Flow in Channels

Macmillan Engineering Hydraulics Series

Flow in Channels
Robert H. J. Sellin, B.Sc., Ph.D.
Lecturer in hydraulics, Department of Civil Engineering, The Queen's University
of Belfast, Northern Ireland.

Macmillan
St. Martin's Press

ISBN 978-1-349-00197-2

ISBN 978-1-349-00195-8 (eBook)

DOI 10.1007/978-1-349-00195-8

R. H. J. Sellin 1969.
Softcover reprint of the hardcover 1st edition 1969 978-0-333-02822-3
First Published in 1969.
Published by
MACMILLAN AND CO LTD

Little Essex Street London wc2

and also at Bombay Calcutta and Madras
Macmillan South Africa (Publishers) Pty Ltd Johannesburg
The Macmillan Company of Australia Pty Ltd Melbourne
The Macmillan Company of Canada Ltd Toronto
StMartin's Inc New York

Preface

This book is intended mainly for students of hydraulic engineering

who have already gained a basic knowledge of fluid mechanics and
who wish to study the hydraulics of open channels as a whole. At
the same time it is hoped that the inclusion of a full range of selected
references will make the book of value to postgraduate students and
practising engineers looking for an introduction to the subject.
Although flow in channels constitutes only one part of hydraulics,
it is today too large a subject for it to be treated easily in detail in a
single small volume. Here, an introduction to alluvial or loose
boundary flow as well as one to unsteady channel flow is given,
together with the more complete treatment of uniform and nonuniform flow in fixed boundary channels.
A student using the early chapters of this book will be expected to
be familiar with the application to fluid mechanics of such basic
concepts as energy, momentum and continuity as well as with the
simple properties of fluids. An understanding of the boundary layer
concept and the existence of shear stresses in fluids will also be an
advantage. The references given at the end of each chapter have been
selected on two criteria: they either describe important original
work or thinking on the topic under discussion or else they refer to
more complete treatments of the subject than is possible in the
present volume. It was the author's intention to make the second
category as up to date as possible bearing in mind that the best in
v

vi

Preface

hydraulics appears to have an enduring validity that far exceeds

many other disciplines in science and technology.
In conclusion, it is necessary for me to express my indebtedness to
all those from whom I have learnt both personally and through
print. The foundations of the subject were most carefully laid by
Dr. E. F. Gibbs of the University of Bristol without whose interest
and encouragement I might never have pursued the subject further.
Finally I would like to thank Professor T. M. Charlton for the
opportunity to write this book and his continuing interest in it.
Department of Civil Engineering,
The Queen's University of Belfast

R. H. J. Sellin

Contents

Preface
1 Uniform flow

1.1 General characteristics of open channel flow

1.2 Definition and occurrence of uniform channel

flow

1.3 Empirical formulae

1.4 Normal depth of a stream

1.5 Velocity distribution in a cross-section

1.6 Fundamental flow relationships
1.7 Hydraulic efficiency of cross-section
1.8 Free surface flow in closed conduits
1.9 Ice covered streams

2 Gradually varied flow

Total energy and specific energy
2.2 Critical depth and critical velocity
2.3 General equation of gradually-varied flow
2.1

vii

23

Contents

viii

2.4 Varied flow surface profiles

2.5 Solution of gradually varied flow equation

3 Rapidly varied flow

49

3.1
3.2
3.3
3.4

Control sections
Transition through critical depth
Analysis of the hydraulic jump
Flow through a sluice-horizontal force on the
structure
3.5 Flow past a submerged obstacle
3.6 Changes in channel width

4 Control and measurement of open

channel flow

73

4.1 River utilisation

4.2 Channel regulation
4.3 Flow measurement

5 Flow in erodible material

5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12

Introduction
Sources and types of sediment
Modes of sediment transport
Different approaches to the problem of sediment transport
Sediment movement as bed load
Suspended sediment movement
Total sediment load
Regime behaviour of channels
River bends and meandering
Formation of alluvial plains
The design of stable canals
Sedimentation in reservoirs

98

Contents

Unsteady flow in open channels

6.1
6.2
6.3
6.4

ix

121

Outline and classification

Hydrographs and flood waves
Flood routing through open channels
Flood routing through reservoirs

References

136

List of Notations

142

Index

145

1 Uniform Flow

1.1 General characteristics of open channel flow

Fluid flow in open channels is characterised primarily by the

exposure of a free surface to atmospheric pressure. For this reason the
fluid concerned is always a liquid and nearly always water. Indeed
rivers and canals, which together make up most instances of this type
of flow, form a frequent and clearly visible part of the landscape.
Problems connected with river and canal flow represent a high
proportion of the work of the hydraulic engineer and the ease of
access resulting from a free surface does not compensate for the
increased complexity of such flows by comparison with flow in
closed conduits. Water flowing in open channels is acted upon by
all the forces that affect pipe flow with the addition of gravitational
and surface tension forces that are the direct consequence of the free
surface.
The free surface of channel flow should more properly be considered as an interface between two fluids, the upper usually* a
stationary or moving gas and the lower a moving liquid. The only
forces constraining the form of the free surface are gravity and
surface tension. These will resist any other forces tending to distort
this interface which will always constitute a flow boundary over
which the engineer has only partial control.
In common with all other instances of fluid flow, the movement of
water in an open channel can be either laminar or turbulent but the

* This interface can exist under particular circumstances between two gases
or two liquids differentiated by some sharp change in density, viscosity or some
other important physical property. Examples of this type of "channel flow"
include stratified air currents, density currents in reservoirs and marine currents
such as the Gulf Stream.

Flow in Channels

relatively large dimensions of most channels, combined with the low

viscosity of water, make laminar flow extremely unusual. Even if the
average velocity in a channel is low enough to permit laminar flow,
secondary factors such as surface wind disturbances usually produce
local velocities or currents that greatly exceed the laminar limiting
velocity in conduits of such a size. The only genuine occurrence of
laminar flow that can be thought of as open channel flow is the
drainage of rain-water from roofs and road pavements because of
its very small depth. In the case of natural river channels the
boundary roughness is normally so great that even the smooth
turbulent flow observed in pipes rarely occurs.
The forces acting on water flowing in an open channel are, in
addition to surface tension and gravity (associated particularly with
the free surface), the component of the water's weight acting in the
direction of the bed slope, shear stresses developed at the solid and
free surface boundaries, internal inertial forces due to the turbulent
nature of the flow, normal pressure at the walls and bed, more
particularly in regions of changing channel geometry, and occasionally sediment movement forces. The mutual interaction of these
forces is responsible for the complexity of open channel hydraulics
and it is only by simplification and generalisation that an understanding of its mechanics can be gained. The factors excluded in the
simplification process can then be reintroduced one at a time as
knowledge of the subject is built up.
1.2 Definition and occurrence of uniform channel flow

The most important quantities that describe the flow of water in a

channel are the depth, width and slope of the channel and the
velocity and volumetric rate of flow, or discharge, of the water. In
general all these can vary with distance and time although some are
not independent of others. A further important factor is the roughness of the walls and bed which cannot be measured directly but
only by its effect on the flow. When these variables have particular
values which are constant with time at different points in the channel,
the flow is termed steady. Conversely, in unsteady flow the velocity,
depth and discharge at, for example, a particular cross-section
change their values with the elapse of time. Some problems involving unsteady flow in open channels are considered in chapter 6;
elsewhere in this book flow is treated as steady.

Uniform Flow

The variation of these quantities with distance (spatial variation)

is more important and will be treated at greater length. When the
velocities at corresponding points in identical cross-sections along a
conduit are the same, the flow is termed uniform. In non-uniform
flow this condition is not present. The uniformity, or lack of it, can
be imposed upon a fluid flow by the form of the conduit, whether
pipe or open channel. If water is made to flow through a pipe whose
diameter changes from section to section, the resulting flow can never
be uniform whatever other conditions obtain. Similarly water that
flows through a channel of rectangular cross-section but varying
width provides an example of non-uniform flow, although in this
case the presence of an unconstrained free surface will make the
cross-section of flow indeterminate if no further details of the flow
are available. Finally, water flowing in a rectangular channel of
constant width, or any other partially full conduit of constant crosssectional shape, can be either uniform or non-uniform depending
upon factors which determine the form of the free surface.
When considering uniform channel flow in which conditions at all
cross-sections are identical, the consideration of a single cross-section
will be representative of the whole. If the behaviour of a finite
volume of water is of importance, then a short length of the channel
flow must be studied. In reality it is difficult to find lengths of
channel flow which are strictly uniform although the flow in long
canals, sewers, small concrete channels and certain stretches of
rivers is sufficiently close to uniform to be treated as such. The
results obtained for uniform channel flow can be used when considering non-uniform flow to allow for the effects of channel friction
over long lengths. Non-uniform flow over short channel lengths
(rapidly varying flow) does not normally involve significant energy
loss from channel friction and its effect is frequently omitted from
the analysis.

1.3 Empirical formulae

1.3.1 The Chezy velocity formula. To the engineer the most important property of an open channel is its capacity to carry water. If
the mean velocity for a given cross-section of flow can be found,
then the discharge through that section is known. Historically this
need led engineers to develop empirical formulae for the velocity of
flow in an open channel and the first practical formula based on

Flow in Channels

sound experimental measurements was published in 1775 by a

Frenchman, Antoine Chezyl*.
The Chezy formula can be developed as follows on two initial
assumptions. The first states that the force of resistance to the flow
per unit area of stream bed is proportional to the square of the
velocity. The second is that for flow to be uniform (in mechanical
equilibrium) the component of the gravity force acting on the water
in the direction of flow must be equal and opposite to the total
resistance to flow.
Consider the uniform flow of water between cross-section I and 2

Flo. 1.1 Derivation of Chezy formula for flow in a sloping

channel.

shown in Fig. I. I. Since the resistance to flow is proportional to

the square of the mean velocity v per unit area,
resistance force = KPLv2
in which K = a friction coefficient,
P = the length of the wetted perimeter in a cross-section

and

L = the length of channel between sections I and 2.

If the channel bed slopes at an angle () to the horizontal, then

sin () = hjL, h being the fall in level over the length L. The slope of
the channel beds is defined ass = sin (} = h/L and the water weight
component causing the flow will then be Ws where W is the weight of
the volume of water between sections I and 2. Hence, if w is the

* References indicated by small superscript numbers are listed at the end of

the book.

Uniform Flow

specific weight of the water and A is the cross-sectional area of the

channel flow,
Ws

= ALws

Equating this force with the resistance to flow gives

ALws

= KPLv2

Rearranging:
v2

w A

=--s
K p

now writing the hydraulic mean depth, m

J(i)

= p'

and putting

= C, a new constant, gives

Cv'(ms)

(1.1)

Eq. (1.1) is known today as the Chezy formula although the form
in which he originally proposed it is somewhat different2. Many
attempts have been made to evaluate Chezy's C which, it was soon
realised, is not a constant (neither is it a pure number) since it has
the dimensions LiT-1, or y(acceleration).
The best values for Chezy's Care given by a formula published in
1869 by two Swiss engineers Ganguillet and Kutter in which Cis
expressed in terms of the bed slope s, the hydraulic mean depth m
and a coefficient of roughness n. In S.I. units this formula is

23+ 000155 +-1

C=
s
n
1 + (23 + 000155) _!!__
s
v'm

(1.2)

The coefficient n in this formula is known as Kutter's n.

1.3.2 The Manning formula. In 1888 Robert Manning an Irish

engineer presenteda a formula based on the greater amount of
experimental data then available. Later simplification of this
formula effectively gave Chezy's C the following value
C = 1 ml/6 in S.l. units
n

(1.3)

Flow in Channels

6
and

.
B .. h
149 m116 m
ntis uruts
= -n

(1.3a)

in which C is recognised to be a function of the hydraulic mean

depth and also of a roughness coefficient n. This roughness coefficient
is for practical purposes identical with Kutter's nand the same values
are normally used in both Eq. (1.2) and (1.3). Substitution in the
Chezy formula gives what has now come to be known as the Manning

formula*:

= ! m213s112 in S.I. units

1.49 m213s1'2 in British units

(1.4)
(1.4a)

The Manning formula has found favour with engineers for a long
time because of its simplicity and the satisfactory results that it gives.
The Ganguillet and Kutter formula Eq. (1.2) is seldom used because
of its complexity, except in tabulated or graphical form. 4 The powers
to which the slope and hydraulic mean depth in the Manning formula
are raised were arrived at quite empirically, the exponents adopted
being convenient mean values of many experimental results. Accordingly, if Eq. (1.4) is used to evaluate n for an experimental channel,
the values will usually be found to vary with the depth of flow. In
general the value of n is not constant for a particular channel and the
factors affecting it will be outlined in the following section.

1.3.3 Evaluation of roughness coefficient n. The value to be given to

the roughness coefficient n will depend upon the "hydraulic condition" of the channel under consideration. All scales of roughness
are significant, ranging from the size of the grains comprising the
bed on the one hand to the degree of meandering developed in the
case of rivers on the other. The level of the water surface or stage
It will be seen from either Eq. 1.3 or Eq. 1.4 that the dimensions of n are
TL-l. While no rational basis for this can be given it will be seen that the
conversion of Eq. 1.3 from S.I. units to British units is consistent with these
dimensions, i.e. 328081 = 1486, there being 32808 ft in 1m; the conversion
factor 1486, when used, is normally rounded off to 149 as being more in keeping
with the expected accuracy of the values of n. Using the appropriate form of the
equation (1.3 or 1.3a) makes the value of n independent of the system of units
used. The S.I form of the Manning formula, or its derivatives, will be used
henceforth.

Uniform Flow

will also affect the value of n which, once determined, should not
therefore be assumed the same for all seasons and times.
Giving an accurate value to n for a particular channel will be
essentially a matter for fine engineering judgement. However, the
Table 1.1. Values of Roughness Coefficient n for
different channel conditions

Description of channel

Glass, plastic, machined metal

Fabricated steel channels
Planed timber, joints flush
Sawn timber, joints uneven
Concrete, trowel finished
Concrete, shuttering
Brickwork
*Excavated channels:
earth, clean
gravel
rock cut, smooth
rock cut, jagged
*Natural channels:
clean, regular section
some stones and weeds
some rocks and/or brushwood
very rocky or with standing timber
Flood plains:
short grass pasture
mature crops
brushwood
heavy timber or other obstacles

Range of values
Minimum Normal Maximum

0009
0011
0010
0011
0011
0012
0012

0010
0012
0012
0013
0013
0014
0015

0013
0017
0014
0015
0015
0017
0018

0016
0022
0025
0035

0022
0025
0035
0040

0030
0030
0040
0060

0025
0030
0050
0075

0030
0035
0070
0100

0040
0045
0080
0150

0025
0025
0035
0050

0030
0035
0050
0100

0035
0045
0070
0160

* Values given are for straight channels. For non-straight channels these
values should be increased by up to 30 per cent, depending upon degree of
meandering present.
margin of error to be expected in its evaluation by an inexperienced
engineer can be materially reduced by the use of tabulated values.
Table 1.1 gives a range of values for n set against a classification of
the channel bed type and condition.

Flow in Channels

1.3.4 Calculation of uniform channel discharge. The foregoing

empirical formulae enable the mean velocity of uniform flow to be
computed if the necessary data is available. The discharge Q
(volumetric rate of flow) of the channel is therefore the product of
the mean velocity and the cross-sectional area of flow A.
Using the Manning formula the discharge can be expressed:
(1.5)

In this equation the variable group Am1 is known as the section

factor and represents the geometrical properties of the cross-section.
It is convenient to evaluate it against depth of flow when carrying
out discharge computations especially when dealing with irregular

cross-sections. The group

~n

Am1 is called the conveyance K and

is frequently used in more advanced computational work. The

value of K for a channel necessarily involves the roughness coefficient.
A more recent and fundamental approach to the problem of
computing the velocity and discharge in open channels is based on
the theory of turbulence and will be discussed in section 1.6. The
Hydraulics Research Station* of the U.K. has published reports by
P. Ackers 5 developing this method and giving extensive design
charts and graphs.
1.3.5 Effect of errors in values of m and s. Considering the Chezy
formula v = Cv'(ms)
or
Hence substituting for C
dv

dm

from which it can be seen that an error of x% in the value of m will

result in an error of! x% in the calculated velocity. A similar analysis
gives the same result for errors in the bed slope s.

* Now the Institute of Hydraulics.

Uniform Flow
Now considering the discharge,
A~

Q = CAv(ms) =Cpl. s*
dQ
A~
-dP = -1Cs*
2
p~

from which errors in A and P will lead to proportional errors in Q

15 and 05 times as great, respectively. If these errors are all small,
their cumulative effect will be equal to their sum under the worst
circumstances (all acting in the same direction). This can be illustrated by assuming errors of p, q and r per cent in the values of s,
A and P respectively. The maximum possible error in Q will then be

(p; + 3J)%
r

Reasonable values that may be attributed to p, q and r will depend

upon the circumstances under which the measurements are made.
They may vary from 5% for field measurements to 05% for
well-conducted laboratory experiments. In most circumstances the
final error in Q will be attributable largely to the value adopted for
the roughness coefficient n.
1.4 Normal depth of a stream

In section 1.2, uniform flow was defined with reference to open

channels. In the case of a channel of rectangular cross-section the
depth of water corresponding to uniform flow for a specified discharge and bed slope will be constant at all points and is known as
the normal depth do for these conditions. The index 0 will in general
be used to designate quantities associated with uniform flow. In the
case of channels of non-rectangular cross-section the normal depth
is defined with reference to the maximum depth of the cross-section.
More generally, when uniform flow is not present in a channel, a
discharge Q can flow at any depth, the exact value being determined
by the end conditions. Fig. 1.2 shows a number of possible water
surfaces for a specified discharge of which one (indicated by a heavy

10

Flow in Channels

line) fulfils the requirements for uniform flow:

:~ = 0.

= constant,

This is at the normal depth. Under special circumstances

there may be two alternative normal depths for a limited range of

discharges (see circular conduits and
flood-plain flow, sections 1.8 and 5.10).
The ease with which the normal
depth can be calculated depends upon
the shape of channel cross-section. For
a channel in which the width is very
FIG. 1.2 Relationship of nor- great with respect to the depth, do can
mal depth in a channel to be computed directly from the chosen
other possible depths.
discharge formula. If the Chezy formula
is assumed (C independent of depth)
Q = ACy(ms) = bdoCy(doS)

since the hydraulic mean depth approaches in value the depth of

flow for a wide channel. From this
do

or, putting

af(__iE_)
b sC

\j

(1.6)

r; = q, the discharge per unit width

(1.7)

If the Manning formula Eq. (1.5) is applied to a wide channel

do=

(nq)~
si

(1.8)

For a rectangular channel that has a width b which is not large in

relation to the depth, the hydraulic mean depth is (b !d;do} The
Manning formula then gives the following relationship between Q,
s, band do
1

Q = ~(bdo) b

bdo )~
2do ys
+

(1.9)

11

Uniform Flow

which can best be solved by either trial and error or by graphical

methods.
For the more common case of a non-rectangular cross-section the
Manning formula, for example, can be
written in the form

Q 1
Q
-=-Ami or-= K,
vis
n
vis
the conveyance. In order to determine
the normal depth, compute values of K
for a suitable range of d, the maximum
depth in the cross section, and plot a
graph of conveyance against depth of
K
K=ls
flow as in Fig. 1.3. Then to obtain the
normal depth for the specified values of Fra. 1.3 Graph of the ConQ and s read off the depth corres ond- veyance K plotted ~gainst
'
Q P
depth of flow to.determme the
ing to a value of K equal to - on the normal depth m a channel.
curve.
vis

1.5 Velocity distribution in a cross-section

When a real fluid such as water flows in an inclined open channel,

equilibrium of forces exists when the flow is uniform. The boundary
shear stresses that prevent the unchecked acceleration of the water in
the downhill direction are transmitted
throughout the body of the flow by
either viscous or turbulent shear
stresses generated by a velocity
gradient over the cross-section. This
FIG. 1.4 Cross-section velocity makes the establishment of uniform
distribution in a small straight flow with a uniform velocity distrilaboratory channel (velocities bution an impossibility.
shown as a ratio of the mean).
The distribution of longitudinal
velocity in a cross-section is controlled by the position of the solid boundaries and the free surface
and to a lesser extent by the roughness of the boundaries. Fig. 1.4
shows by means of isovels the distribution of velocity in a small
straight laboratory channel. Fig. 1.5 shows, for comparison, the

12

Flow in Channels

velocity distribution in a typical river cross-section. Isovels are

lines drawn through points of equal velocity and demonstrate the
distribution of velocity in exactly the same way that contours on a
map represent the elevation of the ground surface above some
arbitrary datum.
If the channel is straight and of a constant and symmetrical crosssection for a sufficient length upstream, the velocity distribution will

FIG. 1.5 Typical velocity distribution in a natural clumnel

cross-section.

be symmetrical and the point of maximum velocity some distance

down from the free surface. In Fig. 1.6, curve (a) shows the vertical
velocity distribution on the centreline of a rectangular channel in
which the depth is equal to one half of the breadth. In the same
figure, curve (b) shows the vertical
distribution of mean velocity; each
point on this curve represents the
average velocity in a horizontal line
across the section at that level.
Among the first to consider this
characteristic of the filament of maximum velocity was Gibson 6 who carried
out experiments which showed this point
to lie a distance below the surface
v
varying from 005 to 025 of the depth.
FIG. 1.6 Vertical velocity
The
closer the sides and the rougher the
distribution curves for flow
in a channel of rectangular boundary the lower was the filament of
maximum velocity. Gibson proposed
section.
a mechanism of secondary currents
producing a double spiral circulation in the plane of the cross-section
as in Fig. 1.7. This accounted for both the depression of the maximum velocity filament and the observed movement of floating
material towards the centre of the channel surface.

Uniform Flow

13

More recent views, based on the work carried out in the 1920s by
Prandtl and Nikuradse on flow in non-circular pipes, confirms the
presence of secondary circulation but suggests that it is more complex and indeterminate than was at frst thought. The number of
"cells" of circulation present appears to vary with the proportions of
the cross-section and their relative strength is very sensitive to the
presence of bends and other transitions for a considerable distance
upstream. Flow around a bend gives rise to a single spiral as the
faster moving fluid near the free surface moves towards the outside
of the bend in an attempt to balance its radial acceleration with that
of the slower moving layers near the bed. This action leads to the
characteristic scour pattern observable at river bends which will be
treated more fully in section 5.9.
This spiral flow in bends is known as
secondary circulation of the first kind
and that in straight channel reaches
as secondary circulation of the second
kind. This latter has been explained
in various ways and a satisfactory
explanation is given by Prandtl which
FIG. 1.7 Double spiral sechas been summarised in English by ondary circulation in open
channel flow.
Rouse.7
In most analyses of open channel
flow a single value is used to express the velocity at a cross-section.
This is normally the mean velocity v, defined as the discharge divided
by the cross-section area v = QJA. This simplification leads to an
error in any calculations of kinetic energy head since the mean of the
squares of individual values is always larger than the square of the
mean value. In order to make allowance for this effect an energy
coefficient IX is normally introduced so that the kinetic energy head at
a cross-section is then IXV2j2g in which IX is found to have values
varying between 103 and 136. Low values of IX apply to wide deep
streams and higher values to small cross-sections. For complex
cross-sections, or close to constrictions such as bridge piers and weirs,
the value of IX may be much higher. For a detailed discussion of this
and other velocity coefficients see Chow. s
In spite of the foregoing, it is frequently sufficient to assume a value
of unity for the energy coefficient. This action can be justified by
consideration of the error so introduced in relation to the low order
of accuracy inherent in many other factors involved.

14

Flow in Channels

1.6 Fundamental flow relationships

The practical importance of river engineering led to the early

development of useful empirical relationships for uniform flow
(Chezy 1775 and Manning 1888) but the complexity of the problem
still prevents an acceptable and complete theoretical solution from
being developed. However, boundary layer theory developed by
Prandt1 9 and von Karman in the early years
IOd
of the 20th century has been applied suc--------Tt
y
I
cessfully
to the vertical distribution of
08d
I
---------,velocity in a wide channel and with some
1 I
success to a more three dimensional channel
I
05d
----------r~
cross-section
(see Delleur10). Fig. 1.8 shows
1
OAd
--------7
the
vertical
distribution
of velocity in a wide
I I
_0:._2!! _ _ _ __
I
channel so developed and integration of
the area under this curve leads to the result
'"-=-------that the mean velocity is equal to the
v
velocity that occurs at a distance 037d
FIG. 1.8 Vertical velocity
above the bed, where d is the total depth.
distribution in a wide
This value compares favourably with that
channel from boundary
of 04d for long accepted by river engineers
layer theory.
on grounds of experience. Thus, if the
velocity is measured with a suitable instrument (see section 4.5) at
this depth in a number of verticals, the values can be averaged to give
the mean velocity over the cross-section. Another conventionally
accepted method of determining the mean velocity in a vertical is
to average measurements taken at 02d and 08d. This process can
be carried out using the theoretical boundary layer velocity distribution and again gives reasonable results. The significance of these
findings is enhanced by the sound physical basis on which the
boundary layer theory is constructed.
Again, since this theory showed that the vertical velocity distribution in a wide channel is a function of the surface roughness, Boyerll
and others were able to relate Manning's n (Eq. 1.3) to the velocity
distribution curve. The two point velocities used above (at 02d and
08d) are satisfactory for representing the form of the velocity
distribution and the ratio of their values xis defined by the expression

vo.z

X=V08

(1.10)

Uniform Flow

15

in which vo2 is the velocity l/5th of the depth, and vo.s that 4/5th of
the depth, below the water surface. Hence x will always have a value
greater than unity. From this standpoint it can be proved that
(x-

l)d~

n=-'-----

557(x

+ 095)

(1.11)

in which for an actual channel d is the mean depth. The value of

Eq. (1.11) for enabling n to be determined objectively would be
considerable but it appears that its validity has not yet been sufficiently established.
1.7 Hydraulic efficiency of cross-section
1.7.1 The engineer faced with the problem of designing a channel
cross-section must keep two requirements in mind. His design must
fulfil the function for which it is needed and it must not cost more
than is essential to construct and maintain. The functional requirements that affect the cross-section are its capacity, which will be
dealt with below, as well as special requirements for sewers (see
section 1.8) and channels in erodible material (see section 5.11). The
cost of construction is roughly proportional to the cross-sectional
area of channel and the problem of maintenance is related to the
stability of the sides if the channel is cut through erodible material.
If the excavation cost is assumed* to be proportional to the
cross-sectional area A then for the most economical design it is
necessary to determine the shape which will require the minimum
cross-sectional area for a given discharge. The determination of
this most economical shape, called the best hydraulic section, is
simplified if the problem is transformed. Consider the crosssectional area A fixed and determine the shape that gives the maximum discharge. Using the Manning formula, Eq. (1.4), the criterion

then becomes one of maximising the term ;: and since the value of
A is fixed this condition will be satisfied when P has a minimum
value.
If there is no limitation to the shape of the cross-section then a

* This is strictly true only if the channel is designed to flow completely full
(operating under bank-full conditions). Normally, material will be excavated
above the design water level which causes an addition to the cost that is not
related to the criterion of the best hydraulic section. The most economic section
then may be slightly narrower than the best hydraulic section.

Flow in Channels

16

semicircle is the best hydraulic section. However, a semi-circular

cross-section is not normally practicable either to construct or to
maintain and various limitations are usually imposed on the crosssectional shape.
1.7.2 Rectangular section. This is frequently used for small channels
constructed in concrete. For a rectangular cross-section having the
dimensions shown in Fig. 1.9
A
A = bd, P = 2d + b hence P = 2d + d

FIG.

1.9 Hydraulic efficiency ofrectangular channel cross-section.

Assuming a fixed value for A, :

=2- ~=0

for the best

hydraulic section. This gives b = 2d which is a rectangle that can

be circumscribed around a semicircle (see Fig. 1.9).
2d2
. case m = dbd b = -d
I n t hts
= -.d In WI"de channeIs the vaIue
2
+
4
2
20d2
of m approaches d. For example when b = IOd, m = 22d = 09ld.
1.7.3 Trapezoidal section. Fig. 1.10 shows a trapezoidal crosssection whose sides slope at a gradient of 1 vertical to z horizontal.

P
A

= 2hv(z2 + I) + b' therefore b' = P- 2hv(z2 + 1)

= h(zh + b') = h(zh + P - 2hv(z2 + 1))

Assume a fixed value for P this time and differentiate A with respect
to h. Then obtain a maximum value of A by putting
gives
b'

+ 2zh =

2hv(z2

+ 1)

'!: =

0. This
(1.12)

17

Uniform Flow

and since the length of a sloping side is hy(z 2 + 1) the best hydraulic
section will occur when the
water surface width = 2 x length of sloping side
from which it can be shown that under these conditions the bed and
two sides of the channel are tangential to a semicircle radius h,
having the free surface for a diameter.
Now assume that the side slope z is a variable and that A and h
have fixed values. The wetted perimeter P can be expressed as
P =

h-

zh

2hy(z2

1)

(1.13)

.) f;,;:Y..

