Вы находитесь на странице: 1из 254

KINEMATIC HYDROLOGY

AND MODELLING

DAVID STEPHENSON
Department of Civil Engineering, University of the Witwatersrand, I Jan Smuts Avenue,
2001 Johannesburg, South Africa

and

MICHAEL E. MEADOWS
Department of Civil Engineering, University of South Carolina, Columbia,
SC 29208, U.S.A.

ELSEVIE R

Amsterdam - Oxford - New York - Tokyo 1986


ELSEVIER SCIENCE PUBLISHERS B.V.
Sara Burgerhartstraat 25
P.O. Box 21 1,1000 AE Amsterdam, The Netherlands

Distributors for the United States and Canada:

ELSEVIER SCIENCE PUBLISHING COMPANY INC.


52, Vanderbilt Avenue
New York, N Y 10017, U.S.A.

Library of Congress Cataloging-in-PublicationData

Stephenson, Divid, 1943-


Kinematic hydrology and modelling.

(Developments in water science ; 26)


Bibliography: p.
Includes indexes.
1. Runoff--Mathematical models. 2. Groundwater
flow--Mathematical models. I. Meadows. Michael E.
11. Title. 111. Series.
GBg8O.S74 1986 551.48'8'0724 86-2175
ISBN 0-444-42616-7

ISBN 0444-42616-7 (Vol. 26)


ISBN 044441669-2 (Series)

0 Elsevier Science Publishers B.V., 1986

All rights reserved. No part of t h i s publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means, electronic, mechanical, photocopying, recording or other-
wise, without the prior written permission of the publisher, Elsevier Science Publishers B.V./Science &
Technology Division, P.O. Box 330, 1000 A H Amsterdam, The Netherlands.

Special regulations for readers i n the USA - This publication has been registed with the Copyright
Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained from the CCC about
conditions under which photocopies of parts o f this publication may be made in t h e USA. A l l other
copyright questions, including photocopying outside of the USA, should be referred t o the publisher,
Elsevier Science Publishers B.V., unless otherwise specified.

Printed in The Netherlands


V

PREFACE

Many stormwater design engineers and indeed hydrologists wi I I be


frustrated by the lack of hydraulic principles in some o f t h e c o n v e n t i o n a l
methods of flood calculation. The Rational method and unit hydrograph
methods are easy to apply but limited in accuracy and versatility.
Kinematic hydrology is the next logical step in s o p h i s t i c a t i o n b e f o r e the
full hydrodynamic equations are resorted to. The kinematic equations in
fact comprise the continuity equation and a hydraulic resistance
equation. In many cases solution of t h e s e e q u a t i o n s f o r f l o w r a t e s and
water depths is simple and explicit. In more complicated problems the
e q u a t i o n s m a y b e u s e d to s i m u l a t e t h e r u n o f f p r o c e s s .
U n f o r t u n a t e l y much of the l i t e r a t u r e o n the k i n e m a t i c method h a s been
highly mathematical and often of an e x p e r i m e n t a l nature. The equations,
graphs and models published are therefore of little use to the practical
engineer, and may discourage him from using this method. In fact once

confidence is gained, the method can be applied in simple form to a

variety of catchments. The term kinematic refers to movement where


accelerations are negligible - which is generally applicable to overland
a n d shallow stream flow.
The book i s aimed at both the theoretician and t h e p r a c t i t i o n e r . Thus
the mathematical sections are useful if modelling is required, but the
chapters on design charts could be read with very little mathematical
understanding other than a basic appreciation of the kinematic method.
L i t t l e mathematical background i s required, a n d no computer knowledge i s
necessary for those sections. It is hoped that the peak flow charts will
provide an alternative to the Rational method and the SCS method for
estimating runoff. Simi l a r l y the dimensionless hydrographs are
competitive with unit hydrograph methods. The user will gradually
become a w a r e o f the f a c t that the k i n e m a t i c method i s f a i r l y easy to a p p l y
if simple solutions are required. It also permits consideration of m a n y
more factors than some other methods of flood calculation, which in turn

can only improve accuracy and provide for greater u n d e r s t a n d i i g of the


r u n o f f process.

Of course the k i n e m a t i c method i s not the final answer in h y d r o l o g y .

There are many questions still to be answered, and some degree of


simp1 i f i c a t i o n is still required. Although the method provides a logical
way of visualizing runoff, actual runoff from many catchments comprises
vi

part overland, subsurface and i n t e r f a c e flow. The combined effect cannot


easily be modelled. Also water does not run off rural catchments in a
sheet - i t f r e q u e n t l y forms r i v u l e t s a n d i s d i v e r t e d b y obstacles w h i c h can
be loosely termed roughness. Some of these f a c t o r s can be accounted f o r b y

adjusting the hydraulic factors used in the equations, or calibrating


models.
Results of research and development are now advanced and
experience in application is required before general acceptance of the
k i n e m a t i c method can be hoped f o r . I n p a r t i c u l a r the a b i l i t y to select s o i l
losses, roughnesses and catchment geometry to adequately describe the
h y d r a u l i c s of the system, can o n l y be g a i n e d w i t h experience.
The scope of the kinematic method is therefore unlimited from the
point of view of the researcher with an enquiring mind. Some of the
theoretical considerations are taken further in chapter 2 on kinematic
equations, 4 on assumptions a n d 5 on numerical theory f o r modelling.
On the other h a n d the p r a c t i t i o n e r i s p r o b a b l y more interested i n the
best answer available. He may manage quite sufficiently reading only
chapter 3 on peak flows, chapter 6 with dimensionless hydrographs and
possibly chapter 7 on m a r g i n a l effects and 9 with some examples of the
value of the techniques. Hopefully he will be inspired to go into

modelling, w h i c h may bring i n c h a p t e r 8 on flow i n c o n d u i t s , a n d 10, 11


a n d 12 w i t h examples of computer models of v a r i o u s catchments.

Much of the m a t e r i a l in t h i s book i s d e r i v e d from notes f o r a course


presented by the authors. There is copious reference to previous
research in kinematic h'ydrology, as well as new material arising from
research by both authors. In particular the senior author was the

recipient of a research contract in urban hydrology from the Water


Research Commission.
The manuscript was typed into its final form by Janet Robertson, for
which the authors a r e most g r a t e f u l .
1

CHAPTER 1

I NTRODUCT I ON

HISTORICAL R E V 1 EW

Kinemat i c hydrology provides a method for estimat i n g stormwater


runoff rates and volumes. It is p a r t i c u l a r l y useful for flood calculation.

It is a relatively new term embracing techniques w h i c h h a v e been a p p l i e d


for many decades. Kinematic hydrology is decidedly more hydraulically
correct than some of the more common methods of f l o o d estimation such as
the r a t i o n a l method, t i m e - a r e a methods, the Soil Conservation Service ( S C S )
method and unit hydrograph methods. The kinematic method is based on
the continuity equation and a flow resistance equation, both basic

h y d r a u l ic equations.
It was the American hydrologist, Horton, (generally associated with
i n f i l t r a t i o n ) who i n 1934 c a r r i e d out the e a r l i e s t recorded s c i e n t i f i c studies

of overland flow. Later Keulegan (1945) applied the continuity and


momentum equations conjunctively for overland flow analysis. He investi-
gated the magnitude of the v a r i o u s terms in the dynamic equation of St.
Venant and indicated that a simplified form of the equation, now, termed
the kinematic e q u a t i o n , would be adequate for o v e r l a n d flow.
An in-depth analysis of the differential continuity and resistance
equations was undertaken by Lighthill a n d Whitham (1955) to whom the
designation kinematic waves 'can be a t t r i b u t e d . They a l s o f i r s t s t u d i e d the
phenomenon of k i n e m a t i c shock w h i c h can be a p p l i e d to d i s c o n t i n u i t i e s in
flow and water depth. Although they suggested the k i n e m a t i c approach f o r
overland flow modelling, it was Henderson and Wooding (1964) who
obtained analytical solutions to the kinematic wave equations for simple
p l a n e a n d channel shapes. A g e n e r a l i z a t i o n of the catchment stream model
was a l s o described b y Eagleson (1967).
The f u l I dynamic equations f o r one-dimensional incompressible flow in

open channels were set down b y St. Venant in 1871. These equations were
for gradually varied unsteady flow such as flood waves. The idea of
graphical integration using characteristic I ines was first suggested by

Massau in 1889. On the other hand Greco and Panattoni (1977) indicate
that i m p l i c i t solution by f i n i t e differences i s the most e f f i c i e n t method b y
computer, avoiding instabi I i ty and giving rapid convergence. Various
numerical methods of s o l u t i o n of the k i n e m a t i c equations were i n v e s t i g a t e d
by Kibler and Woolhiser (1970). The step length in finite difference

schemes plays an important role in the stability of the s o l u t i o n (Singh,


2

1977). Non-convergence was investigated for p l a n e cascades b y Croley a n d

Hunt (1981 ) . Brakensiek (1966) used numerical solutions to the k i n e m a t i c

wave equations for the a n a l y s i s of surface runoff from r u r a l watersheds.


He p r o b a b l y d i d not realise the extent to which numerical m o d e l l i n g would

advance i n l a t e r years u s i n g the k i n e m a t i c e q u a t i o n a n d s q u a r e x-t grids.


The latter approach does not warrant appendage of the term 'wave' to
k i n e m a t i c since d i s c o n t i n u i t i e s a r e lost i n the simp1 i f i e d numerical method.
Wooding (1965 and 1966) presented a comprehensive review of the
theory of k i n e m a t i c waves a n d used numerical solutions to d e r i v e equations
for the rising and falling limbs of hydrographs for simple planes and
channel configurations. During the 1970's the equations were applied to
more complicated catchment shapes (Schaake, 1975), in particular the
catchment-stream model, the c o n v e r g i n g catchment a n d cascades of p l a n e s .
Although analytical solutions are available for some cases the m a j o r i t y of

solutions are numerical, a n d dimensionless h y d r o g r a p h s facilitate the use


of the r e s u l t s of the s t u d i e s (Constantinides a n d Stephenson, 1982). Since
the studies b y Henderson a n d Wooding (1964) a n d l w a g a k i (1955) the shock
wave phenomenon has not really received much attention and for this

reason the use of the name k i n e m a t i c theory i s now considered adequate as


it implies a more general applicability than to waves. I n f a c t Borah a n d
Prasad (1982) indicate shock waves may in fact not e x i s t i n some cases
where predicted using the kinematic equations. This is because the
k i n e m a t i c equations may not a p p l y where the s p a t i a l variation i n depth i s
large. Even the St. Venant equations may not suffice to describe rapid
v a r i e d flow, as v e r t i c a l ' a c c e l e r a t i o n s a r e not considered.
Woolhiser and Liggett (1967) investigated the applicability of the

kinematic equations and proposed a dimensionless parameter indicating


whether the equations are adequate for any particular case with simple
geometry. More recent research (Morris and Woolhiser, 1980) has
investigated i n greater detail the a p p l i c a b i l i t y of the k i n e m a t i c equations
to d i f f e r e n t conditions.
The application of kinematic theory has more r e c e n t l y been extended
to problems such as dynamic storms (Stephenson, 1984a), detention storage
(Stephenson, 1984b), urban drainage networks (Green, 1984) and to the

effects of u r b a n i z a t i o n a n d storm r u n o f f (Stephenson, 1983).

There i s as yet l i t t l e general d a t a a v a i l a b l e on s u r f a c e water losses


(infiltration, and retention) to be used w i t h k i n e m a t i c equations. Skaggs
(1982) reviewed infiltration mechanics including the popular Horton model
a n d more advanced Green-AmDt model.
3

The majority of papers differentiate between surface and subsurface

flow, i.e, overland flow is treated independently. Rovey et at. (1977)


developed an interactive infiltration model to account for non-uniform

soil losses. A further development by Freeze (1972) allows for contri-


butions from re-appearing shallow groundwater flow in a saturated
aquifer.

retention u

Fig. 1 .1 S i m p l i f i e d catchment l o n g i t u d i n a l section

Definitions

Some terms used in this text are used in d i f f e r e n t context elsewhere

so to avoid confusion particularly with respect to times, some d e f i n i t i o n s


a r e g i v e n below.

Time to equilibrium (t,) is the time taken from the commencement of


precipitation until the water p r o f i l e down the catchment is in e q u i l i b r i u m
and inflow equals outflow everywhere, i.e. runoff rate i s equal to excess
rainfal I rate, assuming steady p r e c i p i t a t i o n a n d losses.
Time of concentration (tc) is the time from the commencement of precip-
itation until the effect of excess precipitation everywhere in the catch-

ment has appeared at the outlet. It is equal to the time to equilibrium


for steady excess rain using kinematic theory whereas it is equal to
travel time with time area-theory. It is demonstrated later that for a

simple plane, k i n e m a t i c theory yields


t = (Lie’-m/ajl’m (1.1)
c
m-1
where L i s the length of flow path, a n d flow velocity V = ay where
y w a t e r d e p t h a n d m and a a r e c o e f f i c i e n t s d e f i n e d
is by the equation
m
q = ay where q i s the flow r a t e p e r u n i t w i d t h .
4

0 t
td
Fig. 1.2 Catchment water b a l a n c e

Travel time (t,) is the time for a p a r t i c l e of water to proceed from the
most remote p a r t of the catchment to the d i s c h a r g e p o i n t . For a plane it
is not equal to time of concentration a c c o r d i n g to k i n e m a t i c theory since
water moves slower than a hydraulic response which travels at wave

speed. I t i s shown later that for a plane


tc = tt/m (1.2)

L a g time ( t ) i s the time between 50% of p r e c i p i t a t i o n a n d 50% of r u n o f f .


L
I t w i l l be shown that f o r a p l a n e
tL = mtc/(l+m) (1.3)
Storm duration td is the time from commencement of precipitation until it
ceases. Frequently when storm records a r e a n a l y z e d f o r intensity-duration
relationships storm duration i s d e f i n e d as the time during which average
storm intensity is a specified figure, so that storms within storms can

occur.
T i m e of excess r u n o f f (t ) i s the time measured from the commencement of
runoff. It is therefore less than the time t from the commencement of

precipitation by tu = u/i where u is initial abstraction a n d i i s the


p r e c i p i t a t i o n r a t e (see F i g . 3.3 on page 4 9 ) .
Units of time a r e g e n e r a l l y seconds
___ i f the System I n t e r n a t i o n a l (S.I.) units
of metres, seconds and kilograms, or the old English system of foot,
seconds and pounds are adopted. Later herein modifications for more
practical u n i t s e.g. rainfall i n mm/h o r inches per hour, a r e introduced.

CLASSICAL HYDROLOGY

For various reasons flood hydrology has been a f a i r l y s t a t i c subject


for many decades. The r a t i o n a l method which was invented over 100 years
ago, and hydrograph theory, developed over 50 y e a r s ago, a r e s t i l l used
extensively. If we reconsider the assumptions a n d l i m i t a t i o n s b e h i n d these

methods we may be prepared to consider developing new techniques more


5

a p p r o p r i a t e to o u r technology a n d more accurate.


The s i m p l e l i n e a r h y d r o l o g y methods were p r o b a b l y developed f o r ease
of manual calculation, and as many hydrologists do not have a strong
mathematical background. It is true that some of the standard methods
have been programmed for computers. This facilitates the subdivision of
catchments but does not eliminate the limitations of many of the
assumptions b e h i n d the methods.
The c u r r e n t a v a i l a b i l i t i e s of computers to a l l should c o n s i d e r a b l y ease
the next step - breaking away from simple input-output methods and

introducing more sophisticated hydraulic equations in their stead. It is


possible to simulate water flow and water surface profiles with
considerable accuracy with the aid of computers, even micro computers.
There a r e v a r i o u s l e v e l s of s o p h i s t i c a t i o n which can be adopted to s u i t the

problem a n d the machine a v a i l a b l e .


These methods are based on solution of finite difference versions of
the differential equations of flow. Computations proceed in increments of

time a t selected i n t e r v a l s i n space. There h s v e been numerous advances i n


numerical methods in mathematics in parallel with the developments in

computers. On the other h a n d the a p p r o x i m a t i o n of differentials by finite


increments can lead to in'accuracins unless certain rules are complied
with. Some of the common problems a r e instability, numerical diffusion or
accumulating errors. The correct finite increments can be selected to
approximate the differentials to a first order, second o r d e r or greater
order if necessary. There a r e a l s o methods f o r solving i m p l i c i t equations
such as by gradient convergence or successive approximation. Where a-
number of simultaneous equations h a v e to be solved over a g r i d there a r e
m a t r i x methods a n d r e l a x a t ion methods a v a i l a b l e .
One of the greatest aids to the engineer nowadays may be the desk
top micro computer. Whereas practitioners tend to shy away from main
frame computers ( i f they can access one a t a l l ) the problems of j o b control

language, q u e i n g b a t c h jobs, formal programming a n d debugging a n d r i s k


of runaway costs are no longer of concern. The kinematic method is

intermediate level technology applicable to micro computer solutions,


whether a n a l y t i c a l solutions o r numerical model I i n g i s contemplated.
The basis for much of our hydrology probably originated with an

I r i s h engineer, Mulvaney, i n 1851. He proposed an e q u a t i o n f o r r u n o f f , €I


= KA. K allows for a rainfall intensity but t h i s was not a significant
variable in Britain. The method was taken a step further by introducing
an equation for excess r a i n f a l I i n t e n s i t y , e.g. the Birmingham f o r m u l a ,
6

(1.4)

where i is in inches per hour and t is the storm duration in minutes.


No allowance is made for extreme storms and this equation is for a 1
to 2 year frequency storm. T h e 20 w a s accepted b y some a s representing

a time of entry in minutes (equivalent to the defined concentration time


of overland flow).
It was assumed that 100% r u n o f f occurred from impermeable areas
and none from pervious areas. This assumption was not acceptable in
areas of high r a i n f a l l intensity and in t h e U n i t e d S t a t e s w h e r e K u i c h l i n g
in 1889 modified the runoff equation to Q = CiA where the coefficient
C i s a f u n c t i o n of the catchment.
The coefficient C is most strongly associated with the average
p e r m e a b i l i t y of the catchment - t h u s 100% r u n o f f w o u l d o c c u r i f C is unity
a n d no r u n o f f f o r a completely permeable catchment. Modifications to C a r e
made to account for catchment slope, vegetation cover and so on by

various people. It has also been realized that dntecedent moisture


conditions and severity of the storm (represented by the recurrence

interval) can affect C. For instance Rossmiller (1980) proposed the


following empirical equation for C:

.2
1.48(.15-1) ~ + l e 7
(1 3)
C = 7 . 7 ~ 1 0 - ~ C ~ ' R ~ ~ ( . 0 1 C( .001
~ ) -CN)
~~
(T)
where R is the recurrence interval, S is bed slope in percent, I is
rainfall intensity i n inches p e r h o u r , M i s the f r a c t i o n of watershed w h i c h
i s i m p e r v i o u s a n d CN the S o i l C o n s e r v a t i o n S e r v i c e (SCS) c u r v e n u m b e r .
The assumption of a unique 'C' for any catchment can lead to
s i g n i f i c a n t e r r o r s and u n d e r e s t i m a t i o n o f f l o o d r u n o f f . T h i s i s demonstrated
by F i g u r e 1.3. The runoff rate per unit area for case 'a' is Cil. If the
same C i s used f o r case b, where a higher rainfall i n t e n s i t y occurs, the

loss will be greater and the runoff proportional. A loss which is


independent of rainfall intensity however would produce a runoff as for
case c, which is proportionally greater than for case b . The as s um pt ion
f o r case b t h u s r e s u l t s in a n u n d e r e s t i m a t e o f f l o o d r u n o f f .
In general then, it is implied in the Rational method that runoff

intensity is I inearly proportional to rainfall intensity. T h i s a l s o assumes


that the catchment has reached a n equilibrium, so it became n e c e s s a r y to
estimate the 'concentration time' of catchments. Lloyd-Davies developed
this idea in 1905 and proposed that the maximum peak runoff from a
catchment occurred for a storm with a duration equal to t h e c o n c e n t r a t i o n

time of the catchment. A common equation used for concentration time is


7

0.385 (1.6)
tc = (0.87L3/H)
where t is i n hours, L i s the l e n g t h of catchment i n km a n d H the drop
i n metres, or
t = (11 .6L3/H)0.385 (1 -7)

where L i s i n miles a n d H i n f t .

The r a t i o n a l method does not produce a complete h y d r o g r a p h capable


of routing and so unit hydrograph theory was developed. The theory was
based on the assumption that two units of excess rain produce a
hydrograph with ordinates twice those of a hydrograph produced b y one
unit of excess r a i n in the same time. The term linear hydrology i s often
applied to this theory. The time scale is also incremented l i n e a r l y . Two
successive u n i t s of rain a r e assumed to produce two u n i t h y d r o g r a p h s in
succession which can be added together at all points in time. We thus
have the S-curve h y d r o g r a p h which i s caused b y a n i n f i n i t e l y long storm.
Unit hydrographs do not account for the non-linear response of a
catchment to excess rain. Neither is the concentration time of any
catchment area a unique time, it depends on the flow rate, as seen f o r
instance, in the Manning equation ( 2 . 4 7 ) . I n any case the travel time is
not the same as the r e a c t i o n time which i s also a function of flow r a t e .
Non linear hydrograph theory on the other hand has met with limited
response.
To some extent the error in assuming the travel time is the

concentration time is n u l l i f i e d by assuming a f u l l conduit f o r computation


of travel time. The upstream conduits flow at a lower rate than those
downstream. When the design storm i s o c c u r r i n g f o r a downstream c o n d u i t ,
upstream conduits will be flowing at less than design capacity as the
storm duration will be greater than the design storm for the upper
conduits. Thus the assumption of a higher flow and velocity than will
occur makes the resulting r a t e of concentration more nearly t h a t of the
true h y d r o d y n a m i c system.
Another misconception is that the ful I catchment must contribute for
the maximum runoff rate. Besides odd shaped catchments which can by
analysed u s i n g the tangent method (Watkins, 1962) a true a n a l y s i s would
show many catchments do not contribute from the farthest extremity at
peak flow. This i s not shown up b y the r a t i o n a l method w h i c h i n v a r i a b l y
assumes the entire catchment contributes. I t can be demonstrated o n l y if
soi I-dependent losses a r e assumed, not rain-dependent losses (e.g. 'C'). I t
is shown in chapter 3 that if loss is independent of rainfall then a
shorter duration storm in many cases produce a g r e a t e r runoff r a t e than
one which i s of d u r a t i o n e q u a l to the time to e q u i l i b r i u m .
a

rainfall a n d r u n o f f r a t e s
per u n i t a r e a of c a t c h m e n t

rainfall r a t e il
runoff C i ,

loss = (I -C)il=f
1 I
time t

( a ) Medium storm

i
rainfall runof rate

r a i n f a l l r a t e i,

runoff = Ci2

( b ) Intense storm assuming same C as in (a) above

rainfall runoff r a t e
I

I 1 loss. f

( c ) Intense storm with same loss as ( a )

Fig. 1.3 E f f e c t of c o n s t a n t C on r u n o f f
9

HYDRODYNAMIC EQUATIONS

The Navier-Stokes equations for incompressible fluid flow in three


dimensions a r e

a u au
o(-+U-+V-+W-)
au au = x - 32 + I-’(-+-+-)
3 ’ ~ a 2 u azu
a t ax ay az ax ax2 ay2 az2

p ( 4a uv - + vav
- + w - av
) a~ = Y - 3+ azv
p(-+-+-) azv azv
(1.9)
at ax ay a~ aY ax2 ay2 az2

o(a”+ua”+va”+wa”) = z ap + I-’(---a 2 w a 2 w a--)2 w


- - ~

(1.10)
at ax ay az az axz a y 2 a z 2

where p is the mass density of the fluid, u,v,w, are the velocity
components in the x,y,z directions respectively, X,Y,Z, are the body
forces per unit volume, p i s the pressure and p i s viscosity. I n addition
to these t h r e e d y n a m i c e q u a t i o n s we h a v e t h e c o n t i n u i t y equation

__
au + + W
L = o (1.11)
ax aY az

Although these f o u r equations theoretically describe flow in a n y s i t u a t i o n ,


from the point of view of civil and hydraulic engineers they suffer a
number of drawbacks. For instance viscous forces should be replaced by
turbulent momentum transfer or even by a serni-empirical friction drag
equation, e.g. b y Manning o r Darcy.
It is generally possible to work in one dimension in civil engin-
eering hydraulics. Then the Navier-Stokes equations can be replaced by
the St. V e n a n t equations, which also comprise a dynamic equation and a

continuity equation, namely

_l -a v + v a v + a v + fs- so = o (1.12)
g at g ax ax

(1.13)

where S is the b e d slope (positive down in the x direction), Sf i s the


energy gradient, Q is the flow rate, B the surface width, A the cross
sectional area and P the wetted Derimeter. It will be seen on close
inspection that t h e St. V e n a n t equations are similar in m a n y terms to the
Navier-Stokes e a u a t ions.
10

The solution of the St. Venant equation is, however, a difficult


enough task for the hydrologist or c i v i l engineer. The c l a s s i c a l solution
is by the method of characteristics which can easily be portrayed
graphically. Computer solution of the equation in various forms is now
more common. Rapid s o l u t i o n of a finite d i f f e r e n c e form o f the St. Venant
equations in a simplified form can easily b e u n d e r t a k e n on, for instance,
m i c r o computers.
For the majority of overland flow cases and in many channel and
conduit flow situations the St. Venant equations can be replaced b y the

f o l l o w i n g two e q u a t i o n s (see c h a p t e r 2 ) .

Continuity aaxa + , aatv = i


e
(1 .14)

Dynamics S = Sf (1.15)

where i is the input per unit area of surface (e.g. excess rainfall
intensity).
These e q u a t ions are termed the kinematic equations. Equation (1.15)
merely states that the bed slope can be substituted for the energy
gradient in a f r i c t i o n equation.
F o r o v e r l a n d sheet f l o w q p e r u n i t w i d t h these e q u a t i o n s become

(1.16)

m
4 = aY (1.17)

where i i s the excess r a i n f a l l r a t e .


It is further a s i m p l e m a t t e r to transform the kinematic equations (1.14)
a n d (1.15) i n t o e q u a t i o n s a p p l i c a b l e to s t o r a g e r e s e r v o i r s w i t h interlinking
conduits:

ah
AQ + A- = q (1.18)
at

and AH/L = KQm (1.19)

Here A is the reservoir surface area, Q is the net inflow from


connecting pipes a n d q i s the d r a w o f f from a r e s e r v o i r w i t h w a t e r l e v e l h .

The second e q u a t i o n i s a p p l i c a b l e t o c l o s e d c o n d u i t s a n d i n f a c t i s simpler


than t h e open channel k i n e m a t i c e q u a t i o n s i n c e the v a r i a b l e flow d e p t h i s
el i m i n a t e d .
When the common node between conduits is an open reservoir the
continuity equation w i l l predict the rate of change in water level. I f the
conduits or pipes connect at a closed node it is necessary to solve

simultaneously for head at the node and flow in the connecting pipes.
11

Many methods a r e a v a i l a b l e for this, but the l i n e a r method (Stephenson,


1984b) is particularly suitable. That procedure requires minimal data
preparation and solution is faster than the manual node iterative
correction procedure of Hardy Cross because it is implicit, that i s heads
of all nodes are solved for simultaneously. The kinematic method of
continuous s i m u l a t i o n is a versatile technique f o r a n a l y s i s of u r b a n storm

drainage and water supply pipe networks particularly w h e n operation of


storage r e s e r v o i r s i s i n v o l v e d .
The I imiting assumptions b e h i n d the k i n e m a t i c method should however
be recalled. Although the assumption that the x - d i f f e r e n t i a l terms in the
dynamic equation is zero i s certainly valid, the time differential terms
may in some cases not be zero. T h i s effect i s magnified by introducing
closed c o n d u i t s w i t h u n v a r y i n g cross-sectional area. Pressure r i s e s due to
change i n flow r a t e can be l a r g e , g i v i n g r i s e to water hammer.
In such situations, i.e. when rapid fluctuations in flow are
possible, an a l t e r n a t i v e method of analysis, namely e l a s t i c a n a l y s i s , must
be employed. To analyse a network using the water hammer equations
involves simultaneously solving the c h a r a c t e r i s t i c s a n d c o n t i n u i t y e q u a t i o n
at each node. Aspects of f r i c t i o n damping r e q u i r e p a r t i c u l a r a t t e n t i o n w i t h
this method. I n particular the r a t i o of f r i c t i o n head loss to water hammer
head can have an important effect on the speed of solution. When the
analyst is only concerned with steady state heads and flows he can
artificially speed convergence b y suppressing the wave speed i.e. reducing

the numerical v a l u e used i n the computations.


The a n a l y s t i s thus altering the f i t of the mathematical model to the
real system. There are approximations and consequently scope for
adjustment at a number of stages in the modelling. The f o l l o w i n g stages
a r e r e l a t e d b y the a n a l y s t :
Real system ( c o n d u i t s a n d r e s e r v o i r s )
Imagined system ( w h a t can be v i s u a l i z e d )
Mathemat ical model ( d i f f e r e n t i a l equations)
Numerical model ( f i n i t e differences)
Computer model (successive equations)
By adjusting the imagined system one i s a b l e to speed convergence of
the solution. The finite differences have to be limited according to the
Courant criterion (1956) and particularly when friction is involved,
another c r i t e r i o n proposed b y Wiley (1970)

At < :Ax/c) (l-SgAt/2v)”‘ (1.20)


12

Equation (1.20) indicates that friction affects the stabi I i t y of


numerical solutions. This i s however due to the numerical a p p r o x i m a t i o n in
solving the equations explicitly rather than an instability caused by
friction. F r i c t i o n has g e n e r a l l y an important role i n kinematic theory. It
relates water depth to flow rate i.e. it provides the link between the
continuity equation and the hydrograph. Although friction energy loss
relationships a r e well known f o r stream flow which i s f u l l y turbulent and

sub-surface flow which is laminar, the process of overland flow is not


fully appreciated. Flow depths are small and the dimensions of
roughness are comparable with the flow depth. There are complicating
influences such as tortuous flow paths around and over boulders,
vegetation, structures and other surface disturbances. Rain drops are

reported to cause turbulence at lower Reynolds numbers than for conduit


flow. Overton a n d Meadows (1976) i n d i c a t e t u r b u l e n t flow p e r s i s t s f o r sheet
flow if the Reynolds number in terms of precipitation rate, i L / " = 20 to
2000 where i is the p r e c i p i t a t i o n rate (m/s), L i s the o v e r l a n d flow p a t h
length and L) i s the k i n e m a t i c v i s c o s i t y of the l i q u i d ( w a t e r ) . T h i s would
indicate that the energy g r a d i e n t i s proportional to flow r a t e to the power
of m = 5/3 if the Manning equation (2.47) i s assumed together w i t h the
1/6 power law for velocity distribution. Horton (1938) on the other hand

found m was a p p r o x i m a t e l y 2 on n a t u r a l surfaces implying nearly laminar


conditions for uniform flow (constant depth in the direction of flow).
Actually m = 3 f o r p u r e l a m i n a r flow.

TENSION

\
CONTROLS

GRAVITY

_.
TIME

Fig. 1.4 Typical field infiltration curve


13

I NF I LTRAT I ON

A major component of a stormwater model i s the routine to determine


the rainfall excess. Abstractions or losses are subtracted from input

rainfall resulting i n the r a i n f a l l excess which must be r o u t e d to the b a s i n


out l e t .

The losses w h i c h must be a b s t r a c t e d from r a i n f a l l a r e :

1 Intercept i o n - r a i n f a l I caught by vegetation prior to reaching the


ground. The amount caught is a function of (a) the species, age,
and density of vegetation, ( b ) character of the storm, and (c) the
season of the year. It has been estimated that in a rural watershed
as much as 10 to 20 percent of the rainfall during the growing
season i s intercepted a n d r e t u r n e d to the atmosphere b y evaporation.

2. Depression storage-water caught in smal I surface pockets and voids


h e l d there u n t i l it i n f i l t r a t e s o r evaporates.

3. Evaporation-water returned to the atmosphere through vaporization.


Evaporation i s most important when i t i s not r a i n i n g ; it is negligible
during rainfall events when a representative rate is 0.05 mrn/hr
(0.002 in/hr) (Overton a n d Meadows, 1976).

4. Infiltration-water lost to the soil. Typically, infiltration is the

major abstraction during a rainfal I event. Three distinct processes


are involved: (a) the movement of water into the soil across the
air-soil interface (infiltration); (b) the movement of water through
the soil under the influence of gravity and soil suction (percol-
ation); and (c) the depletion of the available volume within the
soil (storage depletion).

There are two basic approaches to modelling rainfall excess. Each


loss can be modelled separately and the models linked together, or a
single model can be developed that lumps the important losses together,
usually into infiltration. This latter approach i s o f t e n followed i n event
Simulation models. Kinematic stormwater models are mostly event models;
therefore, we are mostly concerned with infiltration models f o r the r a i n -
fall a b s t r a c t i o n model.
A typical field i n f i l t r a t i o n curve i s shown in F i g u r e 1.4. Infiltration
begins a t an initial h i g h r a t e a n d decreases with time to a steady final

rate. The forces influencing the movement of water into and through the
14

soil are suction and gravity. During the early stages, the upper soil
layer is "thirsty" and infiltration i s dominated b y suction. With time, the
u p p e r centimetre, more o r less, of the s o i l s u r f a c e becomes s a t u r a t e d a n d
the infiltration r a t e reduces to that rate at which water moves through
the saturated soil. At this point, gravity dominates. As long as the
rainfall rate exceeds the instantaneous infiltration rate, or water is
ponded on the surface, infiltration w i l l continue a t the maximum possible
rate, defined b y Horton (1933) as the c a p a c i t y i n f i l t r a t i o n rate. The effect
of rainfall r a t e on the infiltration curve i s next examined. Three general
cases f o r infiltration during a steady rainfall were proposed b y Mein a n d
Larson (1973):

Case A : i<k . (The r a i n f a l I r a t e , i, i s less than the s a t u r a t e d s o i l


hydraulic conductivity, k .) Under this condition, runoff will not
occur, regardless of rainfall duration, because all rainfall will

i n f i Itrate.

