 26),

The error inherent in the velocity-head terms of equation 64 is often

acknowledged by means of a coefficient (a),

a, of course, is not an independent expression of the influence of

velocity nonuniformity. It can be evaluated only in terms of the
integral or summation definitions of the true average velocity head,
equations 63 and 65. Thus,

«=^J (66)

Equation 65, better than any other, demonstrates that the main
objective of this discussion has been accomplished. The result is a
general energy equation involving separate terms to account for both
effects of shear a as a measure of its influence on the flow pattern,
and HL as a measure of its energy-dissipating effect.
THE MOMENTUM 'EQUATION
Numerous problems arise in applied fluid mechanics which are not
readily or completely defined by the energy equation. The solution
of many of these problems can be accomplished by means of an in-
dependent form of the equation of motion which does not require
knowledge of the energy losses due to internal viscous shear. This
is known as the momentum equation. As both the momentum equa-
tion and the energy equation are derived from Newton's fundamental
equation of motion, it is necessary to examine the derivations care-
fully in order to establish their independence.
A common starting point for the derivation of both the energy
equation and the momentum equation is the differential equation
given on page 38,

{54)
 SELECTED TOPICS OP FLUID MECHANICS 43

Equation 54 is a statement of Newton's second law as applied to the

accelerated motion of a fluid element. The left-hand member, for
the conditions represented by plate 1 (a-c), is the resultant accelerative
force, and the right-hand side is the total inertial reaction in the
direction of the total-velocity vector. Equation 54 is a vector equa-
tion because force and inertial reaction are vector quantities.
It has been demonstrated that the work-energy significance of the
Bernoulli equation resulted from the fact that, in its derivation, both
sides of equation 54 were multiplied by the distance (ds). This opera-
tion converted force to work and inertial reaction to change in kinetic
energy. Both work and energy are scalar quantities. Thus, the
Bernoulli equation resulted from the integration of a work-energy
equation over a streamline distance without regard for the change hi
direction of the s distance.
It is pertinent to recall that the effects of the shear forces were
ignored hi the derivation of equation 54. Subsequently, however, in
the derivation of the general energy equation, a term which would
compensate for the work done by the internal viscous shear was
included.
In the derivation of the momentum equation, it is convenient to
replace the left-hand member of equation 54 with one symbol which
represents the resultant of all external forces. Thus, in the form

<M, (67)

the quantity dFa is assumed to include the boundary shear force as

well as the boundary pressure force and the weight force. It does not
include internal shear forces, because these are mutually canceling.
Apparently, therefore, the independence of the energy and momentum
equations is related to the fact that the first involves the internal
viscous shear forces and the second involves the external boundary
shear force.7
From the calculus, d^/ds=2v(dv/ds'), where dvfds is the gradient
of the total velocity in the s direction. When this is substituted in
equation 67,
£p rWI j dA=pv
-j- I <k jA [/dv\ j dA,
j- ) ds JA
2[_dsj \ds/
or,
dF,=pv dv dA=p dQ dv, (68)
in which dQ is the rate of flow in the infinitesimal streamtube which
contains the fluid element and dv is the change in velocity over the
1 The boundary shear force Is not considered in the derivation of the work-energy equations because the
relative velocity of the fluid at the boundary is zero, and therefore the work done by the shear force is also
44 RIVER HYDRAULICS

differential distance (ds) which is the length of the element. Equation

68 is still a force-inertia equation. Therefore, it is still a vector
equation.
Because motion along a streamline can involve a change in the
direction as well as the magnitude of the force and velocity terms in
equation 68, an adequate algebraic description of accelerated motion
involves similar equations in each of the fixed coordinate directions.
Thus,
dFx =p dQ dvx, (69a)
dFv=p dQ dvv, (69b)
and dFs=pdQdvs. (69c)
The quantity p dQ dv in equations 68 and 69 has been described
as the inertia! reaction of an elementary fluid particle. The product
p dQ is equivalent to dM/dt, where dM is the mass of the particle. If
this substitution is made in the x form of equation 69, for example,