-:-:-:-:-:-:-:-;-:-:-:-::::::::::::-::.~y
l-b~

FIG. 1.10 Hydraulic efficiency of trapezoidal channel cross-

section.

by eliminating b' between the expressions for A and P above. Equating the differential ~; to zero will now give the minimum value of P:
dP _ -h

dz -

2h

+ v(z2 + l) . 2z -

2zh

y(z2 + 1)

_ h0

'

or 2z = y(z 2 + 1) giving z = 1/y3 and therefore () = 60. Since

the value of z was not specified in the previous analysis (treated as
a non-variable) the best hydraulic section which has () = 60 will
also have its surface width equal to twice the sloping side. It will
therefore have the form of half a regular hexagon or three adjacent
equilateral triangles of side 2hf y3. This section will have a hydraulic
mean depth m = h/2.
When a channel section has to be designed for construction in
erodible material the criterion will then be the critical tractive force
on the bed material (see section 5.7) rather than that of the best

18

Flow in Channels

hydraulic section. However, if a trapezoidal cross-section is acceptable the condition of circumscription around a semicircle will still
remain feasible although the side slope value chosen will be controlled by the sediment transport aspect of the constituent material.
1.8 Free surface flow in closed conduits

Closed conduits carrying surface drainage and sewage normally

operate with a free surface. Their hydraulic behaviour is therefore
essentially similar to that of open channels. Fig. 1.11 shows some

oooo
(a)

(b)

(d)

(c)

(e)

(f)

FIG. 1.11 Typical sewer cross-sections.

sewer cross-sections of which (a), a circular conduit, is the most

common being prefabricated cheaply in concrete or, for the smaller
sizes, in glazed earthenware. For sewers larger than 2m in diameter,
reinforced concrete sections constructed in situ are usual; the
horseshoe section (b) being commonly used. The egg-shaped
section (c), constructed in brickwork, was formerly common but the
comparatively high cost of this section by modern criteria more than
offsets its hydraulic advantage. This advantage is the high velocity
achieved at low flows, to keep solids moving, compared with a
circular cross-section of the same maximum capacity. The sections
(d), (e) and (f) are variants of the egg-shaped sewer.
The variation of velocity and discharge with depth of flow in a
circular conduit can be derived as follows:
The circular pipe flowing partly full shown in Fig. 1.12 has a

19

Uniform Flow

radius R and depth of water at the centre line d. The angle subtended by the wetted perimeter is 20. The mean velocity vis assumed
to be in accordance with the Manning formula, Eq. (1.4).
The wetted perimeter

= 2RO

_ 20 R 2 sin 20
A -R
-2

the cross-sectional
area of :flow
.

the hydraulic mean depth m =

pA = 2R -

R sin20

40

FIG. 1.12 Free surface flow in closed conduits of circular section.

From the Manning formula Q

Ail

si
= Av = -n . ---.
P

To obtain the maximum discharge put

!; =

(As)

d p2
0. Hence-- =0
dO

Differentiating with respect to 0:

AS dP
5A4 dA
p2 . dO - 2 p3 . dO

=0

Simplifying this gives

dP
dA
5P- -2A- =0
dO
dO
Substituting for A and P
IORO(R 2 - R2 cos 20) = 2 ( R20 - R 2 s~n 28 ) 2R

50(1 - cos 20) = 20 - sin 20

30 - 50 cos 20 + sin 20 = 0

20

Flow in Channels

This equation is satisfied when () = 151 12'. The maximum discharge through the section will therefore occur when the depth of
water d = 0938D, the diameter of the pipe.
Substituting for m and A in the Manning formula, Eq. (1.5) gives

Q = ~ (D2() _ D2 sin 2()) (D _ D sin 2())* s*

8()
4
8
n 4
which simplifies to the expression

Q=

D~

n 1008

(e _

sin 2()) ( 1 _ sin 2())* 81

2()
2

( 1.1 4)

When the water surface reaches the top of the pipe the depth d = D
and() = TT. Substituting these values in Eq. (1.14) gives the discharge
QF of the pipe when running full.
TTD~s*

D~s*

QF =ln1008 = 32ln

(1.1 5)

For the maximum discharge put d = 0938D,

TTDisl

Dlsl

Qmax = n938 = 298n

(1.16)

and when the pipe runs half-full

TTD~s 1

ntsl

QtF = n2016 = 641n

(1.17)

from which it can be seen that QF = 2Q1F. This result must follow
as in both cases m =

~ and the ratio of their respective flow areas

is 2: 1. The ratios of the discharge for full, maximum and half-full

conditions is then 1: 108 :05.
By considering the variation of m with depth of flow it can be
shown in a similar way that the maximum velocity occurs when
d = 081D and, as noted above, the velocity at pipe full condition is
equal to that for flow at half depth:
VF

1 D*s*

= -. - - =
n 252

VtF

(1.18)

and when d

Uniform Flow

= 081D
Vmax

21

1 D 1s1

=-.
-- = 114vp
n 221

In Fig. 1.13 values of _g_ and

Qp

(1.19)

.!:.. are plotted against the relative

VF

depth d. It can be seen from this figure that for depths in excess of
D

081D there will be two possible depths (normal depths) for each
discharge up to the maximum, while_ a:similar situation will occur for
mean velocity values above a depth of 05D. It has been found
experimentally that the value of n is not constant for a circular pipe
running part-full and that consequently
the maximum discharge (equal to 103Qp lO
approximately) occurs at d = 097 D ~
while the maximum velocity (105vp)
occurs at d = 094D.
051-----71~----K---'The above discussion applies to any
closed conduit in which the width of the
free surface reduces gradually as the roof
10
ViV;'
is approached. The positions and magnitudes of the maxima will depend upon FIG. 1.13 Flow characterthe shape of the cross-section, but, istics for a circular section
flowing part full.
theoretically at least, these will always
occur at depths below the maximum.
Cross-sections with flat horizontal roofs, such as rectangular
conduits and the U-shaped section shown in Fig. l.ll(d), will show
sharp discontinuities in their velocity and discharge relationships at
maximum depth. The instability that this introduces makes it
unsatisfactory to operate such cross-sections in this region.

1.9 Ice covered streams

In winter, rivers and canals in many northern countries will become
ice covered. This results in a corresponding increase in the wetted
perimeter P of the cross-section and hence a reduction in the hydraulic mean depth. The roughness of the underside of the ice sheet will
vary considerably but in general this surface is very rough when
recently formed in moderately flowing streams, becoming extremely

22

Flow in Channels

smooth with the passage of time as the projections are melted by the
slightly warmer river water.
A reduction in the hydraulic mean depth and an increase in the
average channel roughness will both lead to a reduction in the carrying capacity of the channel at that stage. In the resulting increase in
the river level the ice sheet will, because of its flexibility, remain in a
condition of normal flotation over most of the water surface and
only close to the banks will it be held below the hydrostatic equilibrium level by its attachment to the bank at the former level. In this
region of high curvature cracks will form which will allow the water
to rise above the ice here and this will quickly freeze to restore a
horizontal upper surface to the ice sheet.
The presence of a solid fixed boundary at the former free surface
will significantly alter the velocity distribution in the cross-section.
The point of maximum velocity will now be found close to the half
depth point and the empirical relationships between depth and
velocity frequently used to shorten river gauging will be invalid (see
section 4.3). All gauging of ice covered rivers should be based on
full measurement of vertical velocity profiles.

2 Gradually Varied Flow

2.1 Total energy and specific energy

2.1.1 General. Gradually varied flow is non-uniform flow in which

the change of depth in the channel occurs but gradually, in the

direction of flow. As a result the streamlines in any restricted
locality can be considered straight and parallel and the resulting
pressure distribution will therefore be hydrostatic. This restriction
on the flow conditions also enables the Bernoulli equation to be used
to evaluate the fluid energy. It is known that the Bernoulli equation,
as well as the hydrostatic pressure distribution, breaks down when the
streamlines become significantly curved. The criterion of streamline
curvature will be used to distinguish between gradually varied flow
and rapidly varied flow although in practice the exact point of
distinction is indeterminate.
Because gradually varied flow involves small changes of depth it is
concerned with long lengths of channel. Energy losses due to
channel boundary friction which control uniform channel flow will
therefore be the only significant means of energy dissipation. The
results obtained for uniform flow are then valid if used to describe
conditions locally in gradually varied flow.
In natural river channels changes in cross-section and profile are
frequent or even continuous and, as a result, flow in such channels is
nearly always of this type.
Consider the channel cross-section
shown in Fig. 2.1. The velocity vis assumed to be uniform over this
section, the depth of flow being d and the elevation of the bed z

2.1.2 Total energy of flow.

23

24

Flow in Channels

above an arbitrarily chosen horizontal datum. The Bernoulli

equation gives the total energy H of the streamline at the bed as
v2
H = - +d+z
(2.1)
2g
This expression will be correct for any other streamline in the crosssection since the appropriate reduction in the pressure energy head
d will be balanced by an identical increase in the potential energy
head z.
If a length of channel is considered in profile a line may be drawn
above the water surface to represent the total
energy of the flow. This is known as the total
energy line. At each section it is a distance v2f2g
above the water surface. If there were no
energy losses in the flow (ideal fluid) the total
energy line would be horizontal. For uniform
flow of a real fluid the total energy line is
parallel with the the water surface and channel
bed. In gradually varied flow these three lines,
total energy, water surface and channel bed, may
FIG. 2.1 The total
all
have different slopes with the condition that
energy of the flow
the
total energy line must always slope downin a channel.
wards in the direction of flow unless energy is
added from some outside source.
2.1.3 Specific energy of flow. The energy of a cross-section related to
the bottom of the channel is known as the specific energy or specific
head. The concept of specific energy is a very important one in the
analysis of non-uniform flow. By putting z = 0 in the total head
expression Eq. (2.1), the specific energy (see Fig. 2.2) becomes
v2

E=- + d
2g

(2.2)

In order to remove the velocity as such from the equation, put

v=

~in which A is the cross-sectional area (a function of the depth

d) and Q is the total discharge. The specific energy now becomes

E=__g:_+d
2gA 2

(2.3)

25

Gradually Varied Flow

If the cross-section is rectangular, the kinetic energy term in Eq. (2.2)

may be written in terms of ~ in which q, the discharge per unit
width, is equal to ;. Then
(2.4)
Eq. (2.2) shows that the specific energy is the sum of two parts:
the one being the kinetic energy of the water and the other its depth.
Hence, if the value ofE is fixed, the resulting flow rnay occur in various
ways. On the one hand the depth could be low
and the velocity high, while on the other the
12
depth high and the velocity low. It should be
v
g
noted that the depth can never be greater than
I
E as v2 will always be a positive quantity. The
Ej
above variation of d and v is dependent on the
f
discharge q, equal to the product vd, being . .,....... . . .. . ... . . . . .. . .. . . . . . . :.:.:.:. . . . . . . . . .:.:
the same at all sections (principle of continuity). FIG. 2.2 The specific
Since the datum for specific energy is the energy of the flow in
a channel.
channel bottom, the specific energy of a particular discharge may vary from one point to
another along the channel if the elevation of the bottom is not
constant. Changes in the specific energy from this source then will
not necessarily be accompanied by changes in the total energy of the
flow. The two quantities are, of course, related at any section as
follows

--f--!;-I .,.

H=E+z

(2.5)

Both Eqs. (2.1) and (2.2) must be modified in the case of steeply
sloping channels to take into account the weight component of the
flow normal to the channel bottom. Such modifications are only
necessary when the bottom slope exceeds l in 10 and at such a slope
the flow will become unsteady and possibly entrain air. Under these
conditions the gradually varied flow analysis will not normally be
valid.
For a channel in which the velocity distribution is strongly nonuniform the energy coefficient a. (see section 1.5) must be used to

26

Flow in Channels

modify the kinetic energy term in Eq. (2.2). The specific energy is
then
(2.6)
2.2 Critical depth and critical velocity
2.2.1 Relationship between specific energy and depth. From Eq. (2.4)
it can be seen that for a given channel section and discharge the
specific energy is a function only of the depth. The relationship

Emin.

Flo. 2.3 Variation of specific energy with depth of flow for a

constant discharge q.

between E and d may be shown graphically (Fig. 2.3) by plotting the

specific energy against the depth of :flow for a constant discharge q
per unit width. This is called the specific energy curve. It can be
seen from this figure that at any depth d the specific energy is the
sum of the kinetic energy and the potential energy shown by the two
broken lines. From Eq. (2.2) the potential energy line is straight
and passes through the origin making an angle of 45 with either
axis, as long as the slope in the channel is not excessive. The kinetic
2

energy curve represents the term 2!d2 and is asymptotic to the two
axes. The specific energy curve has two limbs; the upper for large
depths is asymptotic to the line E = d while the lower, for small
depths, is asymptotic to the axis d = 0.
For a particular value of the specific energy E there will be two
possible depths of :flow d1 and d2. These are known as alternate

27

Gradually Varied Flow

depths. The greater d2 is the alternate depth of the lesser d1 and

vice versa. This is subject to the specific energy having a value

greater than the minimum Emin for that particular discharge. At

the point in the curve where the specific energy is a minimum the
alternate depths are coincident and this is referred to as the critical
depth de. The critical depth in a channel is therefore defined as the
depth at which a specified discharge possesses the minimum specific
energy. At the critical depth the Froude number* of the flow is
unity and the mean velocity is referred to as the critical velocity ve.
At depths greater than the critical, the flow is termed sub-critical
Subcritical
flow range

FIG. 2.4

Subcritica/ and supercritica/ flow in relation to the

specific energy curve.

because the velocity v is less than the critical velocity Ve. Similarly
at depths less than de the velocity is greater than ve and the flow is
termed supercritical. Subcritical flow is sometimes referred to as
streaming or tranquil (see Bakhmeteff12) while supercritical flow has
been referred to as shooting or rapid. These alternative terms, which
were originally based on the appearance of the flow, have now
generally gone out of use in favour of sub- and super-critical. It is
most important to remember that the "critical" terms refer to the
velocity of the flow and not the depth.
Fig. 2.4 shows that for a particular channel different discharges
(expressed on a unit width basis) lead to a family of specific energy

* The Froude number, F = vfv'(gd) is a dimensionless group of variables

which has a special significance to all fluid problems involving a free surface. It
will be found to equal unity at the critical depth by differentiating Eq. (2.4) and
dE
putting dd = 0.

28

Flow in Channels

curves. The minimum point (critical depth) on each curve lies on a

straight line called the locus of critical points. This line makes an
angle with the -axis of tan-1 f. The area lying above the line is the
subcritical flow range and the area below it the supercritical flow
range. Any pair of alternate depths for a discharge q will always lie
one in each flow range.
2.2.2 Specific energy and changes in bed elevation. If, in a short
length of channel, the total energy H is assumed to remain constant,
Eq. (2.5) can be used to relate changes in the elevation of the channel

dj

dl -----------------.-',

d3 -----------------,' '

dz

de

----------+-- :
- - - -

,....~..

,~'

: : Crilicol
- - - , -. - - ,-

d~-#{~: :_:_)~-/= ::

:::ti-- :
'

Fro. 2.5 The specific energy curve used to determine changes in

depth produced by changes in channel bed elevation.

bed to changes in the specific energy of the flow. Fig. 2.5 shows a
longitudinal section through a short length of channel in which two
changes of bed elevation occur. The accompanying specific energy
curve is constructed for a discharge per unit width q which the
channel carries. Since the total energy is the same at all three sections
Eq. (2.5) can be developed as follows
(2.7)

hence the change in specific energy between sections (1) and (2) is
(E2 - El)

= (Zl -

Z2)

(2.8)

and between sections (2) and (3)

(2.9)

Gradually Varied Flow

29

The horizontal axis (d = 0) of the specific energy graph is drawn

at the bed level in section (1) and so the values on the d-axis correspond with actual depths at this section only. In order to transfer
depths directly at the other sections it would be necessary to shift
the origin of the axes to coincide with the changed level of the bed.
The vertical line on the graph corresponding to E = E1 = H - z1
shows that the flow can occur at section (1) at either depth d1 or d~,
the former being subcritical and the latter supercritical. The
corresponding depths d2 and d; at section (2) can be obtained from
the intercepts with a second vertical line E = E2. Since in this
example z2 is greater than z1, Eq. (2.8) gives a negative value to the
difference term (E2 - E1) enabling the point E2 to be located on the
E-axis. The alternate depths at section (3) can be found in a similar
manner.
From Fig. 2.5 it can be seen that when subcritical (deep) flow
passes on to a raised portion of the channel bed the water surface
falls but that when the initial flow is supercritical {shallow) the
water surface rises. In either case the situation is reversed when the
channel bed drops. Under all conditions the vertical distance
between the water surface and the total energy line represents the
kinetic energy head at that section. In this figure the flow conditions
at the two transitions are only represented for the sake of continuity
and no particular significance should be placed on the smooth water
surface curves in regions of apparently strongly curvilinear flow. In
reality the horizontal scale could be expanded sufficiently to allow
the design of suitable channel transitions without seriously invalidating the intial assumption of insignificant channel friction
energy losses.
2.2.3 Determination of critical depth and critical velocity. The
critical depth is defined as that at which the specific energy of a flow
is a minimum. Therefore differentiation of the appropriate form of
the specific energy equation will evaluate de.
Consider the general case of a channel of irregular cross-section
(Fig. 2.6) carrying a discharge Q at a maximum depth d and crosssectional area A. Eq. (2.3) gives the specific energy as

30

Flow in Channels

and differentiating this expression with respect to the depth:

dE
dd

Q2

2g

2) dA

- A3

dd

+l

dE
When dd = 0, d = de and the other variables will have values

corresponding to critical flow. Hence

Q2

dA

--=1
g . A~ dd

FIG.

2.6 Relationship between change in depth and change in

cross-sectional area.

From Fig. 2.6

dA

dd

= b.

~A

= b . ~d

therefore

~A

~d

=b

and at the limit

The general equation for critical flow then becomes

Q2bc
gA~

The critical velocity is obtained by putting

(2.10)

Vc = ~ in the above

giving

Vc = J(g~c)

(2.11)

In an irregular cross-section the mean depth can be defined as :

dm = Afb, therefore for critical flow the mean depth will have a
value dmc and

Vc =

v'(gdmc)

(2.12)

Gradually Varied Flow

31

In a cross-section of constant depth (rectangular or approximately

so) the discharge can be expressed on a unit width basis and Eq. (2.4)
developed as follows

dE
q2
-=--+1
dd
gd3

Putting

~=

0 to give critical flow then leads to the relationship

de=
Ve

and
hence

(2.13)

J(~)

q
= {l(gq)
de

=-

~
v~
de=- =2.- but Ee
g
2g

therefore

(2.14)

y(gde)

(2.15)

~ + de = 3 -~
= -2g
2g

de= iEe

(2.16)

At critical flow the depth is equal to twice the kinetic energy head and
therefore two-thirds of the critical specific energy Ee.
2.2.4 Maximum discharge under fixed specific energy conditions. So
far it has been assumed that the discharge in a channel has been
fixed and the resulting depth-specific energy relationship has
been explored. Consider now that the specific energy is fixed and
the variable discharge q is therefore a function of the depth of flow.
It follows as a corollary to the definition of critical depth that under
these conditions the discharge must be a maximum at the critical
depth. This can be verified as follows:

Rearranging Eq. (2.4) gives

q = y'[2gd2(E- d)]

!~ =
putting~ =

v(2g) [ v(E- d) - 2

0 to obtain the conditions for maximum discharge gives

(E-d)=
or

-v<:_d)]

(2.17)

d= fE= de

32

Flow in Channels

which shows that the maximum dischargeqmax =!

at the critical depth de = fE.

vJ:) E~

occurs

The discharge-depth relationship is shown by the curve in Fig. 2. 7

and the limits of this curve may be more easily understood from
the following. When d = E the velocity of flow v (equal to
v[2g(E- d)] from Eq. (2.2)) will be zero. The discharge at d = E
will then be zero. When at the other limit d = 0 there will clearly be
no discharge whatever value the velocity has. Between these two
limits the discharge will be real and
variable and will therefore possess a
maximum value at some point as shown
above.
2.2.5 Critical channel slope. It has been

shown that a channel of known crosssection carrying a specified discharge will

have a determinable critical depth. The
critical depth is independent of the
---q- channel slope and therefore of the normal
depth of flow in that channel. A value
Fro. 2.7 Depth-discharge of the channel slope s may therefore be
curve under fixed specific
chosen at which the critical and normal
energy conditions.
depths coincide. This is known as the
critical slope s0 If a channel has a slope s less than Sc, the flow at
normal depth will be subcritical and the channel slope is referred to
as mild. Conversely, when s is greater than sc the flow will be supercritical at normal depth and the slope steep.
The critical slope may be determined by equating the normal
depth to the critical depth. Since the resulting flow at critical
depth will be uniform, the bed and the total energy line will have the
same slope sc = ic. Therefore the Manning formula may be used

~ m~i!. Since the

n
v(gdc). Then, by eliminating v

to express the mean velocity in the form v =

flow is also critical, the velocity v
between these two expressions

gdcn 2
Sc = - . mc

(2.18)

33

Gradually Varied Flow

If the channel is wide in relation to its critical depth me ~ de and

gn2

Sc

= df

(2.19)

It should be noted that the use of the Chezy formula v = Cy'(mi) is

unsatisfactory in this instance, as the power ! given to the hydraulic
mean depth in this relationship would result in the critical slope for
a wide channel being independent of the discharge and depending
only on the value of the Chezy C, which is not the case. However,
this consequence of using the Chezy formula suggests that the value
given to sc by the Manning formula should be used with caution as
it is so strongly dependent on the empirical factor n and so little on
the critical depth de.

2.3 General equation of gradually-varied flow

Consider the total energy head at a point in a channel carrying
gradually-varied non-uniform flow:
v2

H=z+d+-

2g

(2.1)

..

The channel shown in profile in Fig. 2.8 carries a discharge Q and

at the point under consideration has a cross-sectional area A and
surface breadth b. The limitations implied
Tolol energyHorizon1ol
in the term gradually-varied are explained
~
in section 2.1.1 and in addition the slope
vZ/2g lin e
surface
Wo1er
s of the channel bed is assumed small so
Flow
that the depth d may be considered as either
the vertical or normal distance without disd
tinction (assume cos() = 1). In the analysis
that follows, the velocity distribution across -~.n_onnel berl J (l Horizon1ol
a section will be assumed uniform but sub- -~1
sequently the energy coefficient IX (see section --- ~-- z--- ~~
Horizontal do1um
2.1.3) will be introduced into the appropriate
FIG. 2.8 Graduallyform.
varied flow profile.
The total energy head H and the bed
elevation z are measured positive above a
horizontal datum. By convention, the slope i of the total energy
line is considered positive, sloping downwards in the direction of

-- l

. , . . .. .

flow, and consequently i = sin() = -

c;:: where xis measured along

Flow in Channels

34

the channel bed in the downstream direction. A similar sign convention applies to the bed slope s = sin () = entiating Eq. (2.1) with respect to x:
dH = dz
dx
dx

+ dd + ~
dx

:!:

Now, differ-

(v2)

dx 2g

(2.20)

and, substituting the slope of the total energy line and channel bed,

(v2)

s - i = dd + dd ~
dx dx. dd 2g

(2.21)

from which the rate of change of channel depth can be isolated on

the left hand

dd

s- i

dx = 1 +dv2)
-d 2g
The kinetic energy head can be differentiated by substituting

(v2) = ddd ( 2gA2

Q2) = Q22g (dA)-2
Q2 dA
dd
= - gAa dd = -

d
dd 2g

since from Fig. 2.6

'!.; = b. Eq.

(2.22)

~=

v:

Q2b
gAa

(2.22) now becomes

dd
s- i
dx = 1 _ Q2b

(2.23)

gAa

This equation is known as the general equation of gradually varied

flow. In a channel of rectangular cross-section A= b. d and
Q = q . b giving:
dd
s- i
dx= 1
gd3

_!f:...

(2.24)

The basic assumption for this analysis, discussed in section 2.1.1,

allows the results obtained for uniform flow to be applied locally
to gradually-varied flow. If this is the case a suitable empirical

35

Gradually Varied Flow

formula (section 1.3) can be used to evaluate i the slope of the

total energy line. If the Chezy formula is used

and

v2

dd s- C2,;;
dx =
q2

(2.25)

l--

gd3

in which m is the hydraulic mean depth of the cross-section and C

the Chezy constant. If the Manning formula is used

Eliminating q from Eq. (2.25) gives

dd
dx

v2
s--C2m

v2
gd

(2.26)

1--

a form of the general equation useful in computational solutions.

2.4 Varied flow surface profiles
2.4.1 Special values of varied flow equation. Without considering any
particular problem it is possible to define certain types of surface
profile and significant points by inspection of the varied flow
equation, Eq. (2.23).

dd

(1)- = s

dx

The limiting case when the velocity is zero. There

will be no flow in the channel and the water surface
will be horizontal. Whilst its occurrence does not
constitute open channel flow it forms a limit which
the flow often approaches asymptotically. It can
also occur as a transitory value at lower depths.

36

Flow in Channels
dd

(2)- = +ve

dx

Depth increasing downstream. In channels of mild

slope in which d > do the resulting profile is
Bockwoler curve

-----~~~~~~~~
-.;.;.;.;.:.;.;.;.;.;.;.;.;.:: :.::::;.;,~.:-.;:.;:;;:.;~;.::.;.:.;.; .;.;.;.;.; .;.

;.;.;.;.;.;.

FIG.

';.;.;.;.;.;.;.:.;.;.;.;,,;.;.;.:.;.;

2.9 Backwater curve flow profile.

called a backwater curve (Fig. 2.9). It is produced

by obstructions, such as weirs and bridge piers,
changes of bed slope (increasingly mild) and channels discharging into reservoirs under certain
conditions.
(3) dd = 0

Constant depth, uniform flow. This condition is

obtained when s = i and this special case of the
general equation is discussed in Chapter 1.

dx

FIG. 2.10 Drawdown curve at a free overfa/1.

dd

(4)- = -ve

dx

dd
(5)- =

dx

oo

Depth decreasing downstream. When the channel

slope is mild the resulting profile is called a drawdown curve. It occurs at a free overfall as in Fig.
2.10 or where the channel is approaching a point
at which the slope increases.
Water surface vertical. The equation yields this
value when Q2bfgA 3 = 1 which from Eq. (2.10)
occurs at the critical depth. Hence, except when
the channel slope is critical the water surface must
be vertical at the critical depth. This implies that
when the water surface has to pass from above the

37
Gradually Varied Flow
critical depth to below, or vice versa, it does so
rapidly. This result is borne out in reality, thus
invalidating the basic assumption of graduallyvaried flow. Although the detailed shapes of the
surface profiles in this region close to critical
depth are not therefore predicted accurately by the
varied flow equation, their forms approximate fairly
closely to the theoretical profiles in many cases.
When the surface profile has to pass from d < de to
d > de a hydraulic jump is formed. This represents
the extreme case of the breakdown of the varied
flow equation as there is a sudden and turbulent
increase of depth
N.O.L
accompanied by
concentrated -~~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . ..
energy losses. The
hydraulic jump
Fro. 2.11 Hydraulic jump.
in
represented
Fig. 2.11 will be considered in detail in section 3.3.
A decrease of depth through the critical depth is
known as a hydraulic drop. Such a hydraulic drop
terminates the drawdown curve shown in Fig. 2.10
and although the water surface is not vertical at
that point it is very steeply inclined relative to
normal channel slopes and the drop occupies a
relatively short length of channel. The energy loss
in a hydraulic drop is small and of the same order
as that in uniform supercritical flow.
2.4.2 Classification of surface profiles. The first comprehensive
classification of surface profiles in English was by Bakhmeteff13
This classification is based on the slope of the channel and the
"zone" in which the profile lies. In addition to the slope categories
that have been described already, namely: Mild (M), Critical (C)
and Steep (S), it is necessary to introduce: Horizontal (H) and
Adverse (A).
For any particular channel and discharge, the possible range of
depths of flow 0 < d < oo are divided into three zones by the critical
depth line (C.D.L.) and the normal depth line (N.D.L.). Which
of the two lines is the higher will depend upon the particular

38

Flow in Channels

circumstances. Bearing this in mind, the three zones are designated

as follows:
Zone: (1) Depth greater than both do and de.
(2) Depth lying between do and de.
(3) Depth less than both do and de.
The possibility of three zones in each of five slope categories suggests
15 different profile types. Several of these do not exist in reality for
reasons that will appear in the following descriptive notes to Fig. 2.12.
This figure shows the form of the surface profiles arranged in groups
corresponding to the slope categories. They cannot of course exist
in superposition, as shown on the left hand side of this figure for
convenience. The right hand side shows possible examples of channel flow in which all the profiles are illustrated at least once.
(H) Horizontal bed (s = 0). Uniform flow cannot exist here
(do- oo) and so there can be no H1 profile. In the example
showing flow under a sluice gate on to a short horizontal
apron the two profiles H2 and Ha are joined by a jump.
H2 constitutes a drawdown curve and the profile passes
through the critical depth again at the concluding hydraulic drop.
(M) Mild slope (s < Se). M1 is called the backwater curve, a term
extended by some authors to include all profiles in which
the depth is increasing. This is perhaps the most commonly occurring profile being formed in channels at
changes of section and at junctions with tributaries and
lakes. It is asymptotic both to the normal depth line
upstream and the horizontal downstream if it is not
terminated earlier. The length of the M1 profile can
frequently be measured in kilometers and a practical
limit to its extent is normally assumed when (d- do) =
002 m. Fig. 2.12 shows two examples of the M1 profile.
The M2 profile will occur at the junction with a reservoir
if the water level there is below the level of the normal
depth in the channel exit. It may or may not pass through
the critical depth at the junction, depemling upon the
reservoir level. The Ma profile occurs below steep spillways and sluices and normally terminates in a jump.
Both M2 and Ma profiles are short by comparison with
MI.