Case B: k s < i< f


P
. (The rainfall rate is less than the capacity
i n f i l t r a t ion rate, fp, but is greater than the saturated hydraulic
conductivity.) For a short duration rainfall, where i remains less
than f all the rain infiltrates. But for a rainfall of long
P'
duration, the infiltration capacity will decrease until it equals i,
a n d surface p o n d i n g occurs.

Case C: k < f <i: (The r a i n f a l l rate i s greater than the infiltration


-__ s p
capacity.) Under t h i s c o n d i t i o n , r u n o f f occurs.

Cases 0 and C can be considered as two distinct cases; however,


infiltration often occurs as a two-phase process combining the two cases.
Bodman and Colman (1943) evaluated soi I water distribution during
infi Itration into a uniform, relatively dry soi I under surface ponding
conditions and established that the typical p r o f i l e can be divided into

four zones as shown in F i g u r e 1.5. The uppermost zone i s the s a t u r a t i o n


zone and varies little in thickness, regardless of the total depth of
infiltration. Immediately below this zone, there is a zone of rapid
decrease in the water content, which Bodman and Colman called the
transition zone; and below it, there occurs a zone of nearly constant
moisture called the transmitting zone. T h i s zone increases in length in
direct proportion to the volume of infiltrated water. Next, there is the
wetting zone w h i c h moves downward w i t h a constant shape as infiltration
continues. The wetting zone ends at the wetting front, which is the
15

boundary between water penetration and soil at the initial moisture

content.

Soil Physics Models

There are two approaches to modelling the infiltration process, soil


physics models and hydrologic models. Soil physics models are deter-
ministic models based on the physics of soil moisture movement, while
h y d r o l o g i c models a r e conceptual a n d a r e based on a die-away rate until
the final steady rate is reached. The advantage of soil physics models
is that the parameters are understood and are measurable; the dis-
advantage is that soil p h y s i c s models t y p i c a l l y require a large amount

of data, i n c l u d i n g s i t e measures of soil porosity, hydraulic conductivity,


soi I layering, etc. I n comparison, h y d r o l o g i c models g e n e r a l l y h a v e fewer
parameters, require less data and are easier to solve; however, the

parameters are not a l way physically interpretable and cannot be


measured, hence they must be established by calibration. A further
criticism of hydrologic models is that they oversimplify the infiltration
process, particularly during periods of unsteady rain and rainfal I less
than the soi I s a t u r a t e d h y d r a u l i c c o n d u c t i v i t y .
The g o v e r n i n g equations for infiltration a r e the conservation of mass
a n d an e q u a t i o n of motion.

MOISTURE CONTENT MOISTURE CONTENT

8; -
8,
I
I
I
I
I I
II TRANSMITTING I I SATURATED
r I ZONE I-
a I ZONE
I-
a I W I

3
W n
n I I
I
I WElTINGZONE I

I I
BODMAN AND COLMAN GREEN AND AMPT

Fig. 1.5 Comparison of Green a n d Ampt s o i l moisture p r o f i l e w i t h


Bodman-Colman p r o f i l e
16

The c o n s e r v a t i o n of mass e q u a t i o n i s

av+ao=o (1.21)
az at

where v is the specific discharge (velocity) vertically, 0 is the vol-


umetric moisture content.

The equation of motion is based on Darcy's law for a saturated,


homogeneous soi I ,

v = - k -dh (1.22)
dz

where v is velocity as defined previously, k is hydraulic conductivity,


h is the hydraulic head, dh is the change in head i n the d i r e c t i o n of
flow over the length dz. The negative sign indicates flow is in the
d i r e c t ion of decreasing head.
Darcy's law can be generalized to unsaturated flow by expressing
the h y d r a u l i c head as a f u n c t i o n of soil tension o r suction, and gravity.
During the initial stages of infiltration when the water content is low,
the tension force is much larger than the gravity force and the flow
process is controlled by tension. As the pores fill, tension is reduced
and gravity becomes important. The hydraulic head is then equal to
tension $ plus g r a v i t y , z .
h = $ + z (1.23)
a n d D a r c y ' s law as a p p l i e d to u n s a t u r a t e d flow is

( 1 .24)

By combining Eqs. 1.23 and 1.24, we get the governing equation

for one-dimensional, vertical, unsaturated flow, known as Richard's

equation.

(1.25)

where k and $ a r e both functions of 0 . Due to the n o n l i n e a r r e l a t i o n s h i p


between hydraulic conductivity, suction and soil moisture, there is no
general a n a l y t i c a l s o l u t i o n to Eq. 1.25.
The problem is further complicated by hysteresis in that the
relationship between suction and moisture content is not unique and
single valued. The relationship depends on whether the soil is wetting
(infiltration is occurring) or drying (drainage is occurring). These
relationships are shown in Figure 1.6. Generally, for a given water
content, suction is lower during wetting than during drainage and

minor hysteretic loops can occur between the main hysteretic loops. The
h y s t e r e t i c effect i s a t t r i b u t e d to (1) geometric nonuniformity of individual

pores, (2) variations in contact angle in wetting and drainage,


17

(3) entrapped a i r , and ( 4 ) swelling (Hillel, 1971). Conductivity likewise


e x h i b i t s a h y s t e r e t i c effect.

MOISTURE C O N T E N T
Fig. 1.6 Typical soil suction - moisture relation

Green a n d Ampt Model

A conceptual model utilizing Darcy's law was proposed by Green


a n d Ampt (1911). M a n y s t u d i e s , i n c l u d i n g those b y Mein a n d L a r s o n (1973),

have demonstrated the usefulness of the Green and Ampt model for
model l i n g infiltration. As methods for measuring the model parameters

are made easier, it can be expected the model will be more widely
applied.
D a r c y ' s law c a n be w r i t t e n as
v = f
- = k(h + Lf + $f)/Lf (1.26)
n
where f is the infiltration rate and v i s equal to the vertical velocity,
h is the surface ponding depth, Lf is the depth to the wetting front,
and qf i s suction at the w e t t i n g f r o n t .
Several assumptions were n e c e s s a r y to write Darcy's law in the form
of E q . 1.26, namely:

1. There e x i s t s a d i s t i n c t a n d p r e c i s e l y d e f i n a b l e w e t t i n g f r o n t .

2. Suction at the wetting front, ii, f, remains essentially constant,

r e g a r d l e s s of time a n d depth.
18

3. Above (behind) the wetting front, the soil is uniformly wet and of
constant hydraul ic conductivity k.

4. Below (in front of) the wetting front, the soil moisture content is
r e l a t i v e l y unchanged from i t s i n i t i a l moisture content, 0 _.
These assumptions, when checked against the actual soi I m o i s t u r e p r o f i l e
of Bodman and Colman illustrate the approximate nature of the Green
a n d Ampt m o d e l . T h i s i s s h o w n i n F i g u r e 1.5.
The accumulated infiltration depth, F, can be obtained b y integrat-
i n g Eq. 1.26.
f = dF/dt = k(h + Lf + $ f ) / L f (1.27)

o r more d i r e c t l y f r o m
F = (0 - Oi)Lf = AOLf (1.28)

where 9 i s the s a t u r a t e d moisture content and 0 i s the initial moisture


content. The measure of moisture content, 0, is a volumetric measure,
t h e r e f o r e A0 i s c a l c u l a t e d w i t h t h e r e l a t i o n s h i p
A0 = (0 - 0.) = r $ ( l - S i ) (1.29)

where 4 is the soil p o r o s i t y and S. i s the initial degree of saturation.


I
Applying the relationships in Eqs. 1.28 and 1.29 to Eq. 1.27 and
integrating to obtain F gives
= F - ( $ AO)In[I + F / ( I ~ ~ A+
O hA0)l (1.30)
kt f
which is a nonlinear equation implicit in F and t. An explicit formu-
lation to solve for , the incremental infiltration volume during an
incremental time interval, At, is obtained by rewriting Eq. 1.30.
This gives
kAt - 2Ft 12F
t
- kAt
AF =
2
t/( ) + 2 k A t ( h t A 0 + J, A 0 + F )
f t
(1.31)

where AF is the increment in total infiltration from time t to time


t+At, and F and ht a r e t h e infiltration and p o n d e d d e p t h , respectively,
t
at time t. Therefore, the total infiltration after the time increment is
= F + AF; i f AF < i A t + ht (1.32a)
Ft+At t
or
= Ft + iAt + ht; i f AF > iAt + ht (1.32b)
F t + At
where i is the rainfall intensity. I f AF<iAt for a time step then excess
intensity, ie, o c c u r s .

The incremental cumulative infiltration equation, Eq. 1.32, was developed


assuming uniform soil properties. However, it can be applied to layered
19

soils, assuming each layer has uniform properties. The required soil
properties, i.e. K,, if,
4, a n d Si, a n d the thickness, d, must be k n o w n
for each layer. After computing the infiltration during each time

interval, the cumulative infiltration volume, F, is compared with the


storage capacity of uppermost layer not yet saturated. Once a layer

becomes saturated, the infiltration rate is control led by the conditions


in that layer or the next lower layer, whichever gives the s m a l l e r r a t e .
Bouwer (1966) defined the Green and Ampt parameter k to be "the

actual hydraulic conductivity in the wetted zone," which is less than


the saturated hydraul ic conductivity, ks. He concluded, based on
previous work, that k may be taken as about 0.5kS. The saturated
h y d r a u l ic conductivity can be determined by several standard laboratory
tests.
E f f e c t i v e s a t u r a t i o n i s d e f i n e d as
0 -0
(1.34)
r
where Er is the residual moisture content, Brooks and Corey (1966)
observed a s t r a i g h t I ine r e l a t i o n s h i p

se = (vJ
b
/ J Ic ) ' " ; f o r J, > jib (1.35)

where J, is capillary pressure head (suction) at a given soil moisture


content, 0 ; $ i s termed b u b b l i n g pressure a n d i s defined at the inter-
b
cept of a straight line plot of effective saturation and capillary pressure
head; and B is an index of the pore size distribution. Porous m e d i a
composed of single grain, material have primary porosity (porosity
consisting only of spaces between the grains) and tend to have small
v a l u e s of 8. Media h a v i n g secondary porosity ( p o r e spaces a l s o a v a i l a b l e
f o r flow w i t h aggregates) have large values (>1.0).
The w e t t i n g f r o n t suction i s estimated u s i n g the f o l l o w i n g r e l a t i o n s h i p

f = __
11 'b
- (1.36)
r l - 1 2
where rl = 2+3/B

H y d r o l o g i c I n f i l t r a t i o n Models

Horton (1939) proposed an infiltration equation to represent the


typical infiltration curves observed in double-ring i n f i ltrometer tests.
In these experiments, the water is continuously ponded above the soil;
therefore, the supply is not limiting and infiltration proceeds at the
maximum potential rate. He observed that the infiltration rate was
initially h i g h a n d decreased in time to a s t e a d y f i n a l rate. The d i e - a w a y
followed a n e g a t i v e exponential very closely. His equation i s
20

-k t
f = f + (fo - fc)e (1.37)

where f is the capacity infiltration rate at time t, fo and f are the


C

initial and final infiltration rates, and k is the infiltration constant


which is allegedly a function of soil and vegetation. In theory this
equation assumes the air-soil interface is saturated at all times. In
practical terms t h i s means that it i s assumed the rainfall rate is always
greater than infiltration capacity rates, and hence some ponding will
always result. This is a major disadvantage in the u s e of H o r t o n ' s model
since natural rainfal I rates are highly v a r i a b l e a n d therefore frequently
fall below the capacity rates. This may not be a problem with high
intensity design rainfalls or rainfalls distributed in time to always

exceed t h e c a p a c i t y i n f i l t r a t i o n rates.
Holton (1961) proposed a conceptual model of infiltration backed by

substantial field experimentation. He recognized from soi I physics as


the pores fill, the infiltration rate dies away and approaches a steady
final rate. The final rate of infiltration fc was associated with the

gravity force at field capacity (and is assumed to equal the soil


saturated hydraulic conductivity, ks). He then formulated a model to
relate capacity infiltration rate to the available soil moisture storage
volume r e m a i n i n g a t a n y time, F as
P'
f = aF" (1.38)
P + fc
The parameters a a n d n were determined e x p e r i m e n t a l l y from i n f i ltrometer
plot data. The exponent was found to b e a b o u t 1.4 for all plots studied
and the coefficients varied from 0.2 to 0.8 for the soil-cover complexes
s t u d i ed.

REFERENCES

Beven, K . , Dec. 1982. O n s u b s u r f a c e s t o r m f l o w . Predictions with simple


k i n e m a t i c t h e o r y f o r s a t u r a t e d a n d u n s a t u r a t e d flows. Water Resources
Res. 18 ( 6 ) p p 1627-33.
Bodman, G.B. a n d Colman, E.A. 1943. Moisture a n d energy conditions
d u r i n g d o w n w a r d e n t r y o f w a t e r i n t o s o i l s . P r o c . S o i l S c i e n c e SOC. o f
A m e r i c a , Vol. 7 , pp 116-122.
B o r a h , D.K. a n d P r a s a d , S.N., 1982. Shock s t r u c t u r e i n k i n e m a t i c w a v e
routing. In Rainfall-Runoff Relationships, Edt. Singh, V.P., Water
R e s o u r c e s Pub1 i c a t i o n s , C o l o r a d o , 582 pp.
B o u w e r , H. 1966. R a p i d f i e l d m e a s u r e m e n t o f a i r e n t r y v a l u e a n d h y d r a u l i c
c o n d u c t i v i t y of soi I a s s i g n i f i c a n t p a r a m e t e r s i n f l o w s y s t e m a n a l y s i s .
W a t e r R e s o u r c e s R e s e a r c h , Vol. 2, No. 4 , pp 729-738.
B r a k e n s i e k , D.L., 1966. H y d r o d y n a m i c s o f o v e r l a n d f l o w a n d n o n - p r i s m a t i c
c h a n n e l s . T r a n s . ASAE 9 ( 1 1 , pp 119-122.
B r o o k s , R.H. a n d C o l e y , A.T. 1966. P r o p e r t i e s o f p o r o u s m e d i a a f f e c t i n g
f l u i d f l o w . J o u r n a l o f t h e I r r i g a t i o n a n d D r a i n a g e D i v i s i o n , ASCE, V o l .
92, No. I R 2 , pp 61-88.
21

C o n s t a n t i n i d e s , C.A. and S t e p h e n s o n , D., 1982. D i m e n s i o n l e s s h y d r o g r a p h s


using kinematic theory, R e p o r t 5/1982. Water Systems Research Pro-
gramme, U n i v e r s i t y o f t h e W i t w a t e r s r a n d , J o h a n n e s b u r g .
C o u r a n t , R., F r i e d r i c h s , K. and L e w y , H., 1956. On t h e p a r t i a l d i f f e r -
e n c e e q u a t i o n s o f m a t h e m a t i c a l p h y s i c s . N.Y. U n i v . I n s t . M a t h s .
Croley, T.E. and H u n t , B., 1981. M u l t i p l e v a l u e d and n o n - c o n v e r g e n t
s o l u t i o n s i n k i n e m a t i c c a s c a d e m o d e l s , J. H y d r o l . , 49, pp 121-138.
Dunne, T., 1978. F i e l d s t u d i e s of h i l l s l o p e f l o w p r o c e s s e s . Ch. 7, H i l l -
s l o p e H y d r o l o g y , E d . K i r k b y , M.J., John W i l e y , N.Y.
E a g l e s o n , P., 1967. A d i s t r i b u t e d l i n e a r m o d e l for p e a k c a t c h m e n t d i s -
charge. I n t l . H y d r o l . Symp., Colorado State Univ., Fort Collins, pp
1-18.
F r e e z e , R.A., 1972. R o l e o f s u b s u r f a c e f l o w i n g e n e r a t i n g s u r f a c e r u n o f f .
2, U p s t r e a m s o u r c e a r e a s . W a t e r R e s o u r c e s R e s e a r c h , 8 ( 5 ) , pp 1272-
1283.
G a l l a t i , M. and M a i o n e , U., 1977. P e r s p e c t i v e o n m a t h e m a t i c a l m o d e l s o f
f l o o d r o u t i n g , in M a t h e m a t i c a l Models f o r S u r f a c e Water H y d r o l o g y , E d t .
C i r i a n i , T.A., M a i o n e , U. and W a l l i s , J.R., W i l e y I n t e r s c i e n c e , 423 pp.
Greco, F. and P a n a t t a n i , L., 1977. N u m e r i c a l s o l u t i o n m e t h o d s o f t h e S t .
Venant e q u a t i o n s . I n Mathemat ical Models f o r Surface Water H y d r o l o g y ,
E d t . C i r i a n i , T.A., M a i o n e , U. and W a l l i s , J.R., W i l e y I n t e r s c i e n c e , 423
PP .
Green, I .R.A., 1984. WITWAT s t o r m w a t e r d r a i n a g e p r o g r a m . R e p o r t 1/1984,
Water Systems Research Programme, U n i v e r s i t y of the Witwatersrand,
Johannesburg.
Green, W.H. and Ampt, G.A. 1911. S t u d i e s o f s o i l p h y s i c s , 1 . T h e f l o w o f
a i r and w a t e r t h r o u g h s o i l s . J. o f A g r i c u l t u r e S c i e n c e , V o l . 4, No. 1 ,
pp 1-24
Henderson, F.M. and W o o d i n g , R.A., 1964. O v e r l a n d f l o w and g r o u n d -
w a t e r f l o w f r o m s t e a d y r a i n f a l l o f f i n i t e d u r a t i o n . J. G e o p h y s . Res. 69
( 8 ) pp 1531-1539.
H i l l e l , D. 1971. Soi I and w a t e r - p h y s i c a l p r i n c i p l e s and p r o c e s s e s , A c a d e m i c
Press
Holton, H.N. 1961. A c o n c e p t of i n f i l t r a t i o n e s t i m a t e s in w a t e r s h e d
engineering, U.S. Dept. of A g r i c u l t u r e , Agric. R e s e a r c h S e r v i c e , No.
41-51, W a s h i n g t o n , D.C.
H o r t o n , R.E. 1933. T h e r o l e o f i n f i l t r a t i o n i n t h e h y d r o l o g i c c y c l e . T r a n s .
o f t h e A m e r i c a n G e o p h y s i c a l U n i o n , H y d r o l o g y P a p e r s , pp 446-460
H o r t o n , R.E., 1938. T h e i n t e r p r e t a t i o n and a p p l i c a t i o n o f r u n o f f p l o t
e x p e r i m e n t s w i t h r e f e r e n c e t o s o i l e r o s i o n p r o b l e m s . P r o c . S o i l S c i . SOC.
Am. 3, pp 340-349.
H o r t o n , R.E. 1939. A p p r o a c h t o w a r d a p h y s i c a l i n t e r p r e t a t i o n o f i n f i l t a t i o n
c a p a c i t y . P r o c . S o i l S c i e n c e SOC. o f A m e r i c a , V o l . 5, pp 399-417.
H o r t o n , R.E., L e a c h , H.R., and V a n V I i e t , R . , 1934, L a m i n a r s h e e t f l o w .
Amer. Geophys. U n i o n , T r a n s . , P a r t I I , pp 393-404.
Iwagaki, Y., 1955. Fundamental studies on t h e r u n o f f a n a l y s i s by
c h a r a c t e r i s t i c s . D i s a s t e r P r e v e n t i o n R e s e a r c h I n s t i t u t e , B u l l e t i n 10, K y o t o
U n i v . 25 pp.
K e u l e g a n , G.H., 1945. S p a t i a l l y v a r i e d d i s c h a r g e o v e r a s l o p i n g p l a n e .
Amer. G e o p h y s . U n i o n T r a n s . P a r t 6, pp 956-959.
Kibler, D.F. a n d Woolhiser, D.A., 1970. The k i n e m a t i c cascade a s a
h y d r o l o g i c a l m o d e l . C o l o r a d o S t a t e U n i v . p a p e r 39, F o r t C o l l i n s , 25 pp.
Kouwen, N., L i , R.M. and S i m o n s , D.B., 1980. F l o w r e s i s t a n c e i n v e g e -
tated waterways. M a n u s c r i p t , Colorado State U n i v e r s i t y , F o r t Collins.
L i g h t h i l l , F.R.S. and W h i t h a m , G . B . , 1955. On k i n e m a t i c w a v e s , I , F l o o d
m e a s u r e m e n t s i n l o n g r i v e r s . P r o c . R o y a l SOC. o f L o n d o n , A, 229, pp
281-31 6.
22

Lloyd-Davies, D.E., 1905. T h e e l i m i n a t i o n o f s t o r m w a t e r f r o m s e w e r a g e


s y s t e m s . M i n . P r o c . I n s t n . C i v i l E n g n r s . , 1 6 4 ( 2 ) pp 41-67.
Massau, J., 1889. L ' i n t C g r a t i o n g r a p h i q u e . Assoc. l n g e n i e u r s S o r t i s des
E c o l e s S p C c i a l e s d e s G a r d , A n n a l e s . 435 p p .
M e i n , R.G. and L a r s o n , C . L . 1973. M o d e l i n g i n f i l t r a t i o n d u r i n g a s t e a d y
r a i n . W a t e r R e s o u r c e s R e s e a r c h , V o l . 9, No. 2, pp 384-394.
Morris, E.M. and W o o l h i s e r , D.A., 1980. U n s t e a d y o n e - d i m e n s i o n a l flow
over a plane: P a r t i a l e q u i l i b r i u m and r e c e s s i o n h y d r o g r a p h s . W a t e r
R e s o u r c e s R e s e a r c h , 16 ( 2 ) , pp 355-360.
Overton, D.E. and Meadows, M.E., 1976. S t o r m w a t e r m o d e l l i n g , A c a d e m i c
P r e s s , 358 pp.
R ossmiller, R.L., 1980. T h e R a t i o n a l f o r m u l a r e v i s i t e d . P r o c . I n t l . Symp.
Storm Runof f , U n i v . of K e n t u c k y , L e x i n g t o n .
Rovey, E.W., Woolhiser, D.A. and S m i t h , R.E., 1977. A distributed
k i n e m a t i c model of u p l a n d watersheds. H y d r o l o g y P a p e r 93, C o l o r a d o
S t a t e U n i v . , F o r t Col t i n s , 52 pp.
Schaake, J.C., 1975. Surface waters. R e v i e w o f g e o p h y s i c s and s p a c e
p h y s i c s 13 ( 1 3 ) pp 445-451.
Singh, V.P., 1977. C r i t e r i o n t o c h o o s e s t e p l e n g t h f o r some n u m e r i c a l
m e t h o d s u s e d i n h y d r o l o g y . J. H y d r o l . , 33, p p 287-299.
S k a g g s , R.W., 1982. I n f i l t r a t i o n . Ch. 4, H y d r o l o g i c a l M o d e l l i n g o f S m a l l
W a t e r s h e d s , E d t . H a a n , C.T., Johnson, H.P. and B r a k e n s i e k , D.L., ASAE
Monog. 5.
scs ( s o i l C o n s e r v a t i o n S e r v i c e ) 1972. N a t i o n a l E n g i n e e r i n g H a n d b o o k ,
Secn. 4. H y d r o l o g y , W a s h i n g t o n , D.C.
Stephenson, D., 1983. T h e e f f e c t s of u r b a n i z a t i o n . Course on M oder n
S t o r m w a t e r D r a i n a g e P r a c t i c e , SAICE, C a p e Town.
Stephenson, D., 1984a. K i n e m a t i c s t u d y o f e f f e c t s of s t o r m d y n a m i c s o f
r u n o f f h y d r o g r a p h s . W a t e r S.A. V o l . 10, No.4, O c t . 1984. p p 189-196.
Stephenson, D., 1984b. K i n e m a t i c a n a l y s i s f o r d e t e n t i o n s t o r a g e . EPA/Users
g r o u p meeting. Detroit.
W a t k i n s , L.H., 1962. T h e D e s i g n of U r b a n Sewer S y s t e m s , R o a d R e s e a r c h
T e c h n . p a p e r 55, HMSO, L o n d o n .
W o o d i n g , R.A., 1965a. A h y d r a u l i c model f o r t h e c a t c h m e n t - s t r e a m p r o b l e m
1 , K i n e m a t i c w a v e t h e o r y . J. H y d r o l o g y , 3. pp 254-267.
Wooding, R.A., 1965.b. A hydraulic model for the catchment-stream
p r o b l e m , I I , N u m e r i c a l s o l u t i o n s , J. H y d r o l . 3. pp 268-282.
Wooding, R.A., 1966. A h y d r a u l i c m o d e l f o r t h e c a t c h m e n t - s t r e a m p r o b l e m
I I I , C o m p a r i s o n w i t h r u n o f f o b s e r v a t i o n s , J. H y d r o l o g y , 4, p p 21-37.
W o o l h i s e r , D.A. and L i g g e t , J.A. 1967. U n s t e a d y o n e - d i m e n s i o n a l f l o w o v e r
a p l a n e - T h e r i s i n g h y d r o g r a p h . Water Resources Research, 3 ( 3 ) , pp
753-771.
W y l i e , E.B., 1980. U n s t e a d y f r e e s u r f a c e f l o w c o m p u t a t i o n s . P r o c . ASCE, 96
( H Y 1 1 ) pp 2241-2251.
23

CHAPTER 2

ANALYSIS OF RUNOFF

INTRODUCTION

In this chapter a simplified description of the rainfall - runoff


mechanism i s presented, i.e. one which can be described i n equation form.
The concept of mass b a l a n c e whereby i n p u t equals outflow p l u s change i n
storage, is applied to simple catchments. The build-up of water depth
over the catchment when a storm occurs is described as well as the
mechanism whereby runoff occurs. The relationship between water depth
and flow r a t e forms an important part in the prediction of flow so the
equation of motion ( i n fact only a simple flow resistance equation in the
case of k i n e m a t i c f l o w ) i s introduced.
This simple analysis is confined to a rectangular plane catchment
sloping uniformly down in the direction of flow, and flow is assumed

overland. The equations of continuity and flow are thus particularly


simple. Nevertheless the origin of and the assumptions behind the

simplifications are presented. A simple demonstration of the a p p l i c a b i l i t y


of the kinematic equations is also given. Later other components of flow
e.g. sub-surface flow (Beven, 1982) a n d a more p r a c t i c a l assessment of the
contribution to streamflow are introduced with model I i n g . The different-
iation of surface and subsurface flow is often more complicated than
assumed h e r e (Dunne, 1978).

DYNAM I C EQUATI ON5

The equations governing unsteady, one-dimensional overland and open


channel flow are derived by applying the principles of conservation of
mass a n d momentum to elemental fluid control volumes. One-dimensional
equations actually describe the change in streamflow in two dimensions:
vertical and longitudinal. They are classified as one-dimensional since
only one s p a t i a l v a r i a b l e occurs as an independent v a r i a b l e .

The important assumptions a r e :

1. The water surface profile varies gradually, which i s equivalent to

stating the pressure distribution is hydrostatic, i.e., vertical


accelerations a r e smal I ;

2. Resistance to flow can be approximated b y steady flow formulae;


3. The velocity distribution across the wetted area can be represented

w i t h the cross-sectional average velocity;


4. Momentum c a r r i e d to the strearnflow from lateral inflow is negligible;

and
5. The s1oDe of the channel i s small.

I n addition, for this derivation, the channel i s assumed r e c t a n g u l a r .


This simplifies the mathematics i n v o l v e d a n d h a s l i t t l e effect on the f i n a l
form of the g o v e r n i n g equations.

Conservation of Mass

The continuity p r i n c i p l e states that the net mass inflow to a control


volume must equal the r a t e of change of mass stored within the control
volume. Consider the elemental fluid volume shown i n F i g u r e 2.1, where Q
i s the v o l u m e t r i c f l o w r a t e i n m3/s o r cfs, q. i s the l a t e r a l inflow r a t e i n
rn3/s p e r rn o r cfs p e r foot l e n g t h of channel, y a n d A a r e depth a n d cross
sectional area of flow in metres and square metres (feet and square
feet), respectively, 0 i s the slope of the channel w i t h respect to the

Fig. 2.1 D e r i v a t i o n of c o n t i n u i t y e q u a t i o n

horizontal measured as an angle, and x and t are the space and time
coordinates i n metres ( f e e t ) a n d seconds. The t o t a l i n f l o w to the section i s

I n f l o w = Q + qiAX (2.1 1
and the t o t a l outflow is

Outflow = Q + 22 AX (2.2)
ax
25

The change in volume stored in the section i s equal to the change in

cross-sectional a r e a of flow m u l t i p l i e d b y the l e n g t h of the section.

aA
Change i n volume stored = (2.3)
=Ax

Combining these quantities according to the above stated principle,


d i v i d i n g by A x , a n d r e a r r a n g i n g , y i e l d s the c o n t i n u i t y e q u a t i o n

(2.4)
ax at

Conservation of Momentum

This second equation is given by Newton's second law of motion


which states that the r a t e of change of momentum i s e q u a l to the a p p l i e d
forces. The applied forces, as seen in F i g u r e 2.2, are (1) pressure, (2)
gravity, a n d ( 3 ) r e s i s t i v e f r i c t i o n a l forces.

wt
Fig. 2.2 D e r i v a t i o n of momentum e q u a t i o n

Consider forces in the downstream d i r e c t i o n as positive. The p r e s s u r e


downslope acts opposite to the pressure upslope a n d upon summing, the net

pressure force becomes


-pgA(aY/ax) A X

where p i s the mass d e n s i t y of water and g i s the g r a v i t a t i o n a l acceler-


a t ion.
Similarly, i t can be shown that the g r a v i t y o r weight force a c t i n g on
the volume of water in the section i s g i v e n by pgAAx tan0
26

where, for g r a d u a l ly v a r i e d flow, tan !! closely corresponds to the channel


slope, So, a n d may be expressed as such. This i s c a l l e d the small slope
a p p r o x i mat ion.
Finally, the friction force acting to r e t a r d the flow i s expressed in
terms of an average shear stress
- TPAX
where T i s shear stress a n d P i s wetted perimeter. From the r e l a t i o n s h i p
formed by equating head (energy) loss to the work done by the shear
force w e know that T = YRSf, where Sf i s the f r i c t i o n slope a n d Y i s the
unit weight of liquid. Substituting for T , and recalling that R=A/P, the
fol lowing expression f o r the f r i c t i o n force i s o b t a i n e d .

- VRS PAX = - YS AAx (2.5)


f f
The r e s u l t a n t force on the f l u i d volume i n the d i r e c t i o n of flow i s the
summat ion of the three a p p l ied e x t e r n a l forces.

PgAAx [-(ay/ax) + so - sfl


The change i n momentum consists of two p a r t s , the local or temporal
momentum change and the spatial or convective momentum change. The
local momentum of the f l u i d is pAAx v , a n d the local change i s j u s t the
time d e r i v a t i v e

The s p a t i a l change i n momentum i s the r a t e of momentum change across t h e


control surface. The momentum flux through the control surface is pvZA,
a n d the s p a t i a l change i s the x - d e r i v a t i v e
a (Pv’A) L x
- = ( 2 Av .E + vz2)Axp (2.7)
ax ax ax
The total momentum change is the sum of the temporal and spatial
momentum changes.

AXP(A .LY + v
at
z),+
at
V A X P ( V -a A
ax
+ ZA 2)
ax
S u b s t i t u t i n g the f o l l o w i n g equivalence from c o n t i n u i t y
* a v + v - aA 5 q , - - aA (2.8)
ax ax I at
a l l o w s the r a t e of momentum change to be w r i t t e n as

Equating this expression with the summation of external forces gives

the f o l l o w i n g f a m i l i a r form f o r the conservation of momentum equation.


a v + v -a v + g & =
- ax g ( S o - S ) - Vqi (2.9)
at ax f A

where 5 i s bed slope, S i s f r i c t i o n slope a n d R i s h y d r a u l i c r a d i u s and


f
i s equal to A/P.
27

Eqs. 2. 4 and 2.9 can be made applicable to any cross section for
both overland and open channel flow, though strictly they apply to

rectangular channels only in t h e p r e s e n t f o r m .


These equations are nonl inear, hyperbolic, partial differential
equations and represent a nonlinear, deterministic, distributed, time
variant system. They are sometimes referred to as the St. Venant
e a uat ions.

S l M P L l F I ED EQUATIONS

Equations 2.4 and 2 . 9 a r e accepted as f u l l y d e s c r i p t i v e of one dimen-

sional overland and open channel flow routing. These e q u a t i o n s d e s c r i b e


both the forward or downstream w a v e p r o p a g a t i o n c h a r a c t e r i s t i c s as we1 I
as the backward or upstream characteristics. It is assumed that flood
waves in streams move downstream and since h i llslope runoff is always

downhi I I , the backward characteristics are simply b a c k w a t e r effects, and


in some flow routing instances, they can have substantial impact and
control on the flow. As such, these e q u a t i o n s a r e k n o w n g e n e r a l l y a s the
dynamic wave equations. As a r u n o f f h y d r o g r a p h passes t h r o u g h a c h a n n e l
reach, the combined effects of channel irregularity, pool and riffle
patterns, n a t u r a l and m a n m a d e r o u g h n e s s a n d g r a v i t y f o r c e s a c t t o r e d u c e
the hydrograph peak assuming lateral inflow is insignificant while
lengthening the time base. That is, the peak of the hydrograph is
attenuated while the shape is dispersed in time (also in space). The
dynamic wave equations account we1 I f o r h y d r o g r a p h a t t e n u a t i o n . However,
two drawbacks to the wholesale general use of these equations are the
l a r g e d a t a r e q u i r e m e n t s and the necessity for numerical integration. Very
often, based on channel geometry and alignment and flood wave
characteristics, it is p o s s i b l e to make valid simplifying assumptions that

allow one to utilize appr-oximations to the dynamic wave equations. When


this is possible, advantages in terms of ease of solution and data
r e q u i r e m e n t s arc? o f t e n r e a l i z e d .
Two approximations that have found wide application in engineering

practice a r e the diffusion and kinematic wave models. The d i f f u s i o n wave


model assumes that the i n e r t i a terms in the equation of motion, Eq. 2.9,
are negligible compared with the pressure, friction, and gravity terms.
Thus, the diffusion model equations are continuity, Eq. 2.4, and the
f o l l o w i n g s i m p l i f i e d form of t h e c o n s e r v a t i o n of momentum e q u a t i o n .