P dQdvx=jt (Mvx\ (70)

the right-hand member acquires a new significance. In this form it

is commonly described as the rate of change of x momentum or the
change of z-momentum flux in the flow distance ds. This is the origin
of the name "momentum equation."
For practical applications the momentum equation must be ex-
tended to a finite reach and a finite cross section. The operations
required for this extension involve double integration of the differ-
ential equation of motion first with respect to distance along the
streamline and then with respect to the area of the cross section.
Similar operations were involved in the derivation of the general energy
equation. This tune, however, a vector relationship is involved, and
the integration procedures must be performed for each of the signifi-
cant coordinate directions. In the following steps the x direction has
been selected as typical.
Because dQ, in equation 69 is constant, integration with respect to
distance is readily accomplished. Thus, if the 3-distance integral of
dFx is represented by &FX,

A^=p dQ vf2 -p dQ vXl. (71)

Because all internal forces are canceling, &FX is simply the resultant
of the external forces on the finite reach of streamtube between
sections 1 and 2.
 SELECTED TOPICS OF FLUID MECHANICS 45

The next step is to integrate equation 71 over the total areas at

sections 1 and 2. Substituting dQ=v dA,
r|A*>=pJCA*vXzv2 dA2-p\CAiVsfidAi, (72)

it is apparent that integration depends again (see page 40) on an

integrable relationship between velocity and area. Thus, a general
solution of equation 82 is prevented by the fact that explicit analytical
relationships between v and A are seldom attainable. The left-hand
member, however, can be simplified to 27^, a symbol which represents
the resultant of the external forces acting in the x direction on the
total volume of fluid between sections 1 and 2. Thus, general momen-
tum equations in the fixed coordinate directions are
CQ
r Vx2 dQ p I vxi dQj (73a)

S*V= P JX <Z£-P JX dQ, (73b)

and
2F,=p {Qv,2 dQ-p fX dQ- (73c)

The integrations implied by equation 73 can be approximated by

the numerical summation

(74)

in which the quantity pvx A<2 represents the momentum flux through
a constituent subsection assumed to be located in a zone of essentially
uniform velocity.
Further paralleling the various forms of the general energy equa-
tion, a typical one-dimensional form of the momentum equation is
ZF^QpfaV^-QphV^, (75)

in which V is the average velocity and 0 is a coefficient which accounts

for the nonuniformity of velocities in each cross section. From
equations 73, 74, and 75,

The energy equation and the momentum equation describe the

same state of motion, and they are derived from the same differential
46 RIVER HYDRAtfl/ICS

equation. They are independently useful, however, because the

energy equation is independent of the external boundary shear, and
the momentum equation is independent of the energy losses due to
internal viscous shear. Thus, equations 65 and 75 can be shown to
be equivalent only when the relationship between the boundary shear
force and the internal, energy-dissipating viscous shear forces is
explicit. This is true only when the flow is truly uniform (see below).
Many complex phenomena (such as the hydraulic jump, abrupt
enlargements, and short transitions, all involving rapidly accelerated
motion) are readily and accurately described by means of the momen-
tum equation simply because the boundaries involved are short and
the boundary shear forces can be ignored. The energy equation can
be applied to these phenomena, too, but it includes a term of primary
importance which compensates for the work done by the internal
viscous shear forces. Ordinarily, the energy-loss term is evaluated
by experimental means. However, because of their peculiar inde-
pendence in these phenomena, the energy and the momentum equa-
tions can be solved simultaneously to yield analytical solutions for
tne energy loss.
UNIFORM AND GRADUALLT VARIED FLOW
CHARACTERISTICS OF UNIFORM FLOW
In most of the foregoing discussion, generality was maintained by
assuming that the flow under consideration was accelerated. The
specific problem of uniform flow, illustrated in plate 2 (m, n), will be
examined next,
By definition, the total convective acceleration at all points in a
uniform flow is zero; in other words, a,=0 and aB=0. It follows
that the velocity is constant along each streamline. Because velocity
is a vector quantity, this implies that the streamlines are straight
lines and, therefore, that the boundaries are prismatic.
Uniformity does not require that the velocities in a transverse
section be equal. However, it does require that the transverse
velocity distribution, which is a function of boundary drag and
internal shear forces, be identical at all successive sections. As a
corollary to the last requirement, uniformity implies that the rate of
loss of energy is also constant. It follows that the average velocity
and the velocity-head coefficient (a) are the same at successive sections.
Thus, from the one-dimensional energy equation (65),
Jn-h^-M^Ht, (77)
which states that, as a distinctive characteristic of uniform flow, the
loss of piezometric head (or the negative change of piezometric head)
is equal to the loss of total energy head.
 SELECTED TOPICS OF FLUID MECHANICS 47