39

Gradually Varied Flow

(C) Critical slope (s = sc). C1 and Ca are close to being straight

horizontal lines. Zone 2 is reduced to a line: d

Classificat ion

= do = de

Examples

~~~~ental =~=,...---J7ll...2H~Ho~:i'='zo"'n'-'-'to,!!.l_
---. _-;.p~t... ................... ,_ ..'~-- -

~-

:.:...:::::-.......: ..;.;.-..-.;.;.-.-.;.:-:-::::::::::::::X :-:-.-: ..... .

s=O

Mild
slope

Mt

~""'-~-~-

-'>l

M
_,

~-

u:ll..

--

Mz--......- N:o1.-.--- ---- .. --- .... ; . .. --.

~-

we1r

_ J!

~<o"l..-..

f:

--~-----

, Lake

::-:-:-:-:-:-:-:-:-:-:-:-: -.: - -~... ;-~-~----.-:::-:.-:

. . ..-:. ...:-"71 ~~
=1- - - t!-O...L.- - ' _........

M! . . . . . . ""

Critical
slope

v
~C c--'Ct
2 .D.L.-N'O:r - - - - -

- ---c;-----

~
C'o.L:-iV.o.t.Lake

;:--

-~
- -~:!J~( . O.L.' ~
----,S.C!.i..~./!!.Q.L.

:-ii::::::::::::;-:;;;:~c.-::::.-:.-:: -:. .....;.-.;;.;.;.;.;0:0. ~~

s~s'"~

Steep
slope

------

NDL

s-'- .. " ...

-sc

'S7

,:;::::==t,== ~
- s_ . - -Lake--

-'tOL -..-~----_,
________c_o.~,

--

--~!~~ ------"-=

~~~-NOt_
~uoo

'"

-~- -- C.Ol

-~~
-~
-__-~"---'
....... Sluice :._-.:_

Adverse
slope

==-==='~
1:.--- -~

~~
...\~

FIG.

s.::: 0

..

2.12 Classification of gradually-varied flow surface profiles.

and therefore theoretically C2 corresponds to uniform

flow at the normal depth. At the critical slope the water
surface oscillates about do for uniform flow, giving rise to

40

Flow in Channels
a train of stationary surface waves (Froude number equal
to unity).

(S) Steep slope (s > sc). These flow conditions are found on spillways and chutes. The profiles are fairly short and all
asymptotic downstream (velocity of flow exceeds gravity
wave propagation velocity). The S1 profile is normally
found downstream from a jump and with it forms the
counterpart to the M1 profile in mild channels. The Sz
profile is the continuation of a hydraulic drop formed
when the channel steepens or when the flow leaves a
reservoir. The Sa profile is formed below a sluice gate or
below the junction with a steeper channel.
(A) Adverse slope (s = -ve). Uniform flow cannot occur in the
direction indicated (do has a negative value) and so there
can be no A1 profile. The Az and Aa profiles frequently
occur together, joined by a jump on the apron below a
spillway as the apron may be given a negative slope to
ensure that the jump forms on it. The Az profile will also
be found in front of the Sz profile where a steep channel
leads out of a reservoir; a vertical reservoir side being
regarded as ar. extreme example of an adverse slope
channel. The A profiles are all extremely short.
These 12 profiles (excluding Cz) cover every possible form of
gradually varied flow. The ability to classify correctly the flow in
a particular problem is an essential prerequisite to the computation
of the profile details.
2.5 Solution of gradually varied flow equation

In order to enable the engineer to calculate the detailed water surface

profile for gradually-varied flow, the general equation (Eq. 2.23) must
be solvable for a particular case. Section 2.4 outlines the various
types of surface profiles that can occur of which the M and S types
are the most frequent and important.
2.5.1 Direct integration. Under certain circumstances the varied
flow equation is a straightforward differential equation that may be

Gradually Varied Flow

41

integrated. Thus, from Eq. (2.23), substituting fori from the Manning formula:

X=

J(]~!)
s

dd

(2.27)

A2m~

which, since A, b and m are functions of d, is of the form

X=

(2.28)

jf(d) dd

If this equation could be solved it would give x directly in terms of

d so that, conversely, the depth of flow at any point in the channel
would be determined. The function f(d) is normally too complex for
any but numerical integration and many such solutions have been
published. The most useful of these solutions is due to Bresse (1860)
in which he assumed a wide rectangular cross-section, rearranged
the equation into a dimensionless form and prepared tables of the
resulting Bresse function (for details and tabulated values of the
function see Morris14). A comprehensive list of direct integration
solutions is given by Chow.l5

2.5.2 Graphical integration method. From the differential form of

Eq. (2.28):

:~ =

f(d) it is possible

dx
to compute values of dd corre-

~~I

==-f(d)

sponding to a chosen range of

values of d. These can then be
dx
plotted to form a curve of dd
(reciprocal of water surface slope)
against depth. Fig. 2.13 shows the
form of such a curve for the M1
profile. From Eq. (2.28) the distance X1-2 along the channel between depths d1 and d2 is
Xl-2

FIG. 2.13 Graphical solution of

gradually-varied flow equation.

f (~)

dd

(2.29)

Flow in Channels

42

the value of which is given by the area under the curve between the
ordinates d1 and d2. This area can be measured by planimetering,
use of Simpson's rule, counting squares or any other convenient
method which gives the required accuracy. The value of the graphical
integration method depends upon the relative ease with which the
function ( ~;) can be computed. In general this method is more
tedious and less satisfactory than the numerical step methods
described below.
2.5.3 Numerical step methods. In all step methods the channel is
divided up into short reaches, each reach being bounded by crosssections of known hydraulic properties. Starting from the end
where the conditions are known (control section) the computation is
carried from one cross-section to the next throughout the length of
channel under consideration. These reaches
bz need to be short enough to reduce, within
-+- t...,.=,=-.---.----"-._- -',.a"O""x permissible limits, the error in approximating
0 0/ energy line .
the actual water surface slope through the
z
reach, to the slope corresponding to the
average of the hydraulic properties within the
reach. The control section from which the
computation commences is fully explained in
section 3.1 but in the present context it is
sufficient to know that the computation
commences at the downstream limit of the
FrG. 2.14 The direct surface profile when the bed slope is mild
step method.
and at the upstream limit when the bed
slope is steep.
Of the great number of step methods available, the direct step
method described below is perhaps the most simple, although
applicable only to prismatic channels. Fig. 2.14 illustrates a short
channel reach of length ~x. The water surface is omitted from this
figure and the vertical distance at any point between the channel bed
and the total energy line (both sloping) is equal to the specific energy
of the flow:

v2

E=d+-

2g

assuming uniform velocity distribution within a cross-section. The

Gradually Varied Flow

43

vertical distances at either end of this reach between the horizontal

lines a1a2 and b1b2 can be equated:
E1

+ s6.x =

+ i6.x

(2.30)

in which s and i are the average slopes over this reach of the channel
bed and total energy line respectively. The slope s is considered
small enough for the horizontal distance a1a2 to be equated with 6.x
although x is more correctly measured along the sloping channel bed.
Rearranging Eq. (2.30) gives an expression for the length 6.x between
the sections 1 and 2.
6.x = 2 - E1 = ~
(s- i)
(s- i)

(2.31)

in which 6. is the change of specific energy over the length 6.x. In

the case of a prismatic channel the bed slope s will remain constant
throughout but the total energy line will change its slope from one
point to another so long as the flow is non-uniform. The average
value of i can be computed using the Manning formula based on
average channel properties over the reach concerned.
.

Q2n2

l=--.

A 2m"

(2.32)

The detailed application of the direct step method based on Eq. (2.31)
can be shown best by considering an example:
Example of Backwater Profile Computation (in British units). A
straight prismatic irrigation channel has a trapezoidal cross-section
(Fig. 2.15) with bottom width b' =20ft and equal side slopes of
2 horizontal to 1 vertical. The surface of the channel is smooth
concrete having a Manning's n value of 0014 and the channel bed
slope s = 00002. At a certain point regulation works produce a
depth of 6 ft in the channel when the discharge is 350 cusec. Determine the resulting water surface profile and find its total length if the
limit is defined when the depth equals (do + 005) ft. It is convenient
to proceed as follows:

(I) Determine do and de.

(2) Classify the profile type.
(3) Divide change of depth between d1 (6ft) and (do
convenient steps.

+ 005) into

Flow in Channels

44

(4) Tabulate hydraulic properties of channel at each section.

(5) Compute mean m and i values between adjacent sections.
(6) Compute Llx values between adjacent sections.
(7) Sum Llx values to obtain distance of each section from regulation works and total length of profile.

Steps (4) to (7) inclusive can conveniently form one table. Determination of normal and critical depths will require trial-and-error
computations which can also be arranged conveniently in tabular
form. Before this procedure is started it is necessary to express the
hydraulic properties of the cross-section (Fig. 2.15) in terms of the
single variable d.

....._I

1~

f
b

Surface width, b
Area,
A
"'Wetted

l-b~l

FIG. 2.15 Channel cross-section.

perimeter,
Hydraulic
mean depth,

= 20 + 4d
= d(20 + 2d)
= 20 + 2y'5d
= 20 + 447d
A

m=p
d(20

+ 2d)

m=---(20 + 447d)
(1) At the critical depth

Q 2~ =

gA

Q2(20
g[d(20

1 Eq. (2.10). Hence:

+ 4d)
+ 2d)]3 = f(d) = 1

In order to obtain an approximate value of de assume flow in

a rectangular cross-section 25 ft wide:
350

q = 25 = 140 cusec/ft

de=~(~)= ~G~.~) =183 ft

Now for trapezoidal cross-section compute f(d) for d
19, 20, etc. (Table 2.1)

18,

45

Gradually Varied Flow

TABLE

2.1

2d

4d

1-8
19
20

36
3-8
40

72
76
80

197 3-94 7-88

198 3-96 792

f(d)

(20 + 4d) (20 + 2d) d(20 + 2d) [d(20 +2d)]3

Q2b

gAa

272
276
280

236
23-8
24.0

425
452
480

768. 104
923. 104
1106. 104

135
1-14
096

27-88
2792

23-94
23-96

472
474

1052. 104
1065. 104

1007
0996

from which by interpolation de = 198 ft.

At the normal depth the Manning formula gives
Q

Qn
= Ao . -1"n49 most f rom wh.tch - t = Aom{J = f' (d)
149s

Again, to find an approximate value for do assume a rectangular channel of width 25ft in which q = 3; 5 = 14 cusecjft.

-qn
- - d '
149s1 - 0

Now

do

qn )~ ( 14 x 0014 )~
"
= ( 149s1 = 149v(00002) = 9"42" =

3"85 ft.

Therefore, for the trapezoidal cross-section, compute f'(d) for

d = 38, 39, 40, etc. (Table 2.2)

__2!:_ _ 350 X 0014 _ 2 .

- 149st - 149y'(00002) - 35 5

f'(d) _

Hence: Am1 = 2355 at normal depth.

Flow in Channels

46

TABLE

2.2

38
39
40
405
410

2d 20

76
78
80
81
82

+ 2d

d(20

276
278
280
281
282

from which do

+ 2d)

441d 20

+ 441d

170
174
179
181
183

370
374
379
381
383

1049
1084
1120
113-8
1156

f'(d)

m2

283
290
296
299
302

801
841
876
894
912

mi Ami

200
203
206
208
209

210
220
231
236
241

= 405 ft.

(2) Since do > de the channel slope is mild and as the profile
depths will lie between 6 ft and do (i.e. all greater than do)
the profile will be of the M1 type.
(3) Profile depth limits: 600-410 ft. Take depth increments of
02 ft giving dvalues of: 60, 58, 56, 54, 52, 50, 48, 46, 44
42, 41 ft.
(4)-(7) Tabulate as in Table 2.3 for

~x = (S~E.)
from Eq. (2.31).
-l

From right hand column, total length of water surface profile

is 22,600 ft.
It should be remembered that when the depth of flow
is close to the normal depth in the channel, very large
changes occur in the ~x values for small changes in the
depth or in any of the depth dependent variables such as
A and m.
For channels of irregular (non-prismatic) form a modified numerical method called the standard step method is more suitable. Reference is made to the elevation of the water surface at each section

300

29-6

292

288

284

282

50 100

48 9-6

46 92

44 88

42 8-4

41

221

Q2n 2

+ 2d)

ll6

ll9

127

134

142

150

158

166

175

183

192

d(20

v2

d
2g

v2

+ 2g
-

= 10'9

302 0142 4242

294 0134 4334

276 0118 4518

261 0106 4706

2-47 0095 4895

233 0085 5085

221 0076 5276

211 0069 5-469

200 0062 5-662

191 0057 5857

183

188

197

206

215

224

232

241

250

259

268

4-47d 20

383

388

397

406

415

424

432

44-1

450

459

468

+ 447d

303

307

3-20

330

3-42

354

366

376

389

399

410

A
p

m~

,4'2

Q2n2*

= 00002)

s-i

157
179

305 442 118 13800

336 503 138 19000 0000ll4

314 460 123 15100

97

3-48 528 146 21300

132

0000086

83

3-60 552 154 23700

325 4-81 131 17200

103

73

3-71 574 162 26200

21

43

68

117

127

138

62

383 599 171 29200

145

55

394 622 179 32000

0000152

221,f2mt (s

405 645 188 35300 0000048

2.3. Direct step method for computing backwater profile

182 0051 6051

Q
A

(350 X 0014) 2 241

221
= 221

304

52 104

308

54 108

312

56 112

82

316

58 ll6

+ 2d

320

2d 20

60 120

TABLE

0
0194 1280 - 1280
0195 1340 - 2620
0193 1400 - 4020
0193 1520 - 5540
Ol9l 1630 - 7170
0190 1840 - 9010
0189 2200 - 11210
0188 2770 - 13980
0184 4280 - 18260
0092 4380 - 22640

~E

~E (s- i) L~X

~X

48

Flow in Channels

rather than the depth and, correspondingly, the change in total energy
is considered rather than specific energy. A full description of this
and other numerical step methods is to be found in Woodward and
Posey16 while King and Brater17 give results of more general application to non-uniform channel flow in table form.

3 Rapidly Varied Flow

3.1 Control sections

In non-uniform channel flow the longitudinal position of the
resulting surface profile is fixed by the hydraulic properties of a
particular section in the channel called a control section. In some
cases the position is determined by two control sections, one upstream and the other down. The determinate nature of control
sections enables the depth of flow there to be found as well as the
surface proille to be positioned. In order for the depth to be determined, either a simple relationship must exist there between the depth
and the discharge or else the depth must be imposed by external
factors and be independent of the flow.
Control sections can be conveniently classified into the following
four categories:
(1) Control structures. Weirs, floodgates, sluices and other channel
regulation works will cause non-uniform flow to exist both upstream
and downstream and will also have a simple depth-discharge
relationship. For example, the velocity through a sluice is proportional to the square root of the upstream head above it.
(2) Flow through critical depth. This forms a control section when
the depth is decreasing only, as for example, where flow enters a steep
channel from a reservoir or the slope of a channel changes from
mild to steep (see section 3.2.1). Under these circumstances the
surface profile will pass through the critical depth at or near the
channel transition and the relationship d~ = q2fg (Eq. 2.14) will
49

50

Flow in Channels

hold in a wide channel. A free overfall (Fig. 2.10) also comes in this
category.
(3) Change of bed slope. When the channel slope changes, but
critical flow is not produced, a control section is nevertheless normally formed. If the slopes before and after the change are both mild

(b) Critical depth and change of bed slope

'<7
_.,Control
~eciJOn

A2

M1

,r-==-=-- ----- c:o~c:- - - - -- --"1~~~ -- I

- ----.::.-

Control
section

Reservo~,. .....

'//::0..-.:..::.:.-.::..-:::::::::::::~:::~:;t::::::.:::::::::.-.::::..-..-:.-:..-.-.-.l

FIG.

.X.

Reservoir

(c) Flow between reservoirs

3.1 Examples of control sections.

then the control section depth is the downstream normal depth.

When the upstream slope is steep, the upstream normal depth is the
control section depth, whatever slope exists below the transition. A
transition from mild to steep produces critical depth and comes in
(2) above. If normal depth is not existing at the slope transition due
to the presence of another varied flow profile, then the transition
does not constitute a control section and conditions there will be
dependent on the relevant adjacent control section.
(4) Flow into reservoirs. The known water level in such reservoirs
will determine the depth of flow at the channel termination. This
point is therefore a control section. If the reservoir water level is

Rapidly Varied Flow

51

below the critical depth at the junction then the control section will
be at the point where the depth is critical, as in (2) above, rather than
at the lower level.
Fig. 3.1 gives examples of some of the possible control sections in
relation to gradually varied flow. Fig. 3.1(a) deals with category (1)
above and it will be noted that two apparently so dissimilar structures
as a sluice and a weir lead to identical surface profile types both in
the upstream and the downstream directions. Had the slope in this
channel been mild rather than steep the profiles would have been
M 1 and Ma instead of S1 and Sa. A hydraulic jump would occur in
either case.
Fig. 3.l(b) shows two examples of control sections occurring at
the critical depth and one where the channel slope changes from
steep to mild. The depth will normally pass from below to above
the critical by means of a hydraulic jump (section 3.3) which cannot
constitute a control section.
In Fig. 3.1(c) the upstream reservoir level is above the normal
depth at the channel inlet by an amount corresponding to the
kinetic energy head of the channel flow. The A2 profile therefore
terminates prematurely at the point where d = do, at the channel
entrance (control section). This should be compared with the
occurrence of the A2 profile in Fig. 3.1(b) in which it is fully developed
up to the critical depth. At the downstream end of the channel the
M1 profile only becomes truly horizontal at the limit d-+ oo so that
if there is a sudden fall in bed elevation at the junction of the reservoir
there will also be a very small change in surface slope, probably
imperceptible. The kinetic energy of the channel flow can safely be
assumed lost at the reservoir entrance. If the channel shown in
Fig. 3.1(c) is not long enough to allow the full development of the
M 1 profile then the upper control section disappears and the upper
reservoir level will then reflect changes in the downstream water
level. As the M1 profile is normally long this may well be the case
when a number of basins or lakes are joined by relatively short
lengths of channel.
3.2 Transition through critical depth
3.2.1 High stage to low stage. This occurs at the entrance to steeply
sloping channels from reservoirs or at a transition in channel bed

52

Flow in Channels

slope from mild to steep (Fig. 3.2). Through the transition the flow
is accelerating smoothly, resulting in a steady smooth surface profile
and comparatively little turbulence or loss of energy. Rouse18
shows that if the change of bed slope is too great at a sharp transition
the flow may separate from the bed in the immediate downstream
Subcriticol
flow

Supercri tical
flow

~ _:~_:::~::- .... -~ ..SQ!-~ ..

m
-:::-~:::3::::~
::::3
:~::~~'-"::~: :~::-: :-:-:~:: : :::.~
:-;.;.;:.;.:;:.;.::;::;;:i!"':.
;.;.;.""
:::::,\-::::x::;~

FIG.

3.2 Transition from subcritical flow to supercritica/ flow.

region. The resulting space below the separated streamline will be

filled with eddying water.
At the change in bed slope the M2 upstream profile will meet the
S2 downstream profile tangentially. The water surface at this point
will not be vertical (critical depth) because the pronounced flow
curvature in this region will invalidate the basic assumption in the
development of the varied flow equation (section 2.1.1).
3.2.2 Low stage to high stage-the hydraulic jump. A change of flow
from supercritical (low stage) to subcritical (high stage) will not
normally occur at any fixed point in a channel, unlike the opposite
flow transition described in section 3.2.1 above. Unless specific
measures outlined in section 3.2.3 are taken the change will take the
form of a hydraulic jump which is a stationary surge or shock wave
in which the speed of advance of the wave front (celerity) is exactly
equalised by the velocity of the flow in the opposite direction. In
any particular case the position of the hydraulic jump will depend
upon a combination of the upstream and downstream flow condition.
The overall form as well as the detailed performance of the jump
depends upon the upstream and downstream depths and the velocity
of the approaching flow. The conditions are best described quantitatively by the Froude number (see section 3.3.1) of the flow; F =
v1jy'(gd1) in which (see Fig. 3.3) v1 is the velocity and d1 the depth
immediately before the jump. Because the critical depth relationship
is inherent in this form of the Froude number the inclusion of the

Rapidly Varied Flow

53

upstream flow characteristics only is sufficient to define conditions

at the jump (section 3.3).
The form of the hydraulic jump can be conveniently classified
as follows:

------- --~
f

d, v,:::::::::::::::::::::::::::::::::..:.::::::.:.:.:.:.:.::.:.:.::...:.::.o.-:...-..;.,.::;::
(c) Undulor

(b) Tronsilionol

(c) Direcl

FIG. 3.3 Types of hydraulic jump.

(1) Undular jump (Fig. 3.3a). The flow expands smoothly and then
oscillates so that the jump has the form of a smooth initial wave
followed by a train of waves of decreasing amplitude. Little turbulence is induced by such a jump and the energy loss is therefore small.
An undular jump is formed when the F value lies between 10 and
17. When F = 10 (the limiting case) no jump occurs and the flow
is critical (d1 = d2 = de).
(2) Transitional jump. For Froude numbers between 17 and 25
( 18 <

~: < 30) the

hydraulic jump changes progressively. The

increased rate of energy dissipation (section 3.3) results ih (he first

few wave crests "breaking" and isolated reverse rollers form on the
upstream facing slope of these waves (see Fig. 3.3b) just below the
crest. Under certain conditions it is the second wave that is the first
to break as the F value is increased, followed by the leading wave.
Once one of the waves has broken any waves formerly downstream
of it are lost in general surface agitation.
(3) Direct jump. This form (Fig. 3.3c) of the hydraulic jump occurs
when F > 25 and is characterised by severe turbulence and high

54

Flow in Channels

associated energy losses. The jump now consists of a single monoclinal wave covered by an intensely turbulent reverse roller normally
showing as "white water" due to air entrainment at other than
laboratory scale. The nature of the reverse roller and its function in
the hydraulic jump is described well by Rouse19 in the following
passage:
"Those who have observed carefully the behaviour of the
roller through the glass walls of an experimental flume will
recall vividly enough the fact that the roller is not the idealized,
Limi t of upword expansion
of stream

Reverse
roller

r
of down word di !fusion of eddies

FIG. 3.4 The direct hydraulic jump.

distinctly outlined region of flow shown schematically in

illustrations, but an active participant in the diverging flow.
In other words, no portion of fluid involved at any instant in
the reverse motion of the roller remains in the roller for more
than a brief space of time; instead, violently rotating masses
of this roller are torn away constantly into the live stream
below, this continuous process resulting in the high degree of
turbulence . . . which is characteristic of the phenomenon."
The hydraulic jump is principally used by engineers as a means
of dissipating excess energy in, for example, the flood discharge
carried from reservoirs by spillways. Its further uses include:
(a) an efficient means for mixing added chemicals in water
treatment;
(b) a sure indication that the upstream flow is critical (of importance in discharge measuring installations);
(c) ensuring that a sluice upstream does not operate in a drowned
condition; and

Rapidly Varied Flow

55

(d) raising the downstream water level which could be of value in

irrigation and water abstraction schemes.
Analytical aspects of the hydraulic jump are dealt with in section 3.3.
3.2.3 Low stage to high stage-smooth profile. When energy conservation and surface level recovery are at a premium, a transition
section can be designed for a specified channel and discharge, to
change the flow from supercritical to subcritical without the interposition of a hydraulic jump. Fig. 3.5 shows the general form of
such a transition and superimposed on it the critical depth line
(C.D.L.), the total energy line, and the minimum energy line which
Supercri ticat flow- t -

Subcritical flow

:!<;>l_oi_~~~.O:.QY_.U..~L-.-~- -- ---- _
M tnimum ~

,.,., -

_-

......

~n~r~y ~i~e.., ~-- ...... -__.: .. ' ...... :'.. , ____ _

_ f,!l_l,. ____ . - '

',_f..P"L

FIG. 3.5 Transition through critical depth to higher stage without

a hydraulic jump.

is a profile of the specific energy at critical flow. At the crest of the

transition section (bed tangential to the horizontal) the flow passes
smoothly through the critical depth. King and Brater, 20 who give
a description of the design procedure, state that at the crest the
minimum energy line must be tangential to the total energy line for
a jump to be avoided. However, if the discharge through such a
structure were to change from the design figure, this condition
would no longer be satisfied and a hydraulic jump would form either
upstream or downstream of the crest.
3.3 Analysis of the hydraulic jump*
3.3.1 The momentum equation and specific force. The relationship
governing the depths and average velocities in a hydraulic jump has
been satisfactorily determined by considering the changes in longitudinal momentum undergone by the flow. Since the channel length
occupied by the jump is short in relation to lengths required for
normal channel energy losses (bearing in mind that the jump is

* For a comprehensive review of studies carried out on the hydraulic jump

prior to 1950, see reference 21.

56

Flow in Channels

responsible for a very large energy loss through a different mechanism) the shear force at the channel walls causing this energy loss is
omitted from the analysis.
Under these circumstances, the application of the momentum
equation to a jump in a horizontal prismatic channel of any crosssection gives:
(3.1)
when applied between the two cross-sections immediately upstream
and downstream of the jump (see Fig. 3.6). The term z is the depth
of the centroid of the cross-section
below the water surface and A is
---~--
the corresponding cross-sectional
area. The left hand side of Eq.
(3.1) represents the resultant force
v,
on the flow through the jump,
~
being the difference between the
FIG. 3.6 Momentum analysis of
hydrostatic pressure forces at the
the hydraulic jump.
two sections, while on the right
hand side is the rate of change
of momentum in the direction of flow. Since the velocity v is the
average velocity in the cross-section, the equation is only strictly
true for uniform velocity distribution. Where marked non-uniformity is known to exist the velocity term may be modified by a momentum coefficient f3 which will have a value slightly greater than unity.*
The flow passing section 2 contains a large quantity of random
rotational kinetic energy which cannot be included in any energy
term based on v2 This rotational motion is generated in the jump
primarily by the reverse roller and is irrecoverable, being eventually
dissipated as heat in the downstream channel. Thus it will be seen
that although the mechanism for energy transfer is concentrated in
the compact hydraulic jump the ultimate dissipation occurs over a
much greater length of channel.
Eq. (3.1) may be rewritten in the form:
Q2

Q2

- + A2Z2 =gA1
- + A!Zl
gA2

(3.2)

* For a given velocity distribution the value of fJ will always be less than the
corresponding value of IX: 10 < {J < IX (see section 1.5 and reference 8).

57

Rapidly Varied Flow

Bakhmeteff22 suggested that since the two sides of this equation are
similar and, Q being a constant, are a function of depth only, they
represent different but numerically equal values of a function f(d),
now known as the specific force of a channel flow. Hence for any
discharge Q:
Q2

f(d) = -

gA

+ Az

(3.3)

The specific force function may usefully be plotted as a graph

against depth of flow and, when considering conditions at a hydraulic

--y--~ +Az
gA

Fro. 3.7 Specific energy and specific force relationships at a

hydraulic jump.

jump, it is also of value to plot the specific energy curve on the same
axes (Fig. 3.7). It will be seen from this figure that the specific force
function, like the specific energy, has a minimum value at the
critical depth. Any vertical line drawn on this graph, corresponding
to a real value of

(;~ + Az). cuts the specific force curve at two

points, one (d1) corresponding to a depth less than the critical and
the other (d2) to one greater than the critical.
On momentum considerations a stable hydraulic jump may exist
between two such points which are known as conjugate depths. The
specific energies E1 and 2 corresponding to these conjugate depths

58

Flow in Channels

can be obtained from the d1 and d2 intercepts with the specific energy
curve as shown in Fig. 3.7. The difference 11E between the two is
therefore the specific energy loss attributable to the hydraulic jump.
It should be noted that E1 (supercritical flow) will always be greater
than 2, resulting in a negative energy increment 11E, since energy can
only be dissipated in a hydraulic jump and never added.
As the height of the jump decreases ( ~: approaches unity) the
energy loss 11E approaches zero which is in agreement with the
physical conditions observable in such a jump (see section 3.2.2).
When a hydraulic jump occurs in a wide channel it is possible to
represent the flow conditions by a two-dimensional analysis without
significant error. Rewriting Eq. (3.1) in this form gives
2
-w (d22 - dl)
2

(1 1)

= -w q2 - - 2

dl

d2

(3.4)

in which q is the discharge per unit width. It is now possible to

express one of the two depths in terms of the other and the discharge
directly from Eq. (3.4):
(3.5)
From this relationship the height of the jump is a function of a
2

dimensionless ratio .!!_3 similar to the square of the Froude number

gdl
of the upstream flow
vz
q2
2=_1 = gdl
gdf
and substituting F 2 into Eq. (3.5) gives
a2 =
dl

HvCI

+ 82)- Il

(3.6)

This theoretical relationship between d2/d1 and F has been found to

agree closely with experimental results by Bakhmeteff and Matzke 2 3
and others.
The analysis so far has been restricted to a jump in a horizontal
channel. Bakhmeteff and Matzke 24 investigated the case of a sloping
channel and showed that a term must be added to the momentum

59

Rapidly Varied Flow

equation to account for the component of the weight of water in the

jump resolved in the direction of the sloping channel bed. This
leads to the following relationship for a rectangular cross-section:
wb 2
-(eli
2

wQ

- d'f) = -

(v1 - v2)

+ Ws

(3.7)

in which W is the weight of water contained between the two sections

and s the slope of the channel bed (when the slope is small, s = sin()
= tan 0). For values of s less than 1/10 the weight term may be
omitted from Eq. (3.7), giving the former result (Eq. 3.1) without
serious error.