(2.10)
28

For prismatic channels, Eqs. 2.4 and 2.10 are often combined into the

s i n g l e equation

a Q + c -aQ
- = a2Q (2.11)
at ax DaxZ
where c is the wave celerity in m/s (fps) and D is a hydrograph
dispersion coefficient i n m’/sec (ft*/sec). Because Eq. 2.11 i s of the form
of the classical advection-diffusion equation, it i s commonly called the
d i f f u s i o n wave model.
The kinematic model further assumes the pressure term is negligible,
r e d u c i n g Eq. 2.10 to
so = Sf (2.12)
which means the e q u a t i o n of motion can be approximated b y a u n i f o r m flow

f o r m u l a of the general form


b
Q = ay (2.13)
where a , b a r e constants.
Although approximat ions, both the d i f f u s i o n a n d kinemat i c models h a v e
been shown to be f a i r l y good d e s c r i p t i o n s of the p h y s i c a l phenomemona in

a variety of open channel a n d o v e r l a n d flow r o u t i n g cases. The k i n e m a t i c


model has been successfully applied to overland flow, to small streams
draining u p l a n d watersheds, and to s l o w - r i s i n g flood waves. This latter
case occurs both i n major streams such as the M i s s i s s i p p i R i v e r when long
d u r a t i o n f l o o d h y d r o g r a p h s r e s u l t i n g from, as an example, s p r i n g snowmelt
in the U.S. Midwest and Canada, and in small streams where the
streamflow h y d r o g r a p h nesul ts p r i n c i p a l l y from l a t e r a l stormwater inflow.

THE KINEMATIC EQUATIONS

For o v e r l a n d flow and in many channel flow situations, some of the


terms i n the dynamic e q u a t i o n ( 2 . 9 ) a r e i n s i g n i f i c a n t . Neglecting the qi
component one can w r i t e the e q u a t i o n as

(2.14)

The order of magnitude of each of the f i v e terms i s e v a l u a t e d below


for a shallow stream. If the bed slope ( 2 ) i s 0.01, the longitudinal rate
of change of water depth ( 3 ) i s u n l i k e l y to exceed O.lm/lOOm = 0.001. The

longitudinal velocity gradient term (4) will also be less than


(lm/s/lOm/s‘)(lm/s/lOOm) = 0.001, and the time rate of change in

velocity term (5) will in all probability be less than (l/lO~(l/lOOs~=


0.001.
29

Terms ( 3 ) , ( 4 ) a n d ( 5 ) a r e therefore a t least an o r d e r of magnitude


less than (2) for depths up to lm, and for flow depths less t h a n O.lm
they will be two orders of magnitude less. Those terms can therefore be
neglected for the majority of overland flow problems. The inaccuracy in
solutions orni t t i n g these terms for runoff hydrographs was evaluated by
v a r i o u s researchers:
Woolhiser and Liggett (1967) investigated the accuracy of the
kinematic approximation a n d found it to be v e r y good i f the dimensionless
parameter for planes SoL/yLFL’ is greater than 20 and reasonable if
greater than 10. yL i s the depth at the lower end of the p l a n e of length
1
L and slope So and FL is the Froude number VL/(gyL)?. i.e. gSoL/VL2
> 10. M o r r i s and Woolhiser (1980) and Woolhiser (1981) later found the
a d d i t i o n a l c r i t e r i o n S L / y > 5 i s also r e q u i r e d .
O L
The r e s u l t i n g s i m p l i f i e d dynamic e q u a t i o n o m i t t i n g terms ( 3 ) , ( 4 ) and
( 5 ) s i m p l y states that the f r i c t i o n g r a d i e n t i s equal to the bed gradient.
The friction gradient can be evaluated using any suitable friction
equation, e.g. that of Manning. The two equations referred to as the
k i n e m a t i c equations are thus the c o n t i n u i t y e q u a t i o n which per u n i t w i d t h
of o v e r l a n d flow becomes

(2.15)

a n d a f r i c t i o n e q u a t i o n of the form q = a y m (2.16)


where m is a coefficient and a is a function of the water properties,
surface roughness, bed slope and gravity. Equations (2.15) and (2.16)
apply to a wide f l a t bottom channel o r o v e r l a n d flow. The flow q is per

u n i t w i d t h a n d flow depth i s y .
The quasi-steady flow approximation was originally termed the
k i n e m a t i c wave a p p r o x i m a t i o n s i n c e waves can o n l y t r a v e l downstream a n d

are represented entirely by the continuity equation. Since the dynamic


forces a r e omitted, the Froude number F = v / J ( g y ) is i r r e l e v a n t , and i n
fact that the wave speed c i s not given by C = b u t may be d e r i v e d
by f i n d i n g d x / d t f o r which d y / d t = ie

(2.17)

From the f r i c t i o n e q u a t i o n (2.16) aq = d y = mcly rn- 1 (2.18)


-TT z a2
aY
ax
Substituting i n t o the c o n t i n u i t y e q u a t i o n y i e l d s

(2.19)

but Since ie = dy/dt, the left h a n d side of t h i s e q u a t i o n must also equal


dy/dt.
30

m- 1
Therefore dx
dt
= c = may (2.20)

which i s the speed at which a wave of unvarying amplitude (if i = 0)


t r a v e l s down the plane.
m- 1
Since v = ay i t may , be deduced that the w a v e speed i s related to
water velocity v b y the equation; c = mv. (2.21)

K I NEMATl C FLOW OVER IMPERMEABLE PLANES

The kinematic wave equations h a v e an important advantage over the


d y n a m i c and d i f f u s i o n w a v e e q u a t i o n s ; a n a l y t i c solutions a r e possible f o r
simple watershed geometries. In this section, the kinematic solutions are
developed for runoff from an impermeable r e c t a n g u l a r plane. Under these
conditions, we are not concerned with estimating rainfall loss due to
infiltration, nor with routing flows first overland and then through a
complex stormwater drainage system. Numerical models generally are
r e q u i r e d w h e n i n f i l t r a t i o n i s i m p o r t a n t or mu1 t i p l e r o u t i n g s a r e i n v o l v e d .

R i s i n g H y d r o g r a p h - G e n e r a l Solution
For t h e c a s e o f a long impermeable p l a n e , A = by, Q = bq and R =
y, where q is the flow per unit width, hence Eqs. 2.4 and 2.13 can be
written

(2.15)
and
m
q = aY (2.16)

where ie i s the r a i n f a l I excess intensity. Substituting Eq 2.15 i n t o Eq.


2.16 and p e r f o r m i n g t h e d i f f e r e n t i a t i o n y i e l d s

(2.22)

Eq. 2.22 s t a t e s t h a t t o an o b s e r v e r m o v i n g a t t h e s p e e d

--
dx
dt am’
m-1
(2.23

the depth of flow changes w i t h the r a i n f a l I r a t e

a
dt
=‘ e . (2.24

Eqs. 2.23 and 2.24 provide the b a s i s f o r a method o f c h a r a c t e r i s t i c s


solution to surface runoff. For steady rainfall excess intensity, Eq. 2.24
can be integrated to obtain
y = yo + iet (2.25)

where Yo i s the initial water depth when r a i n f a l l begins. E q . 2.25 i s the


31

equation for depth along each c h a r a c t e r i s t i c as that characteristic moves


from some initial position toward the downstream end of the plane. The
p o s i t i o n on the c h a r a c t e r i s t i c a t a n y instant i n time i s determined w i t h Eq
2.23. For an i n i t i a l l y d r y surface y = 0, hence y = i e t . Substituting this
relationship into E q . 2.23 g i v e s

-dx - - am(ietlm-’ (2.26)


dt
which i n t e g r a t e s to
m-1 m
x = x + n i t
(2.27)
or more s i m p l y
m- 1
x = x + a y t (2.28)
which specifies the downslope p o s i t i o n of the depth y after time t. x is
the p o i n t from w h i c h the f o r w a r d c h a r a c t e r i s t i c s emanate, i.e., the o r i g i n
of the c h a r a c t e r i s t i c s at t = 0, and i s measured from the upslope end of
the plane.
The discharge at any point along a characteristic is given by the
r e la t ionsh i p
m
q = a(i t) (2.29)
Two c h a r a c t e r i s t i c p a t h s a r e shown i n F i g u r e 2.3. The f i r s t emanates
at a point interior to the plane and t r a v e l s the d i s t a n c e L-x d u r i n g the
time to. The depth and discharge at each point (x,t) along this
characteristic is determined from Eqs. 2.25, 2.28 and 2.29. The second

characteristic begins at the upslope end of the plane and travels the
length of the plane durin.g the time tC. In this case, the depth at the
upstream end is zero, yo = 0, for all t. Therefore, as long as the
rainfal I intensity remains constant, once this initial characteristic has
reached the downstream end of the plane, the depth profile along the
plane will remain constant r e g a r d l e s s of how long the rainfall persists,
i.e., an equilibrium depth profile will be e s t a b l i s h e d . The time required

for this to happen is the concentration time tC. At equilibrium no


additional rainfall is being added to surface detention storage, and the
r a t e of outflow equals the r a i n f a l l r a t e .
Recognizing that generally what is required is the r u n o f f h y d r o g r a p h
at the end of the p l a n e catchment, the concept o f an e q u i l i b r i u m time and
f l o w r a t e suggests a way to s i m p l i f y the use of Eqs. 2.25, 2.28 a n d 2.29.
I n the f o l l o w i n g sections, solutions a n d examples a r e g i v e n f o r the time to
equilibrium, e q u i l i b r i u m depth p r o f i l e a n d r i s i n g o u t f l o w h y d r o g r a p h .
32

0 i

Fig. 2.3 Kinematic s o l u t i o n domain f o r p l a n e catchment

Time of Concentration

One can solve for the time of concentration from Eq. 2.27 u s i n g the
conditions that at t =. tc, x - x = L . S u b s t i t u t i n g a n d r e a r r a n g i n g to
solve for concentration time tc which i s equal to time to e q u i l i b r i u m t
m'
tC = ( L/cciem-l ) 1 /m (2.30)

For Manning-kinematic flow, time of concentration in minutes i s

t = (6.9/ie0'4)(nL/So0*5) O e 6
(2.31 )

for i i n mm/hr a n d L i n metres a n d

t = ( 0 . 9 2 8 / i ~ O . ~( )~ L / S ~ O * ' ) O e 6 (2.32)

for i i n in/hr and L i n feet

EXAMPLE 1. Estimate the time of concentration for a rainfall rate of


25 mm/hr on an asphalt parking lot 50 metres long and sloped at 1%.
Assume n = 0.023.
Using E q . 2.31, we f i n d

Oe60 = 8.2 minutes


33

Hence, a rain intensity of 25 mm/hr will bring the parking lot to


equilibrium i n 8.2 minutes.

Equilibrium Depth P r o f i l e

An expression for the equilibrium depth profile is found b y solving


Eqs. 2.25 and 2.28 simultaneously, and recalling that at xo = 0,
yo = 0.
The r e s u l t i n g e x p r e s s i o n is

y ( x ) = ( i e x / a ) 1/m (2.33)
which f o r Manning-kinematic flow i n S I u n i t s becomes

EXAMPLE 2. Estimate the e q u i l i b r i u m depth at the end of the asphalt


p a r k i n g lot i n Example 1.
We n e e d to be careful with units. The r a i n f a l l r a t e i s in mm/hr; but
the units implicit in the Manning equation
metres are and seconds.
Therefore,
6
we n e e d to d i v i d e t h e r a i n f a l l r a t e b y 3 . 6 ~ 1 0 .
i = 25/(3.6xlO 6 ) = 6.9~10-~m/sec

F r o m Eq. 2.34
0.6
0-6 )
0.023( 6 . 9 ~ 1 (50)
Y(L) =
[ ] = 0.0034 metres
(0.01
or y ( L ) = 3.4 mm

The Receding H y d r o g r a p h

H e n d e r s o n and W o o d i n g (1964) derived the kinematic equations for the


falling hydrograph. There are two cases involved: I. when the rising
hydrograph i s at e q u i l i b r i u m , and I I , when the r i s i n g h y d r o g r a p h is at a
flow less t h a n e q u i l i b r i u m , i.e., partial equilibrium.
Case I. D u r a t i o n o f r a i n f a l I, t
d
> = t . A f t e r t h e r a i n f a l l stops, from
2.24, i t c a n b e seen that on a characteristic
dy/dt = 0 (2.35)
which integrates to y = c, where c is some constant. Substituting this
relationship into Eq. 2.23 reveals that the corresponding characteristic
trajectories are lines parallel to the p l a n e and that the depth, discharge
and wave speed dx/dt, remain constant along a characteristic. This
means t h a t b e g i n n i n g w i t h a p o i n t o n t h e e q u i l i b r i u m p r o f i l e and r e a l i z i n g
that the future coordinates of that depth will lie on a single
characteristic, Eq. 2.23 can be used to locate the point in s p a c e a t any

f u t u r e time. T h i s p r i n c i p l e i s i l l u s t r a t e d i n F i g u r e 2.4.
34

The e q u i l i b r i u m depth p r o f i l e a t the cessation of rainfall is i n d i c a t e d as


the line A-B -C3. After some time At the depth profile i s A-B2-C2. The
1
depth at point Bl, yl, has moved a l o n g a constant c h a r a c t e r i s t i c p a t h to
the p o i n t B
2'

A L

Fig. 2 . 4 Water depth p r o f i l e

The distance moved i s g i v e n b y


m-1 (2.36)
Ax = a m y At
T h e new x co-ordinate is

x = x + A x (2.37)
1
m-1
= x + amyl (t-td) (2.38)
1
where x 1 was the position for point B1. Note that if the storm duration

tl > tc the time to equilibrium, then x = x That is, once the e q u i -


l e-
librium depth profile is established it will remaln constant as long as
the rainfall continues at a steady rate. From Eq. 2.33, the e q u i l i b r i u m
depth can be expressed as

Y, = i ieX1/"1 1/ m (2.39)

S u b s t i t u t i n g t h i s r e l a t i o n s h i p i n t o Eq. 2.38 gives

x = x (m-1 ) / m ( t - t ) (2.40)
1 + amIi,xl/a] d
A t the downstream end of the p i a n e x = L a n d q = ay
1
= ieL.
After substituting these identities into Eq. 2.40, we obtain the f o l l o w i n g
relationship between discharge and time for the recession hydrograph
(t-t ) = 0 (2.41 )
d
35

Case I I. Duration of rainfall, td < tc. f the rain stops prior to


reaching equilibrium, then the depth p r o f i l e a t t = td would correspond to
one similar to A-B1-Cl in Figure 2.4. That is, an equilibrium depth
profile will b e developed from the upslope e n d f t h e p l a n e a t x = O t o some
point x 1 given b y

(2.42)

The d e p t h a t p o i n t B1 w i l l m o v e a t a c o n s t a n t r a t e and w i l l r e a c h t h e e n d
of the p l a n e a t t i m e t,. T h i s time i s e v a l u a t e d a s

L - x,
t:; = td t
dx/dt (2.43)
I n c o r p o r a t i n g Eqs. 2.26, 2.27, and 2.30, Eq. 2.43 becomes
o1 im-l m aim-l m
e t c - e td (2.44)
t, = td t
.m-1 m
a mi
e td

which can b e s i m p l i f i e d to
1
t, = td 1 + - [(tc/tdIm - 1 } (2.45)
m
The d i s c h a r g e at the e n d of the p l a n e will r e m a i n c o n s t a n t b e t w e e n
td 5 t S t, and w i l l be

mi
q = a(i t (2.46)
e d
After t,, the recession proceeds according to Case I and E q . 2.41 applies

EXAMPLE 3. Determine the runoff hydrograph from the parking lot in


Example 1 for the same r a i n f a l l r a t e b u t of 10 m i n u t e s d u r a t i o n . Use t h e
Manning kinemat ic solution.
The solution requires that we first determine the time to e q u i l i b r i u m
which was done in Example 1. The next step is to generate the rising
hydrograph. If td tc the rising hydrograph will be an equilibrium
hydrograph. Finally, we must determine w h i c h case f o r recession a p p l i e s
a n d then determine t h e r e c e s s i o n graph a c c o r d i n g l y .
From Example 1 we know t = 8.2 minutes, therefore this event

satisfies the conditions for an e q u i l i b r i u m r i s i n g h y d r o g r a p h a n d Case 1


recession. Because td >= t
C’
an equilibrium profile will exist on the p l a n e
during the time interval f r o m t=8.2 m i n u t e s u n t i l t=10 m i n u t e s . During that
time runoff from the plane will b e constant and e q u a l to the peak rate.
The rising graph is given by Eq. 2.29 and t h e r e c e s s i o n graph b y Eq.
2.41.
36

First determine the equation for the rising graph. The coefficient
in Eq. 2.29 is
c1 = Soe5/n = (0.01)0.5/0.023 = 4.35

The depth in metres i s determined b y


y = iet/(6x10 4 )
3
where t i s in minutes; and t h e d i s c h a r g e in m /sec/m-width of p l a n e i s
6 5/3
q = 4.35 [(iet ) / ( 6 x 1 0 ) ]
Next, determine the equation for the recession hydrograph. After
the appropriate substitutions and units conversion, Eq. 2.41 becomes
25(50) 25
q -- (5/3) (4.35)0*6(q)0‘4(6~) (t-lo) = o
3.6~10

TABLE 2.1 Runoff H y d r o g r a p h Ordinates

D i s c h a r g e , m3/sec
Time, Minutes Depth, mm

5
0.0 0.0 0.0 x 10
1 .o 0.42 1 .o
2.0 0.83 3.2
3.0 1.25 6.3
4.0 1.67 10.2
5.0 2.08 14.8
6.0 2.50 20.0
7.0 2.92 25.9
8.0 3.33 32.4
9.0 3.42 33.7
10.0 3.42 33.7

Fig. 2.5 K i n e m a t i c h y d r o g r a p h s h a p e f o r s i m p l e p l a n e w i t h td = tC
37

FR I C T I ON EQUAT I ON

One of the kinematic equations is a friction energy loss equation.


There a r e many f r i c t i o n equations in use i n h y d r a u l i c engineering a n d a
generalized comparison i s made below. The most p o p u l a r e q u a t i o n relating
flow rate to friction energy gradient is perhaps t h a t of Manning, which
may be w r i t t e n as

(2.47)

where Kl is 1 in 5 . I. units (metre-kilogram-seconds) and 1.486 in


English u n i t s (foot-pounds-seconds), n i s the M a n n i n g roughness, A i s the
cross sectional area, R the h y d r a u l i c r a d i u s A/P, P the wetted perimeter
a n d S the energy slope. The 5 . I. system of units i s adopted below b u t it
should be noted t h a t the equation i s not dimensionless, a n d the roughness
factor n i s a f u n c t i o n of gravity. Written i n terms of flow per u n i t w i d t h
of a wide r e c t a n g u l a r channel, ( a s f o r an o v e r l a n d flow p l a n e )

q = JS 5/3 (2.48)
n Y
1

since hydraulic radius R yb/b = y a n d a r e a yb = y. Hence ~1 = S'/n

a n d m = 5/3. The M a n n i n g roughness coefficient n is reputedly a constant


for any surface roughness. This holds for l a r g e Reynolds numbers and
fully developed turbulent flow, but comparison with the Darcy Weisbach
equation i n d i c a t e s that n a c t u a l l y increases f o r low Reynolds numbers ( y v /
v<lOOO where v i s the k i n e m a t i c v i s c o s i t y of water, about 10-6mz/s). The

Manning equation may be compared with the Darcy equation employing


S t r i c k l e r ' s e q u a t i o n for roughness, n = 0.13K k'/6/gi where k i s a l i n e a r
1
measure of roughness analogous t o the N i k u r a d s e roughness f o r pipes ( i n
metres i f K, = 1). Substituting into Manning's equation yields

-21
Q = 7.7(R/k)'/6A(SRg (2.49)

I f t h i s e q u a t i o n i s compared w i t h D a r c y ' s equation in the form


1 1
Q = (8/f)' A(SRg)' (2.50)

it will be seen Stri < l e r in effect used a Darcy friction a c t o r equal to

0.135(k/R)'/3. (Note B r i t i s h p r a c t i c e i s to use h i n p l a c e of as they use f

f o r a d i f f e r e n t f a c t o r . ) According to Colebrook a n d White,

(2.51 )
1
__ = -2 log ( - + 7 2.5- 1
14.8R Re f
Ji
38

where Re is the Reynolds number, for pipes VD/v, or 4VR/V for non

circular cross sections. Whereas the Colebrook-Whi te equation predicts

higher values of the Darcy friction coefficient f for low Reynolds number
a n d any r e l a t i v e roughness k/R, the S t r i c k l e r e q u a t i o n assumes f depends
only on the r e l a t i v e roughness k/R. The S t r i c k l e r a n d M a n n i n g equations
can therefore be expected to underpredict roughness for low Reynolds
numbers. Higher values of n should therefore be used for o v e r l a n d flow
than f o r channel flows.
In general, the value of n and hence flow depth has to be deter-
mined by trial (assuming the Colebrook-White equation to apply a n d not
Strickler's). It i s therefore p r o b a b l y easier to use the Darcy equation f o r
this purpose but since an explicit equation is required for analytical
solutions to the kinematic equations a n d the v a r i a t i o n i n n i s less t h a n
the v a r i a t i o n in f with y, the M a n n i n g e q u a t i o n i s p r e f e r r e d .
Table 2.2 i n d i c a t e s values of n a n d f w i t h v a r y i n g water depths i n a
wide channel with a slope of 0.0025 a n d absolute roughness k = 0.0125 m.
The values of f are calculated from the Colebrook-White equation using a
f i r s t estimate of Re from M a n n i n g ' s e q u a t i o n , a n d then n i s re-calculated

from n = (f/8g)'R'/6 i.e. as f o r S t r i c k l e r ' s e q u a t i o n .

TABLE 2.2 - V a r i a t i o n of f a n d n w i t h depth

Water Depth, m Reynolds No. Darcy f Manning n

1 .o 2 x lo6 0.03 0.02


0.1 50 000 0.09 0.023

0.01 1 000 0.60 0.04

The Chezy equation is often used in preference to the Manning equation


i n American practice. T h i s equation is
v = K,CJ(RS) (2.52)
where C is known as the Chezy coefficient and K, is 1 in ft - second
u n i t s a n d 0.552 in S.I. units. I n fact the Chezy e q u a t i o n i s v e r y s i m i l a r
to the Darcy e q u a t i o n i n the form
V = J(8g/f) J ( R S ) (2.53)
a n d i t w i l l be seen t h a t C = J m / K 2 ,
also f o r t u r b u l e n t flow from (2.51) l/df'2 log (14.8R/k) (2.54)
39

Hence C = ( 2 / K 2 ) flg log (14.8/k) (2.55)


or v J3gRs' log ( 1 4 . 8 / k ) (2.56)
This equation stems from the log velocity distribution across a section
whereas the Strickler equation follows from a 1/6 power law fit to the

ve loc i ty d i st ri b u t ion.
Resistance to rainfall induced o v e r l a n d flow over natural and man-
made surfaces is influenced by several factors including surface rough-
ness, raindrop impact, vegetation, wind and infi Itration. Although there
have been many laboratory and field investigations to determine the

relative importance of these factors, the appropriate resistance formula,


a n d methods f o r parameter estimation, i n p r a c t i c e the convention has been

to use e i t h e r the Darcy-Weisbach equation m o d i f i e d f o r r a i n d r o p impact, or


the t r a d i t i o n a l forms of the M a n n i n g o r Chezy equations.
In laminar flow s t u d i e s of overland flow the approach has been to

assume the Darcy-Weisbach resistance law is a p p r o p r i a t e , i .e.


v = J(+ Sy)

and to estimate the friction factor, f, from the theoretical relationship


between f a n d Reynolds number Re,
f = K/Re (2.57)
where K is a parameter related to the surface roughness c h a r a c t e r i s t i c s
a n d r a i n d r o p impact. The parameter K i s approximated by
b (2.58)
K = K + A i
where KO is the parameter for surface roughness and A and b are
empirical parameters. When. i i s i n inches p e r h o u r , the coefficient A i s of
the o r d e r of 10 a n d the exponent b i s a p p r o x i m a t e l y u n i t y . T y p i c a l v a l u e s
for KO, Manning's n, and Chezy's C are given in Table 2.3. These
values a r e r a n g e s found i n the l i t e r a t u r e a n d were o b t a i n e d u t i l i z i n g d a t a

from control led experiments o r from smal I experimental watersheds.

TABLE 2.3 O v e r l a n d Flow Resistance Parameters

Surface L a m i n a r Flow T u r b u l e n t Flow


Manning n Chezy C
KO

Concrete o r Asphalt 24 - 108 0.01 - 0.013 73-38


Bare Sand 30 - 120 0.01 - 0.016 65-33
Gravel led Surface 90 - 400 0.012 - 0.03 38-1 8
Bare Clay to 100 - 500 0.012 - 0.03 36-1 6
Loam Soil
Sparse Vegetation 1000 - 4000 0.053 - 0.13 11-5
Short Grass 3000 - 10000 0.10 - 0.20 6.5-3.6
Bluegrass Sod 7000 - 40000 0.17 - 0.48 4.2-1.8
40

In the case of turbulent flow, either the M a n n i n g o r Chezy equation


i s used. The M a n n i n g e q u a t i o n i s p r o b a b l y the more p o p u l a r e q u a t i o n a n d

is used more o f t e n in watershed s i m u l a t i o n studies. The reasons f o r this


are obviously its wide-spread acceptance in open channel h y d r a u l i c s and
the a v a i l a b i l i t y of extensive t a b l e s of n-values f o r most channel types a n d
conditions.
Reported research indicates that low flows are laminar and that h i g h
flows are turbulent; but the location of the t r a n s i t i o n a l Reynolds number
is indeterminate which makes it difficult to apply the Darcy-Weisbach
resistance formulation throughout the entire hydrograph. Transition from
laminar to t u r b u l e n t flow has been r e p o r t e d a t Reynolds numbers r a n g i n g
from 20 to 2,000; with the range 300<Rer500 being the most frequently
reported .
Overton (1972) analyzed 214 e q u i l i b r i u m h y d r o g r a p h s from a n e a r l i e r
s t u d y of a i r f i e l d d r a i n a g e conducted b y the U.S. Army Corps of Engineers
(1954). He noted that these hydrographs supported the argument for low
flows being l a m i n a r a n d h i g h flows b e i n g t u r b u l e n t . I n almost e v e r y case,
the r i s i n g h y d r o g r a p h i n i t i a l l y rose v e r y s l o w l y i n d i c a t i n g viscous laminar
flow, and then became t u r b u l e n t as the flow increased. Overton a n a l y z e d
the rising portion of all 214 hydrographs in dimensionless form. He
normalized the discharge by the rain rate, and time by a lag time
parameter, w h i c h he d e f i n e d as the time from the occurrence of 50% of
tL,
the r a i n f a l l to 50% of the runoff volume. The normalized (dimensionless)
hydrographs were p l o t t e d on transparent p a p e r a n d superimposed. It was
apparent, that within a small error, a single dimensionless rising
h y d r o g r a p h c o u l d represent a l I 214 h y d r o g r a p h s . The a v e r a g e dimensionless
rising hydrograph was then plotted against the laminar, Manning and
Chezy dimensionless rising hydrographs as shown in Figure 2.6. Flows
appear to be laminar during the first half of the p e r i o d of rise and
turbulent during the second half. An error analysis indicated a 15%
standard error in fitting the entire r i s i n g hydrograph for the Manning
kinematic solution, a n d 19% f o r both the l a m i n a r a n d Chezy solutions.
To illustrate the effect of using only the Manning equation to
represent the flow equation hence the flow resistance, throughout the
entire hydrograph, consider Izzard's (1946) laboratory experimental run
Nos. 136 and 138. Run No. 136 consisted of two bursts with rainfall
intensity of 3.56 in/hr interrupted by a two minute lull. The f i r s t burst
lasted for 10 minutes and the second, 11 minutes. The rainfall event
produced a maximum Froude number of 0.55 and a minimum k i n e m a t i c flow
number of 156. Run No. 138 consisted of two bursts: the first was 1.83

in/hr for 8 minutes and the second was 3.55 in/hr for 8 minutes. This
41

1. I

KINEMATIC WAVE MODELS

~ I
i 2.6~ . C o m p a r i s o n o f T u r b u l e n t and L a m i n a r K i n e m a t i c W a v e
Solu ions w i t h Observed R i s i n g H y d r o g r a p h s

event produced approximately the same maximum Froude number and

minimum kinema ic flow number as Run No. 136. The runoff s u r f a c e was
an asphalt p l a n e w i t h the following physical characteristics:
L = 72 f t ; M a n n i n g n - v a l u e = 0.024; and So = 0.01.
Blandford and Meadows (1983) analyzed these events with a finite
element formulation of the kinematic overland flow model and obtained
the results shown in Figure 2.7. For Run No. 136, the predicted r i s i n g

and f a l l i n g limbs of the hydrograph l a g the observed l i g h t l y , while there


is near perfect agreement at the higher flows. From Run No. 138, only
the simulated falling limb lags the observed; the r e s t of the hydrograph
matches the observed v e r y well.

REFERENCES

B e v e n , K., Dec. 1982. On s u b s u r f a c e s t o r m f l o w , P r e d i c t i o n s w i t h s i m p l e


k i n e m a t i c t h e o r y f o r s a t u r a t e d and u n s a t u r a t e d f l o w s . W a t e r R e s o u r c e s
Res. 18 ( 6 ) p p 1627-33.
B l a n d f o r d , G.E. and M e a d o w s , M.E. 1983, F i n i t e E l e m e n t S i m u l a t i o n o f
Kinematic Surface Runoff, Proceedings, Fifth I n t e r n a t i o n a l Symposium
on F i n i t e Elements i n Water Resources, U n i v e r s i t y o f Vermont, B e n n i n g -
ton, Vermont .
D u n n e , T . 1978. F i e l d s t u d i e s o f h i l l s l o p e f l o w p r o c e s s e s . C h . 7, H i l l s l o p e
H y d r o l o g y , E d . K i r k b y , M.J. J o h n W i l e y , N.Y.
H e n d e r s o n , F.M. and W o o d i n g , R.A. 1964 ,
O v e r l a n d F l o w and G r o u n d -
water from a Steady R a i n f a l l of F i n i t e D u r a t i o n , J o u r n a l o f Geo-
p h y s i c a l R e s e a r c h , 69 ( a ) , pp. 1531-1540.
42

I z z a r d , C.F., 1946. H y d r a u l i c s o f R u n o f f f r o m D e v e l o p e d S u r f a c e s , H i g h -
w a y Research B o a r d , P r o c e e d i n g s o f t h e 2 6 t h A n n u a l M e e t i n g pp.
1 29-1 50.
M o r r i s , E.M., a n d W o o l h i s e r , D.A., A p r i l 1980. U n s t e a d y o n e - d i m e n s i o n a l
f l o w o v e r a p l a n e : p a r t i a l e q u i l i b r i u m a n d recession h y d r o g r a p h s .
W a t e r R e s o u r c e s R e s e a r c h , 1 6 ( 2 ) pp 355-366.
Overton, D.E., 1972. A V a r i a b l e Response O v e r l a n d F l o w M o d e l , Ph.D.
D i s s e r t a t i o n . Dept. o f C i v i l E n g i n e e r i n g , U n i v . o f M a r y l a n d .
O v e r t o n , D.E. a n d Meadows, M.E., 1976 S t o r m w a t e r M o d e l l i n g , A c a d e m i c
P r e s s , New Y o r k .
St. V e n a n t , A.J.C. B a r r e de, 1848. E t u d e s T h e o r i q u e s e t P r a t i q u e s s u r
l e Mouvement d e E a u x C o u r a n t e s . ( T h e o r e t i c a l and P r a c t i c a l S t u d i e s
o f Stream F l o w ) , P a r i s .
U.S. A r m y C o r p s of E n g i n e e r s 1954. D a t a R e p o r t , A i r f i e l d D r a i n a g e I n v e s t -
igations, Los Angeles District, O f f i c e o f t h e C h i e f of E n g i n e e r s ,
A i r f i e l d s Branch Engineering Division, M i l i t a r y Construction.
Woolhiser, D.A. 1981. Physical Il y b a s e d m o d e l s of watershed runoff,
pp. 189-202 in S i n g h , V.P., (Ed.) R a i n f a l l Runoff Relationships.
W a t e r R e s o u r c e s P u b l i c a t i o n s , C o l o r a d o , 582 pp.
Woolhiser, D.A. a n d L i g g e t t , J.A., 1967. U n s t e a d y o n e - d i m e n s i o n a l flow
over a plane. The r i s i n g h y d r o g r a p h . Water Resources Research,
3 ( 3 ) pp 753-771.