Because the velocity pattern is uniform, the momentum coefficient

(j8) and the momentum flux are also constant. Thus, from equation
75,
S -0. (78)

Equation 78 states that, in a uniform fluid motion, the resultant of

the external forces is zero.
Most of the uniform-flow characteristics described above are illus-
trated in plate 2 (m, n) for both open and enclosed conduits. It is
apparent that the most distinctive characteristic of uniform flow in
open channels is the requirement that the axis of the channel be paral-
lel to the energy and piezometric grade lines. This is because the
piezometric line and the upper, free surface of the flow are coincident.
THE CHEZY EQUATION
The Chezy equation is the basis for many empirical equations for
uniform flow in both open and enclosed conduits. One of its deriva-
tives, the Manning equation, is widely accepted in the U. S. Geological
Survey as a practical formula for open channels, ^.s an equation of
motion, the Chezy equation must be either a form of the energy
equation (77) or the momentum equation (78). In fact, as will be
shown, it can be considered to be both.
Most frequently, perhaps, the Chezy equation is derived from the
one-dimensional momentum equation. Thus, from equation 78, the
resultant of the external forces acting on a free body of fluid in uniform
motion is zero, or,
0. (79)
In this equation s denotes the direction of the total velocity vector.
The effective forces in the s direction are Fv, the force due to boundary
pressure; Fg, the force due to weight; and Fr, the boundary shear force.
From page 14 the sum of the pressure and weight forces on an
element of volume (As &A) can be expressed as a function of the piezo-
metric gradient. Thus,

bs AA,
UO

or, over the finite volume of length L and area A,

(80)
From equations 79 and 80, considering the fact that the boundary
shear force is negative in the s direction,
(81)
48 RIVER HYDRAULICS

Dimensional analysis, substantiated by experiment, indicates that

FT can be expressed by the relationship
Fr^CrpPLV2, (82)
in which <7r is a dimensionless drag-force coefficient, a form of the
Euler number and a function of the Reynolds number and boundary
geometry, including roughness. P is the perimeter of the wetted cross
section.
From equations 81 and 82,

-Ah=Cr - LV2=Cr > (83)

in which R=A/P is the hydraulic radius of the wetted cross section.

If equation 83 is solved for velocity,

(84>
The quantity M/i in this equation is the L gradient of the piezo-
metric head. This quantity is unfortunately described as the hy-
draulic "slope" ($). For uniform flow in open channels, of course,
S is also equivalent to the free-surface elevation gradient and the total-
energy head gradient. Thus, if S is substituted for the piezometric
head gradient and <7cft is substituted for the coefficient radical in the-
right-hand member of equation 84,
(85>

This is the Chezy equation.

From equations 84 and 85,

It is apparent that <7cft, unlike Cr, is not a dimensionless coefficient. It

is, nevertheless, a drag-force coefficient and a function of the Reynolds-
number and boundary geometry.
ALTERNATE DERIVATION OF THE CHEZY EQUATION

An alternate derivation of the Chezy equation is based on the one-

dimensional energy equation. In equation 77 the head-loss term
(Hi) represents the amount of work done by the external forces in.
overcoming the viscous resistance to fluid deformation. Thus,
whether the flow be laminar or turbulent, uniform or nonuniform, the-
energy equation must contain a term which represents the flow energy
 SELECTED TOPICS OF FLUID MECHANICS 49

dissipated by the internal viscous shear forces. It is customary to

express this loss of total energy as a function of the mean kinetic
energy of flow, or, in general,

(87)

where CL is a dimensionless energy-loss coefficient which, like Cr, is a

form of the Euler number and a function of the Reynolds number and
boundary geometry. Unlike CT, of course, CL is not a drag-force
coefficient.
From equation 77, as one of the most distinctive characteristics of
uniform flow, Ah=HL. It follows that, for this condition, the inde-
pendent momentum and energy relationships expressed by equations
83 and 87 are equivalent, or,

from which,

From equations 86 and 89, therefore,

from which it is concluded that the Chezy coefficient is proportional

to either a drag-force coefficient or an energy-loss coefficient. Thus,
the Chezy equation can be regarded as a form of either the momentum
equation or the energy equation.
It is pertinept t& observe that 0 is not included in tfae.bonmdory-
drag equation (82) and a is not included in the energy-loss equation
(87). Thus, the influence of velocity distribution is apparently
ignored in both of the alternate derivations of the Chezy equation.
This procedure is justified by the observation that ft and a are func-
tions of the Reynolds number and boundary geometry variables
which also govern the value of CT in equation 82 and CL in equation 87.
Thus, the Chezy coefficient takes full account of the influence of
velocity distribution because it depends on the variables which govern
the normal velocity pattern in uniform flow.
If a velocity nonuniformity is caused by an added variable, such as
a local boundary irregularity, a and ft will depend on this local condi-
tion as well as the normal resistance characteristics of the channel.
For this condition, of -course, the flow would not be uniform, the
50 RIVER HYDRAULICS