3.3.2 Physical characteristics of the jump. In addition to the relationship between the two depths at a hydraulic jump it is useful to establish relationships predicting other physical characteristics. The
more important of these are:
Energy loss. The specific energy loss l:iE can be evaluated from
Eq. (2.4) for a wide channel, substituting for the discharge q from
Eq. (3.5), to give
(3.8)

Height ofjump. Sometimes this is defined as a ratio of depths as in

Eq. (3.6) so giving the result immediately in a dimensionless form.
On the other hand Chow25 and others express it as the difference:
d, = d2 - d1 which is then put in a dimensionless form by dividing
by the upstream specific energy E1. Eq. (3.6) will enable the height
of the jump to be computed while reference 23 will satisfy a need for
more detailed information.
Length of jump. The considerable variation found in the results of
different investigators of this aspect of the hydraulic jump is due to
the difficulty in defining systematically the downstream termination.
Variation in the location of this section does not significantly affect
other properties of the jump but is clearly of primary importance in
the length measurement. The cumulative evidence of experimental
results indicate that under most circumstances the length of the

60

Flow in Channels

jump lies within the limits 5 <

~<

65 in which L is the length of

the jump.
3.3.3 Location of the jump. When a hydraulic jump occurs in a
channel it joins either two gradually varied flow profiles or one such
profile to the normal depth. The normal depth, if it is present, will
form the upstream limit of the jump in steep channels and the
downstream limit in channels of mild slope. In channels with
horizontal beds there is no finite normal depth and therefore any
jump present must join the Ha and the Hz profiles. In channels of
adverse slope a stable jump can only exist if the slope is extremely
small. If the channel slope is critical no jump can occur under any
circumstances.
When a stable jump does occur it will form at a point along the
channel where suitable conjugate depths exist. The procedure
necessary to determine this point is outlined as follows:
(A) Normal depth present. Where the normal depth forms one of
the limits to the jump, as in Fig. 3.8, the conjugate depth to do can
be computed from Eq. (3.5) for steep slopes, or from a parallel form
of that equation for mild slopes. When the varied flow profile that

Flo. 3.8 Location of hydraulic jump with normal depth downstream.

forms the other limit of the jump (Ma in this illustration) has been
computed the jump is located by selecting the point on that profile
where the depth is the same as the required conjugate depth.
(B) Normal depth not present. A jump joining two varied flow
profiles is illustrated in Fig. 3.9. The two flow profiles are first
computed and drawn to scale over their full lengths. A line de is then
drawn representing the corresponding conjugate depths to the
upstream M 3 profile. If the length of the jump, as discussed in
section 3.3.2, is insignificant by comparison with the horizontal

Rapidly Varied Flow

61

dimensions of the surface profiles involved the jump will occur at

the section where the conjugate depth curve cuts the downstream
M2 surface curve. However, in the example illustrated in Fig. 3.9,
the length of the jump is considered to be significant and is represented to scale by some length x. The exact location of the jump must
now be found by trial-and-error. A vertical line aa', representing
the upstream limit of the jump, is drawn cutting the conjugate depth

========,=,======..=e

M".;;;;. . . . . . . . . :. .,. ,.,.;.

========:::::==:===.......~............................

===== xiJ;;

Control
sect1on

FIG.

3.9 Location of the hydraulic jump between varied flow

profiles (length ofjump significant).

curve de at some point a. For this position to be correct, a horizontal

line drawn through a must cut the downstream flow profile (M2) at
a point b such that the length ab is equal to x, the known length of
the jump. If the length ab is found to be incorrect the position of
the original vertical aa' must be adjusted suitably. When this condition is satisfied the jump is represented on the channel profile by
the sloping line a'b (note the exaggerated vertical scale). It will be
seen that conditions at two control sections determine the location
and character of the jump in this example, as compared with only
one in that illustrated in Fig. 3.8.
3.4 Flow through a sluice-horizontal force on the structure
A sluice is a common form of control structure used in river and
canal regulation and as such will be considered further in Chapter 4.
It consists of an opening in a water retaining wall or dam, built
across a channel through which water flows due to the head on the
upstream side where the depth is always in excess of the height of the
opening (see Fig. 3.10). On the downstream side the jet discharging
from the sluice may be exposed to the atmosphere, partially submerged by a hydraulic jump or completely submerged. The discharge through the sluice in the latter two cases will be a function of

62

Flow in Channels

the downstream as well as of the upstream depth. A sluice therefore

only forms a control section in a channel if it is discharging freely, as
in Fig. 3.10(a). The downstream conditions will depend in any
particular instance on the depth imposed there by the downstream
control section.
It is of interest in the present context to apply the methods of
specific energy and specific force, used in the analysis of the hydraulic
jump, to free discharge through a sluice gate. The rapidly varying
flow is accelerating and therefore, unlike the hydraulic jump in
which the action is one of deceleration, little energy is dissipated in
the generation of turbulence consequent upon flow separation. It
follows then that the total energy before and after the sluice may be

(o) Free discharge (b) Partially submerged (c) Wholly submerged

FIG. 3.10 Flow under a sluice.

equated to give a relationship connecting the upstream and downstream depths and the discharge. On the other hand, the application
of the momentum equation to the flow is complicated by the existence
of an indeterminate horizontal force exerted on the flow by the sluice
structure (sluice gate). This force, F' in Fig. 3.11, is represented as
the area under the pressure distribution diagram against the upstream
face of the structure. The pressure distribution is modified from the
linear hydrostatic relationship, p/w = h, by the local kinetic energy
head of the water adjacent to the structure as it approaches the
opening.
Considering unit width of sluice the resultant longitudinal force
acting on the flow between sections 1 and 2 is
Wd2
wdr
--F-- z

This can be equated to the rate of change of fluid momentum in this

reach
(3.9)

63

Rapidly Varied Flow

Assuming no energy loss, the specific energy before and after the
jump may be equated from Eq. (2.4),

q2

dl

+ 2gdr =

d2

q2

+ 2gdi =

1,2

from which
(3.10)

IicJ

iJfl.

:
I

12 Specific energy'

E=d +

22

Specific force:

~+9...2
2

gd

FIG. 3.11 Application of the specific force method to flow under

a sluice.

Eq. (3.9) can be rearranged in the specific force form similar to Eq.
(3.2),

(d~ q2 ) F'
q2 )
( d'f
2 + gdl = 2 + gd2 + w

(3.11)

which is represented graphically in Fig. 3.11. It will be seen that

the line drawn between the d1 and d2 intercepts with the specific
force curve is now inclined and its horizontal component represents
the quantity F' by which the two specific force terms differ. These

64

Flow in Channels

two depths, however, intercept the specific energy curve at points

which are joined by a vertical line, demonstrating that no energy
loss occurs in this flow.
The force F' can be evaluated in terms of the two depths by
substituting for q into Eq. (3.11) from Eq. (3.10). This gives
F' = ~ (d1 - d2) 3
d2)
2 (dl

(3.12)

The downstream depth d2 should be taken at the "vena contracta",

at which point the contraction of the jet subsequent to its passage
through the sluice is completed. The dimensions of the "vena
contracta" and the discharge relationships for a sluice under various
downstream conditions are dealt with in section 4.2.3.3.
The disadvantage of the momentum analysis method for determining the force on a sluice structure is that it only gives the magnitude of the force and not its point of application. This can be found
approximately by assuming a hydrostatic distribution of pressure on
the upstream face of the structure (see Fig. 3.11) and the point of
application is then a distance below the upstream surface equal to
two-thirds of the immersed depth of the structure. The magnitude
of the force is also given approximately by the area under the
triangular hydrostatic pressure distribution curve. The only accurate means of determining both the magnitude and location of this
force in a particular instance is from measurements of the pressure
distribution on a hydraulic model of the sluice system.

3.5 Flow past a submerged obstacle

A solid object, whatever its shape, will exert a retarding force on a
stream in which it is immersed. By considering the momentum
changes in the flow past the obstacle it is possible to relate this
force to the flow depths up and downstream. Fig. 3.12(a) represents
subcritical flow over a line of rectangular blocks arranged across the
bed of a channel. The appropriate specific force diagram (see Eq.
3.11) is constructed beside the stream profile. Since the force
exerted by the blocks opposes the stream it must reduce the specific
F'

force of the flow by an amount - where F' is the force exerted per
w
unit width of channel. It will be seen from the specific force diagram

65

Rapidly Varied Flow

that this must reduce the downstream depth in the case of subcritical flow. Fig. 3.12(b) illustrates the effect of an obstacle on
supercritical flow in which case the downstream depth is seen to
increase.
If a submerged obstacle is introduced into a channel where
normal flow would otherwise exist changes will occur as follows:
(a) mild slope. Normal depth will be maintained downstream of the
obstacle with an appropriate increase of water level upstream. Here

. . . -- _!l_L,_
!------------~-

--

___

f --

-c--

dtd.dz --

k -: -::.1. . . . . . .-;. .J. .:-.:.

do

d2

F'

Flow

(o) Subcri lical flow

-.----..../ ...... l
F'

t'

-~-- - - .

~2

:::::::::::::::::::::::::::::.;::;::::::::::.-::.:::.-.-.

Flow

FIG.

(b) Supercrj fical flow

I
I

I
I

I
I

---! f.: f- "

Specofoc
force

Specofic
force

3.12 Effect of submerged obstacle on depth of.flow.

the profile will be of the M 1 type leading to normal depth again

further upstream.
(b) steep slope. Because no changes in depth can be propagated
upstream in supercritical flow, the depth there must be normal.
Downstream the water level will be raised and then follow the Sz
profile until the normal depth is regained.
In no case can a force be exerted against a stream greater than
that which would reduce the specific force of the downstream flow to
the minimum value possible (occurring at critical depth). Under
circumstances where this might appear to be the case the upstream

66

Flow in Channels

depth will always be raised the necessary amount by the obstacle, to

give critical depth downstream. Where the upstream flow is supercritical this process will produce a hydraulic jump a short distance in
front of the obstacle and so reduce the velocity past it and therefore
the force exerted on the stream. Depending upon the downstream
conditions the flow may then revert to supercritical, the obstacle
forming a control section.
The above property of an obstacle in supercritical flow is frequently
used to stabilise the position of a hydraulic jump where this is
required in a channel. An array of concrete blocks of various shapes

dt d' ----- _cl ___

dz

____b

de

~-::::::::::::-:-::::::::::-:c:::"(:::..:::::::::::::

Flow

"'-Blocks

- l ~ 1- 1- .d ---1 "'

Spec1flc
energy

Spec1fic
force

FIG. 3.13 Stabilisation ofhydraulic jump by blocks on channel bed.

and sizes are normally constructed on the bed of the channel at the
required position. It will be seen from the specific force diagram in
Fig. 3.13 that the downstream depth d2 corresponding to the intercept b on the curve is less than the depth d~, at c, which would exist
if the jump had formed unaided; the horizontal component of the
scale distance be being equal to the magnitude of the force term
F'jw. It can also be appreciated from the specific energy curve in
this figure that the energy head t!.E dissipated in the jump is increased
by the presence of the blocks, a result that is in agreement with the
general principles of fluid mechanics.
Finally, in this section it is necessary to consider the methods
available for solving problems involving flow past submerged
obstacles. The principal variables are the upstream and downstream
depths and the force per unit width exerted on the stream by the
obstacle. The discharge will be assumed constant and known. If

67

Rapidly Varied Flow

the specific force

form ( w~

(d2 + gd

+ ;~)

q2 )-from Eq. 3.3-is left in its original

as derived from the momentum equation, this

term will now have the dimensions of a force per unit width and can
be considered as the "force" of the stream F. Considering the case
of a submerged pipeline suspended above the bed of a channel
shown in Fig. 3.14, the "forces" controlling the flow can now be
equated thus
(3.13)
in which F1 and F2 are the "force" of the stream at the two sections
indicated.

F,

Pipe

-I

p-0 F;

FIG.

3.14 Graphical determination of.flow conditions in vicinity of

submerged transverse pipe.

It is convenient in this example to put F on a dimensionless basis

by dividing the relationship

wd2 wq2
F=-+-

gd

(3.14)

right through by (wd;) where d; = q2 fg (de is the critical depth).

This eliminates q from the "force" expression giving

(d)2

F = t - +(de)
wd;
de
d

(3.15)

Eq. 3.13 can now be written:

(3.16)

68

Flow in Channels

from which F' can be evaluated directly if d1 and d2 are known. If,
however, the unknown quantity is one or other of the depths, the
expression to be solved takes the form of a cubic equation including
the known depth, the critical depth and the force F' among its
constant terms. This can most conveniently be solved graphically
by plotting values of (Ffwd~) against the depth d as in Fig. 3.14.
First the critical depth can be computed from the discharge q and,
assuming for the present that it is the downstream depth d2 that is
unknown, the next step is to evaluate
Plan view
;vena con!rac!a" (F1jwd~) from Eq. (3.15). This value
is now plotted on the graph as the
t
b,bz
point a. It is only necessary to comI
,-----t
pute and plot the relevant portion
t
I
FIG. 3.15 Flow in a sharply
of the depth-"force" relationship
convergent channel.
which in this case will have its upper
limit at a and its lower limit somewhat above the critical flow point. From the known force F'
the distance (F'fwd~) can be set off on the "force" axis locating the
downstream point b on the curve. This operation can be expressed
algebraically by the relationship:

, 'r---

F2
wd;

F1
F'
wd; - wd;

(3.17)

The position of the downstream point b on the curve corresponding

to the vertical (F2/wd~) will give the required value of the downstream depth d2.
If it is the downstream depth that is known and the upstream that
is required, the procedure is identical and the construction of the
relevant steps on the "force" diagram reversed.
3.6 Changes in channel width
3.6.1 General. Changes in channel width that occur within a
relatively short distance will produce rapidly varying flow in which
gravitational forces predominate and frictional forces play only a
secondary role. The change may occur as either an increase or a
decrease in width, relative to the direction of flow, and it may occur
sharply (Fig. 3.15) or form a smoother transition. In general terms
a change in the width of a channel is only one of a number of different possible transitions of which a change in bed elevation is

69

Rapidly Varied Flow

possibly the most common (section 2.2.2). Frequently, channel

transitions are designed to combine simultaneous changes in width
and bed elevation.
Where the channel width decreases in the downstream direction
the transition length is referred to as a convergent channel and where
the width increases it is termed a divergent channel. Flow through a
convergent channel is accelerating and, since the total energy of the
water cannot increase, the hydrostatic pressure of any particular
streamline will decrease in the direction of flow. Both theoretical
and experimental studies of fluid mechanics indicate that under such
conditions energy losses are small and the flow will not "separate"
from the solid boundaries. Conversely, in a divergent channel flow
separation is a possibility and, particularly when this occurs, energy
losses are relatively large.
Channel transitions are used to join channels of differing crosssectional shape as well as similar cross-sections of different size.
They are also required where the flow from a channel enters a
tunnel or syphon and under the opposite circumstances where, for
example, a turbine draft tube discharges into its tail race. Most
control structures involve changes in channel width as well as other
sources of rapidly varying flow.
3.6.2 Subcritical flow. The characteristic property of transitions
involving a change of width is the resulting change in q, the discharge
per unit width. There is therefore a corresponding change in the

critical depth of the channel

(de

~~)

through the transition.

Consider the convergent channel shown in Fig. 3.16. It is assumed

that the energy lost in this transition is negligible and therefore the
specific energy E is constant. The critical depth can be computed
before and after the contraction (since, q1 = Q' etc.) and this

enables ~de' the change in

de

bl

to be evaluated. It will be seen from

this form of the specific energy diagram, shown in Fig. 3.16, how
the unknown depth may be determined if either d1 or d2 are known, by
using this value of

~ ~.

When a channel undergoes a sharp reduc-

tion in width (Fig. 3.15) the flow separates at the projecting corners
and experiences a further contraction to a point of minimum width

70

Flow in Channels

known as the "vena contracta" from which it expands to the full

downstream dimension, b2. There is a significant loss of energy
associated with this final expansion which invalidates the method
of solution illustrated in Fig. 3.16. The results of experiments by
Formica26 carried out on energy losses in sharp contractions are
summarised in English by Chow. 27
If the channel shown in Fig. 3.16 has its downstream width
further reduced until (E/dc)2 = 15, the depth at this section and
therefore the flow, will become critical. Further reduction in b2
will have the effect of increasing the upstream depth by an amount
sufficient to maintain critical depth at the narrow section. This

~I

-------.--1

bz

~Plan view

::-----~f~-f---ll
!:X..
ll"F<~-----

--- -----L ___ _

:.,;,;;,;,.:...:...-;:;;;:::::;:::.::..;.::...::::.:,1::.:::.:::

Elevolion

I I

15

Flo. 3.16 Flow in a gradually convergent channel.

is a parallel case to that considered in section 3.5 where a large

obstacle placed across a channel is shown to produce a similar
result.
Subcritical flow through a divergent channel is not so readily
predictable by analytical methods. The proportion of the kinetic
energy head (vrf2g) that is recovered as an increase in surface
elevation depends primarily on the angle of divergence of the sides.
If the expansion is sudden (Fig. 3.15, but with the direction of flow
reversed) little of this energy is recovered and the water surface may
be assumed to undergo no rise in level at the transition. The excess
energy: (vi - v~)/2g is dissipated downstream in the high degree
of turbulence associated with an uncontrolled stream expansion of
this type. As the length of the divergent channel increases (from
zero) so does the proportion of the excess kinetic energy that is
restored to potential energy in raising the water level. The

Rapidly Varied Flow

71

performance of the transition in this respect is further improved

by replacing the sharp junctions of the convergent sides with the
upstream and downstream channels by a suitable radius. Formica 26
carried out extensive experiments on the relationship between angle
of divergence and energy head recovery. In general, divergent
angles greater than 7 from centreline induce flow separation at the
walls accompanied by relatively low head recovery downstream.
The design and behaviour of open channel transitions is widely
reviewed by Ippen. 2s
Channel flow between bridge piers consists essentially of a convergent channel, followed immediately by a divergent one, and is
therefore normally considered as a single occurrence of rapidly
varying flow (flow at a constriction). The formula of d'Aubuisson 29
(3.18)

has been found useful in evaluating ha the drop in water surface

level past the piers. In this expression:

Q is the total channel discharge;

b2 is the aggregate width of the openings between the piers;
da is the downstream depth;
v1 is the mean upstream velocity;
Ka is a discharge coefficient that must be determined experimentally for each configuration of piers considered (see Yarnell30).
Values of Ka normally lie in the range 096-125.
3.6.3 Supercritical flow. Supercritical flow passing through a convergent channel will, in accordance with the momentum and energy
principles used above, increase its depth of flow. It will, however,
be characterised by strong oblique stationary shock waves which
form in the transition and have the effect of increasing the depth in
steps, as in Fig. 3.17. The angle of these oblique shock waves will
depend upon the upstream Froude number (see Ippen and Dawson3 1).
In the case illustrated in Fig. 3.17 the convergence is so designed that
the "positive" waves produced at the re-entrant corners meet the
opposite sides at the projecting corners where their reflections there
should theoretically cancel out the "negative" wave generated at
these points. However, even when this occurs the downstream
channel surface is normally criss-crossed with stationary oblique
waves, although these are usually weaker and less sharp-fronted than

72

Flow in Channels

those in the transition. If the reflected positive wave does not

coincide with the negative wave, the downstream surface will be far
more disturbed.
Supercritical flow through a divergent channel leads to a reduction
in depth although on spillway aprons where the civil engineer
normally meets this type of flow it is followed closely by a hydraulic
jump. An investigation into this problem has been made by the
U.S. Soil Conservation Service and the results are published by
Blaisdell. 32
If supercritical flow in passing through a channel convergence
or constriction approaches too closely to critical flow a hydraulic

Weak
downstreorr
waves

Pion view

---------------

.~P.L.: - -- - --- - -w~i;,--s~;i~~;

/centreline

at

--=~~''-~----_-__,f--< Water surface

of channel wall

Eleva tion

FIG.

3.17 Supercritical flow through a convergent channel.

jump is likely to form in the upstream channel. This may be followed

by a return to supercritical flow at the transition, which now forms
a control section. The change to the alternate depth is quite likely
at flat-bottomed transitions of this type, but it tends to be unstable
and difficult to predict. The actual flow conditions at the transition
will depend upon the upstream and downstream conditions of flow
and the instability of the flow will depend on how close the Froude
number is to unity, values away from unity resulting in more stable
predictable flow profiles.

4 Control and Measurement

of Open Channel Flow

4.1 River utilisation

At this point it should be made clear that open channel hydraulics
is concerned with two principal types of channel, canals and rivers.
The term canal is taken here to include all artificial channels regardless of their size and purpose and while the flow that takes place in
these may be relatively complex the purpose for which they are
designed and built is normally clear. On the other hand, a river*
may fulfil a number of functions of use to man and it does not
follow that those who are concerned at any one time with one
particular function may fully appreciate all the others. Unfortunately, human interference with a natural river that is aimed at
benefiting one of its functions will often affect many of the others,
frequently adversely.
The principal uses to which man can put a river may be summarised by the following headings:
(1) Water supply (domestic, industrial and agricultural)
(2) Waste disposal
(3) Land drainage
(4) Power generation
(5) Navigation
(6) Amenity
(7) River life preservation

* The term river is here used to describe all channels of interest to the hydraulic
engineer that are of natural origin. Some aspects of open channel hydraulics
used elsewhere as a basis for classification (e.g. bed movement) cut right across
this distinction as it is used here.
73

74

Flow in Channels

The dangers that could possibly develop from a mismanaged river

are, in addition to the more negative failure to develop the above
listed potentialities,
(1) Urban and rural flooding
(2) Pollution, possibly leading to disease
(3) Erosion of land
The order of importance given to both the above lists will depend
largely on local or national conditions. To give some examples; the
generation of hydro-electric power is of little significance on English
rivers while it is of extreme importance in Italy, the Scandinavian
countries and others where alternative forms of energy are not
internally available. In arid countries where the river channels are
frequently dry the ability of a channel to carry safely the occasional
severe flood discharge is of much greater importance than its
navigational or waste disposal functions. In such densely populated
countries as Britain the economic value of rivers has become in
recent years more and more closely linked with water supply and
waste disposal. The much improved efficiency of both water treatment and sewage purification works now enables rivers to serve all
the public water needs of the communities along their length.
In the past these various functions of a river system have been on
the whole, the responsibility of different bodies, both public and
private, and frequently the concern of men with widely differing
training, interests and sometimes, loyalties. It is only in the last few
years, with one or two notable exceptions,* that some countries
have passed legislationt setting up public authorities with wide
reaching powers to control river basin development for the optimum
good.
The cost to the community of any particular scheme must be set
against an accurate assessment of the economic value of the benefits
to be derived from it. Such benefits may be capable of solely economic evaluation or, especially in the more developed countries,
contain a measure of social benefit linked more or less directly with
categories (6) and (7) above. It is often the case that the older
countries have a great deal of progress to make up in the matters

* For example, The Tenessee Valley Authority, set up in 1933 in the U.S.A.
See Hoyt and Langbein. 33
t The Water Resources Act 1963 of Great Britain gives much wider powers
and responsibilities for reconstituted River Boards than formerly.

Control and Measurement of Open Channel Flow

75

of amenity and preservation before these aspects of their rivers can

match others of more direct economic benefit.
Development of a river requires control of the following variables:
(1) Discharge
(2) Water level (stage)
(3) Biological and chemical quality
(4) Sediment load

The present book does not deal with quality control, which is the
more particular responsibility of the Public Health and Sanitation
Engineer, while the whole subject of sediment transport is reviewed
in Chapter 5. The control of stage and discharge involves the
engineer in both regulation and measurement. Frequently these two
operations, the one as important as the other, are carried out simultaneously by the same hydraulic structure. In the ensuing sections
of this chapter, hydraulic structures will be considered under channel
regulation or flow measurement on the basis of their principal rather
than their sole function.

4.2 Channel regulation

4.2.1 Canalisation. Rivers in which navigation for the purpose of
inland water transport is important normally need control structures
at regular intervals to ensure an adequate water depth at all points
and at all seasons. Such a form of channel regulation is known as
canalisation and the principles are illustrated in Fig. 4.1. A weir
incorporating one or more navigation locks is constructed across the
river raising the upstream water level to form a back-water M1
profile. The adjacent upstream weir must be sited at the point on
this backwater profile where the channel depth becomes equal to
the minimum permissible depth for navigation.
For a river which in the dry weather is extremely small, the backwater profile will approximate closely to a horizontal line and this
condition must be considered in order to ensure sufficient depth for
navigation through the worst dry period. This will still require that
enough water enters any particular reach to make up the total
losses. Of this outflow the major part is normally due to working
the locks (locking) and leakage at the gates and weirs. This water is
therefore passed on to the next downstream reach to offset similar
losses in that section. Consequently these outflows have to be made

76

Flow in Channels

up only once in a canalised channel and that at the highest point.

Losses due to evaporation, abstraction and groundwater escaping
from the channel can occur continuously over the length of the
channel and may therefore be the more important under certain
circumstances. In a canalised river sufficient water is normally
available at the highest point and the weirs will carry any excess
flow but in an artificial canal provision must be made for introducing
sufficient water economically at suitable points.
The canalisation of a river will undoubtedly upset the processes,
known as "sediment transport," by which a flowing river moves solid
material in the downstream direction. This material consists of
particles of clay, silt, sand and gravel of natural origin of which
Backwater \ .
profile
'\

FIG.

Minimum permissible
depth for navigation

4.1 Canalisation of a river channel for navigation.

the sides and bed of the channel are normally formed. The extent
to which sediment transport takes place in any particular channel
varies greatly and the whole matter is dealt with in Chapter 5. In
canalised channels the possibility of excessive deposition of sediment
upstream of the weirs must be considered.
Where economic and other conditions are favourable it is possible
to develop a river for both navigation and hydro-electric power
generation in a single scheme. The weirs will now incorporate
power generating stations as well as navigation locks. This system
has been exploited successfully in Europe on both large and small
scales in recent years.
4.2.2 Use of reservoirs. In the previous section it has been shown
that canalisation of a river channel can control the level of the water
surface which is of direct concern to navigation interests and of
indirect benefit to other river functions as for example irrigation,
waste disposal and amenity. The more important and usually the
more difficult aspect of river regulation is the control of discharge.
Maintaining a predetermined minimum discharge in a river is of

Control and Measurement of Open Channel Flow

77

primary concern to bodies responsible for the abstraction of water

for domestic, industrial, irrigation and other uses, whilst restricting
the maximum discharge during periods of flood is of obvious social
and economic importance.
The topographical, geological and vegetative character of the
catchment area will influence greatly the condition of the river
during drought and flood conditions but these factors offer only
limited opportunity for beneficial control. Examples of river control
by such "natural" means include the upland planting of forests,
which has the effect of retaining rainfall and so improving river flow
in both wet and dry periods (see Roa3 4), and a limited control over
the agricultural uses of the lower catchment area to minimise
evaporation losses, soil erosion and quick run-off after storms.
A more comprehensive degree of discharge control can only be
achieved by the construction and operation of reservoirs within the
river catchment area. In a small catchment, adequate regulation can
often be obtained by a single reservoir but in larger or more highly
developed river basins a multi-reservoir system will be needed for a
satisfactory solution. The design and operation of multi-reservoir
systems is extremely complex and the reader is referred on this point
to the bibliography for this chapter on p. 139. In the present section
the behaviour of a single reservoir will be considered.
The principal functions of a reservoir are:
(1) Flood alleviation

(2) Water conservation

(3) Power generation

In general, flood alleviation and water conservation can be served to

varying degrees by any reservoir while power generation will only be
a function of a reservoir, if specific provision is made for it. The
order of importance of the above functions will depend upon the
particular conditions for any specific scheme.
(1) Flood alleviation. If an empty or only partially full reservoir

intercepts a river flood resulting from heavy rainfall or snow melt

in the catchment, the storage space available can be used to detain
the flood water, or some of it, and the peak discharge in the downstream river will be correspondingly reduced. Such a reservoir can
be designed and operated so that, for an estimated maximum
probable flood (see Koelzer35), the downstream discharge is held

78

Flow in Channels

back to a level that does not produce unacceptable economic damage.

Fig. 4.2 illustrates this principle by showing graphically the outflow
hydrograph from a flood detention reservoir superimposed on the
inflow hydrograph. The integrated difference between inflow and
outflow over the storage period (graphically indicated by the
shaded area) will give the storage capacity required to protect the
downstream valley from the flood indicated by the inflow hydro graph.
The form of the outflow hydrograph will depend upon the manner
in which the reservoir discharge is controlled. Curve (a) in Fig. 4.2

Required flood storage capac<y

in reservoir (controlled oultlowl

--Time

FIG.

4.2 Principle of operation offlood-detention reservoir.

shows a gate-controlled discharge which is held at the maximum

permissible value for as long as possible, which reduces the required
storage to a minimum. Curve (b) shows the uncontrolled discharge
from a fixed weir or opening in which the discharge is proportional
to the amount of water in storage. This arrangement will require
a greater storage volume for the same degree of flood protection.
If the area requiring flood protection is not immediately downstream of the reservoir then the outflow may be permitted to exceed
the safe downstream discharge for a limited time because of the
natural attenuation or "flattening" experienced by a flood wave
passing down an open channel.