-OBSERVED
4r
c
L

\
e
w
(3

0 4 8 12 16 20 24 28 32
TIME( m in 1

Fig. 2.7 Outflow Hydrograph f o r I z z a r d ' s R u n No. 136


43

CHAPTER 3

HYDROGRAPH SHAPE AND PEAK FLOWS

D E S I G N PARAMETERS

Knowledge of the runoff process enables flow rates and volumes


to be predicted. H y d r o g r a p h c h a r a c t e r i s t i c s a r e of interest to researchers,
planners, designers and managers of drainage systems. The drainage
engineer will be most concerned with peak flows for design purposes.
It is also frequently useful to have the hydrograph shape especially
if detention storage or routing can reduce the peak flow. Expressions
for hydrograph shape and peak flows as a function of excess rainfall
intensity can be derived as follows for overland flow off simple planes.
The k i n e m a t i c equations summarized below a r e used for t h i s purpose.
Continuity

m
Energy 9 = CLY (3.2)
where x is the direction of flow, t is time, i is the excess rainfall
rate i-f, f is the loss rate and q is the discharge rate per unit
catchment width. For the present a l l u n i t s must be assumed consistent.
Later units will be introduced in order to render the numerical values
more meaningful. It will be assumed in the following analysis that i
and f are uniform in time’ a n d space f o r the duration of the storm t
d‘
q is the flow rate per unit width of plane, y i s flow depth, CL is a
coefficient a n d m i s an exponent.

SOLUTION OF KINEMATIC EQUATIONS FOR FLOW OFF A PLANE

The kinematic equations can be solved a n a l y t i c a l l y for some simple


cases. . In particular the runoff from a rectangular plane catchment
subject to uniform excess rain can be studied i n d e t a i l a n d expressions
for the time to e q u i l i b r i u m a n d h y d r o g r a p h shape can be d e r i v e d .
The fol lowing analysis demonstrates the simp1 i c i t y of arriving at

an equation for runoff for the catchment from a simple p l a n e catchment


sloping in the direction of flow. The analysis i s h a n d l e d more r i g o r o u s l y
i n c h a p t e r 2.
One starts with the general ized (one-dimensional) kinematic

equations f o r o v e r l a n d flow namely 3.1 a n d 3.2.


44

If the M a n n i n g e q u a t i o n i s assumed to h o l d then Q = K S $/n a n d m


l o
= 5/3 where So i s the slope of the p l a n e i n the d i r e c t i o n of f l o w , and n
i s the M a n n i n g roughness. K i s 1 .O i n 5.1. (metre) u n i t s a n d 1.486 in ft -
sec u n i t s .
After r a i n f a l I commences, the water depth near the downstream end of
the catchment will increase at a rate i the excess r a i n f a l l r a t e . The
e’
water surface profile then will be parallel to the plane at the

downstream end before e q u i l i b r i u m i s reached, which i s assumed to occur


before the r a i n stops, i.e. t < tc 5 td, where t i s the time to e q u i l i b -
rium, u s u a l l y r e f e r r e d to as the concentration time of the catchment.
Starting at the top or upstream end of the catchment where water
depth and discharge rate will be zero, a negative surge due to a non-
zero d y / d x will travel down the catchment o v e r l a n d i n c r e a s i n g i n depth as
r a i n continues to f a l l . Then a t any p o i n t i n time downstream of the s u r g e
the water surface is increasing in depth at a rate i b u t upstream the
water depth i s a t e q u i l i b r i u m since aq/ax = ie(see F i g . 2.4 l i n e ABIC1).
Eventually the whole catchment w i l l reach an e q u i l i b r i u m w i t h input
i L p e r u n i t width equal to d i s c h a r g e qL. At the i n s t a n t the catchment
reaches e q u i l i b r i u m

(3.3)

t is the c o n c e n t r a t i o n time of the catchment, which i s a f u n c t i o n of the


catchment length L, slope So, roughness n a n d excess r a i n f a l l r a t e i The
e‘
latter effect (i,) rarely appears in time of concentration formulae
associated w i t h the r a t i o n a l method.
D u r i n g the time of flow build-up the water depth a t the e x i t i s i t,
a n d the corresponding d i s c h a r g e r a t e q L = a ( i e t ) m (3.4)
The speed at which the reaction from upstream travels down the
catchment before e q u i l i b r i u m , i s o b t a i n e d from the c o n t i n u i t y equation. At
the wave f r o n t the r a t e of increase i n flow depth i s

(3.5)

where d x / d t i s the r a t e of t r a v e l of the wave f r o n t .

dy/dt = ie a t the wave front point (and downstream of it). One also
has from the c o n t i n u i t y equation b y e x p a n d i n g the a q / a x term

-
3 Y + 3 dy = i

*
at ay dx e
By comparing w i t h aatv + dx
-
dt ax
- i
e
one must h a v e -
dx = a t the wave f r o n t
dt ay
45

m- 1
a n d from 3.2, d x / d t = may (3.7)
Since y = iet,

-
dx = mcr(iet) m-1 (3.8)
dt
which i s the speed of the wave front at any time t 5 tC. Also during
equilibrium the discharge r a t e at any point x from the upstream water-
shed i s q = i x ( l i n e ABlC3 i n Fig. 2.4). Hence from 3.2
e
x = aym/i (3.9)
An expression for the discharge rate after the storm stops, which
is assumed to be after the time to equilibrium (t2t zt ) , i s obtained
d- c
by considering the water depth profile along the catchment a g a i n . A f t e r
the rain stops the effect of all upstream depths travels down to the e x i t
at a speed dx/dt given by 3.7. To predict when the depth at the e x i t
is 'y', imagine a series of waves travelling from the water p r o f i l e c u r v e
in a downstream direction at a constant speed d x / d t = aq/ay = mayrn-'
Integrating, x = x + maym-'(t-td) (3.10)
l-I/m l / m
= q / i e i mq a (t-t 1 (3.11)
d
since y = (q/a)l/m. (3.12)
I n p a r t i c u l a r a t the e x i t ,

L = q / i e + mq 1-1 / m a t / m b t d ) (3.13)
which i s an imp1 i c i t expression for the falling limb of the hydrograph.
The f u l l h y d r o g r a p h shape i s thus as i n F i g . 3.1.

HYDROGRAPHS FOR PLANES

Expressions for the rising and falling limbs of the hydrograph


off a simple rectangular catchment were derived previously. The
discharge at the mouth before time t or td is reached, .is given by
q = a(iet)"' (3.14)
If rain continues after t = tc i.e. td > tc, then the h y d r o g r a p h top is
h o r i z o n t a l as i n d i c a t e d i n F i g . 3.1 case I l l .
I f on the other hand rain stops at t = t then the hydrograph
falls immediately after t (case 11). In either case it may be shown

that the falling limb of the hydrograph is obtained from the implicit

(3.13)

The total depth of excess rain has been kept constant in each case
in Fig. 3.1 so that i = p/td where P is the depth of precipitation.
46

Flow
q
I c-

n u d

+
Time 1

Fig. 3.1 Outflow h y d r o g r a p h shape f o r d i f f e r e n t storm


d u r a t i o n s b u t s i m i l a r t o t a l excess r a i n .

Also illustrated in Fig. 3.1 is the case of the hydrograph for a


s h o r t storm (t < t ) (case I ) . A f t e r time t d t h e downstream d e p t h ( a n d
d c
hence flowrate) r e m a i n s constant u n t i l the i n f l u e n c e of the upstream e n d
reaches the e x i t .
The upstream l i m i t of y = i t is at
e d
m-1 m
x = q/ie = aym/ie = a i t (3.15)
e d
I f t h i s p o i n t t r a v e l s a d i s t a n c e L - x a t a speed
dx/dt = maym-' (3.16)
i t will r e a c h t h e e x i t in time
rn-l
L / a ( ietd) -td
At = Ax/ma(ietd) m-1-- ( L - a i e r n - l t d m ) / m a i e m-1 t d m-1 -
m (3.17)
Assume i i s t h e excess r a i n f a l l r a t e f o r a storm d u r a t i o n e q u a l to t
ec C'
Since i e t d = iectC f o r equal volume of r a i n , and

At = (tc-td)/m (3.18)

This is the duration of the flat top of the hydrograph I in Fig. 3.1.
It should be noted that the f a l l i n g l i m b s of the hydrograph in Fig. 3.1
omit losses after rain stops. If infiltration ( f ) continues the h y d r o g r a p h
will look l i k e those in F i g . 3.2. It i s generally necessary to model such

system n u m e r i c a l l y to get the h y d r o g r a p h shape (see Wooding, 1965).


47

I D

-
9
iL

0.5

0 I I J 1 5

t/f,

F i g . 3.2 Effects of i n f i l t r a t i o n o n catchment d i s c h a r g e


48

DERIVATION OF PEAK FLOW CHARTS

If it can be assumed that the rainfall intensity-duration relation-

ship for a specified frequency of exceedance can be approximated by a


f o r m u l a of the form
a
I = (3.19)
(C+td)P
where i is the rainfall rate in mm/h or inches per hour and td i s the
storm duration in hours, then a simple estimate of peak flow can be
derived. It IS assumed that i is constant during the storm duration t
d
and uniform over the catchment. The storm is also assumed stationary.
c is a time constant unique for a particular rainfall region and p is
an exponent, also a unique function of the region. Thus for temperate
regions in South Africa it was found that c = 0.24h a n d p = 0.89 while
for coastal r e g i o n s c = 0.20h a n d p = 0.75 (Op ten Noort a n d Stephenson,
1982).
a i s a f u n c t i o n of rainfall region, mean a n n u a l precipitation, MAP,
in mm and recurrence interval T in years e.g. a = (b+e.MAP)Toe3where
b and e are regional constants. a is not dimensionless and if c and
td are in hours, then i is in mm/h or inches per hour. An areal
reduction factor is also necessary f o r l a r g e catchments (e.g. Stephenson,
1981).
Losses are subdivided into two components, an initial loss u in
mm or inches and a uniform infiltration loss rate f in mm/h or inches
per hour. A typical rainfalI IDF (intensity-duration-frequency)
relationship and the corresponding hyetograph with losses and excess
r u n o f f i n d i c a t e d i s presented i n F i g . 3.3.
The r a t e of excess r a i n f a l l i s :
i = i - f (3.20)
where f i s the i n f i l t r a t i o n r a t e a n d the d u r a t i o n of excess r a i n i s :
te = td - tu = td - u/i (3.21)
where u is the initial abstraction (measured in terms of a depth of
rain).

For small catchments the maximum peak runoff rate occurs when
the duration of excess rain equals the concentration time, t . For p l a i n
rectangular catchments the concentration time is a function of excess
rainfall rate

where t = te ( b o t h i n hours here) (3.23)

a = rS/n (S.1 u n i t s ) o r 1.486 &/n (fps units) (3.24)


49

hyetogroph

{I I
\>(
ntial
abstrac-
tion

t
4
~-t+----4
Fig. 3.3 Excess flow hyetograph d e r i v e d from IDF curve

i is excess rainfall intensity, L is catchment length, 5 is the


downstream slope, m is an exponent, 5/3 i n M a n n i n g equation, n i s the
Manning roughness, q i s the d i s c h a r g e r a t e p e r u n i t w i d t h .
The following expression may then be derived for ie/a from
equations (3.19) to (3.22):

(3.25)

Thus the maximum r u n o f f r a t e p e r u n i t w i d t h of catchment


q = i L (3.26)
may be o b t a i n e d in terms of a. Equation (3.25) may be solved i t e r a t i v e l y
e.g. using the Newton Raphson procedure, for ie/a, or solution may be
obtained with the aid of graphs of ie/a plotted against ( L / a am-').
The storm duration td corresponding to the peak runoff may be o b t a i n e d
from (3.21) a n d (3.22).

Long Catchments

For very long catchments the theoretical concentration time tc i s


high. I n such cases the corresponding excess r a i n f a l l rate for a duration
equal to tc is low and in fact could conceivably be less than the
infiltration rate f. It is thus apparent that i n such cases the maximum
runoff rate may coincide with a storm of shorter duration than the
concentration time of the catchment. The e n t i r e catchment wil I thus not
50

be contributing at the time of the peak in the (rising) hydrograph.

If a local, intense storm turns out to be the design storm, the areal
reduction factor applied to point rainfall intensity relationships may
be less significant. The factor is generally closer to unity the smaller
the lateral extent of the storm, but on the other h a n d shorter duration
storms have a more significant reduction factor (less than long storms).
These facts will not b e revealed u s i n g the Rational method w i t h rainfall
propor t iona I losses.
Before equilibrium is reached the runoff per unit width at the
mouth of the catchment at any time t after the commencement of excess
r a i n or runoff is
m
q = a (iete) (3.27)
where t td - u / i
- (3.28)
m-1 l / m
<tc = ( L / a i e ) (3.29)

(3.30)

(3.31)

(3.32)

q/aam is plotted against td in Figs. 3.4 to 3.6 (the full lines) for
different values of the dimensionless parameters U = u/a and F = f/a.
For a l l cases of F > 0 the l i n e s e x h i b i t a peak r u n o f f a n d the correspond-

ing storm duration td for an infinitely long catchment. For most catch-
ments it i s necessary to e s t a b l i s h whether t is less t h a n t i.e. whether
the peak occurs before the catchment has reached e q u i l i b r i u m .
I n fact, t = td - t (3.33)
Therefore f o r t =
c te

(3.35)
U

L/aam-l may therefore be plotted against td for selected values of u/a


and f/a as on the right hand side of Figs. 3.4 to 3.6. Each chart is
for a different initial abstraction factor U, and it may be necessary
to interpolate between graphs for intermediate values of U. Lines for
v a r i o u s F a r e p l o t t e d on each g r a p h .
Now the peak r u n o f f w i l l be the maximum of e i t h e r

(a) t h a t corresponding to t = tc f o r short catchments o r


(b) that for t < tc f o r long catchments.
51

In order to identify which condition applies, enter the chart for


m-1
the correct U = u/a with L/aa on the r i g h t h a n d side a n d u s i n g the
dotted line corresponding to the correct F read off the corresponding
+ t on the abscissa. It may occur that the equilibrium tc is
t d = tc
off the c h a r t to the right in which case it is probably of no interest
since the following case applies. Select the full line with the F = f/a
and decide whether its maximum lies at or to the left of the value td

previously established. If the peak lies to the left, read the revised
design storm duration td corresponding to the p e a k , and the correspond-
m
i n g peak flow parameter q / a a on the l e f t h a n d o r d i n a t e .

M o d i f i c a t i o n for P r a c t i c a l U n i t s

The preceding equations assume dimensional homogeneity. Unfortun-

ately both the Manning resistance equations and the I-D-F relationships
are empirical and the coefficients depend on the u n i t s employed. I n the
Manning form of equation (3.241, q is in m2/s if iete is in metres.
a is 6 / n in S.I. u n i t s where S is the dimensionless slope a n d n i s the
M a n n i n g roughness.
It is most convenient to work with td in hours and i and a in
mm/h. The numbers are then more realistic. I n equation (3.32) if q is
in m3/s/m, a i n m-s units, a in mm/h and t in hours then the leff
d
h a n d side should be replaced by
q
- - l ~ 1~ 0 ~~ m
~ -
aam 5/3
oa

= I O ~ Q
5/3 (3.36a)
Baa
where Q i s total runoff r a t e off a catchment of width B metres. Note t h a t
the right hand side of (3.32) is in hm i f td is i n h, so no c o r r e c t i o n
is made to the above factor to convert a to secs, only to convert mm
to m. This is what the left hand axis of Figs. 3.4 to 3.6 represent if
a is in mm/h. It is referred to as the runoff-factor, QF. Similarly the
left hand side of equation (3.35) is L/aam-l in homogeneous units, or
if a is in mrn/h, L in metres and a i n m-s units, then it should be
replaced b y :

-
L a -looom
- - L
a 3600000 m 236aa
/ 3 (3.37a)

This i s termed the l e n g t h f a c t o r LF.


If q is in ft’/s/ft, a i n ft-sec units, a in inches p e r hour and td in
hours then the expression f o r Q should be r e p l a c e d b y

63q ( 3.36b )
aa ’I3
52

... LENGTH W l O A
T t 9.8

Y. Y

u= 0.004 q.0
c= 0.zL.l0
P= 0.890
3.6

3.2
L
-1
2.8
0
I-
v
2.q R
LL

I
t-
2.0 9
W
-I

1.6

I .I

I1.B

0. I 0.9

a. I 6.0
0 I 2 3
5JDRl bCRAlIOU IN HA5

F i g . 3.4 Peak r u n o f f f a c t o r s f o r U = 0.00


53

-- AUNDff Ffl(1UR .. . LENGTH fRClOR


9.6

q.O

3.6

3.2
L
-.l
2.8 fy:
0
I-
U
2.q [L
r

I
I-
7.0 D
z
4

1.6

1.2

0. H

0.q

0.0
E I 2
51ORR DURATION IN HR5

Fig. 3.5 Peak r u n o f f f a c t o r s f o r U = 0.20


54

-- RUNUFf FR(1UR .. . LENETH FACTOR


:. 2 Y.B

I .a 4.0

1.9 3.6

0.9 3.2

LL
L
13 -I
2.9
K 0.7
~y:

0 n
I- I-
LJ V
21.6 2.Y a:
L
LL
L
I
I-

g 0.s
LL
2.0 irr
z
7 w
-I
ct:
O.Y I .6

0.3 I .2

0.2 0.9

0. i 0.q

a.n 0.0

Fig. 3.6 Peak r u n o f f f a c t o r s f o r U = 0.40


55

a n d the l e n g t h f a c t o r expression (3.37b) is replaced b y

687a a‘’’
Figs. 3.4 to 3.6 are plotted in terms of the dimensioned expressions for

L F a n d QF, i.e. use t d i n h o u r s a n d the o t h e r terms i n the chosen m e t r i c


or English units above. Further charts were published by Stephenson
(1982).
EXAMPLE
Consider a p l a n e r e c t a n g u l a r catchment with the f o l l o w i n g c h a r a c t e r i s t i c s :
o v e r l a n d flow length L = 800m

w i d t h 0 = 450 m
slope S = 0.01
M a n n i n g roughness n = 0.1
inland region, MAP = 620mm/annum
20 y e a r r e c u r r e n c e i n t e r v a l storm
r a i n f a l l f a c t o r a = (7.5 + 0.034 x 620)20°‘3 = 70mm/h

c = 0.24h, p = 0.89 i n i mm/h = a / ( c + t )p


d
i n f i l t r a t i o n r a t e f = 14mm/h
i n i t i a l a b s t r a c t i o n u = 14mm
a= JT/n = 1.0
F = f/a = 0.2
u = u/a = 0.2
800
2/3 = 1.30
LF=*= 36 x 1 x 7Q
From F i g . 3.5 c o n c e n t r a t i o n time t = 3.0h
I t will b e noted t h a t there may be two s o l u t i o n s f o r storm d u r a t i o n
td. The longer one corresponds to a very low precipitation rate and is
of little interest. Even the shorter time to equilibrium is longer than
the storm resulting in the peak runoff. I n t h i s case i t appears that the
peak runoff corresponds to a 1.3 hour storm (shorter than the time to
equilibrium) a n d the corresponding
QF = 0.30
= 1 0 561/~aa5/3

therefore Q = 0.30 x 450 x 1 x = 1 .60m3/s

The c o r r e s p o n d i n g p r e c i p i t a t i o n r a t e i s :
i = a / ( c + t )’ = 70/(0.24 + 1.3)o’89 = 48mm/h
d
The e q u i v a l e n t r a t i o n a l c o e f f i c i e n t C i s
1.60/(450 x 800 x 48/3600000) = 0.33

I t may also be confirmed that the storm duration corresponding to time


to e q u i l i b r i u m of t h e catchment i s 2.8 h o u r s :
56

If i = 70/(0.24+3.0)0.89-14 = 10.5mm/h = 2.9x1OP6m/s


m-1 1-m
t = (L/aie )
= {800/1 x (2.90 x I O - ~ ) ~ / ~ = ~ 9050s
~ / ~ = 2.5h
t = u / i = 14/48 = 0.3h
U -
Storm d u r a t i o n t = 2.8h
d
The corresponding r u n o f f would be o n l y :
-6
i BL = 2.9 x 10 x 800 x 450 = l.Om’/s
i.e. the peak runoff corresponds to a storm of shorter duration than
to time to e q u i l i b r i u m of the catchment.

EFFECT OF CANALIZATION

The charts presented are for the case of overland flow. It fre-
q u e n t l y occurs t h a t runoff reaches channels, a n d t h u s flows to the mouth
of the catchment faster than if overland. The critical storm duration
may thus be s h o r t e r a n d the peak flow h i g h e r t h a n w i t h no c a n a l i z a t i o n .
An estimate for the concentration time of a catchment with a wide
rectangular channel down the middle may be made using this chapter
if overland flow time can be neglected. The effective catchment width
is taken as b, the stream width, and both rainfall rate i and losses
f and u should be increased by the factor B/b where B is the true
catchment width. The charts herein can then be applied as in the
example.
I n many situations both o v e r l a n d flow a n d stream flow are signifi-
cant and the problem cannot be solved as s i m p l y as herein. The h y d r o -
logist must then resort to trial and error methods using dimensionless
h y d r o g r a p h s f o r catchment - stream systems as presented l a t e r .

‘channel

F i g . 3.7 Rectangular catchment w i t h c e n t r a l c o l l e c t i o n channel.


57

EST I MAT I ON O F ABSTRACT IONS

The losses to be deducted from precipitation include interception


on vegetation and roofs, evapotranspiration, depression storage and
infiltration. The remaining losses may be divided into initial retention
and a time-dependent infiltration. The losses a r e r e a l l y f u n c t i o n s of many
variables, including antecedent moisture conditions and ground cover.

Infiltration is time-dependent and an exponential decay curve is often


used. The infiltration typically reduces from an initial rate of about
50 mm/h down to 10 mm/h over a p e r i o d of about an hour. The rates,
especially the terminal loss rate, will be higher for coarse sands than
for clays.
The time-decaying loss rate could be approximated by an initial
loss p l u s a uniform loss over the d u r a t i o n of the storm. Values of initial
and uniform losses are tentatively suggested below. The mean uniform
loss rates are average f o r storms of 30 minutes d u r a t i o n a n d the i n i t i a l
losses include the initial 10 minutes rapid infiltration or saturation
amount, In the case of ploughed lands, and other especially absorptive
surfaces an additional loss of up to lOmm or more may be included.
Allowance must also be made for reduced losses from covered areas

(paved or roofed).

TABLE 3.1 Surface Loss Parameters

I n i t i a I a b s t r a c t ion Infiltration rate


Max Soil
Surface r e t e n t i o n moisture d e f i c i t

-
mm -
inches -~
mm inches mm/h inches/h

Paved up to 5 0.2 0 0 0 0
Clay up to 5 0.2 15 0.8 2 - 5 0.1 - 0.2
Loam up to 7 0.3 20 1.2 5 - 15 0 . 2 - 0.6
Sandy up to 10 0.4 30 1.5 15 - 25 0.6 - 1
Dense up to 15 0.6 5 0.2 5 - 15 0 . 2 - 0.6
vegetation
58

REFERENCES

Op ten Noort, T.H. and S t e p h e n s o n , D., 1982. F l o o d P e a k C a l c u l a t i o n


i n S o u t h A f r i c a . W a t e r Systems R e s e a r c h P r o g r a m m e , U n i v e r s i t y of t h e
W i t w a t e r s r a n d , R e p o r t No. 2/1982.
Stephenson, D., 1981. Stormwater Hydrology and Drainage, Elsevier,
A m s t e r d a m . pp 276.
S t e p h e n s o n , D., 1982. "Peak F l o w s f r o m S m a l l C a t c h m e n t s U s i n g K i n e m a t i c
Hydrology,'' Water Systems Research Programme, Report 4/1982.
U n i v e r s i t y of t h e W i t w a t e r s r a n d , J o h a n n e s b u r g .
Wooding, R.A., 1965. A H y d r a u l i c M o d e l f o r t h e C a t c h m e n t - s t r e a m P r o b l e m ,
I I , N u m e r i c a l S o l u t i o n s , J o u r n a l o f H y d r o l o g y , 3, p 268-282.
59

CHAPTER 4

K I NEMAT I C ASSUMPT IONS

NATURE OF K I NEMATIC EQUATIONS

The kinematic flow approximation has proved to be very useful


in s t o r m w a t e r m o d e l l i n g and in the development of a better understanding
of the runoff process. Kinematic models are deterministic models and
represent a distributed, time-variant system. They can, therefore, be
coupled with other process models to investigate the effects of l a n d use
change, temporal and spatial variations in rainfal I and watershed
conditions, and pollutant washoff.
Starting with the formulation of the kinematic wave theory by

Lighthill and Whitham (19551, kinematic overland flow models have been
utilized increasingly in hydrologic investigations. The first application
to watershed modelling was by Henderson and Wooding (1964). The
conditions under which the kinematic flow approximat ion holds for surface
runoff were first investigated by Woolhiser and Liggett (1967); they
found it is an accurate approximation to the ful I equations for most

overland flow cases. Since then, analytical solutions h a v e been o b t a i n e d


for runoff hydrographs during steady r a i n f a l I on s i m p l e geometric shaped

watersheds; and numerical models have been developed for application


to more complex watersheds and unsteady rainfall. With the easy
availability of micro-computers, the numerical models a r e r e a d i l y access-
ible. Successful use of these models r e q u i r e s a familiarity with computers
a n d an u n d e r s t a n d i n g o f k i n e m a t i c o v e r l a n d f l o w .

K I NEMAT I C APPROXIMATION TO OVERLAND FLOW

Kinematic overland flow occurs when the dynamic terms in the


momentum equation are negl i g i b l e . There is no appreciable backwater
effect and discharge can be expressed as a unique function of the depth
o f f l o w a t a l l d i s t a n c e x and t i m e t. That is,

Q = bay" (4.1
where Q is the discharge, y i s the depth of flow, b the w i d t h a n d ci , m
a r e constants.
The latter conclusion can be established by normalizing the
momentum equation by the steady uniform discharge Qn. The momentum

e q u a t i o n t h e n becomes
60

where So is the bed slope, v is flow velocity, g is gravity and A is


cross sectional area. If the sum of the terms to the r i g h t of the minus
sign i s much less t h a n one, then

Q Qn (4.3)
which means that gradually-varied flow may be approximated by a
uniform flow formula such as Manning’s equation. If one writes
Manning‘s equation for a wide rectangular cross-section such as an
overland flow plane, since the hydraulic radius can be approximated
by the depth of flow, one obtains the following expression (SI units)

Q = -1 b y y 2/3s f (4.4)

or Q/b = A S y5I3
(4.5)

where Q/b i s discharge per u n i t width. For r u n o f f from a p l a n e s u r f a c e


with uniform roughness and slope, n and So are constant. Eq. 4.5 can
1
therefore be written in the same form as Eq. 4.1 with cx = S “n and
rn = 5/3; a n d for the c i t e d conditions d i s c h a r g e can indeed be expressed
as a u n i q u e f u n c t i o n of the depth of flow.

Governing Equations

The governing equations for the kinematic overland flow approx-


imation a r e Eq. 4.1 and the equation f o r c o n t i n u i t y

where q. i s the i n f l o w p e r u n i t length x .

Conditions for t h e Kinematic A p p r o x i m a t i o n

The conditions under which the kinematic approximation holds for


overland fiow can best be illustrated by applying the fufi equations
to runoff from a long, uniformly sloped plane of unit width as shown
in Figure 4.1. The plane i s of length L and slope So. Rainfall occurs
over the p l a n e a t the r a t e i ( x , t ) , and i n f i l t r a t i o n i s at the r a t e f ( x , t ) .
By writing the rainfal I and infiltration rates in terms of x and t, we
include the effects of spatial and temporal variations in rainfall and
soil. The c o n t i n u i t y a n d momentum equations a r e w r i t t e n as

(4.7)

and
61

vi
av
-
at + v z
av
+ g=
ay -
- g(S0 - 5f ) - e
Y
(4.8)

where ie(x,t) is the rainfall excess rate at distance x a n d time t and

the other terms h a v e been d e f i n e d p r e v i o u s l y .

UNIFORM RAINFALL

x F
L -x

Fig. 4.1 Uniform r a i n over a long impermeable p l a n e

For the purpose of this discussion, Sf is conveniently defined by


the Chezy e q u a t i o n

(4.9)

C being the Chezy coefficient which equals Jf/sg where f i s the Darcy
friction factor. By writing Eqs. 4.7 and 4.8 in dimensionless form, the
number of parameters are reduced from f i v e to two with obvious advant-
ages. Woolhiser and Liggett (1967) first presented the following dimen-
sionless equations

' H + " E + H - au
- = 1 (4.10)
aT ax ax
and

(4.11)

where
H = y/yo, U = v/vo, X = x/L, T = tvo/L (4.12)
and y and v are the normal depth and velocity, respectively, at the
end of the p l a n e f o r a given steady r a i n f a l l excess r a t e , i The normal-
e'
i z i n g parameters a r e r e l a t e d b y :
i L = v y (4.13)
0 0
62

and
2 2
vo/coYo = so (4.14)

The two independent parameters in Eqs. 4.10 and 4.11 are the normal
flow Froude number, Fro, vo/ J ( g y o ) , and the kinematic flow number,
k (Woolhiser a n d L i g g e t t , 1967).

k = - (4.15)
2
yoFro

Woolhiser a n d L i g g e t t (19671, Brutsaert (19681, Morris (1979) and


Vieira (1983) solved Eqs. 4.10 and 4.11 for the rising hydrograph for
a range of Fr and k values under normal depth, critical depth and
zero depth gradient downstream boundary conditions. The solutions were
started using an analytic solution for simple cases and numerical
solutions in the other three characteristic solution zones. The results
of al I studies were quite similar. Sample results are shown in Figures
4.2 and 4 . 3 .

0 .I 2 3 .4 .5 .6 .7 .8 9 1.01.1 1.21.31.41.51.6
T

Fig. 4.2 Effect of v a r y i n g k on dimensionless r i s i n g h y d r o g r a p h


(Woolhiser a n d L i g g e t t , Water' Resources Research,
3 , 7 6 4 , 1967, American Geophysical U n i o n ) .
63

As seen in Figure 4.2, as the kinematic flow number increases,


the solution converges very rapidly toward the solution for k equal to
infinity. Woolhiser a n d L i g g e t t noted t h a t for Fro = 1 the maximum e r r o r
between the r i s i n g h y d r o g r a p h s f o r k = 10 a n d k equal infinity i s about
10 percent. The effect of varying Fro while holding k constant i s shown
in Figure 4.3. Similar to the results obtained for increasing k, as F r
increases, the solution converges to the s o l u t i o n f o r k equal to i n f i n i t y .
What is the significance of k equal to infinity? If one divides
Eq. 4.11 by k, the momentum e q u a t i o n reduces to the f o l l o w i n g e x p r e s s i o n .
1 - U2/H = 0 (4.16)
Hence
U2 = H (4.17)

Substituting Eq. 4.17 into the dimensionless continuity equation, Eq.


4.10, the k i n e m a t i c wave equation i s obtained.

-
aH a H 2/3 = 1
aT + x (4.18)

S o l v i n g Eq. 4.18 f o r an i n i t i a l l y d r y surface, we get


H = T (4.19)

a n d from Eq. 4.17

U = T1/ z (4.20)
Thus, the r i s i n g h y d r o g r a p h i s g i v e n b y

(4.21)

I .5

Q* :$
AC k =I0
/ PA RAME TER :Fro

.I
'0
1 .I .2 .3 .4.5 .6 .7 .8 .9 1.01.1 1.2 1.31.41.51.6
T

Fig. 4.3 Effect of v a r y i n g F r on dimensionless r i s i n g h y d r o g r a p h


(Woolhiser a n d Liggc?tt, Water Resources Research, 3 ,
7 6 4 , 1967, American Geophysical U n i o n ) .
64

where Qc is discharge normalized b y the excess rainfall intensity. The


result i n Eq. 4.21 suggests that all r i s i n g h y d r o g r a p h s f o r steady excess
rain on uniform planes can be represented by a single dimensionless
hydrograph. This result also suggests there is a unique relationship
between depth and discharge, and the depth is the normal depth for
uniform flow at that discharge. When k i s l a r g e the s o l u t i o n to the f u l l
equation can be closely approximated by the kinematic solution. This
is the kinematic approximation which has been described in detail by
several investigators (Lighthill and Whitham, 1955; Wooding, 1965;
Woolhiser and Ligget, 1967; Morris and Woolhiser, 1980; Vieira, 1983).
Woolhiser and Liggett (1967) stated that the kinematic wave
approximation may be used instead of the ful I equations if k > 2 0 and
F r o > 0.5. Overton and Meadows (1976) noted that Eq. 4.21 i s applicable
to c h a r a c t e r i s t i c zone A (Fig. 5.1) a n d of the solution shown i n F i g u r e 5.1,
zone A constitutes substantially all of the solution for kinematic flow
numbers of 10 o r greater. Therefore, they recommend the k i n e m a t i c wave
approximation be used only when k > 1 0 , regardless of the Froude number
value. Morris and Woolhiser (1980) re-evaluated Eqs. 4.10 and 4.11 and
2
suggested that Fr k '5 if the kinematic approximation is used. It is
0
interesting to note this i s equivalent to the o r i g i n a l criteria suggested
by Woolhiser and Liggett, except that it allows the kinematic approxi-
mation to be used for F r < 0.5, provided k >20. With these c o n d i t i o n s in
mind, and using the results obtained in his own study, Vieira (1983)
developed the plot in Figure 4.4 as a guide to determine when the
k i n e m a t i c a n d d i f f u s i o n wave a p p r o x i m a t i o n s may be used.

k
'
m
KINEMATIC APPROX.