derivations leading to the Chezy equation would be invalid, and the

question of including a and /8 in the Chezy equation is not pertinent.
GRADUALLY VARIED FLOW IN OPEN CHANNELS

Truly uniform flow seldom occurs in open channels. Thus, most

practical problems involving both artificial and natural channels are
concerned with either rapidly varied or gradually varied nonuniform
motion. In either kind of flow, the nonuniformity may represent
either an acceleration or a deceleration. Accelerated fluid motion is
characteristically stable, but decelerated motion, especially if it is
rapidly varied, is usually accompanied by separation, extreme turbu-
lence, and, therefore, a large energy loss.
The distinction between rapidly varied and gradually varied flow
is conveniently associated with the relative magnitudes of the quanti-
ties HL and FT in the energy and momentum equations, respectively.
A rapidly varied flow pattern, by definition, occurs in a relatively
short length of channel. Thus, the boundary shear force (FT) is
usually negligible, and the momentum equation, free of this trouble-
some term, provides an adequate solution for many rapidly varied
flow problems. If the flow is accelerating as well as rapidly varied,
the energy loss might also be negligible. For" this condition the energy
equation provides useful solutions. On the other hand, if the flow is
rapidly varied and decelerating, the energy loss is usually of major
importance, even when the boundary shear is negligible.
Gradually varied flow, by definition, occupies a comparatively long
length of channel. Thus, both the boundary shear force and the
energy loss are significantly related to the flow pattern. The two are
explicitly related to each other only in the limiting condition of uniform
flow, as described on page 49.
Between the extremes of rapidly varied flow arid uniform flow,
gradually varied flow in open channels is a complex function of many
variables. Neither the boundary-shear mechanism nor the,energy-
loss mechanism for this kind of flow is fully understood. In fact>
neither F* nor HL can be satisfactorily evaluated.
In view of the foregoing arguments, it would appear to be ridiculous
to propose that the Chezy equation, or any other uniform-flow equa^-
tion, could be used to describe gradually varied flow in open channels.
Nevertheless, lacking: a .better solution, it is common practice to
assume that a gradually varied nonuniform flow reach can be repre-
sented as the sum of constituent subreaches within which the flow is
essentially uniform'. It is apparent that such reasoning leads to the
conclusion that the energy-loss and boundary-shear mechanisms .are
the same in nonuniform flow as in uniform flow. This must surely be
incorrect and increasingly so «,s the nonuniformity increases.
 SELECTED TOPICS OF FLUID MECHANICS 51

In general, the velocity distribution as well as the average velocity

is different at successive cross sections in nonuniform flow. Thus, if
the one-dimensional energy equation is applied to a gradually varied
flow reach, the velocity-head coefficients at the end sections (04 and
o2) are different. In addition, if an equation like the Chezy equation
is used to evaluate the total head loss (HL) in a nonuniform reach it
must be adjusted to account for the effect of channel shape and for the
effect of using average values of the conveyance characteristics at the
ends of the reach. In other words, if uniform-flow equations are to
be used satisfactorily to describe a reach of gradually varied flow they
must be adjusted to take into consideration the effect of various kinds
and degrees of nonuniformity. These adjustments are ordinarily
included in estimates of the Chezy coefficient, and the estimates are
J>ased on analyses of field and laboratory measurements made in a
variety of open channels; Needless to say, the procedure is not
dependable.
Application of the one-dimensional momentum equation (75) to a
gradually varied flow is also complicated by changes in the velocity
distribution at successive sections. Thus, the momentum coefficients
(j8i and j82) are different, and the value of the coefficient CT in equation
82 must be adjusted to account for the variations in the velocity
pattern within the reach. Evaluation of boundary pressure forces is
also complicated by the fact that the boundary is not prismatic. There
is clearly no advantage in applying the momentum equation instead
of the energy equation to a gradually varied flow.

U. S. GOVERNMENT PRINTING OFFICE : 1958 O 483667