Control and Measurement of Open Channel Flow

79

The above description of the design and operation of flooddetention reservoirs is necessarily much simplified and fuller information should be sought in the following references: Hoyt and
Langbein,33 Gilcrest,36 Kuiper, 3 7 Meinzer3B and Linsley et af. 3 9
Also see section 6.4.
(2) Water conservation. A reservoir designed to maintain the
downstream river discharge at a specified minimum value in order
to meet water abstraction requirements must be kept full as long as
the inflow to the reservoir is sufficient to meet the current demands
on the system. When the reservoir inflow drops below the demand
level, the balance is supplied from the reservoir storage as long as a
certain minimum remains in the reservoir. This minimum is termed
rationed storage and it is only discharged to make up the outflow to
some predetermined proportion of the demand. The reservoir
capacity in excess of the rationed storage from which the full demand
is met is termed normal storage. The adoption of a rationed storage
method of operation delays the time at which the reservoir will run
dry in a severe drought and so alleviates the hardship caused. It is
also possible to allocate priorities to the uses to which the demand
discharge is put and thereby, for example, ensure the continuance
of domestic water supply at full capacity for as long as possible.
(3) Power generation. Hydraulic power generation is outside the
scope of the present book but it should be pointed out here that
water so used is available for abstraction downstream for other
purposes. The restriction that power generation places upon a
reservoir fulfilling other functions is that a minimum amount of
storage must be retained at all times to ensure sufficient head for
the turbines. This portion of the reservoir capacity is referred to as
power storage.

From the foregoing it can be seen that a "multi-purpose" reservoir

can be designed to serve the three named functions. Such a reservoir
will need, in common with all reservoirs, capacity for the storage of
sediment gradually deposited on the floor of the reservoir from the
inflowing water. The amount of sediment storage to be allowed for
will depend upon the sediment load of the inflowing streams, the
percentage trapped by the reservoir and the anticipated operational
life of the reservoir. Sediment storage is essentially "dead storage"

80

Flow in Channels

from the point of view of the reservoir operation. Methods for

removing sediment deposited in reservoirs cheaply have so far had
little success but it is possible to remove the sediment load from the
inflow before it reaches the reservoir by the use of preliminary
settling tanks or other means (see Linsley and Franzini40).
A multi-purpose reservoir of this type then, will have its storage
capacity subdivided as shown in Fig. 4.3, the various subdivisions
being shown in their correct relationship to each other. Estimates
of reservoir performance under a given sequence of inflow and
"Backwater curve"
flood storage (shaded)
Flood storage

Do;;,- -
FIG.

4.3 Multi-purpose reservoir storage classification.

outflow conditions are normally computed by numerical step

methods. In this way the contents of a reservoir at the end of a
stated time interval (usually one month) are computed, knowing the
initial contents and the total inflow and outflow over this period.
This relationship can be expressed by the following equation which
is known as the hydrologic book keeping equation.
Initial storage

+ Inflow -

Outflow - Losses

= Final storage

(4.1)

The "losses" term covers seepage and evaporation from the reservoir
which, as well as being a function of the climate and geology, will also
vary with reservoir surface area and therefore water level. The
application of Eq. (4.1) to a single reservoir is a simple matter and
can be carried out conveniently by graphical means, but for a multireservoir system in which the outflow from one or more reservoirs
becomes, after due modification en route, the inflow to another, the
problem is normally solved with a digital computer.
4.2.3 Channel control structures. The hydraulic structures considered
in this section are constructed on rivers and canals to provide control
of discharge and water level without significant storage.

Control and Measurement of Open Channel Flow

81

4.2.3.1 BARRAGES: The term barrage was first used in India and

Egypt to describe a low-head dam constructed across a river to

control the river flow and to divert water into an irrigation canal.
Such structures are also known as diversion-dams (or in canals,
regulators) and their most important hydraulic feature is normally a
series of discharge openings or vents capable of passing, when their
control gates are fully open, the design flood for that river. The
control gates are used for regulating the flow during lesser discharges. In many barrages the vents together with their intervening
piers occupy the full width of the structure.

-?,.~'
<!>,..

-P'
"~/'--

Turbines discharge
inlo river

-~--------

tomng- - - - - - ~ I

(a) Power house remote from barrage

(b) Barrage Incorpor at ing power house

FIG.

4.4 River barrages used in low head hydro-electric power

schemes.

It is of course essential that such structures should be designed to

withstand the hydraulic forces normally operating on dams in
general. In particular, adequate provision must be made for a cut-off
wall to prevent or restrict seepage under the barrage, a downstream apron to protect the foundations from the erosive effects of
the high-energy flow through the vents during flood, and side walls
to prevent seepage around the ends of the barrage and consequent
failure of the river banks. The detailed design of such dams is
dealt with in many standard reference works which include those by
the following: Creagor, Justin and Hinds,41 Brown,42 Leliavsky. 43

82

Flow in Channels
Barrages are sometimes employed in low-head hydro-electric
power schemes and can either:
(a) divert water into a canal supplying the power station at a
suitable elevation as in Fig. 4.4(a); or
(b) incorporate the turbine house inside the barrage structure as
in Fig. 4.4(b) and use the regulating gates to generate sufficient
head across the barrage to enable the turbines to extract
energy from the water passing through them.
4.2.3.2 GATES: A gate is a movable structure that can close discharge
openings in hydraulic structures and therefore control the flow
through them; it is also used to control water levels above spillway
crests. Gated discharge openings can either operate with a continuous
free water surface, be submerged on the upstream side only, or
submerged both upstream and downstream. Such an opening
submerged on the upstream side is normally termed a "sluice" and
when submerged on the downstream side as well is further described
as "drowned."
Hydraulic gates are classified on their shape and method of
operation.
(1) Vertical lift gates. This type of gate is a plane rectangular steel

or cast iron structure and its method of operation is shown in Fig.

4.5. It is commonly used to close the
vents in barrages, being raised and
lowered from above by a movable
gantry that traversed the top of the
barrage on rails to serve all the gates.
The frictional resistance to moving

_ Rollers against the gate, due to the transmission of

;
abutments
the hydrostatic pressure force to the
abutments, can be reduced by the use
: >-Discharge
of roller wheels as shown in Fig. 4.5.
The gate hoists, in gates with machined
face seals, are so arranged that at first
Sealing sill
movement the seal is broken and the
FIG. 4.5 Vertical lift gate
load transferred to the rollers. Hyfitted with rollers.
draulic cylinders may be used instead
of mechanical hoists depending upon economic and maintenance
factors. Where a barrage contains a sufficient number of vents it is
I

Control and Measurement of Open Channel Flow

83

frequently possible, and indeed more satisfactory, to operate the

individual gates either fully open or closed. Under partial opening,
such gates are frequently set in vibration in the vertical direction
due to instability of the flow past the lower edge.
(2) Radial gates. The upstream surface of a radial gate is cylindrical
in form and the gate rotates about trunnions situated on the centreline of the cylindrical surface.
To wnch
Because of the hydrostatic properties of a cylindrical surface, the
resultant hydrostatic force on the
Trunnions
gate always acts through the
trunnion pins. Consequently the
force required to lift the gate is
due to the combined effects of the
weight of the gate and frictional
resistance at the trunnions. This FIG. 4.6 Radial or Taintor gate.
type of gate, sometimes known as
a Taintor gate, is sketched in Fig. 4.6. It is frequently fitted to
barrages and also to spillway crests to give additional reservoir water
level control.
(3) Rolling gate. This gate is in the form of a steel cylinder spanning
between spillway crest piers in which are mounted inclined toothed
racks, as shown in Fig. 4.7. The gate is
To winch
hoisted by rolling it up these racks in
which teeth arranged around the periphery of the ends of the gate engage. On
account of the potentially great strength
of a cylindrical structure (with suitable
internal bracing) rolling gates can be
used economically over greater spans
than other end-fixing gates. Frequently
a longitudinal steel skirt is attached to a
suitable
point on the periphery so that it
Seal
forms a seal with the fixed spillway crest
FIG. 4.7 Rolling gate.
when the gate is in the lowered position.
(4) Drum gate. This gate is in the form of a watertight steel plate
structure which is hinged to the spillway crest in such a way that

84

Flow in Channels

when it is in the lowered position it occupies a recess in the concrete

structure giving an uninterrupted spillway profile (Fig. 4.8). If
water is now admitted into this recess the gate lifts due to its buoyancy and effectively raises the spillway crest. Drum gates vary
considerably in their design but all utilise water pressure in their
operation. They therefore have certain advantages over other kinds
of gates, principally in needing no spillway superstructure involving
winches and cables.
Other types of gates are used from time to time. Some of these are
automatic in so far as they open when the upstream water surface
Raised position ........_

......../' "

--~----~~-

, ",

'\

\
\

'

Control valves

FIG.

4.8 Drum gate (lowered position).

reaches some predetermined elevation. Such automatically opening

gates may or may not be designed to close automatically when the
flood is past.
4.2.3.3 SLUICES: It is difficult to define the term sluice satisfactorily
as most of the gates described above operate over at least part of
their range with a discontinuous free water surface. It would
perhaps be better to think of a sluice as being sufficiently deeply
submerged on the upstream side, under normal operating conditions,
that the occurrence of flow through it with a continuous free water
surface is of no significance.
Sluices are most commonly found, controlled by vertical lift gates,
in navigation locks and barrages. Because of the depth of submergence, the discharge through a sluice is relatively insensitive to
changes in the upstream head. They must, therefore, be subjected
to constant observation and control, unless operating in conjunction

Control and Measurement of Open Channel Flow

85

with an overflow weir, in order that control of upstream water levels

may be maintained.
Fig. 4.9 shows two-dimensional flow under a sluice gate. It
should be noticed that the supercritical discharge from the sluice
continues to reduce in depth for a short distance downstream. The
point at which this contraction of the flow is complete is called the
"vena contracta." The depth at the vena contracta d2 is related to the
height of the opening y by a coefficient of contraction Cc: d2 = Ccy.

- _j L ~f~-~~~-- ---

- - _Io.!.~l_~!~d--
_

- Pressure
head on
gale

v~l2g

I
I v2!2g

.--

..

__ ..... p

-/~/

Static pressure distribution

at sluice opening

FIG.

4.9 Pressure and flow conditions for two dimensional

discharge under a sluice gate.

Similarly the discharge per unit width q is related to the upstream

depth d1 as well as y by a coefficient of discharge Cn, as follows:
q

Cny. y(2gdl)

(4.2)

Using the notation of Fig. 4.9 the continuity and Bernoulli equations take the forms
q

and

V1d1

= V2Ccy

vi + d1 = v~g + Ccy respectively.

.

2g

Rearranging Eq. (4.2) and substituting into it from the above yields
the following relationships connecting Cn, Cc, y and d1.
Cn =

q
Cc
yy(2gdl)
y(l + Ccy/dl)

(4.3)

86

Flow in Channels

Rouse 4 4 states that values of Cc are essentially constant for this

type of sluice (close to 061).
In some circumstances a sluice opening or vent may be controlled
by more than one vertical lift gate. These may be arranged so that
the discharge can either pass under the lower one as in Fig. 4.9,
between the lower and the upper, or over
=-;:-.:;.:;:=-t
. . . ..... !i>.tat head
the upper, which then takes the form of a
----- sharp-edged weir (see section 4.3.1). Under
1
~)
such varying conditions the value of the
dl1
(/
..... \-~
coefficient of discharge will alter con.,."' d3
siderably as will also the pressure distriYT ~ j bution on the gates. Except in the simplest
. . . . . . . . . . . . . .-.-..-..-.-.-. .-.-.-.-.-.-.-:-.-.-.-.-.-.-.-:-:-.-:-.-:-.-.-. .-.. :. . . . . . . . case, an accurate solution to these unknown
FIG. 4.10 Flow at a
quantities may sometimes only be obtained
drowned sluice.
after conducting a hydraulic model investigation.
A sluice discharging in any of these ways will constitute a control
section and when underflow occurs, more particularly the downstream flow, if undrowned, will be supercritical. Unless the downstream channel is steep (see Chapter 2) a hydraulic jump will form in
the equilibrium position, a function
Free discharge
of channel depths and discharge. If
the normal depth of the downstream
channel is greater than the conjugate depth of the flow at the "vena
contracta", the hydraulic jump will
not find equilibrium conditions in
the channel and will move upstream
until stopped by the gate and in so
0
doing "drown" the sluice. The
d,
y
velocity through the sluice will now
be a function of the difference in FIG. 4.11 Values of the discharge
level between the upstream and coefficient CD for flow through
downstream water surfaces but Eq.
a sluice (after Henry45).
(4.2) may still be used if the value
of Cn is made to include this effect. Fig. 4.10 shows drowned flow
under a sluice and from this figure the downstream depth of submergence is defined as dafy. Graphical values of Cn are given in
Fig. 4.11 in which Cn is plotted against the upstream depth ratio
d1fy. A family of curves corresponding to different values of the

Control and Measurement of Open Channel Flow

87

downstream depth ratio give values of Cn for the case of drowned

flow. A discussion of the limitation of these results is given by
Ippen. 46
4.3 Flow measurement
4.3.1 Thin-plate weirs. These structures are suitable for measuring
the discharge in small streams and artificial channels and in laboratory flumes. They are too susceptible to damage by water-carried
debris for satisfactory use in large streams or rivers. When used
under the correct conditions they provide the most accurate means available
for measuring open channel discharge.
Fig. 4.12 shows the essentials of a
thin-plate weir: this one consisting of
a rectangular opening or "notch" cut
symmetrically in a smooth thin metal
FIG. 4.12 Rectangular
plate fixed at right-angles across a
thin-plate weir.
channel. The important features of its
construction and installation can best be summarised as follows:
(I) Thin-plate weirs should be constructed from a non-corrosive

(2)
(3)

(4)

(5)

metal such as bronze or stainless steel. The plate must be

thick enough to withstand normal use without damage
(5 mm-10 mm).
The crest of the notch must be accurately horizontal and the
sides vertical.
The edge surface of the notch must consist of accurately
machined plane surfaces perpendicular to the upstream face
and meeting it at a square corner (see Fig. 4.13). The edge
surface must not exceed 2 mm (008 in.) in width and any
excess plate thickness must be accommodated in a 45 chamfered surface at the rear.
Any sediment deposited against the upstream face of the weir
must be removed regularly. An accumulation of sediment in
this region will affect the head-discharge relationship for the
weir.
The water level on the downstream face of the weir must
always be below the crest level of the notch (free flow) otherwise a correction will have to be made to the discharge
relationship.

88

Flow in Channels

(6) The head above the weir crest should be measured a sufficient
distance upstream of the weir to avoid the region of surface
drawdown (see Fig. 4.14). This measurement should preferably be made in a stilling well, or similar structure, having free
hydraulic access to the upstream channel.
These requirements are set out in greater detail, as are all other
aspects of thin-plate weirs, in B.S. 3680, Part 4A, 1965.47
The form of notch illustrated
2mm
in Fig. 4.12, to which condition
---1 ~" Edge surfoce" of nolch
(2) above specifically applies, is
Sharp---... 1
Champfered
== squore
termed a rectangular notch (or
rectangular thin-plate weir). The
description "sharp-edged" used
in many texts is only relative (see
section 4.3.2) and could be misleading in the light of the details
FIG. 4.13 Cross-section through
given in Fig. 4.13. Depending
edge of thin-plate weir.
upon the width of the notch in
relation to the width of the channel in which it is fitted three
possible variations of the rectangular notch can be distinguished.

;.~:~~I

,. . .,

(a) full width weir: This is also referred to as a suppressed weir,

since the absence of any lateral restriction on the flow through
the notch suppresses the normal tendency for the two vertical
sides of the water jet or nappe to contract further after springing free from the plate.
L __
However, normal contracSurface drawdown
tion of the flow occurs
---h
approaching weir
along the lower edge of the
Ven ilation
nappe as illustrated in Fig.
4.14. Such flow contraction occurs when water is . . . . . . . . . . . . . . . . . . . :::::: . . :. .:.............. . . . . . .:. . . . . . . . . . . . . . . .:::
discharged from any
FIG. 4.14 Flow over full-width
thin-plate weir.
sharp-edged orifice (see
"sluices", section 4.2.3.3)
and is due to the inability of the bounding streamline (on
pressure and energy considerations) to turn at a sharp
corner with zero radius of curvature. In a full width weir the
air space below the nappe is isolated from the atmosphere and
the trapped air will be slowly entrained and removed by the

Control and Measurement of Open Channel Flow

89

lower surface of the falling jet. Unless ventilation is provided,

the air will finally be removed from this space and the discharging jet will adhere to the downstream face of the weir, a
condition referred to as a clinging nappe. If the weir is to be
used for accurate discharge measurements this air space must
be maintained at atmospheric pressure by adequate ventilation.
(b) fully contracted weirs: The side and bottom contractions of
the nappe are fully developed if the bed and walls of the
channel are sufficiently remote from the notch for the channel
boundaries to have no significant influence on the contractions
of the nappe. B.S. 3680 gives the relevant limits to these
conditions as

B-L

- 2 - {: 2h' and p {: 2h'

in which B is the width of the channel
L the width of the notch
h the head of the maximum discharge above the
crest (Fig. 4.14)
and

p the height of the crest above the channel bed.

Further limits (minimum values) are also given to the above

dimensions. The notch illustrated in Fig. 4.12 may be considered to fall within this category.
(c) partially contracted weirs: This category covers weirs excluded
by their dimensions from both (a) and (b) above. They must,
however, comply with limits laid down to the minimum
permissible values of head, discharge, etc.
The discharge equation for full width weirs, whose derivation is
given by Webber4s and most standard texts, is

Q = J-\/(2g)CnL. he312

(4.4)

in which Cn, the coefficient of discharge, is given by the Rehbock

equation:
h
CD = 0602 + 0083 (4.5)
p
and the effective head,
he= h

+ 00012m

(4.6)

90

Flow in Channels

The discharge equation for fully contracted weirs is as above

(Eq. 4.4) but the Hamilton-Smith equation for Cn should be used to
take into account the side contraction of the nappe:

Cn

= 0616 (1- 01 ~)

(4.7)

A suitable equation for Cn applicable to partially contracted weirs

is given in B.S. 3680 in which the question of accuracy and errors
is also discussed in detail. The British Standard quotes a figure
of 1 % for the accuracy limits to be expected in the computation

.....-,
h

.."

- 1--- "

FIG. 4.15 Triangular thin-plate weir or V-notch.

of Cn from the above formulae. It then discusses the method of

allowing for estimated errors in the measurement of h, p, L and Bon
the expected accuracy of the discharge value so derived.
Fig. 4.15 shows a triangular thin-plate weir or V-notch. The
same conditions (see above) relating to construction and installation
apply to this form of thin-plate weir as to the rectangular notch but
include the requirement that the bisector of the angle formed between
the two edges of the notch shall be vertical. The discharge equation
for the V-notch is

Q=

f)
8
y(2g)Cn . tan- h5/2
2
15

(4.8)

for which B.S. 3680 tabulates values for the discharge coefficient Cn,
Cn is
for common values of the notch angle (90, 45 and 22!
found to lie normally in the range 058-061.
The advantage that the V-notch possesses over the rectangular
notch is its ability to measure a wide range of discharges with higher
accuracy. The reasons for this can be more easily understood by
considering aspects of the following problem. A thin-plate weir is
required to measure a varying flow of which the maximum discharge
is estimated to be ten times the minimum. If a rectangular notch is
0

).

Control and Measurement of Open Channel Flow

91

used it can be shown that the maximum value of the head hmax =
46 hmin, the head corresponding to the minimum discharge. However if a V-notch is used this relationship becomes hmax = 25 hmin
Thus, for a convenient value of hmax the V-notch gives a larger value
of hmtn than the rectangular notch with a corresponding increase in
the accuracy with which it can be measured.
4.3.2 Broad-crested weirs. A weir behaves as broad-crested if the
width of the crest exceeds approximately half the head. The nappe
no longer springs clear of the crest at the upstream corner but,
instead, parallel flow is established over the crest as shown in Fig.
4.16. If the upstream edge of the weir
is sharp, as in this figure, contracted
flow still occurs there but the small
space below the low bounding stream- Flow seporoti on
line is filled with turbulent water and ot sharp corner
frequently with air bubbles. The upper Fro. 4.16 Flow over a broadboundary (the free surface) starts to draw crested weir with sharp corners.
down some distance upstream from
this corner and accordingly the head h should always be measured
at a point in excess of 25h from the upstream face of the weir.
The discharge equation commonly used for broad-crested weirs is

Q = CLh 312

(4.9)

in which L, the width of the weir, is normally the total length of the
structure. The discharge coefficient C varies considerably with the
breadth of the crest and the head h, tables of values being given by
King and Brater. 49 Values of C usually lie between 30 and 332
approaching the higher limit as h increases (h > 06 m).
Parallel flow over a horizontal weir crest is seldom realised, but
if the crest is given a slight downward slope in the direction of flow
(about 1 in 20) and if the upstream corner is rounded to prevent
flow separation occurring there, the depth of flow on the crest is
found to be sensibly constant. It can be shown (see section 2.2.3)
that the depth of flow on the crest is equal to the critical depth for
that discharge and therefore equal to two-thirds of the specific energy,
referred to the crest level as datum, and from this fundamental
relationship the following discharge equation can be deduced (in
S.I. units).

Q = 171L(h + h,)312

(4.10)

92

Flow in Channels

in which hr is the kinetic energy head of the approaching flow. Values

of C in Eq. (4. 9) under conditions of rounded upstream edge and low
approach velocity are found to approach the theoretical limit of 171
but not to exceed it, thus supporting the validity of the "critical
depth" analysis for flow over this type of weir.
Broad-crested weirs, being robust, are frequently employed to
control and measure river flow. One of their advantages as measuring
devices is their insensitivity to downstream conditions, including the
state of ventilation of the nappe. This in practice is never ventilated
purposely and under low flows the nappe always adheres to the
downstream face of the weir. The validity of the discharge equation
based on a single upstream head measurement holds until the downstream water level rises above the critical depth on the crest.
4.3.3 Throated flumes. A throated flume consists essentially of a
constriction in a channel which produces an increase in velocity and
Total energy line!

- -- --r- --; - -r-- -~ -- -.

I

d1

::

~
--~---- -

I;

,
I

- - - - - - - -
I

'

Hydraulic jump :

. ..-..:::::::::::::: .:.::::::.:::.-::.:::.: : ... .:::::::::::::::::.;:.;.;-:-:-::-:-::-::-:::;.;.::::-:::

.::

(b) Drowned fl ow (Venluri flume)

Throat

'

(c) Plan view

FIG. 4.17 Flow through a throated flume.

a corresponding decrease in water surface level. Its principal

advantage over comparable channel flow measurement structures is
the relatively small increase in upstream water level, or afflux, that

Control and Measurement of Open Channel Flow

93

it produces and also a similarly small energy loss involved in the

flow through it. The geometry of the structure is critical to the size
of the energy loss and has consequently become standardised. The
standard throated flume (Fig. 4.17) consists of a short bellmouthed
contraction followed by a parallel throat and then a long flared
expansion section. If energy recovery in the expansion section is of
little importance, the cost of the structure can be reduced by cutting
short this section. The flume illustrated has a flat bottom and is
particularly suitable for flows carrying quantities of detritus, both
buoyant and heavy. Where the detritus or sediment load is not
significant the bottom of the flume is frequently designed with a
hump at the throat.
If the combination of discharge, flume dimensions and downstream
level permit, the flow will pass through the critical depth at the throat
and a hydraulic jump (standing wave) will form in the expansion
section to restore subcritical flow. When a throated flume operates
in this manner it is frequently referred to as a standing wave flume
(see Fig. 4.17(a)). In this case a single measurement of depth d1
(upstream) will suffice to determine the discharge. The fundamental
equation is essentially similar to that for the broad-crested weir and
is
(4.11)

but is modified in practice to allow the removal of the velocity of

approach term hr (incorporated in a coefficient of velocity Cv) and
to take into account the energy losses and other effects in the contraction section (by introducing a coefficient of discharge CD). The
discharge equation then becomes
(4.12)
The proportions of the standard throated flume as well as tabulated
values of Cv and CD are given in B.S. 3680 Part 4A.47 Values of
the coefficients are such that a simplified discharge relationship
(4.13)
can be used where only an approximate value of the discharge is
required.
Where conditions are such that the critical depth is not reached
in the flume (the presence or absence of a hydraulic jump is a sure
guide on this point) the discharge can still be obtained if depth

94

Flow in Channels

measurements are made at two sections, dt and d2 in Fig. 4.17(b).

The flow at the throat is now said to be "drowned" and the flume
operating in this condition is sometimes referred to as a "venturi
flume." A smooth stationary depression is formed in the water
surface through the flume with a correspondingly low energy loss
and afflux. The discharge equation for a drowned throated flume
can be derived by considering the specific energy at a section upstream of the flume (b1, d1) and also at the throat (b2, d2). Introducing a coefficient of discharge Cn (having a value usually in the
range 096-099) to account for the effects of fluid friction, the
discharge.equation is
Q = Cny(2g)b2d2 l(d _ d)
y(l + m2) v 1
2

(4.14)

. of fl ow areas wh.ICh , of course, IS

. a
. wh.ICh m = bb2dd2 , the ratiO
m
1 1
variable.
On the whole it is less satisfactory to operate a throated flume in
a drowned condition because the term (d1 - d2) usually represents
a small difference whilst including the cumulative percentage error
resulting from two physical measurements. In addition, the flow in
the throat tends to be unsteady, leading to fluctuations in the position
of the section of minimum depth. However, the above relationship
(Eq. 4.14) will be found useful if the situation cannot be avoided.
Throated flumes designed specifically for channels of trapezoidal
cross-section are the subject of a report by Ackers and Harrisonso,
whilst two special flume designs deserve mention. The Crump weir 51
is a weir of triangular section having an upstream facing slope of I
in 2 and a downstream one of I in 5 and is receiving increasing
attention in the U.K., while the Parshall flume5 2 consisting of a
throated flume with a complex bottom form has been in use in the
U.S.A. for many years.
4.3.4 River gauging. When continuous measurement of the discharge
is required on a river the method adopted will depend largely on
considerations of cost. For small rivers and streams some form of
critical depth meter as described above, used in conjunction with
continuous water level recording, will be the most satisfactory choice,
but for medium or large sized rivers the cost of such a structure is
normally prohibitive. Under these circumstances a gauging station

Control and Measurement of Open Channel Flow

95

is normally established at a suitable point on the river in the following

manner.
(I) A site is chosen near the downstream end of a fairly straight

reach of river where the cross-section is as uniform as possible.