10-

FULL S A I N T VENANT

F i g . 4.4 A p p l i c a b i l i t y of k i n e m a t i c , diffusion a n d dynamic wave models


( A f t e r V i e i r a , 1983)
65

Kinematic F l o w Number

The kinematic flow number can be placed in terms of the p h y s i c a l


and hydraulic characteristics of a plane b y eliminating y a n d F r o from
Eq. 4.15 u s i n g Eqs. 4.5 a n d 4.13. The r e s u l t i n g r e l a t i o n s h i p i s
1 .2s0.4L0.2
gn 0
k = (4.22)
. 0.8
e
For r a i n f a l l i n t e n s i t y in mm/hr and length in meters, Eq. 4.22 becomes

1 .2s0.4L0.2
k = 1 . 7 ~lo6 "_ .0.8
(4.23)

e
and for r a i n f a l l intensity in in/hr a n d l e n g t h i n feet

1.2 0.4 0.2


5 " L- (4.24)
k = 10
. 0.8
e
In general, high k values are produced on rough, steep, long planes
w i t h low r a i n r a t e s .
Similarly, the quantity kFr2 can be expressed in terms of the
p h y s i c a l a n d h y d r a u l i c c h a r a c t e r i s t i c s of a p l a n e . From Eq. 4.15

kFr2 = ~ (4.25)
0
YO

I f we write Eq. 4.21 in dimensional form using Eq. 4.12, Manning's


resistance law instead o f ' Chezy's, and the following definition for Q+
Q
Q, = -
(4.26)
e
we o b t a i n the e q u a t i o n
tv
0 5/3
Q = i A (-) (4.27)
e L
where A i s the c o n t r i b u t i n g watershed area. For a steady r a i n f a l l excess
rate, the flow is a maximum a n d equal to i when the terms i n s i d e the
parentheses are equal to one, that is, when time i s equal to the time
of c o n c e n t r a t i o n , tC, or preferably, the time to e q u i l i b r i u m . The q u a n t i t y
L/vo i s one definition for the time of c o n c e n t r a t i o n used in peak runoff
estimates. According to Eq. 4.27, for a steady excess rate, at the time
of c o n c e n t r a t i o n , the r u n o f f r a t e i s a maximum a n d equal to i . In other
words, one d e f i n i t i o n for the time of c o n c e n t r a t i o n i s that it i s the time
required for a watershed to reach equilibrium for a steady rainfall
excess. T h i s o c c u r s when

tC = L/vo (4.28)
66

S u b s t i t u t i n g Eq. 4.28 i n t o Eq. 4.13 we get

y o = i t (4.29)
e c
which, when s u b s t i t u t e d i n t o Eq. 4.25 yields

2 oL
kFo = -
i t
(4.30)
e c
Using the definitions of Eqs. 4.28 and 4.29, and Manning's equation,
one o b t a i n s the desired expression, f o r r a i n f a l I i n mm/hr,

(4.30a)

and

460Sb*3 L O e 4
(4.30b
kF: = 0.6i0.2
e
for r a i n f a l l i n i n / h r .

2
I n general, kFo values are high for smooth, steep, long planes with
low rainfall rates. This result is similar to the expression for k, except
that the effect of roughness on the Froude number suggests the k i n e m a t i c
model may be more a p p l i c a b l e to u r b a n watersheds w i t h smooth impervious
surfaces.
To illustrate the hydrological applicability of these results,
consider an asphalt parking lot with the following characteristics:
L = 50 meters; So = 0.005; n = 0.022. For an a v e r a g e excess intensity
of 50 m m / h r , k = 200 a n d k F r 2 = 31.

K I NEMAT I C AND NON-K I NEMAT I C WAVES

It was noted in Chapter 2 that the diffusion and kinematic wave


models may be used instead of the f u l l dynamic wave equations i f certain
assumptions can be made. In this section, conditions under which the

two models can be applied to flood routing in streams are examined.


The material presented here should give the reader a better under-
s t a n d i n g of the p h y s i c a l n a t u r e of k i n e m a t i c a n d non-kinematic waves.
The physical significance of kinematic and non-kinematic waves a n d

the major differences between the respective models a r e b e t t e r understood


if the wave speed and crest subsidence (hydrograph dispersion) charac-
t e r i s t i c s a r e known.
67

Wave Speed - Kinematic Waves

The k i n e m a t i c wave speed i s determined b y comparing the c o n t i n u i t y


equation w i t h no l a t e r a l inflow

(4.31)

w i t h the d e f i n i t i o n of the t o t a l d e r i v a t i v e of Q
dQ - aQ dx
- +
aQ
(4.32)
dt ax dt
By r e w r i t i n g E q . 4.31 as

-a Q
+ - - -aA dQ d t = (4.33)
ax at dA dx
to an observer moving w i t h wave speed, c,

(4.34)

the flow r a t e would appear to be constant, i.e.,

(4.35)

This result follows from the d e f i n i t i o n of the total derivative, Eq. 4.32,
a n d the e q u a t i o n of c o n t i n u i t y , Eq. 4.31.
For most c h a n n e l s where the flow i s in-bank

(4.36)

where B i s the channel top width in meters (feet); and since Q is a


u n i q u e f u n c t i o n of y
Q = clym (4.37)
the k i n e m a t i c wave speed i s g i v e n as

(4.38)

This relationship is analogous to that of Seddon (1900) who observed


that the main body of flood waves on the Mississippi River moved a t
a r a t e g i v e n b y Eq. 4 . 3 8 .
Eq. 4.38 implies that equal depths on both the leading and
recession limbs of a hydrograph travel at the same speed. Since g r e a t e r

depths move at faster rates, it follows that the leading limb of the
hydrograph w i l l steepen a n d the recession limb will develop an elongated
tail. Eq. 4.38 also shows that kinematic waves are propagated down-
stream only, i.e. Eq. 4.38 is a forward characteristic. Kinematic flow
does not e x i s t where there a r e b a c k w a t e r effects.

Crest Subsidence

Combining Eqs. 4.32 and 4.35, and substituting for Q using Eq


68

4.37 it can be shown that to an observer moving with wave speed c

(4.39)

M a n i p u l a t i n g t h i s equation y i e l d s

-
d Y = V + Q dt = o (4.40)
dx ax at dx
which establishes that theoretically, the kinematic wave crest does not
subside as the wave moves downstream.
These results show that a kinematic wave can alter in shape but
does so without crest subsidence. Further, the maximum discharge rate
occurs with the maximum depth of flow. ( T h i s i s the assumption implicit
in the slope-area method for e s t i m a t i n g flood discharges from h i g h water
marks).

H y d r a u l i c Geometry and Rating Curves

One important aspect of the k i n e m a t i c wave model i s the replacement

of the momentum equation with a u n i f o r m flow formula, which i s nothing


more than a s i n g l e v a l u e d r a t i n g between d i s c h a r g e a n d depth (or area)
at a point in the stream. As discussed p r e v i o u s l y , the fact that natural
channels a r e not prismatic leads to subsidence a n d d i s p e r s i o n of a hydro-
graph, suggesting that the discharge rating relationship is not unique
but varies over the hydrograph. If the dispersive characteristics are
small such that a variable rating relationship does not differ signifi-

cantly from the single valued rating, the conclusion can be d r a w n that
the main body of a hydrograph moves k i n e m a t i c a l l y . I n w h i c h case, the
kinematic model (or the diffusion model) should be sufficient for most

simul at ion purposes. This represents an economy of data and compu-

t a t i o n a l requirements over the dynamic wave model.


It is evident from the relationships between hydraulic geometry

and discharge first set forth by Leopold and Maddock ( 1 9 5 3 ) that the
flow in many streams i s essentially kinematic. The fact that the channel
c h a r a c t e r i s t i c s of natural streams seemed to constitute an interdependent
system which could be described by a series of graphs h a v i n g a simple
geometric form suggested the term "hydraul ic geometry". Subsequent
studies have verified and expanded on this initial work with the r e s u l t
that hydraulic geometry equations may be used to estimate general
channel characteristics at any locat ion within the drainage system.

As a result of their analysis of the variation of hydraulic

characteristics at a particular cross-section in a river, Leopold and

Maddock proposed that discharge be related to other hydraulic factors


69

i n the fol lowing manner.


b
w = aQ (4.41a)
f
d = CQ (4.41b)

v = kQm (4.41~)
where w is width, d i s depth, v i s cross-sectional mean v e l o c i t y , Q is
discharge, and a, b, c, f, k, and m a r e best fit constants. It follows
that since width, depth, and mean velocity are each functions of
discharge, then b + f + m = 1.0; and ack = 1.0. Betson (1979) noted
that a f o u r t h r e l a t i o n s h i p also can be presented
(4.41d)
A = nQp
where A is the cross-sectional area of flow. Betson also noted that
f = p - b and m = 1 - p. The relationship in Eq. 4.41 are for indi-
vidual stations in that they relate channel measures to concurrent
discharge.

The results from several studies are shown in Table 4.1. It is


notable that the values do not vary widely, particularly for the depth
discharge relationship. These results reinforce the use of single valued
r a t i n g c u r v e s a n d simp1 i f i e d r o u t i n g models.

NON-K I NEMAT I C WAVES

The result in Eq. 4.40 frequently does not agree with nature.

Rather, due to previously mentioned factors, flow peaks are seen to


subside which suggests the a p p l i c a t i o n of the k i n e m a t i c model is limited,
and that either the diffusion or dynamic wave model is preferred. It
is important then to examine the non-kinematic wave models and to
establ i s h how they d i f f e r from the k i n e m a t i c model.

Differences between the two non-kinematic models can be invest-


igated by examining the significance of each of the dynamic terms in
the momentum equation. The d i s c h a r g e a t a p o i n t i n a stream i s

Q = vA (4.42
The momentum e q u a t i o n can be r e w r i t t e n as follows:

-
Q aQ Q2 aA + -
- 1 -
aQ - -
Q -
aA + ay = g(s -5 ) - -
qi (4.43
A2 ax A3 ax A at A2 at ax o f A

The p a r t i a l derivative of A with respect to time is removed in terms of


the spatial derivative of Q using the continuity expression. After this
substitution and r e a r r a n g i n g , Eq. 4.43 becomes

28 aQ Q2 aA + -1 a B + a v =
s -sf (4.44)
ax ax gA at ax
gA2 gA3
70

TABLE 4.1 Typical Station Exponent Terms f o r Geomorphic E q u a t i o n s

Exponents

width depth velocity area Reference


LOCAT I ON b f m P
OF BASIN

Midwest 0.26 0.40 0.34 0.66 Leopold, et al.


(1954)
B r a n d y w i n e , P.A 0.04 0.41 0.55 0.45 ditto
158 Stations i n U.S. 0.12 0.45 0.43 0.57 ditto
B i g Sandy R i v e r , KY 0.23 0.41 0.36 0.64 S t a l l a n d Yang
(1976)
Cumberland P l a t e a u , K Y 0.245 0.487 0.268 0,732 Betson (1979)
Johnson C i t y , TN 0.08 0.43 0.49 0.51 Weeter a n d
Meadows (1 979)
T heore t ica I 0.23 0.42 0.35 0.67 Leopold a n d
L a n g b e i n (1962)

At any cross-section Eq. 4.36 holds; a n d f o r most n a t u r a l channels, the


wave speed (celerity) is approximated by the kinematic wave speed.
If Chezy's r e s i s t a n c e e q u a l ion i s assumed

c = -38 (4.45)
2A
D r a w i n g on these two relationships and the definition for Froude number

Fr2 2 = __
g28
(4.46)
gy gA3
the v a r i o u s terms in Eq. 4.44 c a n be r e w r i t t e n as

and

(4.47c)

Tracing back, the contribution of each term in the momentum equation


i s f o u n d (Meadows, 1981).

and
71

_1 -av = (-0.75 Fr
2 aY
- (4.48b)
g at ax
w h i c h allows the momentum equation to be w r i t t e n as
(1 - 0.25 Fr2) a
ax
=
'0 - 'f
(4.49)

An e q u i v a l e n t expression was found b y Dooge (1973).


Examination of Equations 4.48 and 4.49 r e v e a l s that the convective
and temporal acceleration terms essentially are of equal magnitude but
opposite sign, and hence, act to nearly cancel each other. These two
terms are s i g n i f i c a n t for Froude numbers g r e a t e r t h a n 0.60, where s i g n i -
ficance is taken as 10 percent of the coefficient value i n Equation 4.49.
Evidence of Froude numbers less than 0.60 for unsteady events i n s m a l l
streams is documented in the literature, e.g. (Gburek and Overton,
1973). Further, using the theoretical values for hydraulic elements of
Leopold a n d L a n g b e i n (1962), i t was shown b y Meadows (1981) that
Fr oi

demonstrat i n g that Froude number is largely insensitive to increasing


discharge i n most natural streams for flow i n bank. These r e s u l t s suggest
the d i f f u s i o n wave model can be c o n f i d e n t l y applied to most f l o o d r o u t i n g
events.

Wave Speed

Based on the method of characteristics, i t was shown that dynamic


waves propagate both downstream (forward characteristic) and upstream
(backward characteristic). The diffusion wave speed is given by the
kinematic wave speed. As such, the diffusion wave model has only a
forward characteristic meaning that wave forms are propagated only
downstream and that b a c k w a t e r effects are negligible. It is left to the
r e a d e r to confi rm t h i s .

C r e s t Subsidence

Both the dynamic and diffusive wave models simulate a dispersing


hydrograph, hence, a subsiding wave crest. To i I lustrate, consider the

modified diffusive wave equation, Eq. 4.49. For the following develop-
ment, a rectangular cross section is assumed. As with the derivation
of most overland and open channel flow equations, this assumption
greatly simp1 i f i e s the mathematics, yet does not alter appreciably the
f i n a l form of the equations b e i n g developed.
72

Approximating the friction slope with Chezy's equation, Eq. 4.49


becomes

(1 - 0.25 Fr2) a
ax
= S
o
- -
Q2
c2A2R
(4.50)

T a k i n g t h e p a r t i a l d e r i v a t i v e w i t h respect to time

( 1 - 0.25 F 2~ a) (--)
- ay =
Q2_ -1 2
- _ aQ 2 aA 1 aR (4.51)
at ax c2A2R Q a t A a t R at J

From c o n t i n u i t y

(4.52)

or

(4.53)

Generally, over a reach, aq/ax = 0. Thus,

(4.54)

F o r a p r i s m a t i c section
dA =
_ B
dy
such t h a t
-
a A * B d y g B a
ax dA ax (4.55)

which, when s u b s t i t u t e d i n t o Eq. 4.54 yields

(4.56)

I n o b t a i n i n g E q . 4.56, the assumption was made t h a t aB/at = 0;

which is satisfactory if the channel is rectangular or the flood wave


rises slowly. The momentum e q u a t i o n c a n now be w r i t t e n

(4.57)

For a wide rectangular channel (w > lOy), the hydraulic radius, R,


i s approximately equal to the depth of f l o w , y. Using this approximation
a n d c o n t i n u i t y f o r a r e c t a n g u l a r geometry

(4.58a)

(4.58b)

the r i g h t h a n d s i d e of Eq. 4.57 i s rewritten as


73

Combining similar terms and recognizing that the coefficient terms are
merely S
0’
s [ 2- -aQ- - 3 qij
o a a t A a x A
The w h o l e e q u a t i o n t h u s becomes

(4.59)

Multiplying by Q/2

(4.60)

For Chezy’s equation

c = -3Q (4.35)
2A

Making this substitution i n t o Eq. 4.60

(4.61)

which is a convective-diffusive equation for unsteady streamflow. This


equation illustrates the origin of the diffusive wave label. The presence
of the dispersion term (second partial derivative) confirms that the
d i f f u s i v e w a v e model s i m u l a t e s a s u b s i d i n g p e a k .
One very interesting property of the crest region of a diffusive
wave can be derived by rewriting Eq. 4.50 in terms of Q as follows

Q = C y B ,/y[So-(l - 0.25 F r 2;2] (4.62)


ax
where h y d r a u l i c radius has been approximated by y. Taking the d e r i v a -

t i v e w i t h respect to x and equating to z e r o y i e l d s

(4.63)

In the region of the crest, the shape of the hydrograph is concave


downward, and a 2y/ax2 0, and therefore, by Eq. 4.63, ay/ax < 0,
also. That is, the peak flowrate does not occur where depth is a

maximum, but a t a point in a d v a n c e o f the maximum depth.


74

Looped Rating Curves

Eq. 4.62 clearly demonstrates that a single valued rating between


discharge and depth (area) does not hold for non-kinematic waves. An
a p p r o x i m a t e expression for the variable (looped) rating curve is given

by

-Q _ - 4- ( 1 - 0.25 Fr')
ax
(4.64)
'n
where Q is the uniform flow at a given depth. This expression is
rendered more useful if the spatial derivative is replaced by some

alternate quantity, d e d u c t i b l e from i n - s i t u conditions.


Using the kinemat i c r e l a t i o n s h i p

ay - 1 -
ay (4.65)
ax c a t
Eq. 4.64 can be w r i t t e n as

2
( 1 - 0.25 Fr )
(4.66)
at

I t must be noted that Eq. 4.66 i s not strictly correct since the k i n e m a t i c
r e l a t i o n s h i p was included.
A typical looped rating curve is shown in the Figure 4.5. Com-
parison with the associated discharge hydrograph illustrates that as
a flood hydrograph passes a point, the maximum discharge is first
observed, then the maximum depth, and finally a point where the flow
is uniform. The u n i f o r m 'flow occurs when the flood wave i s essentially
horizontal and therefore has a slope, dy/dx, that i s very small relative
to the bed slope. T h i s obviously will occur close to the r e g i o n of max-
imum depth. The occurrence of uniform flow is illustrated graphically
as the point of i n t e r s e c t i o n of the looped rating curve with the single
v a l u e d uniform flow r a t i n g curve.
It should be noted that the scale i s exaggerated for clarity. The
three points in question are more likely to occur much closer together
than i n d i c a t e d b y the f i g u r e .
The usefulness of the looped rating curve compared with a single
valued rating curve is determined by how wide the loop is relative to

the s i n g l e v a l u e d curve. It should be noted however, that most p u b l i s h e d

streamflow data and associated rating curves determined from field


discharge measurements generally are better approximated by a single

valued relationship. Looped curves can be approximated using Eq. 4.64


o r 4.66 and time series records of r i v e r stage a t a s t a t i o n .
UNIFORM FLOW
RATING W R V E d
Q

TIME

F i g . 4.5 Loop stage-discharge r a t i n g c u r v e a n d associated d i s c h a r g e


h y d r o g r a p h f o r a t t e n u a t i n g wave.
76

MUSK I NGUM R I VER ROUT I NG

Flood r o u t i n g r e f e r s to a set of models used to p r e d i c t the temporal


and spatial v a r i a t i o n s of a f l o o d wave (runoff hydrograph) as i t travels
through a channel reach. Routing techniques are classed into two
categories: h y d r a u l ic and hydrologic. The kinemat i c , diffusion and
dynamic wave models a r e h y d r a u l i c r o u t i n g models. The h y d r o l o g i c models
are based on continuity and an empirically derived relationship between
channel storage and discharge; therefore, they are not as rigorous as

the hydraulic models and represent a further simplification to the full


equations for open channel flow.
Perhaps the best known and most widely used of the hydrologic
models is the Muskingum routing model. This model was developed
originally for flood routing on the Muskingum River in Central Ohio,
hence the o r i g i n of the name. T h e model u t i l i z e s c o n t i n u i t y
dS
l + Q L - 0 = - (4.67)
dt
where I is inflow to a river reach, QL i s lateral inflow (=qAx), 0 is
outflow and S is the storage w i t h i n the reach; and the storage r e l a t i o n -
ship
S = K [Zl + (l-Z)O] (4.68)
where K i s a characteristic storage time approximated as the travel time
t h r o u g h a reach, and 2 i s a w e i g h t i n g coefficient.
For a t t e n u a t i n g waves, z<o-5.
Equations 4.67 and' 4.68 are solved using a finite differencing
technique. Defining I 1 = l ( t ) and I = I ( t + A t ) , and s i m i l a r l y , 0 1 , 02, a n d
2
S1 a n d S2, the f o l l o w i n g a p p r o x i m a t i o n to Eq. 4.67 i s written

11+12+&--=-
01+02 s2-s1
(4.69)
2 2 At

where QL i s the a v e r a g e l a t e r a l inflow during the time interval At. The


inflow hydrograph provides I, and 12, and O2 is the desired quantity.
O1 is known from either initial conditions or a previous calculation.
S and S a r e expressed i n terms of I a n d 0 as follows
1 2
S - S = K[Z(12-11) + (1 - Z) (02-01)] (4.70)
2 1
S u b s t i t u t i n g Eq. 4.70 i n t o Eq. 4.69 and s i m p l i f y i n g gives
o2 = C0l2 + C,I, + c20, + c3QL (4.71)

where

-KZ + 0.5At (4.72a)


'0 - K-KZ + 0.5At
77

KZ + 0.5At
(4.72b)
c1 = K-KZ t 0.5At
K-KZ - 0.5At
'2 = K-KZ + 0 . 5 ~ 1
(4.72~)

and
At
(4.72d)
'3 - K-KZ + 0 . 5 ~ t

K and t must h a v e the same time unit, and the first three coefficients
sum to 1.0.

E s t i m a t i o n of Model P a r a m e t e r s

The Muskingum model is quite sensitive to the selection of model


parameters. Historically, K a n d Z h a v e b e e n e s t i m a t e d b y m a t c h i n g model
output with actual inflow-outflow records. The obvious shortcoming is
that t h e model i s limited to g a u g e d streams. Oftentimes, we n e e d t o r o u t e
flood hydrographs along ungauged streams. To do so requires that we
have a means of estimating model p a r a m e t e r s from a v a i l a b l e c h a n n e l a n d
hydrograph characteristics.
Following the technique of Cunge (1969) and using a Taylor series
expansion to each of the terms in Eq. 4.67, it is transformed to an
e q u i v a l e n t e q u a t i o n of the c o n v e c t i v e - d i f f u s i v e form
- -
'a + ?A ?-!? = bx(l-Z)c(Q)- I -
- hXL ]- a % QL
at K ax ' 2 K 2 + T (4.73)
ax
where A x i s the reach length. Comparison o f t h i s equation with Eq. 4.68
shows t h a t

= $Q) (4.74a)

and
2 (4.74b)
= - Q(1-0.25Fr )]
BSoA x c ( Q)

C u n g e (1969) and later r e s e a r c h e r s developed s i m i l a r expressions to Eqs.


4.74. Ponce and Yevjevich (1978) considered the variation of K and
Z with Q; w h i l e Dooge (1973) i n c l u d e d the c o r r e c t i o n f o r d y n a m i c effects,
2
(I-0.25Fr 1, in the equation for 2, but considered c(Q) to be constant
and not a function of 8. Therefore, Eqs. 4.74 are the most general
expressions for K a n d Z (Meadows, 1981).
Another very important feature of Eq. 4.73 is that i t demonstrates
the Muskingum r o u t i n g model i s d i f f u s i v e for Z < O . 5 , a n d o f f e r s t h e same
advantages of the diffusion wave model. If, however, Z=O.5, the
78

Muskingum model predicts pure translation, and is equivalent to the


kinematic wave model. Typical values for Z for natural streams a r e 0.3
to 0.4, a n d for p r i s m a t i c channels, 0.4 to 0.5.

K I NEMATIC AND D I FFUS ION MODELS

We have discussed the kinematic and diffusion wave models as


approximations to the dynamic wave model and have shown they are
applicable for certain flood wave and channel c o n d i t i o n s . . As users, we
need c r i t e r i a or guidelines for selecting which model to use. Two n o t a b l e
works toward establishing such guidelines are Henderson (1963) and
Ponce, et a l . (1978).
Henderson conducted a theoretical examination of the governing
equations similar to t h a t presented in the p r e v i o u s sections. He compared
theoretical results with a limited number of f l o o d h y d r o g r a p h s , and noted
that subsidence is most pronounced in the vicinity of the wave crest.
Generally, he may be credited with efforts to classify flood waves
according to the magnitude of So into waves broadly characteristic of
steep, mi I d and intermediate slopes. However, he cautioned that this
classification is not exhaustive, but should suffice for most floods in
natural waterways. He did not offer specific guidelines to define mild,
intermediate and steep, although he concluded the kinematic model is
applicable in steep channels; the diffusion in mild and steep channels;
and the dynamic to all three. The r a t i o n a l e for this conclusion i s that
he considered Fr2 >> 1 in steeply sloped channels; hence, according to
Eq. 4.49, the momentum equation will become S = Sf. For mild slopes,
Fr2 < < I , and the momentum equation becomes :he strict diffusion wave
model. For i n t e r m e d i a t e slopes all terms in the momentum equations are
si g n i f i cant.
Ponce, et al., applied linear stability analysis in an effort to
propose a theory that accounts for wave celerity as well as a t t e n u a t i o n
characteristics. To do so required they use a linearized, therefore
somewhat simp1 i f i e d , version of the governing equations. Assuming a
sinusoidal wave, they compared the p r o p a g a t ion characteristics of the
kinematic, diffusion and dynamic wave models. As expected, the dynamic
wave model is applicable to the entire spectrum of waves that can be
routed with a one-dimensional model. For F r < 2 , the c e l e r i t y o f a dynamic
wave i s greater than the kinematic wave celerity. For Fr=2, roll waves
will form. Thus, for primary waves (main body of a flood wave), Fr=2
is the threshold dividing attenuation and amp1 i f i c a t i o n . For secondary
waves, Fr=l is the threshold dividing the propagation upstream or
79

downstream; for Fr=l they remain stationary or propagate downstream


only; and for F r <1, they propagate only downstream. A physical
observation by Stoker (1957) e x p l a i n s t h i s conclusion r e g a r d i n g secondary
waves:
"What seems to happen is the following: small forerunners of a
disturbance (wave) travel with the speed fi relative to the flowing
stream, but the resistive forces act in such a way as to decrease the
speed of the main portion of the d i s t u r b a n c e f a r below the values given
by a.. .I'

Ponce, et al., did offer first generation criteria for application


of the k i n e m a t i c a n d d i f f u s i o n wave models:

Kinematic: TBSo > 171

Diffusion : TBSo(e) '30

where TB is the duration of the flood wave, So is the channel slope,


v and y are the initial velocity a n d depth of flow, respectively, and
P
g is gravity. Based on these criteria, the kinematic model applies to
shallow flow on steep slopes (hence the steep channel of Henderson and
surface runoff from hillslopes) and to long duration flood waves (slow
r i s i n g floods on m a j o r r i v e r s as o b s e r v e d b y Seddon). The d i f f u s i o n model
is applicable to these as well as a wider range. When these two models
b r e a k down, t h e d y n a m i c model a p p l i e s .
These criteria are si,gnificant in that they r e l a t e model a p p l i c a t i o n
to channel slope a n d h y d r o g r a p h c h a r a c t e r i s t i c s . The r e a d e r i s cautioned
that these a r e o n l y f i r s t g e n e r a t i o n f o r m u l a e .

REFERENCES

Betson, R.P., 1979. A geornorphic model f o r use i n s t r e a m f l o w r o u t i n g ,


Water Resources Research, V o l . 15, No. 1 , p p . 95-101.
Brutsaert, W. 1968. The initial p h a s e of the rising h y d r o g r a p h of
turbulent free surface flow with unsteady lateral inflow. Water
Resources Research, V o l . 4, p p 1189-1 192.
Cunge, J.A., 1969. O n the s u b j e c t of a f l o o d p r o p a g a t i o n c o m p u t a t i o n
method ( M u s k i n g u m M e t h o d ) . J. H y d r . Res., V o l . 7, No. 2, p p . 205-230.
Dooge, J . C . I . 1973. L i n e a r t h e o r y of h y d r o l o g i c systems. U.S. Dept. of
A g r i c u l t u r e , A g r i . Res. Ser. Tech. B u l l . No. 1968.
Gburek, W.J. a n d Overton, D.E., 1973. S u b c r i t i c a l k i n e m a t i c f l o w in a
s t a b l e stream. J. H y d r . D i v . ASCE. V o l . 99, No. HY9, p p . 1433-1447.
Henderson, F.M., 1963. F l o o d waves in p r i s m a t i c c h a n n e l s . J. H y d r . D i v .
ASCE, V o l . 89, No. H Y 4 , p p . 39-67.
Henderson, F.M. a n d Wooding, R.A., 1964. O v e r l a n d f l o w a n d g r o u n d w a t e r
from a steady r a i n f a l l of f i n i t e duration. Journal of Geophysical
Research, V o l . 69, No. 8, p p . 1531-1540.
80

Leopold, L.B., et al., 1954. F l u v i a l P r o c e s s e s in G e o m o r p h o l o g y . N.H.


Freeman, San Fra ncisco, Cal.
Leopold, L.B. a n d Langbein, W.B., 1962. The concept o f e n t r o p y i n
l a n d s c a p e e v o l u t i o n . U.S. G e o l o g i c a l S u r v e y P r o f . P a p e r 500-A.
Leopold, L.B. a n d Maddock, T., Jr. 1953. T h e h y d r a u l i c g e o m e t r y o f
s t r e a m c h a n n e l s a n d some p h y s i o g r a p h i c i m p l i c a t i o n s . U.S. Geological
S u r v e y P r o f . P a p e r 252.
L i g h t h i l l , M.J. and W h i t h a m , G.B., 1955. On k i n e m a t i c w a v e s : I. F l o o d
movement in l o n g r i v e r s . P r o c . R o y a l S o c i e t y , L o n d o n , Vol. 229, No.
1178, pp. 281 -
Meadows, M.E., 1981. Modelling the impact of stormwater runoff, in
Proceedings, I n t e r n a t i o n a l Symposium on U r b a n H y d r o l o g y , H y d r a u l i c s
a n d Sediment C o n t r o l , U n i v e r s i t y o f K e n t u c k y , L e x i n g t o n , K e n t u c k y , pp.
31 3-31 9.
M o r r i s , E.M., 1979. T h e e f f e c t o f t h e s m a l l s l o p e a p p r o x i m a t i o n a n d l o w e r
b o u n d a r y c o n d i t i o n s on s o l u t i o n s of the S a i n t Venant Equation. Journal
of H y d r o l o g y , V o l . 4 0 , pp. 31-47.
Morris, E.M. a n d Woolhiser, D.A., 1980. U n s t e a d y o n e - d i m e n s i o n a l flow
over a plane: p a r t i a l e q u i l i b r i u m a n d r e c e s s i o n h y d r o g r a p h s . Water
R e s o u r c e s R e s e a r c h , 16 ( 2 ) , pp 355-366.
O v e r t o n , D.E. a n d Meadows, M.E., 1976. S t o r m w a t e r M o d e l l i n g . A c a d e m i c
P r e s s , N.Y.
Ponce, V.M., L i , R.M. a n d Simons, D.B., 1978. A p p l i c a b i l i t y o f k i n e m a t i c
a n d d i f f u s i o n w a v e m o d e l s . J. H y d r . D i v . ASCE, V o l . 104, No. HY3, pp.
353-360.
Ponce, V.M. a n d Yevjevich, V. 1978. Muskingum-Cunge Method w i t h
v a r i a b l e p a r a m e t e r s . J. H y d r . D i v . ASCE, Vol. 104, No. HY12, pp. 1663-
1667.
Seddon, J.A., 1900. R i v e r h y d r a u l i c s . T r a n s : ASCE, V o l . 43, p. 179.
S t a l I , J.B. a n d Y a n g , C.T., 1970. H y d r a u l i c g e o m e t r y o f I I I i n o i s s t r e a m s .
R e s e a r c h R e p o r t No. 15, W a t e r R e s o u r c e s R e s e a r c h C e n t e r , U n i v e r s i t y o f
II I inois, U r b a n a , I I I inois.
V i e i r a , J.H.D., 1983. C o n d i t i o n s g o v e r n i n g t h e u s e o f a p p r o x i m a t i o n s f o r
the S a i n t Venant e q u a t i o n s f o r s h a l l o w s u r f a c e w a t e r flow. J o u r n a l of
H y d r o l o g y , Vol. 60, pp. 43-58.
Weeter, D.W. a n d Meadow's, M.E., 1978. W a t e r Q u a l i t y M o d e l i n g f o r R u r a l
Streams. F i r s t Tennessee - V i r g i n i a D e v e l o p m e n t D i s t r i c t , J o h n s o n C i t y ,
104 p .
Wooding, R.A., 1965. A h y d r a u l i c model f o r t h e c a t c h m e n t s t e a m p r o b l e m ,
I . K i n e m a t i c w a v e t h e o r y . J o u r n a l o f H y d r o l o g y , V o l . 3, pp 254-267.
W o o l h i s e r , D.A. a n d L i g g e t t , J.A. 1967. U n s t e a d y o n e d i m e n s i o n a l f l o w o v e r
a p l a n e - t h e r i s i n g h y d r o g r a p h . W a t e r R e s o u r c e s R e s e a r c h , Vol. 3, NO.
3, pp. 753-771.
81

CHAPTER 5

NUMER I CAL SOLUT IONS

METHODS O F SOLUTION OF EQUATIONS OF MOTION

There are no known general analytical solutions to the hydraulic


equations

(5.1

They must therefore be solved using the method of characteristics or


numerical integration techniques. Available numerical methods include
f i n i t e d i f f e r e n c i n g a n d f i n i t e elements.
F i n i t e differencing techniques a r e founded on the c l a s s i c a l definition
for a continuous derivative term. Use of these methods transforms the
set of partial d i f f e r e n t i a l equations i n t o an equal number of approximate
algebraic equations which then are solved according to the rules of
I inear algebra.
The finite element method is a relatively recent approach to s o l v i n g
partial differential equations that govern hydraulic processes. The basis
of finite element integration is approximating polynomials. In essence,
the polynomial coefficients are adjusted to minimize an error term while
satisfying known b o u n d a r y . conditions. The resulting p o l y n o m i a l s express

the unknown variables in terms of the known (independent) variables.


The detai I s of t h i s method a r e beyond the scope of these notes; however,
application of this method to kinematic overland flow is illustrated in
a later chapter.
Given the many and varied ways of integrating the flood routing
equations, one can logically ask which technique to choose. Some csn
be discarded as being inaccurate or unstable or too time consuming;
others seem to reproduce solutions relatively well. However, there i s no
single answer to which method is "best". Indeed, the answer to that
question depends on the particular application, and perhaps on the

a v a i l a b l e comput i n g equipment.

METHOD O F CHARACTER I ST I C S

The method of characteristics may be described as a technique


whereby the problem of solving two simul taneous partial differential
82

equations (continuity and momentum) can be replaced by the problem

of solving four ordinary differential equations. This method has been


known for many years; it was devised long before the computer as a
means for graphically integrating the unsteady streamflow equations.
The characteristic equations are no longer solved graphically, but are
solved n u m e r i c a l l y u s i n g the computer.

By m a k i n g the s u b s t i t u t i o n
2
c = SY (5.3)
into Eqs. 5.1 and 5.2 and then by writing first the sum, and then t e
difference, of the two new equations, we o b t a i n the two equations

(5.4a

qi
(,-,)a0
ax + a0
at = g(so - S f ) - ( v+c) -
A (5.4b

which are two equations in the form of directional derivatives of


v +- 2c. Recalling the definition of a total derivative, i t can be shown
that for

-z _; - v + c (5.5)

then
dx
- -a(vk2.c)
+- a(V'2.c) = d (v*2c)
(5.6)
ax
~

dt at dt
which gives the desired set of ordinary differential equations to r e p l a c e
the partial differential equations. The characteristic roots (directions)
are given by Eqs. 5.5, and along each direction the respective. total
derivatives i n Eqs. 5.6 hold. The r e s u l t i n g equations can be r e w r i t t e n :

c + . dx
_ - v + c (5.5a)
' dt

-d (v+2c)
- - qi
dt
g(S0 - Sf) - ( v - c ) -
A
(5.6a)

c- . d_dxt = v - c (5.5b)
'

d ( v_- 2_
_ c) - 'i
g(S0 - Sf) - ( v + c ) -
A
(5.6b)
dt
where c
+ and c- symbolically designate forward and backward character-
i s t i c respective1y.

Physically, the characteristic roots represent the path in time

and space followed by a disturbance, e.g. flood wave. The speed of


propagation is given by the slope dx/dt; and the state of the system
(values of the dependent variables) is given by the total derivatives
t h a t h o l d a l o n g the c h a r a c t e r i s t i c p a t h s .
Mathematically, the c h a r a c t e r i s t i c s a r e loci of p o s s i b l e d i s c o n t i n u i t i e s

I n the temporal and spatial derivatives of the dependent variables. Thus,

one may think of a characteristic c u r v e as a l i n e of s e p a r a t i o n between


83

two regions of somewhat different physical conditions. This is important


when modelling unsteady flows where the boundary conditions vary with
time, since the solution at interior points of the (x,t) domain are
dependent on the boundary information. The solution at those points
above a characteristic curve requires more boundary information than
the solution below the characteristic. In fact, the characteristics in
Eqs. 5.5 define f o u r u n i q u e solution zones as shown i n F i g u r e 5.1.

AI

UPSTREAMI t DOWNSTREAM
BOUNDARY BOUNDARY

-x
x=0 X=:L
F i g . 5.1 Zones of solution domain defined b y c h a r a c t e r i s t i c s .
(Woolhiser a n d L i g g e t t , Water Resources Research,
3, 755, 1967, American Geophysical U n i o n ) .

These zones are formed by the intersection of the forward and backward
characteristics ernanat ing from the upstream (x=O) a n d downstream (x=L)
end of the channel reach at the initial time. The solution in Zone A
requires only the initial values (the beginning state of the system at
all x); w h i l e the solution i n Zones B a n d C r e q u i r e s both initial values
and a boundary condition. This is because these zones lie above the
backward and forward characteristics, respectively. Zone B requires the
downstream boundary condition, a n d Zone C , the upstream. Finally, Zone
D, which lies above both characteristics requires the initial values and
both b o u n d a r y conditions.

Numerical I n t e g r a t i o n of C h a r a c t e r i s t i c Equations

The objective of the method of characteristics is to fill the (x,t)


plane with characteristics as shown in Figure 5.2. The unknowns are

determined at the intersections where the f o u r equations 5.5a, 5.5b, 5.6a


84

a n d 5.6b are satisfied. The extent to which solutions can be obtained


over the (x,t) plane i s dependent on the amount of initial value, (x,O),
and boundary condition, (0,t) and ( L , t ) , information that is specified
beforehand. Initial conditions are the velocity and depth of flow at all
x at the beginning of the simulation, u s u a l l y designated time zero. The
upstream boundary condition typical ly is the known inflow hydrograph
that is to be routed downstream. Values for v and y are obtained with
known rating relationships. Usual l y the outflow hydrograph at the down-
stream end (boundary) is the desired result; hence, Q, v and y at the
downstream boundary are unknown. However, if rating relationships are

known, they can be used as the downstream b o u n d a r y c o n d i t i o n .


With the boundary information specified, the solution for v and y
at a sufficient number of intermediate points and at the downstream
boundary is obtained at the intersection of forward and backward
characteristics and at the intersection of forward characteristics and

the downstream boundary, respectively. The number of solution points


must be sufficient to adequately describe the movement of a flood wave
downstream and is determined by the number of c h a r a c t e r i s t i c s inscribed
on the (x,t) plane. Usually, the availability of data limits the number
of characteristics; however, it should be noted that a better solution
generally is obtained when more characteristic curves are involved.
Mathematically, a complete solution is obtained if all the boundary
informat ion i s u t i I ized.
The usual procedure for solving Eqs. 5.5 and 5.6 simultaneously
is shown in F i g u r e 5.2. Consider the points numbered 2, 5, 6, 10, 11
a n d 12. There i s both a f o r w a r d a n d a b a c k w a r d c h a r a c t e r i s t i c emanating
from points 2, 5 and 10. The intersection of the forward characteristic
out of point 10 w i t h the b a c k w a r d c h a r a c t e r i s t i c out of p o i n t 5 specifies
the conditions (values of v and y) at point 11. Similarly, point 6
conditions are determined at the i n t e r s e c t i o n of the respective character-
istics out of points 5 and 2. The forward characteristic out of point 10
is continued downstream in time and space until it intersects with the
backward characteristic out of point 2, thereby determining v and y
at point 12. The procedure continues for a1 I forward and backward

characteristics unt i I they intersect either the downstream or upstream


boundary.
At each intersection there a r e f o u r unknowns x , t, v, and y. These
are uniquely determined by the simultaneous solution of the four
equations given by Eqs. 5.5 and 5.6. At the downstream boundary x
is given and is no longer an unknown. The other three unknowns are
determined b y the simultaneous solution of Eqs. 5.5a a n d 5.6a and r a t i n g
85

F i g . 5.2 Characteristics i n ( x , t ) Plane

curves that r e l a t e v and y.


A method for solving the characteristic equations according to this
procedure is outlined as follows. With reference to F i g u r e 5.3, it is
assumed the values for v and y a r e known a t L and R and a r e desired
a t M. Eqs. 5.5 a r e approximated as

a n d XM = XR + ( t M - tR) (V - c ) ~

These two equations can be e a s i l y solved for the two unknowns XM and
tM. Once these a r e known Eqs. 5.6 a r e solved b y the same approach.

where A = g(So-Sf)L - (v-c), -


qi (5.11)
AL
86

F i g . 5.3 C h a r a c t e r i s t i c Solution f o r Point M

The method outlined by Eqs. 5.7 t h r o u g h 5.13 is linear and can

be solved for vM, cM hence yM. The boundary condition, inflow, initial
values and downstream rating curve must be known. A more stable and
accurate solution can be obtained with a nonlinear formulation (Overton
a n d Meadows, 1976; a n d Mahmood a n d Yevjevich, 1975).
When solving problems using the method of characteristics, check
whether the, flow is subcritical or supercritical. When the flow is sub-
critical, v<c and the forward characteristic has a positive slope dx/dt
in the (x,t) plane while the backward characteristic has a negative
slope, as shown in F i g u r e 5.4a. When, however, the flow is supercritical,
v>c and both characteristics have a positive slope in the (x,t) plane,
F i g u r e 5.4b.

++

F i g . 5.4 C h a r a c t e r i s t i c L i n e s f o r S u b c r i t i c a l a n d S u p e r c r i t i c a l Flows

F I N I T E DIFFERENCE METHODS L

Finite differencing involves replacing the continuous derivative


terms with approximate finite difference quotients, thereby transforming
the set of differential equations into a set of either linear or non-
linear algebraic equations which can be solved more readily. These
87

algebraic equations relate unknown dependent variable values at nodal

points on a finite grid overlaying the continuous solution domain to


known initial v a l u e s a n d b o u n d a r y conditions. Solution of these equations
i s either direct or through a root determining scheme such as the Newton-
Raphson Method. In e i t h e r case, depending on the manner in which the
replacements a r e made, m a t r i x techniques may also be r e q u i r e d .

Difference Quotients

Finite difference quotients are obtained by dividing the difference


between two values of a function by the corresponding two values of
the independent variable. For the case of a function of a single
variable, e.g. f(x), the d i f f e r e n c e quotient i s given b y
f(x+Ax) - f ( x )
A X

The l i m i t i n g v a l u e , as Ax->O, i s the d e f i n i t i o n of the d e r i v a t i v e

df(x)
~- - lim f(x+Ax) - f ( x )
dx Ax (5.14)
a x->o
Thus the finite difference quotient i s an approximation to the continuous
d e r i v a t i v e as l o n g as A x i s kept small.
Several difference quotients can be defined to approximate partial
derivatives. To illustrate some of them, consider a function of two
independent variables, say U(x,t). With reference to the f i n i t e difference
grid in Figure 5.5, the most commonly used difference quotients are
defined as follows. The forward difference approximation to the first

p a r t i a l d e r i v a t i v e f o r U w i t h respect to x is

-
2U = U(x+Ax,t) - U(x,t)
(5.15)
ax Ax
Physically, one can t h i n < of an observer standing at the point (x,t),
looking ahead ( f o r w a r d ) to the p o i n t (x+ Ax,t), and using the elevation
(function value) d i f f e r e n c e between the two p o i n t s d i v i d e d b y the d i s t a n c e
to evaluate the slope ( v a l u e of the d e r i v a t i v e ) . The b a c k w a r d difference
approximation i s

au -
- - u(x,t) - u(x-Ax,~)
ax A X
(5.16)

The centered ( o r c e n t r a l ) difference approximation i s

au
- = U(x+Ax,t) - U(x-Ax,t)
(5.17)
ax 2 ax
88

Fig. 5.5 F i n i t e Difference G r i d f o r x , t Solution Domain

NUMER I C A L S O L U T I O N

There are two basic finite difference schemes used in solving the
streamflow routing equations. They a r e the e x p l i c i t and imp1 i c i t schemes.
Explicit schemes utilize initial value and left hand side (upstream)
boundary information and solve for the remaining grid p o i n t s one a t a
time. They are subject to stability limitations on the allowable grid
interval size which means explicit schemes typically have large data
requirements. However, explicit methods often result in linear algebraic
equations from which the unknowns can be evaluated directly without
iterative computations. Implicit schemes utilize initial value and both
left and right hand side boundary information, and solve for the
unknown grid points at the next time level simultaneously. Therefore,
imp1 i c i t schemes often require matrix techniques. Implicit methods
typically involve nonl inear algebraic finite difference equations whereby
the s o l u t i o n i s attained by iteration. Both schemes can be and h a v e been
used in solving the governing equations for overland and open channel
flow.
89

Most existing methods for numerical solution of equations can be

classified into the fol lowing groups:

(a) E x p l i c i t f i n i t e difference methods


(b) I m p l i c i t f i n i t e d i f f e r e n c e methods
(c) F i n i t e element methods

The use of the first two methods was summarised by Liggett and
Woolhiser (1967). They reviewed different explicit finite difference

schemes. T h e schemes w e r e :
a) method of c h a r a c t e r i s t i c s

b) u n s t a b l e method
c) diffusion method

d) Lax-Wendroff method
e) leap-frog method

The method of characteristics uses an irregular grid following the


characteristic curves while the others use a rectangular grid for the

s o l u t i o n of the equations.
The method of characteristics employs the fact that flow conforms
to certain relationships along characteristic curves and therefore the

solution is performed along the characteristic curves. The main


advantages of the characteristic method is that it i s accurate and fast.

It is the most acurate method for the same initial point spacing of all
methods. Its accuracy is a consequence of following the characteristic
curves which describe the path of the disturbances in the flow. It also
covers the x - t p l a n e f a s t e r t h a n a n y o t h e r m e t h o d w i t h t h e same i n i t i a l
point spacing. The main disadvantage of the method of characteristic

is that data at intermediate points in the x - t plane is difficult to

obtain in an acceptable form, requiring tedious interpolation techniques.


If the method is applied to a two-dimensional problem the use of the

characteristic method becomes even more difficult. More recently more


elaborate methods of characteristics were developed. Abbott and Verwey
(1970) used a four-point method of characteristics, i.e. utilising three

different points in fixing the properties of a fourth point. This solution

could only be used with the dynamic equations a s the k i n e m a t i c equations


do not h a v e n e g a t i v e c h a r a c t e r i s t i c s required for t h i s method.

The implicit method of solution involves simultaneous solution of

all the flow properties by solving a matrix; i t s main advantage is that


the ratio of space to time interval, Ax/A t, is not governed by any
stability criteria and the method is considered to be stable for any

choice of Ax and At. Most previous investigators considered this to be


90

an advantage. Liggett and Woolhiser (1967) report, however, that they


were u n a b l e to make p r a c t i c a l use of this 'advantage'. If they increased
Ax/At ratio more than would be allowed for in an explicit finite dif-
ference scheme, inaccuracy resulted and sometimes stabi I i t y problems
occurred. They suggest that the implicit methods seemed to be more
advantageous when dealing with river problems but pointed out that
a t t e n t i o n should be p a i d to the accuracy of the r e s u l t s obtained.
Only a few i n v e s t i g a t o r s h a v e used f i n i t e element methods i n solving

the St. Venant equations. The main reason for nor being used is that
f i n i t e element programs a r e expensive to r u n a n d accuracy and s t a b i l i t y
c r i t e r i a can become tedious to a p p l y .
Explicit finite difference schemes h a v e been w i d e l y used in the p a s t
for the solution of the one-dimensional St. Venant equations. They differ
from each other in the way they define their discharge and depth
gradients, but they all express the flow properties at a certain time

as a function of the flow properties at a previous time thus permitting


an explicit solution. They a r e simple to use as they use a f i x e d r e g u l a r
g r i d and i t i s easier to follow the variation of the flow properties along
the catchment as the solution is performed explicitly. They have been
found to be accurate and economical when properly used. The main
problems accompanying the choice and the use of an explicit finite
difference scheme are, however, those of accuracy and s t a b i l i t y . Choosing
the most proper scheme a n d using i t accordingly is, therefore, important
i n o b t a i n i n g s t a b l e and a c c u r a t e r e s u l t s .
The main explicit finite difference schemes which have been used
previously are summarised in F i g u r e 5.6 in terms of the p o i n t s used a t
a time i n t e r v a l to p r o p a g a t e i n f o r m a t i o n a t the next time i n t e r v a l .
The properties of the different schemes are summarized by Liggett
and Woolhiser (1967). The unstable method was found to be unreliable
while the rest shcwed signs of instability when used in certain cases.
The Lax-Wendroff scheme tended to dampen out instabilities and produce
better results.

Various other investigators were faced with similar problems when


using such different schemes for the s o l u t i o n of the k i n e m a t i c equations.

Constant i n i d e s (1982), however, argued that as the n a t u r e of informat ion


propagation for the kinematic equations differs from that of the St.
Venant equations alternative difference schemes had to be developed.
Furthermore, he argued that the scheme to be used should propagate
numerically, information in a similar manner as suggested by the k i n e -

mat i c characteristic equations. Using this he developed a scheme shown


to be accurate, stable and fast ( T a b l e 5.1, p. 103).
91

X- AX a x+ &X X-AX X X+AX

( a ) Unstable method ( b ) D i f f u s i n g method

( d ) L a x Wendroff method

Uses f o r d i f f u s i n g scheme
f o r the f i r s t time i n t e r v a l
a n d the leap-frog scheme
f o r subsequent t ime i n t e r v a l s

X- AX X X+hX

( c ) Leap-frog method

P : Point where flow p r o p e r t i e s w i l l be c a l c u l a t e d ( x , t )


+ : Points used f o r d e f i n i n g d i s c h a r g e g r a d i e n t s
: Points used i n the depth g r a d i e n t d e f i n i t i o n
4: D i r e c t i o n a t which information i s p r o p a g a t e d
f o r discharge
-: D i r e c t i o n a t which i n f o r m a t i o n i s p r o p a g a t e d
f o r depth

Fig. 5.6 E x p l i c i t f i n i t e difference scheme used i n the s o l u t i o n of the


one-dimensional St. Venant equations.

E x p l i c i t Scheme

The a p p l i c a t i o n of the e x p l i c i t method to the unsteady flow equations


is primarily the outcome of pioneering work by J.J. Stoker; a complete

description is found in Isaacson, et at. (1956). The explicit scheme


92

shown here is from that report. A rectangular channel with no lateral


i n f l o w i s assumed.
A network of node points is shown in Figure 5.7 for solving the
governing equations using the explicit method. The variables are known
at points L, M and R, and are to be determined for point P. Using a
centered difference quotient to approximate the spatial derivatives and
a forward difference quotient to approximate the temporal derivatives,
the f o l l o w i n g a p p r o x i m a t i o n s a r e made a t p o i n t M:

(5.18)

(5.19)

(5.20)

F i g . 5.7 Network o f P o i n t s f o r E x p l i c i t Method

Simi l a r a p p r o x i m a t i o n s a r e made to the other derivative terms. When


these approximations are inserted into Eqs. 5.1 and 5 . 2 , v ( P ) and y ( P )
c a n be s o l v e d d i r e c t l y a s

(5.21 )

(5.22)
93

The solution procedure is to use the information at time level t and


solve for the unknowns at each of the grid points at time level t t

A t. Once this row of values has been determined, advance the com-
putations to time level t t 2At. The values at time level t + A t become
the initial values for determining the unknowns at this advanced time
level. The solution proceeds in this fashion until all the grid points
in t h e s o l u t i o n d o m a i n h a v e been determined.
To ensure stability, the g r i d sizes Ax and at a r e chosen to satisfy

the c o n s t r a i n t

(5.23)

This criterion for computational step sizes, known as the Courant


condition, insures that the time increment i s selected such that the node
point P lies within the area bounded by the forward and backward
characteristics generated from node points L and R. As discussed

previously, this ensures that point P is within solution zone A a n d can


be fully determined using only the initial value information contained
a l o n g the l i n e from L to R.

Imp I ici t Scheme

A network of node points is shown in F i g u r e 5.8 for solving the


unsteady flow equations using a n implicit method. the centered four point

d i f f e r e n c e scheme i s i I l u s t r a t e d ( A m e i n a n d F a n g , 1969).

pstream Downstream
t Boundary Boundary
: o n d i t . i o n 2 m / Condition

A t
1

Fig. 5.8 Network of P o i n t s f o r I m p l i c i t Method


94

The fol lowing approximat ions to the derivative terms are made:

(5.24)

(5.25)

(5.26)

(5.27)

(5.28)
Sf = 1,
1 [Sf(l) + S f ( 2 ) + Sf(3) + Sf(4)]
(5.29)

These a p p r o x i m a t i o n s are used to replace the respective terms in


Eqs. 5.1 and 5.2. Hydraulic variables at node points 1, 2 and 3 are
known from boundary conditions and initial values, hence the unknowns
are Q(4), v(4), y(4), A(4) and Sf(4). Since y(4) and A(4) are related
by the cross-sectional geometry and Q(4), v(4) and A(4) are related by
continuity, 'there are actually three unknowns and two equations. Since
there is the need for another equation, the difference scheme is written
for all of the distance steps at given time level until the downstream
boundary is reached. In F i g u r e 5.8 there are 12 g r i d boxes, meaning

there will be 24 equations to be w r i t t e n b u t there will be 27 unknowns.


The three additional equations a r e specified b y the downstream boundary
condition which most often is a r a t i n g curve between discharge and area

( d e p t h 1.
The resulting set of algebraic finite difference equations is non-
linear and must be solved using an iterative root-finding scheme. Amein
a n d Fang (1969) found that the Newton scheme c o u l d b e used to linearize
the equations w h i c h they then solved u s i n g m a t r i x techniques.
The solution procedure is to solve for all the unknowns at one
advanced time level before proceeding to the next. All values are
determined simultaneously, and must satisfy al I boundary conditions.
Therefore, this method avoids the stability requirements of the explicit
method meaning that larger x and t grid interval sizes can be used
w h i c h r e q u i r e s less i n p u t d a t a .
95

ACCURACY AND STAB I L I TY OF NUMERICAL SCHEMES

There are two approximations in numerical modelling. One needs

to ask the questions: "How well is the natural system modelled by the
differential equations?", a n d , "How well is the s o l u t i o n to the d i f f e r e n t i a l
equations represented by the computational algorithm?". In the analysis
here more attention is paid to the second question. The first question

can only be answered by studying the behaviour of the natural system

and comparing it to the equations applied to it. Therefore it will be


assumed here that the differential equations a p p r o x i m a t e the system we1 I
despite the fact that it has been noticed that this is not necessarily
the case. Abbott (1974) noticed that a difference scheme considerably
different from the differential equations used to describe a system, can
y i e l d more accurate results than a difference scheme similar to the
d i f f e r e n t i a l equations when compared w i t h experimental results.
There are three possible sources of error associated with finite
difference solutions to partial differential equations. It is important
that one understands these sources, their consequence if not control led,
and means for controlling them. These three sources of error are:
truncation, discretization, and round-off. Truncation error occurs when
a derivative is replaced with a finite difference quotient; discretization

error is due to the replacement of a continuous model (function) with


a discrete model; and round-off error is essentially machine error in
that the algebraic finite difference equations are not always solved
exact I y .
For finite difference solutions to be accurate, they must be con-
sistent and stable. Consistency simply means that the truncation errors
tend to zero as Ax and At - > 0, i.e., as Ax and At - > 0 the f i n i t e
difference equation becomes the original differential equation. This is
examined in the follcwing paragraphs. Stability implies the controlled

growth of round-off error. Stability considerations apply principally to


explicit schemes to be discussed later. Any numerical scheme t h a t allows
the growth of e r r o r , e v e n t u a l l y "swamping" the t r u e s o l u t i o n , is unstable.
96

Generally, to ensure stability requires that limits be placed on the


a l l o w a b l e sizes for Ox and At, The c r i t e r i o n f o r establishing the a l l o w -
able sizes is that they be chosen such that the forward and backward
characteristics will not travel the distance Ax in the time interval At.
This insures that the s o l u t i o n a t the advanced p o i n t i n time can be f u l l y
determined from available initial value information; i.e. the grid point
being solved is in solution Zone A. Generally, if a numerical scheme
is both consistent and stable, its solution will be convergent (accurate)
w i t h the s o l u t i o n of the p a r t i a l d i f f e r e n t i a l e q u a t i o n .
The truncation error is examined with a Taylor's series expansion

for U ( x , t ) a t the p o i n t (x,O), i.e. time i s h e l d constant.

U(x+Ax,t) = U ( x , t ) + Ax%
ax
+ %
2!
2
L
ax
2u
+ * ' *
(5.31)

where the derivatives are evaluated at x,t. Dividing Eq. 5.31 by ,x,
and rearranging, gives the series equivalent to the forward difference
quotients, Eq. 5.8
(5.32)

which shows that replacing a U / ax with the forward difference quotient


introduces an error of approximation equal to those terms on the right
hand side of Eq. 5.32 after a U/ ax. This error is proportional to th?

first power of Ax; we cal I this first order error (or approximation).
Similarly, it can be shown that the backward difference quotient has

first order error, and the centered difference has second order error.
Consider the f o l l o w i n g p a r t i a l d i f f e r e n t i a l e q u a t i o n
aa
- + c - = o
aa
(5.33)
at ax
One f i n i t e d i f f e r e n c e a p p r o x i m a t i o n to t h i s equation is

Q(x+Ax,t+At) + Q(x,t+At) - Q(x+Ax,t) - Q ( x , t )


2 At
Q(x+A.x,t+At) - Q ( x , t + A t ) =
+ C (5.34)
Ax

Examination of the Taylor's series residuals reveals the absolute value


of the t r u n c a t i o n e r r o r i s
2
Error = -. At a 2 Q + O ( C x2 ,At 2 )
-
2
(5.35)
ax

-
where the l a s t term indicates a second order of approximation. On
inspection it appears that Eq. 5.34 is consistent with Eq. 5.33 as A t
0. However, for this particular solution, stability considerations
r e q u i r e that
Ax
c 5 - (5.36)
A t
97

Substituting this inequality into Eq. 5.34 transforms the error term into

(5.37)

which indicates a small error term, but one that can become s i g n i f i c a n t
if At - > 0 faster than Ax2 - > 0. Since A x and At a r e f i n i t e and a r e
not approximately zero, Eq. 5.34 approximates Eq. 5.33 with second o r d e r
accuracy but with a term introducing artificial (numerical) dispersion.
This example was chosen because it illustrates how k i n e m a t i c models
can simulate a dispersing hydrograph. Eq. 5.33 is merely the kinematic
wave equation for no lateral inflow which, theoretically, cannot predict
hydrograph dispersion. Eq. 5.34 is one of the finite difference models

used to solve the kinematic model. Because of the presence of the


truncation error, it simulates a dispersing hydrograph, thereby demon-
strating that a numerical kinematic model can simulate a dispersing
hydrograph.

Numerical dispersion or diffusion is the process in which the Error


is formed. It is the development of the truncation e r r o r . to the error

t h r o u g h the numerical technique used.


Lax's (1954) theory, proved by Richtmyer and M o r t o n (1967) states
that for l i n e a r equations w i t h constant coefficients o p e r a t i n g on u n i f o r m l y
continuous initial and boundary data the following theorem holds. Given
a properly posed initial-value p r o b l e m and f i n i t e difference approximation
to it that satisfies the consistency conditions, stability is the necessary
and sufficient condition for convergence. This is however proved only

for linear equations and according to Abbott (1979) it b r e a k s down when


there a r e discontinuities in f l o w .
Since one is dealing with non-linear partial differential equations

(p.d.e's) there is no rigorous proof specifying stability criteria. For


linear p.d.e's, however, stability analyses exist. Von Neuman (1949)

was first to devise a powerful technique for determining stability criteria


for linear p.d.e's. He m a d e u s e o f the fact that just about any function
can be represented by a Fourier series. The linear stability analysis

method essentially determines how the Fourier coefficients behave (grow,


decay, or stay constant) with time for any term in the Fourier series.
For stability to occur the ratio of a Fourier coefficient of any term a t
any time o v e r the Fourier coefficient of the same term at a p r e v i o u s time
m u s t b e l e s s t h a n one.
The effect of Ax and A t on stability and accuracy are summarized
in Figure 5.9. From Figure 5.9 one can deduce that the main criteria
in the selection of Ax and At values for an explicit finite difference
scheme a r e :
98

solution i s solution is
stable
___)

Accuracy of solution decreases

I
due t o numerical diff sion

For fixed (6X/At), accuracy ncreases


for smaller A X and At

Fig. 5. 9 Effect of v a l u e of A x a n d A t on s t a b i l i t y and accuracy for


a n d e x p l i c i t f i n i t e d i f f e r e n c e scheme.

a) that the scheme shal I proceed u n d e r s t a b l e c o n d i t i o n s


(5.38)

Ax
b, at
Ax s h a l l be close to (-)
At c r
to minimise d i f f u s i o n e r r o r s and obtain
optimal accuracy.
c) the difference scheme’shal I be convergent. T h i s c o u l d be a s c e r t a i n e d
by running the scheme with different Ax‘s and At’s and comparing
w i t h a n a l y t i c a l r e s u l t s i n a simple case.

( A X / At,]cr has been shown to be the speed of wave disturbance or


i n f o r m a t i o n as it i s propagated. T h i s can be demonstrated b y considering

the manner in which information i s propagated along the characteristic


curves. For example, consider a central difference scheme, simi l a r to
the diffusion method, for solving the St. Venant equations. Let i
represent a space interval, and k represent a time interval as shown
in Figure 5.10. The point in question, i.e. where the flow properties
are to be calculated, has the co-ordinates (i,k). Information about the
flow properties i s sought from the p r e v i o u s time interval. I n F i g u r e 5.10
(a) the t r u e p r o p a g a t i o n speed i s smaller than the numerical propagation
speed while in F i g u r e 5.10 (b) the converse is true. Numerical propa-
gation lines a r e lines that have a slope Ax/A t in the x - t plane while
true propagation lines have a slope dx/dt in the x - t plane. I n Figure
99

5.10 (a) information is obtained within the i - I, i + I range by the


true propagation lines. In F i g u r e 5.10 information i s sought by the t r u e
p r o p a g a t i o n l i n e s outside the i - I , i + I range.
Since information outside this range is not propagated by the
numerical scheme, it cannot be found and thus instability will result.
A more d e t a i l e d e x p l a n a t i o n i s g i v e n b y Stoker (1957).
For stabi I i t y of an explicit finite difference scheme the following

must therefore h o l d :
Lix dx
- > - (5.39)
At = dt
This is referred to as the "CFL condition" after Courant, Friedrichs
a n d Lewy (1928), o r s i m p l y the Courant c r i t e r i o n f o r s t a b i l i t y .

t t

1-1 i i+1 1-1 i i+i

(a) ax
At
> dx
dt

Numerical p r o p a g a t i o n lines; slope ( A x / A t )


---- True p r o p a g a t i o n l i n e s ; slope ( d x / d t )

Fig. 5.10 Comparison of numerical and theoretical propagation of


i n f o r m a t i o n i n a c e n t r a l d i f f e r e n c e scheme

I t has been noticed, however, that even if one satisfies the CFL

conditions it is not necessarily true that the solution of the difference


scheme is inherently stable (e.g. by LAX, 1954; Richtmyer and Morton,
1967; Abbott, 1974). There are two possibilities which could give rise

to instability. There could be a physical discontinuity in the flow, e.g.


100

a bore o r a h y d r a u l i c jump or p a r a s i t i c waves c o u l d be generated w i t h i n


the difference scheme.
I n terms of characteristics a physical discontinuity imp1 ies the i n t e r -
section of two or more characteristics. Theoretically this results in
different values of flow properties for a fixed place and time. In a
difference scheme with a fixed grid this theoretical multivaluedness
cannot be accounted for and in the solution is present in the form of
oscillations. If the difference scheme tends to amp1 i f y these oscillations
instability will occur. I f however, these o s c i l l a t i o n s get damped s t a b i l i t y
will result and our scheme is referred to as a dissipative difference
scheme.
The difference scheme being used can also cause oscillations called
parasitic waves. It has been noticed (e.g. by Abbott, 1974) that the
parasitic waves do not only occur when a physical discontinuity occurs
but can arise out of the numerical procedure used. Therefore certain
difference schemes h a v e been found to produce p a r a s i t i c waves w h i l e o t h e r s
do not when c o n s i d e r i n g the same p h y s i c a l problem.
There are two ways these problems can be overcome. If a physical
discontinuity exists, it can be located, the laws governing the discon-
tinuity can be applied, and the laws governing continuous flow can be
a p p l i e d to each side.
It i s also p o s s i b l e to a d j u s t any difference scheme to dampen instead
of amp1 i f y parasitic waves. The solutions o b t a i n e d from these "dissip-
ative difference schemes", are called "weak solutions", as in this way
stability is obtained at the loss of accuracy (see Lax, 1954). Abbott
(1974) describes the dissipative schemes and the amount of accuracy lost
extensively.
If one considers the method of setting u p a d i s s i p a t i v e scheme, one
will also illustrate the p r i n c i p l e of the weighted averages which i s based

on a v e r a g i n g flow properties at a certain time interval by linear inter-


polation according to where the characteristic curves intersect at t =

constant line. Consider f o r example a b a c k w a r d d i f f e r e n c e scheme a s shown


in Fig. 5.11 and the way information about depth ( y ) i s propagated.

Depth at time t = k - 1 i s taken to be as 4 (1


k;l
- r,)y I
k-1
+ v i - 1 (see
Figure 5.11).
Suppose now one wants to propagate the depth at point Q. Then
interpolating linearly between p o i n t s A a n d B one must use depth a t Q a t
kT1 k-1
time t = k-1 as (1-r)y I + ryi-1 ,
where r i s the r a t i o of d i s t a n c e QB
over distance AB i n f i g u r e 5.11.
101

t
T
at

1-1 I i+l

4- -t

Fig. 5.11 The principle of weighted averages for information prop-


agation i n a b a c k w a r d d i f f e r e n c e scheme.