Here the water level is measured and recorded continuously
by some suitable means.
(2) This record of water level, or stage, is related to the discharge
in the river by calibration based on current meter traverses of
the river cross-section at the gauging station.
Calibration normally consists of measuring the discharge by the
velocity area method for a number of different stages, thus building
up a relationship between stage and discharge for that particular
gauging station. The stage-discharge relationship so established
invariably has to be extrapolated to obtain the discharge in times of
flood for the following reasons. Firstly, it is difficult to carry out
precise field work of the quality required for this method during bad
or even violent weather and whilst working on a river in flood.
Secondly, such discharges are seldom anticipated by more than a
few hours and this is frequently a period when communication and
transport is difficult if not impossible.
In order to be certain of the continuing validity of a stage-discharge
relationship it is necessary to make at least a few check discharge
measurements at regular intervals. It is also necessary to re-survey
the cross-section from time to time to ensure that it is not changing
significantly due to deposition or erosion of the bed. A further
complication to river gauging is the dependence of the stagedischarge relationship on the slope of the water surface (see Mitchell53). This will differ from the average bed slope of the reach if
the flow is unsteady (rising or falling stage). It is therefore unwise
to place too much reliance on the simple stage-discharge relationship
when the water level is clearly rising or falling rapidly.
The most comprehensive information on the selection, calibration and operation of river gauging stations, together with an analysis
of sources of error and their effects on the results is given in B.S.
3680: Part 3 : 1964.54
4.3.5 Velocity area methods (current meters). The velocity in a river
can be measured at a particular point by an instrument known as a
current meter. These instruments consist essentially of a helical or

96

Flow in Channels

cup-vaned wheel which the flowing water turns at a speed proportional to the water velocity. A mechanical make-and-break attachment on the rotating spindle of the instrument triggers an electrical
pulse that can be counted or recorded remotely on the river bank or
in a boat. The velocity measured is therefore the average over a
short period of time. Common patterns of current meter are made
by Ott in Western Germany, by Hilger and Watts in the U.K. and
by Price in the U.S.A. The two former are of the helix type while
the last is of the cup-vaned wheel form. All current meters require
+- Location of current meter measurement points

FIG. 4.18 Velocity-area method of discharge measurement.

careful and frequent calibration either in a towing tank or against

a standard to ensure accuracy is maintained.
In order to determine the discharge at a gauging station for a
particular stage, a traverse is made with a current meter and a drawing
of the cross-section constructed showing the velocity distribution.
Fig. 4.18 sho~s such a diagram in which the cross-section is divided
up into a number of vertical strips for each of which a vertical
velocity distribution curve is drawn based on a line of current meter
measurements. One such curve, for the vertical xx, is superimposed
on the cross-section in Fig. 4.18 and the area under this curve
(shaded), divided by the depth x . . . x, gives the assumed mean
velocity for the strip abed. The area of the strip abed can be determined by planimetering, or some other method, and multiplied by
this mean velocity to give the discharge through the strip. The total
discharge of the river is then the sum of these strip discharges over
the full cross-section.
In an alternative velocity area method the current meter measurements are used to construct a velocity distribution diagram of the
type shown in Fig. 1.5. Interpolation between the point velocity
values enables isovels to be drawn for convenient increments of

Control and Measurement of Open Channel Flow

97

velocity. The average velocity in the area enclosed by adjacent

isovels is now taken to be the arithmetic mean of the two isovel
values and, when multiplied by the corresponding scale area, gives the
discharge through this portion of the cross-section. The total
discharge can be obtained as before by the summation of these
partial discharges.
Both the above velocity area methods for determining the discharge in a river from current meter traverses are described in detail
in B.S. 3680: Part 3. 54

5 Flow 1n Erodible Material

5.1 Introduction

In this chapter, consideration is given to the relationship which

exists between the flow in a channel and its boundaries. This
relationship is only static, and therefore of no interest here, when the
channel boundaries are non-erodible under all conditions of flow
(concrete lined channels and canals) and when in addition, no
water-borne sediment is introduced into the channel. In natural
rivers and unlined canals the channel boundary is always erodible to
a greater or lesser extent and sediment is carried into the channel
both from the bed upstream and from the surrounding land. The
flow in such a channel is always adjusting itself to the presence of
this movable material whose behaviour in tum depends upon the
action of the flowing water upon it. The whole will therefore form a
single, highly complex, flow phenomenon: part fluid, part solid, in a
channel of variable geometry.
Because flow in erodible channels is so complex, the theoretical
models used to help understanding of the fundamental mechanics of
the process are still inadequate. There is a large gap between the
river engineer's needs and the tools available to him.
Any particular river will always attempt to adjust its conditions
(plan, slopes, depths, rate of sediment movement) in order to reach
a state of equilibrium or balance of forces. However, the state of
equilibrium of a river depends in tum upon the discharge, slope and
other significant factors and it follows that any river which is not
artificially controlled will always be in a state of change seeking, but
seldom finding, equilibrium. The time scale of river changes varies
considerably, depending upon many factors. Whilst one river may
98

99

Flow in Erodible Material

show a major change of course after the recession of a single flood,

another may undergo channel movements which can only be detected
by comparing different maps of the relevant region covering a time
span of fifty or a hundred years.
The complexity of this subject has resulted in the past in its being
treated separately from rigid-boundary channel flow. It is the
author's intention in the present chapter to outline the subject and
to give references to significant sources on specific points.
5.2 Sources and types of sediment
Any solid material that is, or can be, carried by moving water and
which is heavier than water is termed a sediment. The most important characteristic of a sediment is its size, which can vary enormously.
Under extreme flood conditions boulders weighing tens of tons can
be moved down a steep water course while at the other end of the
scale a particle is classified as sediment if it will settle in still water at
a significant rate. In reservoirs and lakes such very fine material may
have months in which to reach the bottom which indicates the order
of magnitude of the settling velocities concerned.
Sediment is normally of natural origin, being the result of the
weathering of rocks and soil that constitute the surface of the earth.
TABLE

Upper
limit

mm

Grade
Class

0002

0006

002

5.1. Sediment classification by size

(from B.S. 1377: 196155)
006

02

06

Fine Med. Coarse Fine Med Coarse

Clay

Silt

Sand

20

60

Fine Med. Coarse

Gravel

200

>200

-----

Cobbles Boulders

Under special circumstances man-produced sediments may be

significant in streams as, for example, the coal dust and mineral
particles frequently found in streams running through coal-mining
and mineral processing areas. Sediments are classified by size as
shown in Table 5.1.
The chemical composition of natural sediments is closely linked
with size. Gravel grades and larger represent fairly closely the
parent rock, whilst most grades of sand consist principally of quartz

100

Flow in Channels

(silica) and represent the different stages in the gradual breaking

down of the quartz crystals in granite and allied igneous rocks.
Modern sand sediments may alternatively have passed through a
secondary geological stage as sandstone or quartzite, or through a
geologically younger one forming alluvial or glacial deposits. Clay
particles represent the final stage in the weathering of other rock
constituents, notably felspars, while silts are normally mixed in
character, possessing many of the properties, both physical and
chemical, of sand and clay.
Other important characteristics of sediments include shape,
settling velocity, density and electro-chemical bonds (including
flocculation behaviour) some of which are related to each other and
the size of the particle.

5.3 Modes of sediment transport

Sediments are moved by flowing water in two principal ways. Sands,
gravels and larger particles move principally along the bed of a
channel-bed-movement-and form what is termed the bed load of a
river, while silt and clay particles are maintained in suspension, being
distributed more or less uniformly throughout the flowing water.
In the latter case it is the dispersive character of turbulence that
counteracts, on balance, the natural tendency of these particles to
settle to the river bed and so the upper size limit to the suspended
load in any river will depend upon the degree of turbulence of the
flow as well as on the sediment available.
The transition from suspension to bed-movement is not a sharp
one and averagely fine sand, for example, will be transported
alternately by the two mechanisms. Bagnold 56 , as a result of studies
into the movement of desert sand, identified an intermediate mode of
transport which he called saltation. Sand grains are seen to rise
into the air under the action of shear stresses and to fall back to the
surface at a point some distance downwind where their momentum
is transferred to a number of fresh grains in a manner analogous to
a splash. In this way it can be seen that the process will build up
as a chain reaction. Many authors have since proposed saltation as
an important mode of transport for small grains in water but there
is some doubt whether the process being observed was saltation in
the strict sense of the word, as the relatively high density of the water
has a strong damping effect on the impact and momentum transfer
between the sand grains.

Flow in Erodible Material

101

5.4 Different approaches to the problem of sediment transport

The problem of sediment transport by flowing water in rivers and
canals has been of interest to engineers for many centuries but the
earliest published scientific work on the subject is due to a Frenchman, Dupuit57 and dated 1865. From this beginning there quickly
developed a "school" of experimenters, principally French, German
and American, who tackled the fundamental problem of the movement of sediment grains by flowing water.
Shortly after the commencement of this fundamentalist school in
Europe, Anglo-Indian engineers, tackling the enormous task of
irrigating the Punjab and other large areas oflndia, began to assemble
a mass of data from canals they were operating and which were found
to behave satisfactorily as far as sediment transport is concerned.
Such a canal or river, in which neither scour nor deposition of sediment occurs, when considered over a long period of time, is said to
be in regime. Even a regime channel will normally undergo seasonal
variations in discharge, slope and sediment load so that the bed will
scour at certain parts of the seasonal cycle and deposit sediment at
others. It is only when considered over a sufficient length of time
that it is seen to be in regime.
The analysis of data collected from a large number of regime
channels enabled a set of empirical equations to be established,
relating channel slope, cross-section and discharge with sediment
type and load. These empirical equations were then used to design
further canals and modified in the light of the experience so gained.
Blench5 8 describes the development and use of the empirical school
formulae and makes contributions of his own to the regime approach
to river channel behaviour.
The problem of sediment transport is so complex that so far the
most useful contribution to engineering practice has been made by
the empirical school but it would be incorrect to imagine that there
is not a great deal of overlapping and cross-fertilisation of ideas
between the two schools of thought (see section 5.11 ).
5.5 Sediment movement as bed load
5.5.1 River bed form. The form of the bed in sand-carrying channels
has been well explored in laboratory flumes and canals and recent
surveys of sandy rivers using echo-sounding equipment indicate

102

Flow in Channels

similar forms. When the mean velocity in a channel is less than the
critical velocity (see section 2.2.3) the bed form consists of dunes.*
Fig. 5.1 shows a section through a typical sand dune in which
erosion takes place on the gently sloping upstream face and deposition on the downstream face which lies at the angle of repose for
that particular sediment under water. This results in a dune moving
downstream at a velocity that is small compared with the mean water
velocity. The average velocity of the surface sand grains in the
downstream direction is approximately the same as the propagation
velocity of the dunes.
Sand dunes and their intervening troughs normally occupy the
available bed surface but are usually arranged in an irregular or

FIG.

5.1 Section through sand ripple or dune formed under water.

random manner. This leads to considerable variation in the plan

form of individual dunes although they all exhibit the characteristic
profile shown in Fig. 5.1 when sectioned parallel to the local direction of flow. The size of river dunes varies widely but a fairly close
correlation can be observed with the depth of water, proportionately
large dunes having been detected in large rivers, for example in the
Mississippi.
When the mean water velocity exceeds the critical velocity Vc the
dune form of movement is replaced by uniform movement of the
surface over a flat river bed. When the mean velocity exceeds
25 vc a third phase of bed movement can be observed in which the
bed adopts a symmetrical sine curve form of relatively low amplitude.
These sand waves are known as anti dunes and can be observed to
move slowly upstream. The water surface normally reflects the bed
form with a train of standing waves.
Bed forms in gravel rivers are less well investigated but probably

* The structure under discussion is referred to alternatively as a ripple or dune.

The distinction is largely subjective and normally an indication of size. Hence
ripples can be observed occasionally superimposed on dunes. Whatever the name
used the fundamental mechanics of formation and movement is the same.

Flow in Erodible Material

103

show general movement over large areas of the bed. Dunes and
other local structures are poorly developed. A river flowing through
consolidated clay deposits will erode its bed and banks if the velocity
of the water is sufficiently high (see section 5.11) but the eroded
material is carried away in suspension and not normally deposited
under river channel conditions.
5.5.2 Bed load formulae. Many formulae have been developed to
give the rate of bed movement in an alluvial channel. The earliest
is that of du Boys (1879) which is based on the assumption that
movement of the bed occurs in a number of thin discreet superimposed horizontal layers and that their velocity reduces linearly
to zero at a depth below the sand surface depending upon the fluid
shear stress at the bed. This shear stress is a function of the water
surface slope (see section 1.3) and when the bed is just on the point
of moving this shear stress is referred to as the critical tractive stress.
Du Boy's formula is very simple, as is the physical model on which
it is founded, but it has been used in various modified forms for a
long time. The following form applicable to wide channels is
derived from a modification by Straub. 59
g 8 = 1ps7/5 (

.!!....)

6/5

15

q6f5

(in British units)

(5.1)

in which g 8 is the weight of sediment passing a fixed point per

second per unit channel width.

and

s the
n the
q the
1p the
the

surface slope of the channel

Manning's roughness number
discharge per unit width
"sediment characteristic" given approximately by
following empirical formula:
111,000
1p

dm3/4

(5.2)

in which dm is the mean sediment diameter in mm.

Other important bed load formulae are given by Kalinske 60 based
on the turbulent fluctuations in the water velocity at the bed, and by
Einstein61 who uses a statistical function to describe the probability
of a particle being moved as a basis for his formula.

Flow in Channels

104

5.6 Suspended sediment movement

Although the problem of suspended sediment movement by rivers

has received less attention from investigators than that of bed
movement, it is responsible for far larger quantities of sediment being
transported than occurs by bed movement. Since it is the dispersive
nature of the turbulent river flow that maintains the sediment in
suspension it follows that it is the settling velocity of the sediment
that is its most important characteristic. Consequently the fine sediment
Water surface
fractions whose settling velocities are
small in comparison with the magnitude
0
\
...--- Fine silt
\
of the turbulence are found almost
\
uniformly distributed throughout the
'\
depth of flow as is shown by the distri', /
Fine sand bution curve for fine silt in Fig. 5.2.
Material of this size or finer is not likely
to be deposited until it reaches the
relatively still waters of a lake, a
Sediment concentration
reservoir or the sea and is sometimes
FIG. 5.2 Representative
referred to as the wash load of a river.
vertical distribution curves for
Coarser material is carried in sussuspended sediment in rivers.
pension under rather different conditions. The random distribution and strength of upcurrents due
to the turbulence results in such particles sometimes being lifted and
at other times settling at comparable velocities. This mechanism,
when equilibrium conditions are reached, gives rise to a vertical
distribution of sediment of the type shown in Fig. 5.2 for fine sand
in which the concentration of suspended sediment increases greatly
as the bed of the channel is approached. In this case a slight reduction of turbulence at a point along a river's course might lead to
large scale deposition occurring in that region. As has already been
stated, this process merges indefinably with bed movement as the
size of the sediment considered increases or the level of turbulence
decreases.
Mathematical analysis, based on boundary layer theory, has been
attempted by many with varying degrees of success but the only
satisfactory means at present of determining suspended sediment
load is by mechanical sampling. This is both easier and more
satisfactory than the direct measurement of bed load in the field

it
..0

'

Flow in Erodible Material

105

and is normally carried out in conjunction with current meter

traversing. The following equation, due to Straub, 59 uses two
measurements in a vertical to give an average concentration and
hence the load per unit width.
S = (iso-sa + fso-2a)q

(5.3)

is the weight of suspended sediment passing a fixed

point per unit time per unit width,
so-sa the weight of suspended sediment per unit volume
from sampler, taken at 08 depth from surface,
so-21.1 the weight of suspended sediment per unit volume
from sampler, taken at 02 depth from surface,
the discharge per unit width.
q

in which S

This method can be used in conjunction with the two-point method

for determining the mean velocity in a vertical and hence q (see
section 1 6). Thus Eq. (5.3) becomes
d
S = 16 (3so-sa + 5so-2a)(vo-sa

+ vo-2a)

(5.4)

in which vo-sa and vo-2a are the corresponding measured velocities

and dis the depth of flow. More recently however point sampling of
suspended sediments is being replaced by the use of integrating
samplers that measure the average concentration in a vertical
directly.
It has been found that at a particular station on a river a statistical
relationship exists between the discharge and the suspended sediment
load. The resulting equation is entirely empirical and its accuracy
depends upon the amount of data available for its compilation.
While the day-to-day prediction given by such an equation is frequently grossly in error due to variations in a large number of
important factors, the values given for monthly or even longer
periods, based on the discharge record for that period, are within
acceptable limits of accuracy. This so-called silt-discharge curve is
of exponential form
(5.5)

106

Flow in Channels

in which S is the suspended sediment load passing the station

(total mass/sec),
C the constant varying considerably from one river to
another and from one station to another,
Q the river discharge,
and
x the exponent, normally between 20 and 23 but falling
in value at a particular station during periods of extreme
flood.
5.7 Total sediment load

Since it has been frequently stressed that the division of sediment

transport into suspended load and bed load is to a certain extent an
artificial one leading to many difficulties in defining limits, it is not
surprising that attempts have been made to develop an expression
for the total sediment load of a stream. Garde and Albertson62 in
discussing a paper on this subject by Laursen63 , give three possible
approaches:
(1) A rational solution based on modern fluid mechanics. Little
progress has been made as yet in this direction for lack of
adequate fundamental knowledge of the problem.
(2) Additive methods in which an expression developed for the
bed load is used in conjunction with one for the suspended
load. Einstein's approach is of this type and has been developed and used with some success by Colby and Hembree. 64
(3) Empirical relationships have been developed based on dimensional analysis and stream data. (See references 62 and 63.)
Both the additive and empirical approaches are used to a limited
extent but in addition to the normal limitations of such methods a
further variable seems to be of primary importance and has so far
proved difficult to handle. This is the form of the stream bed,
discussed in section 5.5.1. Channels in which bed load is of importance will normally have sand grades forming the majority if not all
of the transportable bed material and the type and scale of the
resulting sand dune structure is found to vary in a complex manner.
Given similarity in other hydraulic conditions, the bed load will be
much higher for a more nearly plane bed compared with one on
which dunes or ripples are well developed. In addition suspended
load formulae, when used, require some value of the sediment

Flow in Erodible Material

107

concentration which is normally taken close to the bed (Einstein's

method) and this also is made more difficult by the complexity of the
bed form.
In appreciating the unsatisfactory state of sediment transport
knowledge, it must be realised that the attempts at solutions outlined
above represent only the first tentative steps towards a solution of
the problem. A complete solution must await a fuller understanding
of the relationship between bed forms and the water flow in the
vicinity, particularly with regard to the turbulence structure and
velocity distribution.

5.8 Regime behaviour of channels

A channel formed in erodible material will only reach equilibrium
conditions when a balance has been established between the channel
discharge, cross-section, slope and sediment load. A channel that
has established this equilibrium is said to be in regime or graded.
One that is not in regime but which is actively changing its slope,
width and depth towards regime conditions is however usually described as a regime channel and given sufficient time free from external interference will approach its regime conditions exponentially.
Such natural processes lie within the scope of geomorphology and
are of interest to geographers and geologists as well as hydraulic
engineers. The study of regime conditions in channels was put on a
quantitative and essentially empirical footing by the Anglo-Indian
irrigation engineers as stated in section 5.4 and their basic regime
equations derived from canal data have since been modified slightly
by U.S. engineers to make them applicable to natural river channels
as well. There has always been some controversy about the regime
approach to sediment transport and river morphology but it is still
widely supported by active engineers.
Since canals are laid to predetermined routes, usually straight,
the meander pattern of alluvial rivers did not initially have to be
considered. The principal variables to be considered are then the
slope, depth and width of the channel, which can be expressed as
three equations in terms of the dominant discharge and the equilibrium conditions of the flow (sediment transport conditions). The
dominant discharge is that which is most significant in determining
the dimensions of a stream channel and is usually taken to be the
bank-full discharge in streams where the flow is variable.

108

Flow in Channels

The development of the regime equations which are outlined

below, is given by Blench5B and many earlier writers. The first
relates the slope to the channel flow resistance in a manner analogous
to that used for pipe flow.

Q175

(...L)

= --:---..---363

vo25

(5.6)

B2d275

in which s is the channel slope (water surface slope)

Q the dominant discharge
v the kinematic viscosity of the water
B the breadth of the channel
and
d the depth (in any consistent system of units).
The depth equation is developed from the empirical relationship
between the sediment load and the Froude number of the flow,
(5.7)

in S.I. units, except that dm is the median size of the sediment in mm.
The breadth equation is wholly empirical in origin, having no obvious
physical basis, and in S.I. units is

B = 483Qt

(5.8)

In applying the above regime equations to natural rivers the constant

in the breadth equation (Eq. 5.8) has been found to vary for individual rivers from the value of 483 derived by Lacey from canal data.
In conclusion certain properties of regime channels, some of them
implicit in the regime equations, can be summarised qualitatively
in the following list. This has been adapted from material published
by Blench58 and Morris. 65
(1) Channel banks are always formed from the smallest sediment
fraction present. Banks formed from silt and clay particles
will be clearly defined while those consisting principally of
sand or gravel will be poorly defined.
(2) Channel width in a natural channel is largely determined by
the discharge at bank-full stage.

Flow in Erodible Material

109

(3) The relative width (with respect to depth) of a regime channel

increases with the dominant discharge. It also increases with
the mean sediment size and quantity present.
(4) The slope of a stream decreases as the discharge increases and
as the bed load particle size and quantity decrease.
(5) The slope of a stream is so dependent on its regime conditions
of load and discharge that if the stream is dammed the bed
level upstream will be raised by deposition until the regime
slope is re-established.
(6) Sediment load is determined not only by bed shear stress and
grain size but also by the form of the bed.
(7) The hydraulic roughness of the bed is affected both by the
grain size and by the bed form, which in tum is affected by the
discharge, slope, etc.
(8) Even ungraded streams behave so nearly as do graded streams
that only extreme conditions of non-equilibrium can be
detected by quantitative measurements over short periods.
The behaviour of natural streams in determining their route or
plan will be discussed in the following section.

5.9 River bends and meandering

Only under exceptional circumstances do river channels follow a
straight course. This natural inclination to sinuosity becomes much
more marked in the lower reaches where the "mature" river crosses a
-------Lm

FIG.

5.3 Sine form meander pattern formed by a channel in sand.

gently sloping wide alluvial plain before entering the sea. It is in this
region that downcutting of the bed is reduced and a greater proportion of the stream energy is directed towards eroding the banks.
When such a channel becomes significantly tortuous it is referred to
as meandering.

110

Flow in Channels

Important laboratory studies of the meandering process were

carried out by Friedkin66 who demonstrated that small scale laboratory streams show the same strong tendency to meander as do rivers.
Under such ideal conditions a meandering stream course consists of
alternate left-hand and right-hand bends of sine wave form (Fig.
5.3), the point of inflection being used normally to define the junction
between consecutive bends. Centrifugal action in the bends results
in a distortion of the velocity distribution in the cross-section so
that the point of maximum velocity moves close to the outer or
concave bank. Thus erosion of the
concave river bank is active and results
in the development or lengthening of
the bend perpendicular to and away
from the general axis of the river. This
process does not proceed indefinitely
as a limit is reached to the width of
the meander Wm (see Fig. 5.3), at which
short channels termed chutes are de(o)
"(b)
FIG. 5.4 Development of a veloped across the inner bank sediment
cut-off and an ox-bow lake. deposits, at high flows. These chutes
form because the resistance to flow
along such paths is less than that around the outside of the bend. At
low flows the stream normally returns to the main channel. Chute
development not only reduces erosion in the vicinity of the bend
extremity but also concentrates erosive activity on a region further
downstream on the concave bank where the chute flow impinges.
This ultimately results in the general movement of the full developed
meander system in the axial downstream direction.
When the bank material of a meandering river is of a cohesive
nature (i.e. contains silt or clay) the resulting bends approximate in
plan to a segment of a circle whilst rivers meandering through
uncohesive sandy material develop bends of the sine form characteristic of laboratory test conditions. Because of random variations in the erodibility of alluvial plain deposits river banks offer a
variable resistance to the erosive mechanism of meandering which is
responsible for the complex and irregular form of most natural
meander patterns. One consequence of variable erodibility is the
formation of cut-offs as illustrated in Fig. 5.4. An area of harder
sediment arrests the normal seaward movement of one limb of a
meander bend and the adjacent upstream limb then closes up on it

Flow in Erodible Material

111

until the separating strip of land is broken through. Normally flow

starts across this neck of land during flood conditions and, because
of the steep slope resulting from the much shortened course, a new
channel is quickly eroded. The old meander loop will now be
particularly susceptible to deposition and will soon be isolated under
normal flows to form an ox-bow lake. These lakes slowly silt up
due to the combined action of vegetation and the regular inflow of
suspended sediment from the river during floods.
Although river meanders have been extensively studied and
measured all over the world their basic mechanics is still only
partially understood. At first empirical formulae were proposed,
relating certain dominant physical characteristics. The Inglis
meander formulae were developed from measurements of Indian
rivers and give the meander length and breadth, as defined in Fig.
5.3, in terms of the maximum probable discharge, Qmax
Lm = CL(Qmax)t
Wm = Cw(Qmax)t

(5.9)
(5.10)

In S.l. units the constant CL ranges in value between 33 and 76 with

a representative "average" value of 50, while the constant Cw is of
the order of magnitude of iCL but varies widely for different types
of rivers. The "maximum probable discharge" corresponds to that
discharge which has a probable recurrence interval of 100 years,
called the 100-year flood.
The measurement of meander wavelength and other characteristics
depends, in this and similar work, on the manner in which a bend is
delimited, for example by the point of inflection method mentioned
above. Such methods, however, prove difficult to apply satisfactorily
to all but the simplest meander patterns. Recently this analysis has
been made on a power spectrum basis which recognises that a
particular river course may contain superimposed meander systems
of different wavelength. Speight6 7 describes such an analysis and
produces meander spectra characteristic of different types of river
channel.
Erosion and deposition of sediment at river bends result in the
excavation of a deep channel close to the concave outer bank and in
the construction of shallows extending out from the inner bank due
to deposition in that region (see Fig. 5.5). Most of the sediment
eroded from the other bank at a bend is deposited initially at the
next point of inflection downstream where a crossing bar is developed.

Flow in Channels
112
From there some of the material will pass to the inside bank deposits
at the next bend.
Erosion at the outside bank of a bend is found to be most active
in the recession period following a peak discharge when the point of
maximum velocity in the cross-section is at its closest to the concave
bank. The channel near this bank is deepened during flood discharges and the crossing bars also reach their maximum development then. During intervening periods of low flow the crossing bars
are lowered by erosion and the deep channels at the bends are

'-

~;~~5~:~-~~ -
Section at x---x
(to enlarged scale)

FIG.

5.5 River bend showing areas of deposition and erosion and

characteristic cross-section.

partially refilled. In river channels that are relatively deep and narrow in cross-section, centrifugal action at the bends sets up a secondary circulation resulting in spiral flow (see section 1.5); the
streamlines near the free surface moving outwards while those near
the bed move towards the inner bank. By this mechanism a proportion of the material eroded at a bend probably gets deposited on the
inner bank shoal of the same bend after travelling diagonally across
the bed of the river. In wide shallow streams it seems probable that
this secondary circulation is not established to a significant degree.

5.10 Formation of alluvial plains

An alluvial plain (flood plain) is a flattish area of land constructed
by a meandering river and consisting geologically of unconsolidated
deposits. The time taken in its formation is short by geological
standards but normally long when compared with the useful life of
engineering undertakings. As has been stated in the previous
section, an alluvial plain is formed by a river when it approaches its
base level, which can be either sea-level or that of a lake.

Flow in Erodible Material

113

The meander belt width ( Wm), which is approximately ten times the
bank-full stream width, is normally much less than the width of the
alluvial plain. Evidence of past meanders is normally apparent at
the edges of the flood plain where former stream courses have
eroded the harder rocks of the adjacent higher ground. One of the
principal reasons for the lateral movement of a meander system
from time to time is the way in which the ground level is raised
locally by overspill from the main channel. Overspill, occurring as
it does in times of flood, will contain a large suspended load which
will be deposited on the flood plain due to the lower velocities there.
This process leads to the gradual development of natural levees,
being the raising of the banks above the general level of the flood
plain. It can be seen from this that a time may come when the river
will break from its established channel and construct a new one at a
lower level.
In addition to the upward building of flood plains from overspill
deposition, the normal progression of meander systems in the downstream direction coupled with their less frequent sideways movements
will result in a slow reworking of the deposits forming the flood plain.
Until a river is fully graded (in regime) fresh material will be added
to the flood plain by this latter process since the material entering the
lower course of a river will be in excess of that discharged at its
mouth.
The principal interest that these processes hold for the engineer
concerns the construction of engineering works on the flood plain.
Man has always built in such areas, more particularly during the
last two or three centuries, and suffered regularly from damage to
life and property caused by inundation. The advantages of fertile
soil, level ground, easy communications and a reliable water supply
for both domestic and industrial purposes have usually appeared to
outweigh the disadvantages of the situation. Today flood protection
can be provided but its cost must be weighed very carefully against
the value of the property being protected and the risk to human life.
In recent years the question of legal control of flood plain usage and
development has been much discussed and .flood plain zoning (as it
is called) is being currently considered in the U.S.A.
Roads and railways have unavoidably to be constructed across
flood plains, usually on embankments, and some knowledge of river
meander behaviour is essential if damage is to be avoided. Firstly,
the opening provided in the embankment must be large enough to

114

Flow in Channels

carry the design flood (usually taken as the "100-year flood")

without raising the upstream water level unduly and secondly the
possibility must be considered of the river meander loops impinging
against the upstream face of the embankment during its useful life.
The rate at which meander loops move downstream varies greatly
depending upon the flood plain material but it should be appreciated
at the design stage that protection works may be required at some
time in the future in addition to the channel training works normally
carried out initially in the vicinity of the embankment openings (see
Blench6B).
5.11 The design of stable canals

The irrigation or drainage engineer when faced with the problem

of designing an unlined canal through erodible soil will need to
consider its performance primarily in terms of sediment transport.
Design considerations based on rigid boundary hydraulics will play
only a secondary role in the matter. In designing a "silt-stable"
canal there are two basic methods of approach:
(1) The canal can be designed so that no appreciable sediment

movement takes place. This approach is only valid if the flow

entering the canal does not carry a sediment load that might
be deposited. The maximum permissible velocity method, of
largely historic interest now, and the tractive force method are
both based on this criterion.
(2) The canal can be designed so that sediment transport does
take place but in such a way that the total erosion and deposition at any point along its length is zero over a full operational
cycle. This is essentially a regime method and the empirical
regime equations are used in a modified form. It will be seen
that this method is the only possible one when the flow
entering the canal carries material comparable in its transport
behaviour with that comprising the bed and sides of the canal.
In current design procedure these two approaches are complementary, both being frequently employed to arrive at a particular design
(except when the canal is to carry silty water, in which case the
regime equations are used alone).
Formerly, canals were designed from published values of maximum
permissible velocity tempered to a large extent by the judgement and

115

Flow in Erodible Material

experience of the engineer. Table 5.2 gives permissible velocity

values based on those first published by Fortier and Scobey in 1925
(see Brown69) and some of the shortcomings of this method are
immediately obvious from this table by its attempt to allow approximations for the influence of the depth of flow. The permissible
TABLE

5.2. Maximum permissible velocities for non-scouring canals

(after Fortier and Scobey*)

Velocity (In It/sec), after
ageing, for canals carrying:

Channel
Material

I.
2.
3.
4.
5.
6.
7.

8.
9.

10.

II.

12.
13.
14.