I f one uses the fact that information i s t r u l y propagated as a speed


of -
dx
dt then the slope of line QP, shown in Figure 5.11 should be the
v a l u e of -
dx a t p o i n t Q (representing a point in space at a particular
dt
time) denoted as ( d x / d t l g

S t r i c t l y s p e a k i n g the v a l u e of r should therefore be

dx AX
r = (3lQ / t (5.40)

A dissipative difference scheme i s one as described above but with


r chosen in such a way as to dampen oscillations. The discrepancy
between r chosen a n d r i n equation (5.40) w i l l result i n loss of accuracy
i n the s o l u t i o n of the d i f f e r e n c e scheme.
102

EFFECT OF FRICTION

Because t h e f r i c t i o n term in the flow equation i s non l i n e a r i t makes


solution of implicit type equations more difficult than without the
friction term. A number of methods of accounting for the friction term

was d e s c r i b e d b y Cunge et a l . ( 1 9 8 0 ) : The friction gradient is assumed


to b e of the f o r m
sf = QIQI/K~ (5.41)
where K = A R Z ' 3 / n ( M a n n i n g , S.I. units) (5.42)
R = A/P (5.43)
If an explicit scheme is not acceptable, for instance i f Sf is large
compared with ay/ax, then some form of averaging of Sf in time is
required. Cunge et al. suggest taking the average Q over the distance
interval and squaring that, rather than the a v e r a g e of the squares of
the Q l s o v e r t h e i n t e r v a l , i.e.

(5.44)

An alternative which produces a linear equation and also yields


the correct sign of Q was suggested by Stephenson (1984) for closed
conduits: ,

+ 1 - 0
[(Q"/K")~ + (Q" /K" 1'3 (5.45)
4
j j j+l j+l
Strelkoff (1970) indicates that the d i r e c t e x p l i c i t scheme i s i n h e r e n t l y
unstable. He indicates the Lax type scheme should satisfy the Courant
criterion. For implicit schemes he suggests that to ensure stability in
friction
KO
At < A g / T (5.46)

where KO = AoC 1% = Qo/ (5.47)

.'. At < (5.48)


gsf
Wylie (1970) suggested that for a simple linear explicit system for
open c h a n n e l s t h a t f o r s t a b i l i t y
At 5 (Ax/c) (1 - gSfAt/2V)'" (5.49)
E v e n t h i s does n o t g u a r a n t e e s t a b i l i t y a c c o r d i n g t o W y l i e .
103

CHOOSING AN EXPLICIT FINITE DIFFERENCE SCHEME FOR THE SOLUTION

OF THE ONE-D I MEN5 I ONAL K I NEMAT I C EQUAT I ONS

Constant i n i d e s (1982) used various schemes for solving the one-


dimensional k i n e m a t i c e q u a t i o n s i n an a t t e m p t t o c h o o s e t h e most s u i t a b l e
scheme. The d i f f e r e n c e schemes mentioned e a r l i e r as well as new proposal
schemes were used. The equations were solved for different problems
which can also be solved with analytical methods. The analytical sol-
utions were then compared w i t h r e s u l t s from the numerical solutions. The

suitability of the various difference schemes was then evaluated on the


basis of accuracy and stability. The choice of a difference scheme was
done by the process of elimination as more complicated problems were
considered. A new p r o p o s a l scheme, s h o w n i n T a b l e 5.1 was f o u n d to y i e l d
extremely accurate results, to b e s t a b l e as long a s t h e Courant criterion
i s s a t i s f i e d and t o b e f a s t and e c o n o m i c t o run. T h e scheme i s s u m m a r i s e d
in T a b l e 5.1 b y d e f i n i n g the d i s c h a r g e r a t e a n d d e p t h a t a time i n t e r v a l .

TABLE 5 . 1 Backward-central e x p l i c i t d i f f e r e n c e schemes

D i f f e r e n c e Scheme Discharge Rate Depth y at

- at t = k - 1 t = k - 1
ax

i-l i i+l

INDEX

x p o i n t s where flow properties a r e to be calculated

+ p o i n t s used f o r c a l c u l a t i n g d i s c h a r g e a t time t = k - 1
0 p o i n t s used f o r c a l c u l a t i n g d e p t h a t time t = k - 1

The explicit finite difference scheme shown in Table 5.1 although


chosen by trial and error as being the most efficient scheme, becomes
apparent when one considers the method of characteristics described
earlier. The schemes propagate information downstream only as is
suggested b y the c h a r a c t e r i s t i c equation.
104

REFERENCES

Abbott, M.B., 1974. C o n t i n u o u s f l o w s , d i s c o n t i n u o u s f l o w s and n u m e r i c a l


a n a l y s i s . J. H y d . Res., 12, No. 4.
A b b o t t , M.B., 1979. C o m p u t a t i o n a l h y d r a u l i c s . P i t m a n P u b l . L t d . L o n d o n
A b b o t t , M.B. and V e r w e y , A., 1970. F o u r - p o i n t m e t h o d o f c h a r a c t e r i s t i c s .
J. H y d . D i v . , ASCE, HY12, Dec. 1970.
A m e i n , M. and F a n g , C.S. (19691, S t r e a m f l o w r o u t i n g - w i t h a p p l i c a t i o n s to
North C a r d l ina Rivers. Report No. 17, Water Resources Research
I n s t i t u t e , U n i v e r s i t y o f N o r t h C a r o l i n a , Chapel H i l l , N o r t h C a r o l i n a .
Constantinides, C.A., 1982. Two-dimensional k i n e m a t i c model l i n g o f t h e
r a i n f a l I-runoff process. Water Systems Research Programme, Report
1/1982. U n i v . o f t h e W i t w a t e r s r a n d .
C o u r a n t , R., F r i e d r i c h s , K.O. and L e w y , H., 1928. U b e r d i e p a r t i e l l e n
D i f f e r e n t i a l g l e i c h u n g e n d e r M a t h e m a t i s c h e n P h y s i k , M a t h . A n n , 100.
Cunge, J.A., Holly, F.M. and V e r w e y , A., 1980. P r a c t i c a l A s p e c t s o f
C o m p u t a t i o n a l R i v e r H y d r a u l i c s . P i t m a n s , B o s t o n , 420 p p .
I s a a c s o n , E., S t o c k e r , J.J., and T r o e s c h , B.A., 1956. N u m e r i c a l s o l u t i o n o f
f l o o d p r e d i c t i o n and r i v e r r e g u l a t i o n p r o b l e m s . I n s t . M a t h . S c i . R e p o r t
No. IMM-235, New Y o r k U n i v e r s i t y , New Y o r k .
L a x , P.D., 1954. Weak s o l u t i o n s f o r n o n - l i n e a r h y p e r b o l i c e q u a t i o n s and
t h e i r n u m e r i c a l a p p l i c a t i o n s . Comm. P u r e A p p l . M a t h . 7.
Ligget, J.A. and Woolhiser, D.A., 1967. Difference solutions of the
s h a l l o w w a t e r e q u a t i o n . J. E n g . Mech. D i v . ASCE, A p r i l .
L i g h t h i l l , F.R.S. and W h i t h a m , C.B., M a y 1955. On k i n e m a t i c w a v e s 1 .
F l o o d movement i n l o n g r i v e r s . P r o c . R o y . SOC. L o n d o n , A, 229.
Mahmood, K . and Y e v j e v i c h , Eds., 1975 , U n s t e a d y f l o w in open c h a n n e l s ,
Vols. I and I I , W a t e r R e s o u r c e s P u b l i c a t i o n s , F o r t C o l l i n s , C o l o r a d o .
O v e r t o n , D.E. and M e a d o w s , M.E., 1976. S t o r m w a t e r Model I i n g . A c a d e m i c
P r e s s , New Y o r k .
R i c h t m y e r , R.D. and M o r t o n , K.W.. 1967. D i f f e r e n c e m e t h o d s o f i n i t i a l v a l u e
p r o b l e m s . 2 n d E d . I n t e r s c i e n c e , New Y o r k .
S t e p h e n s o n , D. 1984. P i p e f l o w A n a l y s i s . E l s e v i e r , A m s t e r d a m , 274 p.
S t o k e r , J.J. 1957. W a t e r Waves. I n t e r s c i e n c e P r e s s , New Y o r k .
S t r e l k o f f , T., 1970. N u m e r i c a l s o l u t i o n o f S a i n t - V e n a n t equations. Proc.
ASCE. J. H y d r . D i v . 9 6 ( H Y 1 ) , 223-252.
Von N e u m a n , J . , 1963. Recent t h e o r i e s o f t u r b u l e n c e . Collected Works
(1949/1963) e d i t e d b y A.H. T a u b , 6 , P e r g a m o n , O x f o r d .
Wyl ie, E.B., Nov. 1970. U n s t e a d y f r e e - s u r f a c e f l o w c o m p u t a t i o n s . Proc.
ASCE, J. H y d r . D i v . , 9 6 ( H Y l l ) , 2241-2251.
105

CHAPTER 6

DIMENSIONLESS HYDROGRAPHS

UNIT HYDROGRAPHS

I n the same w a y that the p e a k f l o w graphs in Chapter 3 can r e p l a c e


the Rational equation, so kinematic theory can b e used t o g e n e r a t e u n i t

hydrographs for larger catchments. The simpl i f y i n g assumptions in the


Rational method and the peak flow charts a r e often inaccurate when it
comes to larger catchments. An extension of the Rational method became
necessary for l a r g e catchments a n d u n i t h y d r o g r a p h theory was d e v e l o p e d .

The h y d r o g r a p h shape was needed f o r r o u t i n g too. A n analogous procedure


is developed below for selecting hydrographs for various catchment
configurations. A n a d v a n t a g e o v e r the u n i t h y d r o g r a p h methods i s that the
hydrographs here are dimensionless and a l low for various simpl i f i e d
catchment configurations. This i s offset b y a s l i g h t l y more c o m p l i c a t e d set
of calculations. As with unit hydrograph procedures however, the
catchment storm d u r a t i o n i s selected b y t r i a l .
The dimensionless hydrographs presented below are synthesized for
selected uniform storm durations. The catchments selected have varying
shape and topography representing the m a j o r i t y of small catchments. The

hydrographs, being dimensionless, are presented as functions of rainfal I


intensity and should therefore find international applicability. The user
must select r a i n f a l I rates corresponding to d e s i r e d r e t u r n p e r i o d s as well
as initial abstraction and i n f i l t r a t i o n rates applicable to t h e c a t c h m e n t in

quest ion.
The hydrographs are intended for use b y d e s i g n e n g i n e e r s where n o t
only the hydrograph peak flow rate but the shape of the hydrograph is

important. The a p p l icat ion to different catchments of varying shape and


topography in developing the hydrographs makes their use more a d v a n t -
ageous o v e r o t h e r t e c h n i q u e s , as e x p l a i n e d below.
The lag effect due to overland flow length, s u r f a c e roughness and
slope is invariably i n c l u d e d i n the g r a p h s p r e s e n t e d . The r e s u l t i s a more

realistic and effective hydrograph for the d e s i g n e r than i s possible w i t h


p r e v i o u s methods. The e f f e c t o f flow concentration i n streams a f t e r f l o w i n g
overland cannot be readily assessed using isochronal methods (or any
other s t a n d a r d method). N e i t h e r c a n the e f f e c t o f c h a n g i n g g r o u n d slope o r
c o n v e r g i n g flow which can a l l be a c c o u n t e d f o r with the k i n e m a t i c models
used h e r e .
For peak discharge computation storms of duration smaller or equal
to the time of equilibrium of the catchment are important, as a storm
could produce maximum peak discharge off the catchment. Higher flood
peaks may result from a shorter storm. The critical storm d u r a t i o n , i.e.
the storm duration that w i I I produce maximum peak discharge, w i l l depend
on two factors, these being the way the catchment responds to storms of
duration less than the catchment's time of equilibrium, the rainfall
c h a r a c t e r i s t i c s a n d the r e t e n t i v e p r o p e r t i e s of the catchment's soi I s . Storms
of durations longer than the catchment's time of equilibrium are also
important, especially i n cases where r u n o f f volume i s of importance.

Neither a s i n g l e v a l u e of peak d i s c h a r g e r a t e nor t o t a l r u n o f f volume


are generally sufficient for a l l the purposes of the d r a i n a g e engineer. The
time the catchment takes to reach its peak discharge as well as the
complete hydrograph shape are generally of prime importance. I n cases

where r u n o f f h y d r o g r a p h s h a v e to be combined from d i f f e r e n t catchments o r


are routed through hydraul i c conduits, the complete runoff hydrograph
shape i s essential for a c c u r a t e design.
The hydrograph shape is also important in designing hydraulic
structures to cope with floods of h i g h e r r e t u r n periods than those which
they were designed to c a r r y . The p a r t of the hydrograph not carried by
the hydraulic conduit structure, if known, can be diverted by suitable
means, while, its backwater effects upstream and the force on the
s t r u c t u r e could a l s o be e v a l u a t e d .
The volume under the hydrograph i s of particular importance when
detention or retention storage are contemplated. The r o u t i n g effect and
peak flow attenuation are particularly s e n s i t i v e to the h y d r o g r a p h shape
as opposed to the peak.
In general the dimensionless hydrographs should be of particular

interest to the u r b a n d r a i n a g e engineer who w i l l w i s h to s t u d y stormwater


management and the effects of u r b a n i s a t ion - changing surface configur-
ation, roughness a n d p e r m e a b i l i t y on flow rates.

DEVELOPMENT AND USE O F GRAPHS

I n developing runoff hydrographs for a catchment it i s important to


u n d e r s t a n d how the catchment w i l l react to d i f f e r e n t storms. The volume of
surface runoff is primarily a function of rainfal I and infiltration
characteristics, while the hydrograph shape is a function of catchment

shape, roughness a n d topographical characteristics.


107

Computer models can account for any time and space variation of
r a i n f a l I and catchment c h a r a c t e r i s t i c s as described l a t e r . T h e i r use e n t a i I s
substantial computer time and the model has to be used in conjunction
with various storm i n p u t s to ensure c r i t i c a l storm input. I n t h i s section,
runoff hydrographs off catchments of fixed shapes and with spatially

varied catchment c h a r a c t e r i s t i c s a r e presented. The r e s u l t i n g h y d r o g r a p h s


are dimensionless, i.e. in terms of catchment size and rainfall rate,
a l lowing the use of different catchment dimensions a n d d i f f e r e n t roughness
and catchment slope parameters. The design engineer can use these
h y d r o g r a p h s for n a t u r a l catchments w h i c h h a v e simi l a r shapes to the model
catchments studied and where the roughness a n d slope c h a r a c t e r i s t i c s a r e
consistent. The design engineer still has to use his judgement in
approximating catchment shapes and in averaging roughness and slope

pa ramet ers.
The k i n e m a t i c equations h a v e been used to p r e p a r e the hydrographs
presented by Constantinides a n d Stephenson (1982). Computer solution of
the finite difference form of the equation of motion and the flow
resistance e q u a t i o n was performed f o r numerous s i t u a t i o n s . With the use of
dimensionless parameters the number of variables i s reduced c o n s i d e r a b l y
and a few graphs present a r a n g e of hydrographs covering the r a n g e of
parameters normal l y encountered.
Runoff hydrographs off three model catchments a r e presented, these
b e i n g the f o l l o w i n g :

( a ) A s l o p i n g p l a n e catchment
( b ) A c o n v e r g i n g surface catchment
( c ) A V-shaped catchment w i t h stream

Design h y d r o g r a p h s may be o b t a i n e d b y comparing dimensional r u n o f f


hydrographs f o r different storm d u r a t i o n s , a n d s e l e c t i n g the one r e s u l t i n g
I

i n maximum flow rate (if the u n a t t e n u a t e d peak i s of concern) o r greatest


volume r e q u i r e d t o a t t e n u a t e the flood i f storage i s to be designed, o r any
other relevant c r i t i c a l parameter.

L i s t of Symbols

x space a x i s a l o n g o v e r l a n d p l a n e (m o r f t )
z space a x i s a l o n g channel (m o r f t )
L length of o v e r l a n d p l a n e (m o r f t )

Ls length of channel o r stream (m o r f t )


So b e d slope of o v e r l a n d p l a n e
108

n roughness coefficient of o v e r l a n d p l a n e s
n roughness coefficient of channel o r stream
0 a n g l e d e s c r i b i n g c o n v e r g i n g surface catchment (radians)
r r a t i o d e s c r i b i n g c o n v e r g i n g s u r f a c e catchment
w w i d t h of o v e r l a n d flow i n c o n v e r g i n g surface catchment ( m or f t )
H depth of channel ( m or f t )
b w i d t h of channel (m o r f t )
yo depth of o v e r l a n d flow ( m or f t )
qo discharge p e r u n i t w i d t h of o v e r l a n d flow (m’/s or ft’/s)
ys depth of channel flow (m o r f t )
Q discharge of channel flow (m3/s or f t 3 / s )
Q discharge of c o n v e r g i n g s u r f a c e ( m 3 / s or f t 3 / s )

Kinematic equations

The one-dimensional kinematic equations for flow have already been


presented and are merely stated here. They consist of the continuity
e q u a t i o n and a n equation r e l a t i n g h y d r a u l ic resistance to flow.

aQ
- + - aA -
ax at - q~ (6.1 1
and q = uym (6.2)

where Q i s the flow r a t e (m’/s or ft’/s), A i s the cross sectional area


(m‘ or ft’), t i s time (secs), x i s the space a x i s (m o r f t ) , q i s l a t e r a l
L
inflow per unit length along the x - a x i s ( m ’ / s or f t ‘ / s ) , q i s the
average discharge across a section per u n i t width (m’/s or ft‘/s) and y
i s the depth of water (m o r f t ) . u , m a r e coefficients dependent on surface

roughness a n d bed slope.

EXCESS RA I NFALL

I n developing r u n o f f hydrographs off the s i m p l e catchments already


outlined, an excess rainfall distribution is required. I n t h i s case, excess
rainfall intensity i s assumed to be u n i f o r m i n space, a n d constant d u r i n g
the storm a n d e q u a l to a n e g a t i v e constant (being a constant infiltration

rate) after the storm. Fig. 6.2 d e p i c t s the assumed excess r a i n f a l l input
and Fig. 6 . 1 shows the assumed rainfall input and loss d i s t r i b u t i o n f o r
o b t a i n i n g the excess r a i n f a l l d i s t r i b u t i o n shown i n Fig. 6.2.
109

(mm/h

U
VfC
i l -
* u- ted t (h)

I
+4
i
* ld ).

*te

F i g . 6.1 Assumed r a i n f a l l i n p u t a n d d i s t r i b u t i o n losses

I n Figs. 6 . 1 and 6 . 2 i i s r a i n f a l l i n t e n s i t y r a t e (mm/h or inches/h),i is

excess rainfal I intensity rate (mrn/h or inches/h), td is storm duration

(h), ted is excess rainfall duration (h), fc is final infiltration rate


(rnm/h or inches/h),f i s u n i f o r m i n f i l t r a t i o n r a t e (mm/h o r inches/h) a n d u

i s i n i t i a l abstraction (rnm o r inches).

The final infiltration rate, fc is a function of soil type and


vegetation cover or land use. The excess rainfall intensity, ie, is a
function of excess rainfall duration, t, which depends on local rainfall
c h a r a c t e r i s t i c s a n d on catchment soi I a n d vegetation cover p r o p e r t i e s .
110

Rainfall intensit:
(mm/h)

t (hi

t
1 -
1 -

t
time runoff stops

Fig. 6 .2 Excess r a i n f a l l input

D I MENS IONLFSS EQUAT I ONS

It is evident from kinematic theory that if any catchment is


subjected to a constant excess rainfall intensity i for a p e r i o d equal to
or longer than i t s time of e q u i l i b r i u m it will produce a peak discharge
equal to i multiplied by the area of the catchment. In deciding on
dimensionless parameters to be used f o r developing runoff hydrographs it
therefore seems logical to plot the ratio of discharge divided by excess
rainfall intensity and area against a r a t i o of time d i v i d e d b y the time of
e q u i l i b r i u m of a s i m p l e catchment, namely the s l o p i n g p l a n e catchment.

Sloping Plane Catchment

For the s l o p i n g p l a n e catchment depicted in Fig. 6.3 the c o n t i n u i t y


e q u a t i o n becomes:

(6.3a)

= -fcfor t ? ted (6.3b)


111

The u n i f o r m flow equation can also be expressed as:


m
16.4)
qo = uoyo
1
where a = Soy/no

and n o,So a r e the M a n n i n g coefficients and bed slope respectively.


Expressing y in terms of qo from equation (6.4), differentiating with
respect to t a n d s u b s t i t u t i n g i n e q u a t i o n ( 6 . 3 ) yields:

The f o l l o w i n g dimensionless v a r i a b l e s a r e then defined:

x =x
LO

P qo
= -

i eLo
T = mte
-
tc 0
mt
TD = ed
~

(6.9)
tCO

F = fc
- (6.10)
I
e
where t is the time of concentration of a sloping plane in kinematic
co
theory a n d i s g i v e n b y :

tco = (--- Lo l/m (6.11)


m-1
'oie
Substituting for qo, x, t, ted a n d f i n equation (6.5) and manipulating
y i e l d s the f o l l o w i n g e q u a t i o n :

_ 1_ _
ap
+ -aP = 1 f o r T < T
4 D
aT ax
= -F for T (6.12)
> TD

Fig. 6 . 3 Sloping p l a n e catchment


P

&

F i g . 6 .& Dimensionless r u n o f f hydrographs f o r the s l o p i n g p l a n e catchment


F = 0.0
F=O. 5

Pmax versus' T
D

F i g . 6 .5 Dimensionless r u n o f f h y d r o g r a p h s f o r the s l o p i n g p l a n e catchment


F = 0.5
114

where ( m - l ) / m = 0.4 f o r m = 5/3

Equation (6.12) is solved for flow P as a function of time ratio T at


the outlet end of the catchment plane. This i s repeated for different T
D
values. Different plots are obtained for different F values in Figs. 6.4
and 6.5. The theory of Overton (1972) was also adapted to cascades of
p l a n e s b y K i b l e r a n d Woolhiser (1970).

Converging surface Catchment

For the converging surface depicted in Fig. 6.6 the continuity


eq. (6.1) becomes ( W o o l h i s e r , 1969):

(6.13)

= -w f for t > ted (6. 4 )


o c
where w = ( L o - x)B
m
and Qo = (6. 5 )
WoaoYo
Expressing y in terms of €Io from equation (6.4), differentiating with

r e s p e c t to t a n d s u b s t i t u t i n g i n e q u a t i o n (6.13) yields:

-a+Q o 5
wo
l-l/rn
= i w for t j t
ax 7- 0 0 at
= -w
e o
f for t > td
ed

o c (6.16)
In addition to dimensionless variables defined in equations (6.8) to
(6.10) the following dimensionless variables are defined (Singh, 1975):

Fig. 6.6 Converging s u r f a c e catchment


115

(6.17)

(6.18)

where (1-r2 )/2 is the area of the catchment and r the ratio of bottom
segment to t h e t o t a l catchment r a d i u s .
For the converging surface tCO i s defined as the time of equilibrium
f o r a s l o p i n g catchment of l e n g t h L ( 1 - r ) , i.e.

la
Lo(l-r) 1 /m
-~
ti0 - Oiem-l] (6.19)

Substituting for x, Qo,wo,t,ted,fc and m in equation (6.16) and


man i p u I a t i n g y i e l d s :

(6.20)

(6.21)

Equations (6.20) and (6.21) were solved numerically to give S as a


f u n c t i o n of T at the o u t l e t for different T,, values. Plots a r e f o r v a r i o u s
r a n d F values as presented in Figs. 6.7 a n d 6.8.

V-Shaped Catchment w i t h Stream

In the V-shaped catchment (Fig. 6.9) the discharge from overland


flow is used as input in the channel. Kinematic theory i s used to route
overland flow runoff through the channel. It is assumed that both
overland flow planes are similar. From kinematic theory the continuity
equation i n the channel would be:

a Q ~+
__
bays
= 2q0L (6.22)
az at
A basic assumption in equation (6.22) is that the natural depth of the
channel is always greater than the water depth i n the channel. Another
assumption i s that the channel area i s small compared to t h e p l a n e a r e a .
The u n i f o r m f l o w r e s i s t a n c e e q u a t i o n f o r t h e c h a n n e l may b e w r i t t e n :

m
Qs = baSYs (6.23)

Expressing y in terms of Q from (6.23), differentiating with respect


to t a n d s u b s t i t u t i n g i n t o (6.22) yields:
....
m

Fig. 6.7 D i m e n sion I e ss r u noff hy d rog r alp h s f o r the con v e r g i n g s u r f ace c a t c hm e nt

R = 0.05 F = 0 .0 0
-
0,

5- R=O .05
F=O. 50

Smax versus T
D
S

F i g . 6 .8 Dimensionless r u n o f f h y d r o g r a p h s f o r the c o n v e r g i n g s u r f a c e c a t c h m e n t
R = 0.05 F = 0.50
118

(6.24)

In addition to the dimensionless variables defined in equations (5.6)


a n d (6.10) the f o l l o w i n g dimensionless v a r i a b l e s a r e d e f i n e d :

Q = Q , / ~ L ~ L ~ ~ ~ (6.25)

2 = z/L!j (6.26)

where t is the same as for the sloping plane, i.e. equation (6.11).
LO
Substituting for Q ,z,t,qo, and rn in equation (6.24) and re-arranging

yields:

(6.27)

where G = (c2L5

5
)
0.6 ba 0.6

2L0
(6.28)

Equation (6.12) is solved to yield P as a inction c T at X .- 1 for


the planes. P is used as input in equation (6.27) 10 solve for Q as a
f u n c t i o n of T at the o u t l e t for different values for F G G a n d the r e s u l t s
appended at Figs. 6.10 and 6.13. The same problem was handled in a
d i f f e r e n t ' way b y Wooding (1965).

H > y s a t a1
C a tc hme n t

F-ig. 6 .9 V - s h a p e d catchment w i t h stream


9.-
d

9-
d
Q -
5-
9-
d

61-
d

3-
d

8-
d

R-
d

3-
d

F i g . 6 .10 D i m e n s i o n l e s s r u n o f f h y d r o g r a p h s f o r the V-shaped catchment with stream


G = 0.5 F = 0.0
>
N
0

Fig. 6 .ll Dimensionless runoff hydrograph for the V-shaped catchment with stream
G = 0.5 F = 0.5
I
Fig. 6 .12 Dimensionless r u n o f f h y d r o g r a p h s for the V-shaped catchment b v i t h stream
G = 2.0 F = 0.0
G=2.0

F i g . 6 .13 D i m e n s i o n l e s s r u n o f f h y d r o g r a p h s f o r t h e V - s h a p e d catchment w i t h s t r e a m
G = 2.0 F = 0.5
123

Fig. 6.14 Example : Catchment w i t h stream


124

X
w

0
0 2 4 6 8 10
EXCESS STORM DURATION (HOURS]

F i g . 6.15 Example : Excess i n t e n s i t y - d u r a t i o n relationship

TABLE 6 .1 Example : M a n n i n g ' s roughness coefficients

a n d b e d slopes

Cover Manning's n Slope

Overland flow Medium g r o w t h 0.15 5%


meadow
Channel flow Medium g r o w t h 0.15 1.2%
meadow
125

USE OF D I MENS I ONLESS HYDROGRAPHS

The procedure for u s i n g the dimensionless h y d r o g r a p h s i s i l l u s t r a t e d


b y means of an example.

Problem

Consider the natural catchment outlined in Fig. 6.14 and the 5


year recurrence interval excess IDF relationship shown in Fig. 6.15.
Obtain the runoff hydrograph producing the worst peak discharge off
the catchment. The excess IDF relationship given allows for the storm
spatial distribution (which has been reduced from the point excess
rainfall IDF relationship) and has been developed using local rainfall
data and catchment characteristics. The average final infiltration rate
of the s o i l (f ) i s 1.5 mm/h.

So Iu t ion

The natural catchment shown in Fig. 6.14 is approximated by a


V-shaped catchment with stream. The main waterway in the catchment
has a length of 1350 metres and subdivides the catchment approximately
in the middle. The other waterways a r e minor a n d most of the catchment
flow is in the form of overland flow flowing perpendicularly to the
waterway. The waterway . i s assumed to be a r e c t a n g u l a r channel 31-17 wide.
The assumed V-shaped catchment with stream is illustrated in F i g . 6.16.

Manning's roughness coefficients a r e shown i n Table 6 .1 w h i l e bed slopes


are averaged using the contour lines from Fig. 6.14 and summarized in
Table 6 . 1 . Parameter G must be e v a l u a t e d u s i n g ( 6 . 2 8 ) :

G = Z(1350)-
2 (308.9)

Figs. 6.10 and 6.11 with G = 0.5 are used for choosing the critical

runoff hydrographs. The infiltration parameter F is a function of the


excess r a i n f a l l r a t e .

Table 6.2 shows the c a l c u l a t i o n s i n choosing a c r i t i c a l runoff hydrograph

a n d dimensioning i t . The t a b l e r e f e r s to F i g u r e 6.10.


126

Outlef
Scale I :7500

F i g . 6 .76 Example : Assumed c a t c h m e n t


T A B L E 6 .2 Exarrple : Choosing a n d dimensioning r u n o f f h y d r o g r a p h w i t h m 9 x l m u m p e a k c i s c h a r g e

-___r----------- - -.

0 .6 ~ 0.4
( 3 . 6 ~ 1O b )
3600
0 1

var j.a bX e
t
CO
a Qs I F a c t o r s t.o d i m e n o j . o n
r u n o f F hydrograph
M u 1t l p l y

a x is

units hours m/hr hours m' Is hours


___~. I I

dirnensionl.ess
source guess excess IDF's hydrographs

1 .o 13.99 0.107 0.995 1.675

0.5 17.55 0.086 0 .go9 1.101


--
1.2 12.70 U.118 1 .034 1.934
I

1
11.66 2.70 1 P.620 I 2.942

1.4 11.63 0.12', 1.071 2.178 11.29


_ "....,

Critical storm h a s an CXCPSF: duration of 1 . j : h o u r s prodiicIny a dlsclrarqc peak of 1 .70 cum~c'i.

~~..---,.--._......-.---I--_ -
3.0

2 -5

n
E

I .o

0.5

0.0 1.0 2.0 3.0 4.0 5.0


Time (hours)
!Fig. 6.17 Example : Critical runoff hydrograph
129

As c a n b e seen f r o m T a b l e 6 . 2 the storm p r o d u c i n g the maximum peak


discharge off the c a t c h m e n t h a s an e x c e s s s t o r m d u r a t i o n o f 1.2 hours and

produces a peak d i s c h a r g e o f 2.70 cumecs. The complete r u n o f f h y d r o g r a p h


is obtained from Fig. 6.10 for a v a l u e of TD = 1.93. The hydrograph is
rendered dimensional by multiplying the two axes of Fig. 6.10 by the
v a l u e s g i v e n i n T a b l e 6.2 a n d i s shown in F i g . 6.17.

REFERENCES

C o n s t a n t i n i d e s , C.A. and S t e p h e n s o n , D., 1982. D i m e n s i o n l e s s h y d r o g r a p h s


using kinematic theory, Water Systems Research Programme, Report
5/1982, U n i v e r s i t y o f t h e W i t w a t e r s r a n d .
Kibler, D.F. a n d Woolhiser, D.A., 1970. The k i n e m a t i c cascade a s a
h y d r a u l i t model. Hydrol. paper 39, Colorado State U n i v e r s i t y , Fort
ColI ins.
O v e r t o n , D.E., 1972. Kinematic flow o n long impermeable planes, Water
Res. B u l l . 8 ( 6 ) .
Singh, V.P., 1975. Hydrid formulation of kinematic wave model of
w a t e r s h e d r u n o f f , J. H y d r o l . 27.
W o o d i n g , R.A., 1965. A h y d r a u l i c m o d e l f o r t h e c a t c h m e n t s t r e a m p r o b l e m ,
I I . N u m e r i c a l S o l u t i o n s . J . H y d r o l . 3.
Woolhiser, D.A., 1969. O v e r l a n d flow o n a c o n v e r g i n g surface. Trans.
Am. SOC. A g r . E n g r . 1 2 ( 4 ) , 460-462.
130

CHAPTER 7

STORM DYNAM I CS AND D I STR I BUT ION

DES I GN PRACT I C E

It is common practice to design stormwater systems for uniform


intensity, uniformly distributed, stationary storms. Lack of data often

makes any other basis for design difficult. There is little information
available on instantaneous precipitation rates, storm cell size and cell
movement. Time average precipitation r a t e or precipitation depth can be
predicted from intensity-duration-frequency curves (e.g. Van Wyk and
Midgley, 1966) o r equations such as that of Bell (1969). The most common
method of abstracting data from rainfall records is to select a duration
and calculate the maximum storm precipitation in that period. The so-
defined storm may include times of low rainfall intensity immediately
preceding a n d succeeding a more intense p r e c i p i t a t i o n r a t e .
Such simp1 i f i c a t i o n s in data render runoff calculation simplistic.
Even when employing numerical models it is simplest to use a uniform
intensity hyetograph for every point on the catchment. Although time
varying storms are sometimes used, the precipitation pattern is seldom
r e l a t e d to the maximum possible r u n o f f rate.
Warnings have been made against simplification in rainfal I
patterns. For example, James and Scheckenberger (1983) indicated that
storm movement can affect' the runoff hydrograph significantly. Eagleson
(1978) has expounded on the s p a t i a l variability of storms a n d Huff (1967)
studied the time v a r i a b i l i t y of storms.
Although much research has been done on storm variability,
relatively little has been published on the resulting effects on runoff
hydrographs (Stephenson, 1984). Research appears to h a v e concentrated on
models of particular (monitored) storms over particular catchments. The
design engineer or hydrologist does not have sufficient guidance as to
what storm pattern to design for. Presumably certain rainfal I sequences,

spatial variations and storm movement will result in a higher r a t e of


runoff than other rainfall patterns for a p a r t i c u l a r catchment. A p a r t from

an i n d i c a t i o n of what storm p a t t e r n produces the worst flood, one needs a n


indication of what storm pattern could be expected for the design
catchment. Such d a t a should be a v a i l a b l e on a frequency b a s i s i n o r d e r to
estimate the likelihood of the worst hyetograph shape, spatial storm
distribution and movements occurring. Although i s o l a t e d catchments have

been studied at many research centres considerably more information is


required for the country as a whole. Analysis and use of such data
131

i n different combinations would r e q u i r e many t r i a l s before the worst storm


patterns would emerge. An alternative approach is a deterministic one.
Before calculating runoff, the analyst determines the following in order
to select the correct design storm:

i) The storm duration. For small catchments this is usually equated


to the time of concentration of the catchment.
ii) Variation i n p r e c i p i t a t i o n r a t e d u r i n g the storm

iii) S p a t i a l d i s t r i b u t i o n of the storm; a n d the


iv) D i r e c t i o n a n d speed of movement of the storm.