Fine sand (non-colloidal)

Sandy loam (non-colloidal)
Silt loam (non-colloidal)
Alluvial silts (non-colloidaD
Ordinary firm loam
Volcanic ash
Fine gravel
Stiff clay (very colloidaD
Graded, loam to cobbles
(non-colloidaD
Alluvial silts (colloidaD
Graded, silt to cobbles
(colloidaD
Coarse gravel (noncolloidaD
Cobbles and shingle
Shales and hardpans

water

clear
water
no
detritus

carrying
colloidal
clay

water

carrying
non-colloidal
silt, sand
gravel or
rock fragments

150
175
200
200
250
250
250
375

250
250
300
350
350
350
500
500

150
200
200
200
225
200
375
300

375
375

500
500

500
300

400

550

500

400
500
600

600
550
600

650
650
500

for non-straight channels decrease values

slightly sinuous
moderately sinuous
very sinuous

Representative
variation of
permissible velocity
with depth of water

Depth
I It

Depth
lOft

100

150

100

150

250

350

300

450

3-50
SOO

500
750

by the following amounts:

S%

13 %
22%

Fortier, S. and Scobey, F. C. Permissible canal velocities,

89, 1926, pp. 940-981.

Trans. A.S.C.E.,

velocity method has now largely been replaced by the tractive force
method. The tractive force method can be used satisfactorily on its
own design canals in non-cohesive soil which scour but do not silt.
The canal section is designed so that the fluid shear stress on the bed
reaches its critical value (on the threshold of bed movement) at at
least one point around the wetted perimeter. If there is no restriction
on the shape of the cross-section for other reasons, a "maximum
efficiency" design criterion can be adopted (see section 1.7.1). In

Flow in Channels

116

terms of sediment behaviour alone this condition will be satisfied

when all points on the wetted perimeter reach the critical shear
stress at the design conditions. Such a section has a cosine form,
shown in Fig. 5.6, expressed by the equation
7TX

d=dpCOSB

(5.11)

.
.
.
B
TTdp
m whrch the half-wrdth 2 = 2 tan cp

where d is the water depth at a

horizontal distance
x from the cross-section
~-~~
centre line
dp is the maximum (centre
I
line) depth
ct.
cp is the angle of repose of
Fro. 5.6 Form of canal cross-section
the bed material underdesigned to reach critical shear stress
water.
simultaneously around wetted
~------8--------~
I
f--x---1
I

perimeter.

The maximum depth will in

tum depend upon the discharge,
available slope and critical shear stress of the bed material and there
being insufficient space for a proper treatment of this complex
subject here the reader is referred to Chow70 for details of the design
procedure which originated in work by Lane71 and the U.S. Bureau
of Reclamation.
The regime theory was developed originally from Indian canal
data and yields a set of equations by which the mean velocity, slope,
cross-section area and wetted perimeter can be calculated (see section
5.8). It has been used for the satisfactory design of canals in India
for many years but early attempts to use it elsewhere revealed that
the so-called constants in the equations were tied to the particular
conditions of soil and discharge with which they were originally
associated. Simons and Albertson72 have reviewed this method of
canal design and re-worked the equations basing them this time on a
much wider range of canal data. In this paper they also compare the
method with the tractive force method outlined above and find that
the two can sometimes be combined in a particular design problem
with advantage. They conclude that the tractive force method seems
to be valid for the coarse non-cohesive range of soils but that the

Flow in Erodible Material

117

regime equations can first be used to establish a more satisfactory

cross-sectional shape for the canal for the required discharge. For
the smaller size range of non-cohesive soils as well as the cohesive
ones the regime equations should be used to establish the depth and
the bed slope as well as the other section characteristics.
As has already been stated, the tractive force method is not suitable
for the design of canals carrying a sediment load and therefore
the regime method must be used under these circumstances. However even this approach appears questionable when an appreciable
suspended load is present, the limiting value being 500 parts per
million by weight. A suspended load exceeding this figure is found
to increase significantly the erosive effect of the flow on the canal
banks.
In conclusion, the essentials of canal design can be summarised
as follows: The economic design of canals through erodible soils is
a very complex problem and a complete unified solution, which
would require a large number of variables being taken into account,
is not yet available. At present there are two different methods of
approach which are partially complementary:
(1) the tractive force method which can be used successfully in
coarse non-cohesive soils when the canal is to carry clear
water only.
(2) the modified regime method which is wholly empirical and
can be used for most types of soil with a mobile bed and
either clear or moderately silt-laden water has been found
satisfactory so long as it is applied to problems which fall
within the range of conditions from which the method was
derived.

5.12 Sedimentation in reservoirs

Any impounding reservoir constructed across a sediment-carrying
river will interrupt the normal movement of that sediment in the
downstream direction. As a result, the reservoir will accumulate
sediment so reducing its storage capacity, the regime of the river
upstream will be disturbed due to the introduction of a new base
level and the river downstream will have its supply of sediment
either stopped or significantly reduced and will therefore degrade
its channel.

118

Flow in Channels

It is essential that the engineer responsible for the design of such

a reservoir, and nearly all reservoirs come in this category, should
take account of this sediment trapping and consequent steady
reduction in storage capacity at an early stage in the project. The
only situation in which loss of capacity is of no importance is in the
case of a river dammed to concentrate head in a run-of-the-river
hydraulic power station. Even under these circumstances the dam
is normally utilised to create some storage in order to maintain a
higher base load power output. Most reservoirs (see Fig. 4.3) are
designed to incorporate a certain percentage of dead storage for
sediment deposition but this approach, to be rational, must be
related to the anticipated life of the system as a whole. In practice
Range of water levels

Cloy

FIG. 5.7 Development of sediment deposits in a reservoir (vertical

scale greatly exaggerated)

it has been found that a reservoir, whatever its function, will have to
be either replaced or supplemented by the time 50% of its storage
capacity has been lost in this way.
Fig. 5.7 shows in diagrammatic form the manner in which sediment deposits build up in a reservoir. The coarser material forming
the bed load in the river is deposited very close to the inlet point in
the form of a delta. When the reservoir water level is variable these
deposits will continually be eroding and reforming to give a multiple
delta system as indicated in the figure. The fine sediment load in the
river will be deposited over a much wider area of the reservoir
because of the action of density currents. These are formed by
denser silty water flowing down the slope of the reservoir bottom
underneath the clear water and result in the deepest areas of the
reservoir (close to the dam) being slowly filled with deposits of clay.
The rate at which a reservoir fills with sediment will depend
primarily on the rate of supply from the inflowing rivers. This can

Flow in Erodible Material

119

be determined by direct sediment load measurement in the rivers

although regrading of the upstream channel following the construction of the reservoir may well change the transport rates for an
appreciable length of time. In addition, the rate of sedimentation
will also depend upon the trap efficiency E of the reservoir, defined as
the percentage of incoming sediment that remains within the reservoir, the complimentary quantity corresponding to (100-E)% being
normally passed out of the reservoir through the discharge gates or
over the spillway in suspension.
The volume of reservoir storage lost per year V8 is given by the
equation
Vs = EQs

(5.12)

in which Q8 is the volumetric rate of sediment inflow (per year).

Values of the trap efficiency E will depend upon the size of the
reservoir relative to the catchment area or, more significantly, to
the yearly inflow. The larger the relative size of the reservoir the
higher will be the value of E, the majority of reservoirs having in fact
values greater than 80 %. The trap efficiency value will depend also
on the detailed shape of the reservoir (proximity of the inflow points
to the discharge structure, etc.) and the character of the sediment
concerned. For example, a tendency for stratified flows or density
currents to form as mentioned above may result in heavy silt-laden
water moving rapidly to the draw-off point at the dam, so reducing
considerably the detention time in the reservoir and therefore the
settling time available for that sediment. In calculating the volumetric rate of sediment inflow Q8 from the sediment load of the
river (normally expressed by weight) the ultimate bulk densities of
the reservoir deposits must be taken into account. This point is of
great importance when clay factions are present because of the slow
and complex manner of their consolidation following deposition.
The reservoir designer can take various steps to reduce the rate
of reservoir sedimentation at the design stage. The most satisfactory
solution to the problem is to site the reservoir where the sediment
load entering it will be a minimum but this degree of freedom is
seldom available. Alternatives include the use of settling basins
through which the inflow passes before entering the reservoir, the
construction of diversion channels whereby muddy storm discharges
by-pass the reservoir altogether, and the rapid passage of muddy

120

Flow in Channels

storm water through the reservoir, taking advantage of density

currents and operating the reservoir discharge facilities accordingly.
Attempts to remove sediment once it is deposited in the reservoir, by
either hydraulic or mechanical means, tend to be unsuccessful or
far too costly respectively.

6 Unsteady Flow 1n Open

Channels

6.1 Outline and classification

6.1.1 Introduction. In considering unsteady flow in channels it
is necessary to add time as a variable to the other combinations
of hydraulic variables considered primarily in the first three chapters.
By virtue of its free surface, unsteady channel flow is essentially
non-uniform flow and it can therefore be conveniently considered in
two categories, gradually varied and rapidly varied unsteady flow.
Such a classification is made on the same basis as that used for
steady non-uniform flow. The first category will have almost
parallel streamlines, gradual changes in depth, vertical accelerations
will play a minor part and channel friction forces a large one. The
second category will be characterised by very pronounced curvature
of the streamlines, rapid changes in depth and sometimes discontinuous water surface profiles, large vertical accelerations and
channel friction forces of secondary importance.
The addition of the time variable makes rigorous analysis very
complex and only the most simple cases will be considered here.
The analytical solution of engineering problems is normally made by
numerical step methods of various kinds.
Unsteady channel flow can be classified on appearance and other
factors as follows:
6.1.2 Solitary waves. These represent a form of gradually varied
flow in which a single smooth profiled wave, with no associated
troughs, is generated at a point in a channel and travels along it
121

Flow in Channels

122

with celerity* c which remains approximately constant. The amplitude or height of the wave is gradually diminished as the energy
contained in it is dissipated by friction. From a consideration of
momentum and continuity the following equation can be shown to
represent the celerity of a solitary wave
(6.1)
in which the symbols used are defined in Fig. 6.1. If the height of

::"
n

..c

Padd le

"

"""
""
""

,,"~ :
u

.:.:::::::_.:::.:-::;.,.::.::::.:::::::::-:::-:.:-::..::::::::.-::::::::-:::::-:-..::-:::::<::::::::::::::::::.::::.:.:::::...-:::::::....... :.:.

FIG. 6.1 Paddle generation of a solitary wave in a channel containing water at rest.

the wave is small compared with the water depth then dz - d1 - d

and
c

v(gd)

(6.2)

Solitary waves can be produced in the laboratory by the sudden

movement of a paddle, as indicated in Fig. 6.1, and in rivers by
sudden floods, by the operation of a control structure or by any
similar disturbance.
In flowing streams the celerity c of a solitary wave, as found by
one of the above equations, must be added to the velocity of the
stream to give the absolute velocity of propagation apparent to a
stationary observer on the bank. It follows from this that such
waves will only be able to move upstream if the stream velocity
v < y(gd), which is the condition for sub-critical flow.
The solitary wave, in common with all the other unsteady channel

* The term "celerity" is the standard name for the velocity of propogation of
any wave form. It is normally greater than any associated particle velocities, in
this case those of the water particles transmitting the wave. It is here considered
to be relative to the velocity, if any, of the water through which the disturbance
is travelling.

Unsteady Flow in Open Channels

123

flow forms described here, is a translatory wave in which the particles

are shifted horizontally as a result of its passage. Orbital particle
movement, where present, is of secondary importance.
6.1.3 Monoclinal waves. This is another form of gradually varied

unsteady flow similar in certain respects to one half of a solitary

wave. The monoclinal rising wave produces an increase in the water
level at a point after its passage and is shown in Fig. 6.2. This
type of wave is produced in a channel by a gradual increase in the

I
------;-t----=;;,;;~,.----_;_~----~
v1.::ll

I.

Q2

d2

Wave !rant after time .::11

--o;---

V2

Vw

d1

v,

FIG. 6.2 Progression of a monoclinal rising wave along a

channel.

discharge which is assumed in the case illustrated to reach a steady

value Q2 in a finite time. If the initial steady discharge in the
channel is Q1 then the propagation velocity of the wave can be
evaluated by continuity considerations and is given by the expression
Vw

Q2- Ql
A2- At

=----

(6.3)

and in a rectangular or very wide channel

(6.4)

In a time interval !lt the wave front advances a distance vwllt which
must be greater than the corresponding distance travelled by any of
the water particles, since Vw must be greater than both v1 and v2.
If the channel is initially empty there will be an exception to this
rule and Vw will be equal to v2. The dynamic equation of the uniformly progressive wave, of which this forms one case, is discussed
in detail by Chow73 and its development enables the profile of the
wave front to be determined. Because of its importance in relation

Flow in Channels

124

to flood wave movement in rivers, this wave form will be discussed

further from the point of view of the river hydrograph in section 6.2.
6.1.4 Surges. When a monoclinal wave has an abrupt change of
surface curvature and slope at its forward end the conditions of
gradually varied flow break down and the resulting rapidly varying
wave profile is termed a surge. The profile length of a surge wave
is very short by comparison with the wave forms discussed above
and the effect of channel friction forces on its behaviour correspondingly small. It is possible therefore to apply the momentum
equation to a surge without having to include a Chezy formula term.

(a)

~"Stationary surge"

_
(v,.- v2)

.... --- -

.::.__

(;,:=:_ v,) =c

-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:.:-:-:-:-:-.-:-:-:-:-:-:-.-:-:-:-:-:-:-.-:-:-:-::-:

(b)

Fro. 6.3 Channel surge wave as it appears to (a) a stationary observer, and (b) an observer moving along the bank with velocity Vw .

The similarity of a surge to a hydraulic jump (section 3.3) will be

seen by considering its appearance to an observer on the bank. In
Fig. 6.3(a) a stationary observer sees the wave travelling downstream
with velocity Vw. The initial velocity v1 and depth d1 are replaced
after the passage of the surge by v2 and d2 both greater than their
corresponding former values. This propagation velocity relative to
a stationary observer Vw, can be deduced by considering momentum
and continuity and related to the celerity c (wave travelling into
still water) by the equation
Vw

+ V1

(6.5)

in which
(6.6)

Unsteady Flow in Open Channels

125

This equation refers to a surge wave travelling downstream in a

wide rectangular channel. Fig. 6.3(b) shows what the observer sees
when he is travelling in the downstream direction at the wave velocity
Vw and the surge now appears
stationary and similar in all respects
to a hydraulic jump. The shallow
- v,
flow is now seen to be approaching
>:-:-.-.-.-:-:-.-:-.-:-;-:-:-:-:-:-:--:-::::..-.:.-.-.::::::::..::.:::.::.-:-.-:-.-:-:-:-:-;-:-:-:-.-;-:-;-:.-..
the surge with velocity c (given by
(a) Steep - fronted surge
Eq. 6.6) and the deep flow retreating from the surge with velocity
(vw- v2).
- v,
The form of a surge will, as in . . . . -..-.-.-...-.-.-. .-.-.-.-..-.- -:-:-.-.-:-:-. . :-:-.-.-.-:-.-:-. . _._._._._._._._._._._._.. . . . . . . . . . . . . . . . . . . . . .
(b) Undular surge
the case of the hydraulic jump,
depend upon the ratio of these two
FIG. 6.4 Alternative forms
of surge.
depths. The two alternatives are
shown in Fig. 6.4, the steep-fronted
surge being associated with a large depth ratio and the smooth
profiled undular surge with a depth ratio close to unity. The intense
turbulence generated in the steep-fronted surge is able to absorb a
proportionately high energy load.
Surges can advance either upstream or downstream depending
upon how they have originated and their passage can produce either

Vw

1__.,----

~
- ~

..

v,

;..-,..::::. ....:.::::..

v:2

........ .

(a ) Positi e surge advancing upstream

{bi"'P~sit;~e

'su.rge

~d ~ancing do~~stream
r - - VK

(c)

r .gah ve surge advanctng upslt earn

Fra. 6.5 Classification of steepfronted surges.

an increase of depth at a particular point (a positive surge) or a

decrease in depth (a negative surge). Fig. 6.5 shows the four resulting
possible combinations from these two variables for a steep-fronted
surge. The same four alternatives may occur as undular surges if

126

Flow in Channels

the depth ratios are appropriate to this form. Positive surge profiles
are always stable while negative surges are unstable and quickly
disperse because the deep water end of the profile travels faster than
the shallow end.
The example shown in Fig. 6.5(a) can occur in canals supplying
hydraulic power plants as a result of a sudden decrease in power
output (a rejection surge) in which case the final velocity v2 will not
reverse direction. However another instance of type (a) is found at
times in tidal rivers in regions of the world where the tidal range is
very high. Such a surge is normally called a tidal bore and its passage
is accompanied by a reversal in the direction of flow so that v2 will
act in the "upstream" direction. This marks the commencement of
the "flood" tide. The type (b) surge will occur when there is a sudden
increase in discharge at some upstream control structure. It has
been produced on a number of occasions on a disastrous scale by
the failure or destruction of an upstream dam. The type (c) surge
will occur in a power supply canal when there is a sudden increase in
power output at the downstream station (a demand surge). The
type (d) surge could be produced by a sudden decrease in the discharge from an upstream control structure. It must be remembered
that neither of these latter two negative surges will maintain a
detectable rapidly varied flow profile far from their point of origin.
6.2 Hydrographs and Hood waves

In discussing unsteady flow in river channels it is necessary to

introduce topics more usually classified with Hydrology. The
downstream movement of flood water resulting from heavy rainfall
on the catchment area of that river is a case in point.
A hydrograph is a record of the variation of river discharge past a
point as a function of time. It is frequently related to rainfall by
hydrologists, thereby enabling river discharges to be predicted for
any specified rainfall on the catchment area. The term flood wave is
commonly used to describe the solitary wave that passes down a
river channel as a result of a period of heavy rainfall on the catchment. When represented graphically it will show at a specified time,
either the depth of flow or the discharge, at all points along the length
of the river channel so covered. A flood hydro graph is a hydro graph
constructed from the derived discharge past a gauging station
during the passage of a flood wave.

Unsteady Flow in Open Channels

127

In calculating river discharges from rainfall records or assumptions, use is frequently made of the unit hydrograph of the catchment
concerned. This represents the discharge-time relationship when
uniform rainfall occurs over the whole catchment for unit time (the
unit of time is frequently taken as one day or a fraction of a day).
The unit hydrograph is normally deduced from an actual rain storm
(not uniform rainfall for unit time) on the catchment concerned, for
which adequate records of rainfall and stream flow exist. It can
also be developed analytically under rather specialised circumstances
(see Horton74).
In all operations relating rainfall and river discharge it is essential
Rainfall!

DISCharge
(ou1flow
from

colchmenr)

Time

FIG. 6.6 The Unit Hydrograph.

to consider the state of the catchment. It is only when the ground is

fully saturated, small surface depressions are filled and evaporation
needs met that excess rainfall will produce immediate runoff of the
type shown by the unit hydrograph in Fig. 6.6. Rainfall remaining
after ground storage and all the other "losses" referred to above
have been met is termed effective rainfall. The hydrograph shown is
related directly to the above ground drainage (surface runoff) of the
rainfall indicated and it would normally (except for the most
impervious and rocky catchments) be superimposed on a much
lower "base flow" hydrograph of much longer duration due to
ground water movement in the downstream direction.
Once the unit hydro graph has been established, the river discharge
can be predicted for any distribution of rainfall on the catchment if
the rainfall record is represented by a block diagram or histogram
which must be based on the same unit of time as the hydrograph.

Flow in Channels

128

The hydrographs corresponding to the rainfalls in each period are

obtained by scaling the unit hydrograph proportionally and these
are then added algebraically, remembering to observe the correct
interval between corresponding points on the individual hydro graphs.
In Fig. 6.7 this method is illustrated and the three hydrographs for
rainfalls (1), (2) and (3) are drawn commencing at a, b, and c respectively and combined as stated above to give the resultant hydro graph
for the rain storm. Sherman75 gives further details of the development and use of the unit hydro graph method in runoff prediction.*
Rainfall

lnfir: :, lt: : :tHl, , , ~ =, , , .l

(I)

'

(2)

(3)

Discharge

Fro. 6.7 Streamflow synthesis from rainfall record using the unit
hydrograph method.

The shape of a flood wave is closely related to the hydrograph at

any particular station along the length of the river and changes in
the shape of the hydrograph between different stations will reflect
modifications that have taken place in the flood wave profile as it
travels downstream. The principal changes that occur in the profile
in this way are a reduction in the elevation of the crest and an increase in the effective overall length. The extent to which such
modification will occur depends upon the changing cross-section of
the river channel and the presence of reservoirs, either natural or
artificial, or the occurrence of channel spill resulting in temporary
overbank storage. These and other factors affecting flood movements will be discussed in the following sections.

* The reader here is also referred to Engineering Hydrology by E. M. Wilson, a

companion volume in this series, which treats the subject in detail.

Unsteady Flow in Open Channels

129

6.3 Flood routing through open channels

The object of flood routing is the prediction of the movement and
resulting modifications of flood waves. This section considers the
movement of flood waves down river channels in which the
interaction of variables is more
complex than in the case of
1
routing through reservoirs
as considered in section 6.4.
Differential equations can be
established analytically for
-IIflood routing but these can
<J.t
only be solved for channels
having a
mathematicallyexpressible shape, both in FIG. 6.8 Length of time element At
cross-section and plan and so shown to horizontal scale of hydrograph.
the solution of a practical
problem necessitates the use of finite difference methods in one
form or another.
The success of such methods depends primarily upon the suitable
choice of a length of channel element or reach. The reach must be
short compared with the flood wave and also in relation to the river
topography. The time interval !!..tat which conditions in the reach
are considered must also be chosen with care so that the shape of
the hydrograph, particularly its crest, can be expressed in adequate
detail (Fig. 6.8). The basic relationship on which flood routing
methods are based is one of continuity. Over a short time interval
it states that in any reach

Disc~arge

average inflow

average outflow

+ rate of change of storage

(6.7)

Referring to Fig. 6.9, I represents the inflow discharge to the reach and
0 the outflow, whileS is the so-called storage or the total volume of
water contained in the reach at any particular time. If the variation
in I and 0 is assumed linear throughout the period !!..t and if all
ground water flows are either ignored or included in the I and 0
terms, then Eq. (6.7) can be expressed simply in terms of these
quantities at the beginning and end of this period.

h + I2 01 + 02
-2-=
2

(S2 - S1)
!!..t

(6.8)

Flow in Channels

130

In solving the above equation over a finite time interval, the conditions in the reach at time t1 will be known (h,01 and S1) as well as
the inflow h at the close of the interval, since the inflow hydro graph
must be known initially or previously determined. There will
remain two unknown quantities 02 and S2. These can be related in
general terms by the expression
(6.9)

Reach___;

. . ......... ~ . . . . :.t,g.

but this relationship cannot be written in explicit

form when referring to river channels
Discharges designated:It and Ot at lime It
for the following reasons.
12 and 0 2 at time 12
The stage-discharge relationship for a
12-lt = Jl
channel is only uniquely defined when the
FIG. 6.9 Definition of flow is uniform, because the discharge in an
flood routing variables open channel is proportional to the t power
for a river reach.
of the slope of the water surface (assumed
parallel with the total energy line for slow
flows). It is quite clear from what has been stated in sections 6.1 and
6.2 that the water surface slope will be much greater during the rising
stage than during the falling stage as a flood wave passes. The type

Slog~

I
I
1~0

1
I
I

(decline) 1

~eo

FIG. 6.10 Stage-discharge relationship during passage of flood

wave.

of stage discharge relationship which results under these conditions is

shown in Fig. 6.10. Apart from stressing the fact that two widely
different discharges occur under the same stage on the two limbs of

Unsteady Flow in Open Channels

131

the curve this figure also shows that the maximum discharge does
not necessarily correspond with the maximum stage occurrence of
the flood, a difficulty not always appreciated when estimating flood
discharges from maximum water level records. The stage-storage
relationship is found to be similar for non-uniform flow and again
a looped curve is developed during the passage of a flood wave.
A single stage-discharge curve of the type shown in Fig. 6.10 is
of little value in establishing an explicit relationship for Eq. (6.9).
It is only based on conditions applicable to a single flood and clearly
for the relationship to be of any value in flood routing all possible
combinations of stage-slope-discharge must be covered or be
Outflow

Storage

FIG. 6.11 Storage-outflow relationships (for different values of

inflow).

available by interpolation. For flood routing purposes the stagefall-discharge data is transformed into an inflow-storage-outflow
relationship which, once determined, could be plotted on a threedimensional basis were this practicable. More conveniently this
relationship is expressed as a family of storage-outflow curves, each
curve having a specified inflow value, as in Fig. 6.11. This method
is explained in greater detail by Kuiper. 76
Channel flood routing computations are of great complexity and
use is normally made of electronic computers, either analogue or
digital. References to the many flood routing procedures that have
been developed are given by Chow77 while he himself describes in
detail a method based on the use of characteristics. Gilcrest7B, in
addition to dealing at length with the mathematical side of flood
routing, gives a full coverage of computational methods including
the use of mechanical computing devices developed before electronic

132

Flow in Channels

computers were available. The description of the programming of

these machines is, however, still relevant to the design of electronic
computer programmes.
6.4 Flood routing through reservoirs
Flood detention reservoirs are constructed to protect the downstream valley from flood damage (see section 4.2.2). To do this
successfully they must restrict the maximum discharge in the downstream river to a value that the channel can contain without overspill. In designing a project of this nature by far the most difficult
Usable prism s!oroge

FIG. 6.12 Flood routing through a large reservoir.

task facing the engineer is deciding on the design flood hydrograph

and normally hydrological methods are used for this purpose.
Assuming now that the design inflow hydrograph has been so
determined, the resulting flow can be routed through a reservoir by
application of the principles introduced in the previous section.
Fig. 6.12 shows the elements used in the method normally adopted
for large reservoirs. The reservoir is assumed to discharge all its
outflow over a fixed crest spillway and the water surface in the
reservoir is assumed horizontal at all places and times. The usable
storage S (above the spillway crest level) is a simple function of the
head H over the spillway, the relationship being determined from a
topographical survey of the reservoir site. The outflow-head relationship will depend upon the type of spillway used and if a broadcrested weir of length Lis assumed this will have the form: 0 =
CLH312. It can therefore be seen that, for the conditions assumed,
an explicit relationship exists between 0 and S to replace the general
one shown in Eq. (6.9).
The basic continuity relationship (Eq. 6.8) can be rearranged in
the form
(6.10)

133

Unsteady Flow in Open Channels

in which, at the commencement of each step in a finite-difference

solution, all the terms on the left-hand side are known (section 6.3).
Using the topographical relationship between water surface area and
elevation, the term

(~ +

0) can be plotted as a function of H

(Fig. 6.13). Also on this graph the 0

drawn and the line

Head H
above
spillway
crest

(~~-

0).

CLH31 2 relationship is

The computational procedure used

/

0 = CLH 312

0 outflow discharge;

!; +0

(, ~

Fra. 6.13 Graphical construction of inflow and outflow relationship

and

(~

0)

term.

in this method is outlined in Table 6.1 and referring to both this

table and Fig. 6.13 the method will now be described step by step.
At the commencement of the computation the initial head HI at
the spillway will be known from the hydrograph. If it is zero then
OI and SI will be zero and so therefore will column (6). If HI has
a non-zero value then OI and Sr are given and column (6) may be
determined. The remaining steps will be discussed in relation to
the nth time interval from the start: h and h at times (n - 1).1t
and n.1t, are known from the inflow hydrograph, their sum giving
2Scolumn (5). Column (6) is next found from the ( ~ line in

0)

the graph and the Ht value, which is the same as the H2 value on
the line above. By adding (5) and (6) a computed value is obtained

nth

(n- 1) t::.t

tst

(n- 1)th

It

(1)

nt::.t

t::.t

t2

(2)

given

It

(3)

given

12

(4)

sum of(3)
and(4)

11+12

(5)

t::.t

t::.t
2

)/

Ht

(8)

_____--;::;7

2S2 + O

(7)

H2

(9)

....------value
from
from (9) line
sum of (5)
from
line graph
above using
and (6)
above
graph
~

2St _ Ot

(6)

6.1. Computational procedure used in conjunction with Fig. 6.13 for routing a .flood through a reservoir

time
interval

TABLE

from
graph

02

(10)

Unsteady Flow in Open Channels

for( 2: :

135

+ 02) in (7) and this is used in the graph with the appro-

priate curve to give a value of H2. At the same time the outflow
value 02 can be obtained from the outflow curve at head H2. It is
now possible to proceed with the (n + I)th step using the value of
H2 above to give the current H1 value and so on until the maximum
value of 02 and its timing have been obtained. From Fig. 6.14 it
can be seen that the interval between the occurrence of the maxima
Di:.chorge
I

tume of maximum
reservoir level :

Time I

maximum storage:
maximum ou tflow

FIG. 6.14 Principal features of inflow and outflow hydrographs for

flood routing through a reservoir.

in the inflow and outflow hydrographs is termed the lag time while
the reduction in their magnitudes is called the attenuation.
When considering flood routing through small reservoirs it may
be necessary to take account of the storage under the backwater
curve (wedge storage) produced by the inflow in addition to the
horizontal surface (prism) storage. It may also be necessary, under
these conditions, to allow for the storage potential in the channel of
the inflowing river below the level of the maximum water surface
elevation in the reservoir (see Gilcrest7B).
A reservoir that discharges through other forms of control
structure will have different outflow-head relationships and these
must be determined before the computations begin. Two or more
different types may be in use at the same time which will result in a
multi-term 0 - H relationship but this may still be plotted as a
single curve corresponding to the one shown in Fig. 6.13 in which
a broad-crested weir spillway is assumed.

References

Chapter 1
1. CHow, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), p. 93 footnote 1.
2. RousE, H. and INcE, S.: History of Hydraulics, Iowa Institute of
Hydraulic Research (1957), p. 119.
3. MANNING, R.: "On the flow of water in open channels and pipes,"
Trans. of l.C.E. Ireland, 20, (1891), pp. 161-207.
4. MORGAN, E. E.: Stream and Channel Flow, Chapman and Hall Ltd.
(1938) (see tables, p. 113).
5. AcKERS, P.: "Resistance of fluids flowing in channels and pipes,"
Hydraulics Research Paper No. l, H.M.S.O. (1958). Also: "Charts
for the Hydraulic design of channels and pipes," Hydraulics Research
Paper No. 2, H.M.S.O. (1963).
6. GIBSON, A. H.: "On the depression of the filament of maximum
velocity in a stream flowing through an open channel," Proc.
Royal Society, Series A, 82, (1909), pp. 149-159.
7. RousE, H.: Fluid mechanics for hydraulic engineers. Engineering
Societies Monographs, McGraw-Hill Book Co., New York (1938),
pp. 266-268.
8. CHOW, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.
New York (1959), pp. 27-28.
9. PRANDTL, L. and TIETJENS, 0. G.: Applied Hydro- and Aeromechanics,
Engineering Societies Monographs, McGraw-Hill Book Co.,
New York (1934), Chapter 4.
10. DELLEUR, J. W.: "The boundary layer development in open channels,"
Proc. A.S.C.E. Eng. Mech. Div., 83, January (1957).
11. BoYER, M. C.: "Estimating the Manning coefficient from an average
bed roughness in open channels," Trans. Am. Geophysical Union, 35,
No. 6, December (1954), pp. 957-961.
Chapter 2
12. BAKHMETEFF, B. A.: Hydraulics of Open Channels, Engineering
Societies Monographs, McGraw-Hill Book Co., New York (1932),
p. 59.
136

References

137

13. BAKHMETEFF, B. A.: Hydraulics of Open Channels, p. 70.

14. MoRRis, H. M.: Applied Hydraulics in Engineering, Ronald Press,
(1963), pp. 120-124.
15. CHow, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), p. 253.
16. WooDWARD, S. M. and PosEY, C. J.: Hydraulics of Steady Flow in
Open Channels, John Wiley, New York (1941), pp. 94--109.
17. KING, H. W. and BRATER, E. F.: Handbook of Hydraulics, McGrawHill Book Co., New York, 5th Edition (1963), pp. 8-48 to 8-66.
BIBLIOGRAPHY
HENDERSON, F. M.: Open Channel Flow, Collier Macmillan, London
(1966), Chapter 5.
Chapter 3
18. RousE, H.: Fluid Mechanics for Hydraulic Engineers, Engineering
Societies Monographs, McGraw-Hill Book Co., New York (1938),
pp. 324--325.
19. RousE, H.: Discussion (p. 651) on paper by Bakhmeteff and Matzke:
reference 23.
20. KING, H. W. and BRATER, E. F.: Handbook of Hydraulics, 5th
edition, McGraw-Hill Book Co., New York (1963), pp. 8-35, 8-36.
21. The Standing Wave or Hydraulic Jump, Govt. of India Central Board
of Water and Power, Publication No. 7, Simla India 2nd edition,
(1950).
22. BAKHMETEFF, B. A.: Hydraulics of Open Channels, Engineering
Societies Monographs, McGraw-Hill Book Co., New York (1932),
p. 234.
23. BAKHMETEFF, B. A. and MATZKE, A. E.: "The hydraulic jump in
terms of dynamic similarity," Trans. A.S.C.E. 101, (1936), pp.
631-647.
24. BAKHMETEFF, B. A. and MATZKE, A. E.: "The hydraulic jump in
sloped channels," Trans. A.S.M.E. 60, (1938), p. 111.
25. CHow, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), p. 396.
26. FORMICA, G.: ("Preliminary test on head losses in channels due to
cross-sectional changes"), L' Energia Elettica, Milano, 32, 7, pp.
554--568, July (1955). (See Chow: reference 27-section 17-4.)
27. CHOW, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), p. 465.
28. IPPEN, A. T.: "Channel transitions and controls," Chapter 8 in
Engineering Hydraulics (Editor: Rouse, H.), John Wiley (1950),
pp. 496-538.
29. d'AuBUISSON de VOISINS: Traite d'hydraulique, Paris (1840).
30. YARNELL, D. L.: "Bridge piers as channel obstructions," U.S. Dept.
of Agric. Tech. Bull. No. 442, November (1934).

138

Flow in Channels

31. lPPEN, A. T. and DAWSON, J. H.: "Design of channel contractions,"

Trans. A.S.C.E. Symposium Paper, 116, (1951), pp. 326-346.
32. BLAISDELL, F. W.: "Flow through diverging open channels at
supercritical velocities," Report SCS-TP-16 U.S. Soil Conservation
Service (1949).
BIBLIOGRAPHY
For full bibliography on the hydraulic jump see CHow: reference 25
(end of Chapter 15).
Chapter 4
33. HoYT, W. G. and LANGBEIN, W. B.: Floods, Princeton University
Press (1955), pp. 241-249.
34. RAo, V. S.: "The role of forests in flood control," Soil and Water
Conservation, India 3, 4, July (1955), p. 163.
35. KoELZER, V. A.: "The use of statistics in reservoir operations," J. of
Hyd. Div., A.S.C.E., Paper 1008, 82, HY-3, June (1956).
36. GILCREST, B. R.: "Flood routing," Chapter 10 in Engineering
Hydraulics (Editor: Rouse, H.), John Wiley, New York (1950).
37. KuiPER, E.: Water Resources Development, Butterworths, London
(1965), pp. 210-221.
38. MEINZER, 0. E. (Editor): Hydrology, Chapter 11, Sections Hand I,
pp. 561-578, Dover Publications, New York (first published, 1942).
39. LINSLEY, R. K., KOHLER, M.A. and PAULHUS, J. L. H.: Hydrology
for Engineers, Chapters 10-13, McGraw-Hill Book Co., New York
(1958).
40. LINSLEY, R. K. and FRANZINI, J. B.: Water Resources Engineering,
McGraw-Hill Book Co., New York (1964), pp. 159-163.
41. CREAGER, W. P., JUSTIN, J. D. and HINDS, J.: Engineering for Dams,
(3 volumes), John Wiley, New York (1947).
42. BROWN, J. G. (Editor): Hydro-electric Engineering Practice (3
volumes), Blackie, London (1958).
43. LELIAVSKY, S.: Irrigation and Hydraulic Design (3 volumes), Chapman and Hall, London (1955-1960).
44. RousE, H.: "Engineering Hydraulics," Chapter 1: Fundamental
Principles of Flow, John Wiley, New York (1950), p. 52.
45. HENRY, H. R.: "Characteristics of sluice gate discharge," M.Sc.
thesis, State Univ. of Iowa (1949). Published as a discussion of
paper by Albertson, M. L., eta/. "Diffusion of submerged jets,"
Discussion: Proc. A.S.C.E. December (1949), pp. 1541-1548.
46. IPPEN, A. T.: "Channel transitions and controls," Chapter 8 in
Engineering Hydraulics (Editor: Rouse, H.), John Wiley, New
York (1950), pp. 536-539.
47. B.S. 3680: Methods of Measurement of Liquid Flow in Open Channels,
Part 4, "Weirs and flumes," 4A, "Thin-plate weirs and venturi
flumes," British Standards Institution, London (1965).

References

139

48. WEBBER, N. B.: Fluid Mechanics for Civil Engineers, Spon, London
(1965), pp. 216-217.
49. KING, H. W. and BRATER, E. F.: Handbook of Hydraulics, 5th
Edition, McGraw-Hill Book Co., New York (1963), section 5-46.
50. AcKERS, P. and HARRISON, A. J. M.: "Critical depth flumes for flow
measurement in open channels," Hydraulics Research Paper No.5,
D.S.I.R., H.M.S.O., London (1963).
51. CRUMP, E. S.: "Moduling of irrigation channels," Punjab Irrigation
Branch, Papers No. 26 and 30A, Lahore, India (1922-1933).
52. CHOW, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), pp. 72-81.
53. MITCHELL, W. D.: "Stage-fall-discharge relationships for steady flow
in prismatic channels," Water Supply Paper No. 1164, U.S.
Geological Survey (1954).
54. B.S. 3680: Methods of Measurement of Liquid Flow in Open Channels, Part 3: "Velocity area methods," British Standards Institution, London (1964).

BIBLIOGRAPHY
Reservoirs and Planning
LEOPOLD, L. B. and MADDOCK, T.: The Flood Control Controversy, The
Ronald Press, New York (1954).
LANGBEIN, W. B.: "Queuing theory and water storage," J. of Hyd. Div.,
A.S.C.E. Paper No. 1811, HY-5, 84, October (1958).
SYMPOSIUM, 1946-47: "Multi-purpose reservoirs," Proc., A.S.C.E. 75,
March (1949). pp. 288-390.
GILCREST, B. R.: "Flood routing," including a bibliography of 34
references (pp. 709-710). Chapter 10 in Engineering Hydraulics,
(Editor: Rouse, H.) John Wiley, New York (1950).
KNAPPEN, T. T., STRATTEN, J. H. and DAVIS, C. V.: "River regulation by
reservoirs," chapter in Handbook of Applied Hydraulics, (Editor:
Davis, C. V.) 2nd Edition. McGraw-Hill Book Co., New York (1952).
pp. 1-21.
MoRRis H. M.: Applied Hydraulics in Engineering, Ronald Press, New
York (1963), pp. 297-313.
Control Structures
LEUAVSKY, S.: Irrigation and Hydraulic Design, Volume 3, pp. 71-149,
Chapman and Hall, London (1955-1960).
BOWMAN, J. S. and BoWMAN J. R.: "Spillway crest gates," Section 8 in
Handbook of Applied Hydraulics, (Editor: Davis, C. V.) 2nd Edition,
McGraw-Hill Book Co., New York (1952).
HENDERSON, F. M.: "Open channel flow," Chapter 6: Channel Controls,
Collier-Macmillan, London (1966).

140

Flow in Channels

Flow Measurement
TROSKOLANSKI, A. T.: Hydrometry, translated from the Polish, Pergamon
Press, London (1960).
Chapter 5
55. B.S. 1377: Methods of Testing Soils for Civil Engineering Purposes,
British Standards Institution, London (1961).
56. BAGNOLD, R. A.: Physics of Blown Sand and Desert Dunes, Methuen,
London (1941).
57. DUPUIT, A. J.: see Rouse, H. and Ince, S., History of Hydraulics,
Iowa Institute of Hydraulic Research, State University of Iowa
(1957), pp. 171-172.
58. BLENCH, T.: Regime Behaviour of Canals and Rivers, Butterworths,
London (1957).
59. STRAUB, L. G.: "Mechanics of rivers," Chapter 13c, pp. 614--636, in
Hydrology, (Editor: Meinzer, 0. E.), Dover Press, New York (1942).
60. KALINSKE, A. A.: "Movement of sediment as bed-load in rivers,"
Trans. American Geophysical Union, 22, (1947), pp. 615-620.
61. EINSTEIN, H. A.: "Formulas for the transportation of bed load,"
Trans A.S.C.E., 107, (1942), pp. 561-597.
62. GARDE, R. J. and ALBERTSON, M. L.: Discussion of reference 63
published as Paper No. 1856, Proc. A.S.C.E., J. of the Hydraulics
Division, November (1958).
63. LAURSEN, E. M.: "The total sediment load of streams," Proc. A.S.C.E.,
Paper No. 1530, J. of the Hydraulics Division, February (1958).
64. CoLBY, B. R. and HEMBREE, C. H.: "Computations of sediment
discharge, Niabrara River, near Cody, Nebraska," Water Supply
Paper No. 1357, U.S. Geological Survey (1955).
65. MORRIS, H. M.: Applied Hydraulics in Engineering, Ronald Press,
New York (1963).
66. FRIEDKIN, J. F.: A Laboratory Study of the Meanderings of Alluvial
Rivers, Report of Mississippi river commission, U.S. Waterways
experiment station, Vicksburg, Miss., May (1945).
67. SPEIGHT, J. G.: "Meander spectra of the Angabunga river," Journal
of Hydrology, 3, No. 1, (1965), pp. 1-15.
68. BLENCH, T.: Regime Behaviour of Canals and Rivers, Butterworths,
London (1957), pp. 81-85.
69. BRoWN, C. B.: "Sediment transportation," Chapter 12 in Engineering
Hydraulics (Editor: Rouse, H.), John Wiley, New York (1950),
p. 810.
70. CHow, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), pp. 164--179.
71. LANE, E. W.: "Progress report on studies on the design of stable
channels by the Bureau of Reclamation," J. of the Hydraulics
Division, Paper No. 280, Proc. A.S.C.E., 79, (1953).
72. SIMONS, D. B. and ALBERTSON, M. L.: "Uniform water conveyance
channels in alluvial material," J. of the Hydraulics Division, Paper
No. 2484, Proc. A.S.C.E., May (1960), pp. 33-71.

References

141

BIBLIOGRAPHY
BLENCH, T.: Regime Behaviour of Canals and Rivers, Butterworths,
London (1957).
HENDERSON, F. M.: Chapter 10 in Open Channel Flow, Collier-Macmillan, London (1966).
LELIAVSKY, S.: An Introduction to Fluvial Hydraulics, Constable, London
(1955).
LELIAVSKY, S.: "Design textbooks in civil engineering: 4." River and Canal
Hydraulics, Chapman and Hall, London (1965).
KUIPER, E.: Chapter 4 in Water Resources Development, Butterworths,
London (1965).
BROWN, C. B.: "Sediment transportation," Chapter 12 in Engineering
Hydraulics (Editor: Rouse, H.), John Wiley, New York (1950).
Chapter 6
73. CHow, V. T.: Open Channel Hydraulics, McGraw-Hill Book Co.,
New York (1959), pp. 531-537.
74. HoRTON, R. E.: "The interpretation and application of run-off plot
experiments with reference to soil erosion problems," Proc. Soil
Science Soc. of Am., 3, (1938), pp. 340-349.
75. SHERMAN, L. K.: "The unit hydrograph method," Chapter 11E in
Hydrology (Editor: Meinzer, 0. E.), Dover Publications, New
York (first published 1942).
76. KUIPER, E.: Water Resources Development, Butterworths, London
(1965), pp. 94-101.
77. CHOW, V. T.: Open Channel Hydraulics, Chapter 20, McGraw-Hill
Book Co., New York (1959).
78. GILCREST, B. R.: "Flood routing," Engineering Hydraulics, Chapter 10
(Editor: Rouse, H.), John Wiley, New York (1949).
BIBLIOGRAPHY
HENDERSON, F. M.: Chapters 8 and 9 in Open Channel Flow, CollierMacmillan, London (1966).
Sections 6.2-6.4
DE WEIST, R. J. M.: Chapters 2 in Geohydrology, John Wiley, New York
(1965).
GILCREST, B. R.: "Flood routing," Chapter 10 in Engineering Hydraulics
(Editor: Rouse, H.), John Wiley, New York (1949).

list of Notations
A
cross-sectional area of flow
A, C, H, M, S
classification of surface profiles
b, B
width of channel; water surface width of channel
b'
channel bed width
c
wave celerity (velocity of propagation in still water)

suffix indicating critical flow conditions

coefficient
Chezy coefficient
coefficient of discharge
Cc
coefficient of contraction
Cv
coefficient of velocity
CL
meander length coefficient
Cw
meander width coefficient
C.D.L.
critical depth line
d
depth of flow; maximum depth of flow in a cross-section
d'
depth of flow (supercritical)
de
critical depth of flow
d1
height of hydraulic jump
dm
mean depth in a cross-section
dm
mean particle diameter
do
normal depth of flow
dp
maximum (centre line) depth in a cross-section
D
diameter of circular pipe
E
specific energy
E
trap efficiency of reservoir
F
suffix indicating conditions in a conduit flowing full
F
Froude number
F, F'
hydrostatic pressure force
g
accleration due to gravity
gs
bed load (weight) per unit volume
h
depth of water
h
vertical distance
ha
drop in water level past bridge piers
he
effective head above weir
hr
kinetic energy head
h, H
head
H
total energy
i
slope of total energy line
I
inflow
K
a friction coefficient
K
'Conveyance' of a channel
Ka
a discharge coefficient
c

C
C
Cn

142

List of Notations
length
width of weir (across channel)
length of meander
m
ratio of flow areas
m
hydraulic mean depth
n
unspecified integer
n
exponent
n
Manning roughness coefficient (also Kutter)
N.D.L.
normal depth line
o
suffix indicating uniform flow conditions
0
outflow
p
height of weir crest above channel bed
p
pressure
p
error value
p.p.m.
parts per million
P
wetted perimeter
q
discharge per unit width
q
error value
Q
total discharge; dominant discharge
Qmax
maximum probable discharge
Q.
volumetric rate of sediment inflow (per year)
r
error value
R
radius
s
sediment load in sample (weight/unit volume)
s
slope of channel bed
S
suspended load (tons/sec.)
S
storage
time
v
velocity
v mean velocity in cross-section
Vc
critical velocity
vw
velocity of propagation of wave in moving stream
Vs
volume of reservoir storage lost per year to sediment
w
specific weight
W
weight of water
Wm
width of meander
x
ratio of velocity values
x
distance along channel bed
x
horizontal distance from centre line of channel cross-section
x
length of hydraulic jump
y
vertical distance
z
horizontal distance in channel side slope ratio
z
vertical distance above datum level
z
depth of centroid of channel section below surface
ex
energy coefficient
fJ
momentum coefficient
L
L
Lm

143

144
(}

r/>

rf>
'1T

1JI

list of Notations
angle; angle of channel side slope
kinematic viscosity
angle of slope of total energy line
angle of repose of sediment (non-cohesive) under water
constant
'sediment characteristic' in bed load formula

Index

ACKERS, P. 8, 94
adverse slope 37, 40
afflux 92
Albertson, M. L. 106, 116
alluvial plains-see flood plains
alternate depths 26, 27
angle of repose of sediment 102
anti dunes 102
attenuation of flood waves 78, 135
d'Aubuisson de Voisins 11

36, 38
backwater curves, storage under 80,
135
Bagnold, R. A. 100
Bakmeteff, B. A. 27, 37, 57, 58
bank-full discharge 107
bank-full stage 15, 108
banks of river channels 101
barrages 81
base flow 127
bed form of channels 101
bed load, definition 100, 102
formulae 103
bed-movement of sediment 100
bends 13, 109, 112
Bernoulli equation 23, 24
best hydraulic section 15
Blaisdell, F. W. 72
Blench, T. 101, 114
bore, tidal 126
boundarylayertheory 14
Boyer, M. C. 14
Brater, E. F. 48, 55, 91
Bresse function 41
bridge piers, flow through 71
broad-crested weirs 91, 132
BACKWATER CURVE

Brown, C. B. 115
Brown, J. G. 81
CANALISATION OF RIVERS 75
celerity of waves 122, 124
change of bed slope 50
channel sections, rectangular 16
semi-circular 16
trapezoidal 16, 17, 94
Chezy, A.-see Chow, V. T. 4
Chezy formula 3-5, 33, 35
Chow, V. T. 4, 13, 41, 59, 70, 94, 116,
123, 131
chutes 40, 110
circular conduits 19
closed conduits 18
coefficient, of contraction for sluices
85
of discharge, flow through a channel
constriction 71
for broad-crested weirs 91
for full width weirs 89
for fully contracted weirs 90
for sluices 85
for throated flumes 93
for V-notch weirs 90
energy coefficient oc 13, 26
Manning roughness coefficient 1/ 6,
14
table of values 7
momentum coefficient {3 56
of velocity for throated flumes 93
cohesive soils 110
Colby, B. R. 106
conduits, circular 19
closed 18
conjugate depths 57

145

146

Flow in Channels

constriction, flow through a 71, 92

control of flow 75, 80
control sections 49, 86
control structures 49, 80
convergent channels 23
conveyance 8
cost of channel construction 15
Creager, W. P. 81
critical depth 27, 29
flow transition through 49
line (C.D.L.) 37
slope 32, 38
tractive force 17
tractive stress 103, 116
velocity 27, 29, 102
crossing bars 111
Crump, E. S. 94
Crump weir 94
current meters 95, 96
cut-offs 110
DAWSON, J. H. 71
dead storage in reservoirs 79, 118
Delleur, J. W. 14
delta construction in reservoirs 118
demand surge 126
density currents in reservoirs 118
density of sediment deposits 119
deposition of sediment, in channels
104
in reservoirs 79, 104
depth, critical 27, 29, 52
hydraulic mean 5
normal 9-11
depths, alternate 26, 27
conjugate 57
design of stable canals 114-117
direct integration of gradually varied
flow equation 40
direct jump-see hydraulic jump
divergent channels 69
diversion channels for reservoirs 119
diversion dams-see barrages
dominant discharge 107, 108, 109
drawdown curve 36, 38
drowned sluice 86
drum gates 83
Du Boy's formula 103
dunes 102, 106
Dupuit, A. J. 101
EFFECTIVE RAINFALL 127
efficiency of cross-section, hydraulic
15, 115

egg-shaped sewers 18
Einstein, H. A. 103
electronic computers for flood routing
131
empirical school (sediment transport)
101
empirical velocity formulae 3-9
energy coefficient ~ 13, 26
energy losses in channels 4, 24
in hydraulic jumps 58, 59
in waves 122
specific 24
total 23, 28
erodible material, flow in 98
error analysis of Chezy formula 8
FLOOD DETENTION RESERVOIRS 78, 132
flood plain zoning 113
flood plains, flow on 113
formation of 112
flood protection by levees 113
by reservoirs 77, 132
flood routing through open channels
129-131
through reservoirs 132-135
flood wave-hydrograph relationship
128
flood waves 78, 126
floods, 100-year flood prediction 111,
114
flow measurements 87
flow relationships, fundamental 14
force diagram 67
force-see specific force
Formica, G. 71
Franzini, J. B. 80
free overfall 36, 50
Friedkin, J. F. 110
Froude number 27, 52, 58
functions of a river 73
fundamental flow relationships 14
fundamentalist school (sediment transport) 101
GANGUILLET AND KUTTER FORMULA 5
Garde, R. J. 106
gates, hydraulic-sluice gates 61, 84
vertical lift gates 82
other types 83
gauging of rivers 94
Gibson, A. H. 12
Gilcrest, B. R. 79, 135
graded channels-see regime theory

Index
gradually varied flow 23
general equation of 33
direct integration of 40
graphical integration methods
gravel rivers 14

41

HAMILTON-SMITH EQUATION 90
Harrison, A. J. M. 94
head, specific-see specific energy
height of hydraulic jump 59
Hembree, C. H. 106
Henry, H. R. 86
Hinds, J. 81
horizontal channel bed 38
Horton, R. E. 127
Hoyt, W. G. 74
hydraulic drop 37
hydraulic efficiency of cross-section
15
hydraulic jump 37, 52, 86
analysis 55
in a sloping channel 58
physical characteristics 59
stabilisation by submerged blocks
66
uses of the 54
hydraulic mean depth 5
hydro-electric power generation 79,
81, 118
hydrograph 78, 126, 127
hydrologic book-keeping equation 80
IcE COVER 21-22

lnce, S.

137, 140

Indian canal data 107, 111

inflow discharge 78, 129
Inglis meander formula 111
Ippen, A. T. 71, 87
isovels 11, 96

JUMP, HYDRAULIC-see hydraulic

jump
Justin, J.D. 81
KALINSKE, A. A. 103
King, H. W. 48, 55, 91
Koelzer, V. A. 77
Kohler, M. A. (see under Linsley)
Kuiper, E. 79, 131
Kutter's TJ 5
LACEY 108
lag time 135

79

147

laminar flow 2
Lane, E. W. 116
Langbein, W. B. 74
Laursen, E. M. 106
Leliavsky, S. 81
levees 113
Linsley, R. K. 79
location of hydraulic jump 60
locus of critical points 28
loss of storage in reservoirs 119
loss of water from canals 76
loss of water from reservoirs 80
losses-see energy losses
MANNING FORMULA 5-6, 32, 35
Manning, R. 5
Manning roughness coefficient TJ 6,
14
table of values 7
Matzke, A. E. 58
maximum permissible velocity in
canals 115
meandering 107, 109-112
meander belt width 110
meander formula, Inglis 111
measurement of flow 87
Meinzer, 0. E. 79
mild slope 32, 38
Mitchell, W. D. 95
momentum coefficient fJ 56
momentum equation-application to
hydraulic jump 55
application to solitary wave 122
application to surge wave 124
monoclinal waves 54, 123
Morgan, E. E. (see Ganguillet and
Kutter) 6
Morris, H. M. 41, 108
multi-purpose reservoirs 79, 80
NAPPE 88, 89, 92
navigational requirements 76
nonuniform channel flow 23
definition of 3
normal depth 9-11
line (N.D.L.) 37
normal storage in reservoirs 79
notches 87
numerical step methods-direct step
method 42
flood routing methods 129-135
standard step method 46

Flow in Channels

148
OBLIQUE SHOCK W AYES

72

orbital particle motion 123

outflow discharge 78, 129
overbank spill 113
overfall, free 36
ox-bow lakes 111

94
Paulhus, J. L. H. (see under Linsley)
79
permissible canal velocities 115
piers-flow through bridge piers 71
Posey, C. J. 48
power generation with reservoirs 79
power storage 79
Prandtl, L. 14
prism storage 132, 135
profiles, longitudinal surface 35-48
pARSHALL FLUME

RADIAL GATES 83
rainfall-discharge relationships 126128
Rao, V. S. 77
rapid flow-see supercritical flow
rapidly varied flow 49
rationed storage 79
reach 129
rectangular channel section 11, 16
rectangular notch-see weirs
regime method of canal design 114,
116
regime, rivers in 101, 107
regime theory-qualitative results
108, 109
regime equations 108
regulation of channels-see control of
flow
regulators-see barrages
Rehbock equation 89
rejection surge 126
reservoir useful life 79, 118
reservoirs-as channel surface profile
controls 50
flow between 51
functions of 76, 132
river, functions of a 73
river gauging 94
ripples 102
rolling gates 83
roughness coefficient 'rJ 6, 7, 14
Rouse, H. 13, 52, 54, 86

SALTATION

100

secondary circulation of the first kind

13, 110, 112
of the second kind 13
section factor 8
sediment-classification and origins
99
sediment deposition in reservoirs 79,
117
sediment removal from reservoirs 120
sediment transport 100--120
semi-circular channel sections 16
settling basins for reservoirs 80, 119
sewers 18
Sherman, L. K. 128
shooting flow-see supercritical flow
silt-discharge curve 105
Simons, D. B. 116
slope critical 32
sluice gates 61, 84
sluice-horizontal forces on the structure 61
solitary waves 121, 126
specific energy 24, 57, 62
specific energy curve 26, 28
specific force 55, 57, 62, 65, 67
specific head-see specific energy
Speight 111
spillway aprons 40
spillways 38, 40
stable canals design 114
stage 95
stage-discharge relationships 95, 130
stage-storage relationships 131
standing wave flume 93
standing wave train on surface 40
steady flow, definition of 2
steep slope 32, 40
storage-definitions 79, 80, 129, 131,
132
stratification in reservoirs 119
Straub, L. G. 103
streaming flow-see subcritical flow
streamlines 112, 121
subcritical flow 27, 28
supercritical flow 27, 28, 71
submerged obstacles, forces on 64,
67
suppression of point of maximum
velocity 12
surface profiles, longitudinal-classification 35-40
worked example 42-48
surges 124

149

Index
surge, demand 126
rejection 126
stationary-see hydraulic jump
suspended sediment load 100, 104,
117
movement 100, 104
TAINTOR GATES 83
Tennessee valley authority 74
throated flumes 92
tidal bore 126
total energy 23, 28
total energy line 24
total sediment load 106
tractive force, critical 17
tractive force design method 115-117
tranquil flow-see subcritical flow
transitional jump-see hydraulic jump
transitions-changes in cross-section
68-72
flow through critical depth 51-55
translatory wave motion 123
trap efficiency of reservoirs 79, 119
trapezoidal channel section 16, 17,
94
two-point method for mean velocity
determination 14, 105
UNDULAR JUMP-see hydraulic jump
undular surge 125
uniform channel flow 2-22
definition of 3
computations 8
uniformly progressive wave 123
unit hydrograph 127

unsteady flow 121-135

definition of 2
VELOCITY-AREA METHODS 95, 96
velocity distribution-in a cross-section
11,96
in an ice-covered river 22
in a vertical 12, 14
velocity formulae, empirical 3-9
ventilation of weirs 88
vents 81
venturi flume--see throated flumes
V-notch weirs 90
WASH LOAD 104
water conservation by reservoirs 79
Water Resources Act 1963 74
waves 121-126
waves, flood 78, 126
monoclinal 54, 123
solitary 121
Webber, N. B. 89
wedge storage 135
weirs-broad-crested weirs 91
full width weirs 88
fully contracted weirs 89
partially contracted weirs 89
thin-plate weirs 87-91
V-notch weirs 90
wetted perimeter 4, 116
width, changes in channel width 6872
Wilson, E. M. 128
Woodward, S. M. 46

YARNELL,

D. L. 71