The above information could be employed in numerical modelling


of the design storm. Alternatively, for minor structures, simplistic
methods such as the Rational method could be employed. Since data
shortage often limits the accuracy of modelling, the latter, manual
approach, is often sufficiently accurate. The guides presented below
may assist both the modeiler by providing information on which design
storm would produce the highest runoff rate and the formula orientated
s o l u t i o n b y p r o v i d i n g f a c t o r s to account f o r storm v a r i a b i l i t y .

STORM PATTERNS

Variation i n r a i n f a l l intensity during a storm

In order to understand the reasons for and extent of variability

(spatial and temporal) of rainfall, it is useful to describe the p h y s i c a l


process of cloud formation and precipitation. Convective storm clouds
originate from rising air masses. The size and shape of the rising air
mass depends on the topography and the air masses will usually b e of
smaller scale than the air mass which has been brought by advection
and which contains sufficient moisture for raindrops to precipitate.
Mader (1979) concluded from r a d a r o b s e r v a t i o n s of storms i n South A f r i c a
that storm areas, duration and movement were related to mean 500 mb
winds, thermal i n s t a b i l i t y a n d w i n d shear.

Most recorded hyetographs indicate that rainfalI intensity is


highest somewhere in the middle of the storm duration. Huff (1967)
presented e x t e n s i v e d a t a on rainfall rates for storms of varying intensity
indicating a time distribution somewhat between convex upward and
triangular. In order to create a hyetograph which could be used for
simple design of interconnecting stormwater conduits, Keifer and Chu

(1957) proposed an exponential distribution termed the Chicago storm.


132

The position of the peak intensity could be varied and was observed

to occur about 0.375 of the storm d u r a t i o n from the s t a r t .

mm/h
i

'
1

Fig. 7.1 Hyetograph w i t h peak n e a r b e g i n n i n g

Spat i a I d i s t r i b u t i o n

The nature of storm cells within a potential rain area has been
documented by many researchers e.g. Waymire and Gupta (1981). The
persistence of storms observed in the northern hemisphere has not been
found i n c o u n t r i e s south of the equator however ( C a r t e , 1979). The l a r g e r
air mass within which storm cells occur is referred to as the synoptic
area (see Fig. 7.2). The synoptic area can last for 1 to 3 days and
the size is generally greater than 104km2. W i t h i n the synoptic area are
large mesoscale areas (LMSA) of lo3 to lo4 km2 which have a l i f e of

several hours. Sometimes small mesoscale areas (SMSA) of lo2 to lo3 k m 2 can
e x i s t simultaneously. Within the mesoscale areas or sometimes on their
own, convective cells, which are r e g i o n s of cumulus convective precipi-
tation, exist. These may have a n a r e a extent of 10 to 30 km2 a n d h a v e
an average l i f e of several minutes to half an hour. These cells a r e of
concern to the hydrologist involved in stormwater design. By comparing

the storm cell size with the catchment size he can decide whether the
133

cell scale is significant in influencing spatial distribution over the

catchment. There may be o v e r l a p p i n g cells which could result i n greater


intensity of precipitation than for the single cells. Eagleson (1984)
investigated the statistics of storm cell occurrences in a catchment and
found the possibility of large storms can be computed assuming over-

l a p p i n g small storms.

Synoptic a r e a 7

Fig. 7.2 Areal d i s t r i b u t i o n of a convective storm

The shape of the storm cell has significance for catchments larger
than the cell. Scheckenberger (1984) indicates that the cells are ellip-
tical which may be related to storm movement. The rainfall intensity
is highest at the centre a n d decreases outwards. The intensity has been
shown to decrease exponentially, radially outwards from the focus, in
various localities as in Fig. 7.3 (Wilson et al., 1979). Generally the
variability i n i n t e n s i t y does not necessarily cause h i g h e r r u n o f f intensities
but on small catchments near the centre of the c e l l the average precipi-
tation can be higher than for a larger catchment, and as a rule, the
r a i n f a l l depth increases the smaller the storm area.

S t o r m movement

Clouds generally travel with the wind at their elevation. As the


rain falls it goes through lower speed wind movements so t h a t the most
significant speed is that of the clouds. The direction of lower winds
can also d i f f e r from the general d i r e c t i o n o f movement of the u p p e r s t r a t a .
This may be the reason Changnon and Vogel (1981) observed slightly
different directions for storm a n d c l o u d movements. Dixon (1977) a n a l y s e d

storm data and indicated storm cells have a circulation in addition to


a general f o r w a r d movement.
134

NUMERICAL MODELS

The effect of storm dynamics and distribution can be studied


numerically and the results for simple plane catchments are presented
below. The kinematic equations are employed in the numerical scheme.
Although these solutions a r e no s u b s t i t u t e f o r d e t a i l e d catchment m o d e l l i n g
when there are sufficient data, they do indicate which variables are
likely to be the most important in storm dynamics. It must be pointed
out that the following studies are simplified to the extent of assuming
constant speed storms with unvarying spatial distribution. True storms
a r e c o n s i d e r a b l y more complex as e x p l a i n e d i n the above reference.

Fig. 7.3 I l l u s t r a t i o n of s p a t i a l d i s t r i b u t i o n of p r e c i p i t a t i o n i n t e n s i t y

Kinematic equations

The one-dimensional kinematic equations are for a simple plane


catchment (Brakensiek, 1967):

The c o n t i n u i t y e q u a t i o n ; aat y + axa q = i and

Flow resistance; q =aym

y i s water depth on the plane, q is discharge rate per unit width of

plane, ie i s excess rainfall rate, t is time, x is longitudinal distance


down the p l a n e , a i s assumed a constant a n d m i s a coefficient. Employ-
ing the Manning discharge equation in S.I. units a = J(So)/n where
So is the slope of the plane, n is the Manning roughness coefficient,
a n d m i s 5/3.
135

The number of variables can be reduced to facilitate solution

by re-writing the equations in terms of the following dimensionless


variables:

x = x/L
T = t/tc
I = i / i
e a
Q = q/iaL
where L i s t h e l e n g t h o f o v e r l a n d flow, i a i s t h e t i m e and s p a c e a v e r a g e d
excess rainfall rate and tc is the time to equilibrium, or time of
concentration, for an average excess rainfall i
a
. Subscript c r e f e r s to
time of concentration, d to storm duration, a to t i m e and s p a c e a v e r a g e
and p to peak. Then the following expression for t can be derived:

In general the dimensionless variables are proportional to the

dimensioned variables. Thus Q is the proportion of maximum flow at


1 -m
equilibrium. Substituting y = (q/a) from the r e s i s t a n c e e q u a t i o n and
for X,T,I and Q from the equations for the dimensionless terms, the
following equation replaces the c o n t i n u i t y equation.

This single equation can be solved for Q i n s t e p s o f T and X f o r v a r i o u s


distributions of I and m = 5 / 3 .

Fig. 7.4 P l a n e r e c t a n g u l a r catchment s t u d i e d w i t h storm

Numerical Scheme

Although it appears a simple matter to replace differentials by


finite difference, there can be problems of convergence and speed of
solution unless the correct numerical scheme is employed. The simplest
136

finite difference schemes are explicit, employing values of Q at a

previous T to estimate new values at the next time T. This method is


not recommended as it i s often unstable when discontinuities in rainfall
intensity occur. Upstream differences are usually taken i n such schemes,
as downstream effects cannot be propagated upstream according to
Huggins and Burney (1982). I t is also necessary to limit the value of

AT/AX to ensure s t a b i l i t y .
Woolhiser (1977) documented various numerical schemes including
very accurate methods such as Lax-Wendroff ' s . Brakensiek (1967)
suggested 3 schemes: f o u r p o i n t , i m p l i c i t and e x p l i c i t . His second scheme
(implicit) is adopted here as it is accurate and r a p i d for the examples
chosen.

I M -'l M *
X
Fig. 7.5 X-T g r i d employed i n numerical solution

Employing the n o t a t i o n i n the g r i d i n F i g . 7.5,

aQ - Q1-Q2
ax AX
Q 1 + Q -Q -Q
-aQ- - 2 3 4
2T 2AT

Since aQ/ aT is not sensitive to Q2/5, ( t h e power is less than one),


Q2/5 i s approximated b y ((Q3+Q4)/2)2/5, i.e. an e x p l i c i t form i s employed
here or else the resulting equations would be difficult to solve. The

f i n i t e difference a p p r o x i m a t i o n to the d i f f e r e n t i a l e q u a t i o n i s t h u s :
137

Q +€I0 . 4 Q2 Q +Q -Q
(3( ) 3 4 2
+

solving for Q1: Q, = 5


- 1 +-)
3 2 AX 2A T

_
2AT
1
_
5/3
AX
- .
~ (-
Q3+Q4
2
)
0.4

Starting at the upstream e n d of t h e c a t c h m e n t w h e r e Q2 = 0 and r e p l a c i n g


Q2 at the next point by Q1 a t the previous point, all the variables on
the right hand s i d e a r e k n o w n a n d o n e c a n s o l v e f o r Q , . The dimension-
less time step used was 0.05. The difference for smaller time steps was
found b y t r i a l to b e u n n o t i c e a b l e .

SOLUT!ONS FOR DYNAMIC STORMS

Time varying storms

One of the most frequently used simplifying assumptions, but a

dangerous assumption, in many r a i n f a l I-runoff models is that of constant


precipitation rate throughout the storm duration. The temporal variation
of precipitation intensity for storms over Illinois was documented by
Huff (1967) whose findings were often extrapolated to other regions. He
suggested identifying the quartile of maximum precipitation and further
employing probabilities of the rains occurring sooner or later than the
median. Huff plotted his results as mass rainfall curves so it is not
easy to discern the shape of the hyetographs unless his curves are
differentiated with respect to time. In general they are found to be
convex upwards. Apart from Keifer and Chu's (1957) synthetic hyeto-
graph, evidence points to convex up hyetographs. The assumption of a
triangular hyetograph is thus extreme as a real storm would tend to
be less 'peaky' than a triangular one. The general triangular-shaped
rainfall rate versus time relationship depicted in Fig. 7.6 is therefore
studied. The time of the peak is v a r i e d ' between the start of the storm
(TP = 0 ) and t h e e n d ( T = 1).
P

'I

T 1 T
P
Fig. 7.6 Temporally v a r y i n g storm
138

Simple models of hyetographs assume a single peak in rainfall


intensity. Storms with m u l t i p l e major peaks can be synthesized from over-
l a p p i n g compound storms. It is a single peak-storm which i s considered
here a n d the time of the peak i n t e n s i t y p e r m i t t e d to v a r y .

Q
.
Constant exce'

0 1 2 T

F i g . 7.7 Simulated dimensionless hydrographs caused by storms with


time v a r y i n g rainfall intensities (Fig. 7.6) but the same t o t a l
precipitation

Design storms for flood estimation generally peak in intensity in


the f i r s t h a l f of the storm. This i s an a l l e v i a t i n g factor i n peak r u n o f f ,
as indicated in Fig. 7.7. That i s a p l o t of h y d r o g r a p h s from the simple
catchment depicted in Fig. 7.4 with various hyetographs imposed, i.e.
a rectangular hyetograph and triangular hyetographs with various peak
times were employed. The ordinate in Fig. 7.7 is the discharge rate
expressed as a fraction of the mean excess precipitation rate, and the
abscissa is time as a fraction of the time of concentration for a uniform
storm with precipitation r a t e equal to the mean r a t e over the storm f o r
each of the t r i a n g u l a r hyetographs.
It will be observed from F i g . 7.7 that if the storm intensity peaks
in the first part of its duration (T 50.5) the peak runoff is less t h a n
P
that for a uniform storm of the same average intensity. This holds for
peaks up to 80% of the duration after commencement of rain. Only for
the peak at the end of the storm (e.g. T = 1.0) does the peak runoff
P
exceed that for a uniform intensity storm. Then the peak runoff is
approximately 10% g r e a t e r than for a u n i f o r m storm of the same d u r a t i o n .
139

Q . Td= 0.4 0.6 0,8 1 1.2

I . _ . . . _ . _ . a

0 1 2 1

Fig. 7.8 Simulated dimensionless h y d r o g r a p h s caused b y late peaking


storms of constant volume a n d v a r y i n g d u r a t i o n

If the storm duration is not equal to the time of concentration for

a uniform storm however, the peak can be higher. Fig. 7.8 is for a
storm of constant volume peaking at its termination (TP = 1) and for
durations represented by Td = 0.4 to 1.2. These hydrographs are for
storms of equal volume so that the shorter duration storms are of a
higher intensity than longer d u r a t i o n storms. Depending on the IDF c u r v e
then a short duration storm may or may not result in a higher runoff
rate than for one of duration equal to the concentration time of the
catchment.
It should be recalled that all other hydrographs plotted are for
a specified excess rate of precipitation. That is, if the hyetograph is
uniform so are the abstractions. In practice, losses will be higher at

the beginning of a storm, resulting in a l a t e peak i n excess r a i n even


for a uniform precipitation rate. This has the same effect as a storm
peaking in the latter part as it increases the peak runoff. The effect
is compounded as a storm which peaks near the end will occur on a
relatively saturated catchment so a greater p r o p o r t i o n of the h i g h e r r a t e

of rain will appear as runoff n?ar the end. This tends to make the
excess rain versus time graph concave upwards if the hyetograph was

a straight-lined triangle. This effect is not modelled here but all the
effects result in a higher peak than for a uniform input. Scheckenberger
in fact indicates peaks up to 30% greater than for uniform storms due
to the sum of these effects.
140

Spatial v a r i a t i o n s

It appears that areal distribution of the storm is less effective


than temporal distribution in influencing peak runoff rate. Fig. 7.10
represents the simulated runoff from a 2-dimensional plane subjected to
various distributions of a steady excess rain. The storm duration was
made i n f i n i t e i n case the time to e q u i l i b r i u m exceeded the storm d u r a t i o n .
The spatial (or longitudinal in this case) distribution was assumed
triangular, the peak v a r y i n g from the top to the bottom of the catchment
as i n F i g . 7.9.

I
peak intensity
'P
=2

I,
_-
=1

F i g . 7.9 Catchment w i t h l o n g i t u d i n a l l y v a r y i n g storm

The same example would apply to a uniform intensity storm over


a wedge-shaped catchment, the catchment width increasing I inearly to
X a n d then decreasing l i n e a r l y towards the o u t l e t where X = 1 .
P

0 1 2 1
Fig. 7.10 Simulated dimensionless hydrographs caused by steady semi-
infinite storms of varying distribution down catchment (Fig.
7.9).
141

Fig. 7.10 depicts the resulting simulated hydrographs which indi-

cate that the runoff n e v e r exceeds that for a rectangular spatial distri-
bution of rainfall. The resulting dimensionless time to equilibrium is
nearly unity for all cases, implying t h e same t i m e o f concentration holds
for uneven distribution as for uniform distribution of rain. There is
therefore not a chance of a shorter duration storm with a higher
intensity contributing to a greater peak than the uniform storm (unless
the intensity-duration curve is a b n o r m a l l y s t e e p ) s i n c e t h e t i m e t o e q u i l -
i b r i u m i s n o t r e d u c e d r e l a t i v e to a u n i f o r m storm.

X=x/L
cotchmcnt lcngth

F i g . 7.11 C a t c h m e n t w i t h a s t o r m m o v i n g d o w n i t

9 1 1

1 2 1

F i g . 7.12 Simulated dimensionless hydrographs caused by unit steady


uniform storms moving down catchment at different speeds
( s e e F i g . 7.11)
142

Moving storms

Fig. 7.12 represents simulated hydrographs from a storm with a

constant p r e c i p i t a t i o n rate and spatially uniform travel I ing down the


catchment. The longitudinal extent of the storm cell is the same as the
l e n g t h of the catchment since i n general smaller area storms are reputed
to be more intense than larger cells. C is X/Tc or the speed divided
by the rate of concentration. For slow storms (CS 1) the dimensionless
hydrograph peak is unity while for faster storms the peak is less. The
faster storms do not fall on the catchment long enough to reach equi-
I i b r i um.

Q
1

cS=

0 1 ' 2 7

F i g . 7.13 Simulated dimensionless h y d r o g r a p h s caused b y steady u n i f o r m


s e m i - i n f i n i t e storms moving down catchment a t d i f f e r e n t speeds

Fig. 7.13 indicates there is also no increased peak for storms of


semi-infinite longitudinal extent ( n e v e r e n d i n g once they enter the catch-
ment). All peaks converge on unity and there is no peak greater than
unity. Thus movement does not appear to result in a hydrograph peak

g r e a t e r t h a n f o r a s t a t i o n a r y storm.
For storms of limited extent travelling up the catchment, the peak
flow was observed ' t o be less t h a n for a stationary storm a n d the f a s t e r
the speed of t r a v e l of the storm the s m a l l e r the peak r u n o f f .
143

I t has been demonstrated u s i n g numerical solutions to the kinematic


equations for simple catchments that non-uniformity in rainfall intensity
can affect peak runoff rates. Temporal variation in excess precipitation
rate can increase runoff r a t e above that for a steady r a t e of rain. Since
storms usually peak sometime after commencing and time diminishing
abstractions tend to cause a later peak in excess rainfall rate, the
assumption of steady rainfall can be dangerous as peak runoff is
u n derest imated.
Uneven spatial distribution of a storm does not directly contribute
to a higher peak runoff unless it results in a shorter duration storm
being the design storm. Storm movement reduces the peak flow unless
the movement is down-catchment, when this model shows no change in
peak runoff rate. A smaller, more intense storm than the one to e q u i l i -
brium for t h e catchment may however result in a h i g h e r p e a k r u n o f f r a t e .

REFERENCES

B r a k e n s i e k , D.L., 1967. K i n e m a t i c f l o o d r o u t i n g . T r a n s Amer. SOC. A g r i c .


E n g s . l O ( 3 ) p 340-343.
Be1 I , F.C., 1969. General ized r a i n f a l I-durat ion-frequency relationships.
P r o c . Amer. SOC. C i v i l E n g r s . 95 ( H Y 1 ) 6537, p 311-327.
Carte, A.E. 1979. Sustained storms o n t h e T r a n s v a a l H i g h v e l d . S.A.
Geogr. J o u r n a l , 6 1 ( 1 ) p. 39-56.
Changnon, S.A. and V o g e l , J.L., 1981. Hydrocl imatological character-
i s t i c s of i s o l a t e d s e v e r e r a i n s t o r m s . Water Resources Research 17(6)
p 1694-1700.
Dixon, M.J., 1977. Proposed Mathematical Model f o r t he E s t i m a t i o n o f
Areal P r o p e r t i e s of H i g h D e n s i t y Short D u r a t i o n Storms. Dept. Water
A f f a i r s , T e c h . R e p t . TR78, P r e t o r i a .
E a g l e s o n , P.S., 1978. C l i m a t e , s o i l and v e g e t a t i o n . 2. T h e d i s t r i b u t i o n
o f a n n u a l p r e c i p i t a t i o n d e r i v e d f r o m o b s e r v e d s t or m sequences. Water
R e s o u r c e s R e s e a r c h 1 4 ( 5 ) p 713-721.
E a g l e s o n , P.S., 1984. T h e d i s t r i b u t i o n o f c a t c h m e n t c o v e r a g e b y s t a t i o n a r y
rainstorms. W a t e r R e s o u r c e s R e s e a r c h , 2 0 ( 5 ) P 581-590.
H u f f , F.A., 1967. Time d i s t r i b u t i o n o f r a i n f a l l in h e a v y storms. Water
R e s o u r c e s R e s e a r c h , 3 ( 1 4 ) p 1007-1019.
Huggins, L.F. and B i r r n e y , J.R., 1982. Surface runoff, s t o r a g e and
routing. I n H y d r o l o g i c M o d e l l i n g of S m a l l W a t e r s h e d s . Ed. H a a n , C . T . ,
J o h n s o n , H.P. and B r a k e n s i e k , D.L., Amer. SOC. A g r i c . E n g r s . Mono-
g r a p h No.5.
James, W. and S c h e c k e n b e r g e r , R., 1983. S t o r m d y n a m i c s m o d e l f o r urban
runoff. I n t l . S y m p . U r b a n H y d r o l o g y , H y d r a u l i c s and S e d i m e n t c o n t r o l ,
L e x i n g t o n , K e n t u c k y . p 11-18.
K e i f e r , C.J. and C h u , H.H. 1957. S y n t h e t i c s t o r m p a t t e r n s f o r d r a i n a g e
d e s i g n . P r o c . Amer. SOC. C i v i l E n g r s . 83 ( H Y 4 ) p 1332-1352,
Mader, G.N., 1979. Numerical s t u d y of storms in the T r a n s v a a l . S.A.
G e o g r . J o u r n a l , 6 1 ( 2 ) p 85-98.
N a t u r a l E n v i r o n m e n t R e s e a r c h C o u n c i I, 1975. F l o o d S t u d i e s R e p o r t , V o l .
1. H y d r o l o g i c a l Studies, London, 5 volumes.
Scheckenberger, R., 1984. Dynamic s p a t i a l l y v a r i a b l e r a i n f a l I models
for stormwater management. M. Eng. Report, McMaster University,
H a m i I ton.
144

Stephenson, D., 1984. Kinematic study of effects of storm dynamics on


r u n o f f h y d r o g r a p h s . Water S.A. October, Vol. 10, No. 4, p p 189-196.
Van Wyk, W. a n d M i d g l e y , D.C., 1966. Storm s t u d i e s i n S.A. - Smal I
area, h i g h i n t e n s i t y r a i n f a l l . The C i v i l Eng. i n S.A., June, Vol. 8
No.6, p 188-197.
Waymire, E. a n d Gupta, V.L. 1981. The mathematical s t r u c t u r e of r a i n f a l l
representations 3, Some a p p l i c a t i o n s of the p o i n t process theory to
r a i n f a l I processes. Water Resources Research, 1 7 ( 5 ) , p 1287-1294.
Wilson, C.B., Valdes, J.B. a n d Rodrigues, 1 . 1 . ) 1979. On the i n f l u e n c e
of the s p a t i a l d i s t r i b u t i o n of r a i n f a l l i n storm r u n o f f . Water Resources
Research, 1 5 ( 2 ) , p 321-328.
Woolhiser, D.A., 1977. Unsteady f r e e s u r f a c e flow problems. I n Math-
ematical Models f o r Surface Water Hydrology. Ed. by C i r i a n i , T . A .
Maione, U. a n d Wal I i s , J.R., John Wiley G Sons, 423 p p .
145

CHAPTER 8

CONDUIT FLOW

K I N E M A T I C EQUATIONS FOR NON-RECTANGULAR SECTIONS

The analysis of flow in conduits is more complicated than for


overland flow on account of side friction. Non-rectangular cross sections
e.g. trapezoids and circular drains are more difficult than rectangular
sections to analyze. Surface width and hydraulic radius become a

function of water depth. The sides of the channel (and top in the case
of closed c o n d u i t s ) increase f r i c t i o n drag. As far as the form of the
basic kinematic equations is concerned the mathematical expressions
become more complicated, and numerical solutions are necessary in the
m a j o r i y y o f cases.
The c o n t i n u i t y equation remains

o r e x p a n d i n g the second term,

where the first term is the rate of rise, the second prism storage and
t h e t h i r d wedge s t o r a g e .
The dynamic e q u a t i o n reduces to
M
Q = aAR (8.3)
where Q is the discharge rate, a IS a function of conduit roughness,
q is inflow per unit length, B is the surface width, A is the cross
sectional area of flow and R is the hydraulic radius A/P where P is
the wetted perimeter. Employing Manning's f r i c t i o n equation,
LIZ
a = K S /n a n d M = 2/3 (8.4)
1
where K1 = l(S.1. u n i t s ) and 1.486 (it-sec units)
n = M a n n i n g ' s roughness coefficient

Owing to the greater depths in conduits i n comparison w i t h overland

flow, lower values of n are applicable. The above equations can be


solved f o r special cases of non r e c t a n g u l a r c o n d u i t s a s i n d i c a t e d below.

PART-FULL C I RCULAR P I PES

The cross sectional area of flow in a circular conduit (Fig. 8.1)


running part full (Stephenson, 1981) i s
146

A = -D 2 a . a
( -0 -cos-sin-)
4 2 2 2

a n d P = DO
2
Thus i f one takes 0 as the variable, the continuity equation becomes

aA -a o
_ aa
-
ax
+

aa at = q ;

and

’; (1 + s i n 2 ~ - cos2 0 ao i@ =
(8.7)
2 %)at+ ax

I n f i n i t e difference form, solving for 0 a f t e r a time i n t e r v a l A t ,

a = o + ( q - -G Q) 8 Gt
AX
2 1 ~ 2 ( l + s i n 2 5- cos’g) (8.8)
2 2
a n d i n terms of t h e new , s i n c e 61 = aARZ3

0 .a
. a cos-sin- 2 3
a -D2 0 - 2 2
Q =
4
(3 cos-sln-)
2 2
{ z(l-
o’} (8.9)

I n order to simulate flow and depth variations in pipes, the latter


two equations are applied at successive points for successive time
i n t e r v a Is.
In addition to analysis of flows in pipes, the methods can be
applied to design by successive analysis. When designing storm drain
collection systems there are many approaches (Yen and Sevuk, 1975).
It is in normal practice not necessary to c o n s i d e r surcharged conditions
i n a d e s i g n u n l e s s a d u a l system ( m a j o r a n d m i n o r c o n d u i t s ) i s employed.
If pipes are designed to run just full at their design capacity, then
they will run part full for any other design storm duration. The higher
up the leg a pipe length is, the shorter will b e the concentration time,
or time to flow equilibrium. The design storm duration will equal the
concentration time of the drains down to the pipe in question. Any
subsequent pipes wi II have larger concentration times and consequently
a l o w e r storm i n t e n s i t y .

Fig. 8.1 Cross section t h r o u g h p a r t - f u l l pipe


147

COMPUTER PROGRAM FOR DESIGN O F STORM DRAIN NETWORK

The preceding scheme was employed in a program for analysing


the flow in each pipe in a drainage network the plan of which is
specified by the designer. The engineer must pre-select the layout, sub-
division of catchment, position of inlets and grades. The grades will
in general conform to the slope of the ground.
It is necessary to simulate overland flow and each upper drain
in order to size any lower drain. Such analysis can only be done
practically by digital computer using numerical solutions of the flow
equations. Many calculations are necessary for complex networks. A
limitation on the maximum time interval for numerical stability implies
many iterations until equilibrium flow conditions are reached for each

pipe design. In addition, a number of different storm durations must


be investigated for each pipe. A s i m p l e and e f f i c i e n t iterative procedure
was therefore sought in order to minimize computer time. The kinematic
form of the flow equation was employed to ensure this. The emphasis

throughout the program is simplicity of data input and minimization of


computational effort. Some accuracy is sacrificed by the simplifications

but the overriding assumption of precipitation pattern is probably more


important.

The design method (Stephenson, 1980) p r o c e e d s f o r successive pipes,


the diameters of which are calculated previously. It is assumed the
network layout is specified,. and the pipe grades are dictated by the
g r o u n d slope. Starting at t h e top e n d s o f a d r a i n a g e system, the program
sizes successively lower pipes. Thereby each pipe upstream of the one
to be designed is pre-defined. It is necessary to investigate storms of
different duration and corresponding intensity of flow to determine the
d e s i g n s t o r m r e s u l t i n g i n m a x i m u m f l o w for t h e n e x t p i p e .
It is assumed that the design storm recurrence interval is pre-
selected. The intensity-duration relationship is then assumed to be of
the form

i
e
= a
b + t d (8.10)

By selecting storms of varying duration td, and simulating the


flow buildup down the drains, the program can select a storm which
will result in the maximum peak flow from the lower end of t h e system.
That discharge is the one to use for sizing the next lower pipe. Thus
the program proceeds from pipe to pipe until the entire network is

designed.
148

The program is limited in application to selection of drain pipe


diameters for a simple gravity collecting system, and uses kinematic
theory and the limitations of the theory should be recalled. It should
be noted that for major pipes it may become necessary to allow for
backwater and routing effects (Barnes, 1967). The program does not
optimize the layout (Argamon et al. 1973; Merritt and Bogan, 1973). Nor
is surcharge (Martin and King, 1981) or detention storage considered
here.

*
\ drain
L s u b c o t c h m e n t boundary

4 ‘drain number

F i g . 8.2 Layout p l a n of d r a i n a g e network sized i n example

Program description

Pipes are assumed to flow initially at a depth corresponding to


a subtended angle of 0.2 radians a t the centre. The corresponding flow
is very low, but this assumption avoids an anomaly for the case of
zero depth when the numerical solution of the explicit equation is
impossi b le.
I n f l o w from subcatchments i s assumed to occur a l o n g the f u l l length

of the respective pipe, i.e. subcatchment breadth i s assumed to be equal


to pipe length. This affects overland flow time to some extent. If
necessary (if flow is sensitive to storm duration) the subcatchment
friction factor could be adjusted to give the correct o v e r l a n d flow time.
The computer program, written i n FORTRAN f o r use i n conversational

mode on a terminal connected to an IBM 370 machine, i s appended. The


149

input format is described below. Data is read in free format and can

be i n p u t on a t e r m i n a l as the program stands.

F i r s t l i n e of d a t a :
M, A, B, E , IN, IR, 11, G.
Second and subsequent lines of data (one l i n e f o r each l e n g t h of pipe):
x(I), s(I), z(I), C(I), SO(I), EO(I), I B ( I ) .
The i n p u t symbols a r e e x p l a i n e d below:

M - The number of pipes: the number of p i p e s should be minimized


f o r computational cost m i n i m i z a t i o n . F o r computational accuracy
the pipes should be divided into lengths of the same order
of magnitude. It i s convenient to make the p i p e lengths equal
to the distance between inlets. Inlets between 10 a n d 200m
apart are normally sufficient for computational accuracy.
There should be a t least two p i p e s i n the system.

A,B - Precipitation rate i is calculated from an equation of the


form i = A/(B + td) where td is the storm duration and B
is a regional constant (both in seconds). A is a function
of storm return period and catchment location and its units
are in m if SI units are used, and f t i f ft-lb-sec units are
used.

E - Pipe roughness. This is analogous to the N i k u r a d s e roughness

and E is measured in m or ft. It i s assumed i n the program


that all p i p e s h a v e the same roughness. A conservative f i g u r e
of at least 0.001 m (0.003 ft) is suggested to account for
surface deterioration with time due to erosion, corrosion or
deDos i ts .
IN, - For each pipe sizing computation various storm durations
IR are investigated, r a n g i n g from IUl to IU2 i n steps of IR (all
in seconds). The smallest storm duration IU1 is set equal

to the overland flow time for an upper p i p e of the previous


pipe design storm duration for subsequent pipes down a leg.
The number of storm durations investigated is specified by

I N and the increment in trial storm duration is specified by


IR. Thus IU2 = IU1 + IN";II. The accuracy of the computations

is affected by the number of trial storm durations. A value


150

of IN between 3 and 10 is usually satisfactory. The upper


limit can be estimated beforehand from experience o r by trial
(if all design storm durations turn out to be less than the
IU2 s p e c i f i e d then the I N selected i s s a t i s f a c t o r y ) .

The computational time a n d cost i s affected by the time i n c r e -


ment of computations II (seconds). The maximum possible
value is dependent on the numerical stability of the compu-
tations. A v a l u e e q u a l to the minimum v a l u e of

w i l l normally be satisfactory (of the order of 60 to 300


seconds) .
Gravitational acceleration (9.8 in SI units a n d 32.2 i n ft-sec
units).

The pipe data are next read in line by line for M pipes.
As the program stands, 98 individual pipes are permitted,
and any number of legs subject to the maximum number of
pipes.

The pipe length in m or ft, whichever units are used. An


upper limit on individual pipes of 200m is suggested for
computational accuracy and a lower limit of 10m f o r o p t i m i z i n g
computer time.'
The slope of the p i p e i n m p e r m o r f t p e r f t .
The surface area contributing runoff to the pipe in m or ft
The proportion of precipitation which runs off (analogous
to the 'C' i n the Rational f o r m u l a ) .
The o v e r l a n d slope of the c o n t r i b u t i n g a r e a , towards the inlet
a t the head of the p i p e .
EO( 1 ) The equivalent roughness of the overland area in m or ft
depending on u n i t s employed.
IB(I) The number of the pipe which is a branch into the head of
pipe I .

For no b r a n c h , put IB( I ) = 0

For a header p i p e a t the top of a leg, put IB(I) = -1.


Only one b r a n c h p i p e p e r i n l e t i s permitted.
More must be accommodated by inserting short dummy pipes

between.
The order in which pipes are tabulated should be obtained
as f o l l o w s :
151

Computer Program for Storm Network P i p e Sizing

L.OOO1
L.0002
L.0003
L.0004 5
L.0005 10
L.0006
L.0007
L.0008
L. 0 0 0 9
L.OO1O
L.OO1l
L.0012 12
L.0013
L.0014
L.0015 13
~ . 0 0 1 6I 5
L.0017
L.OOI~
L.0019
L .0520
1.0021
L.0022
L.5023
L.5024
L.0025 2 0
~.0026
L.OU27
~.502a
L.0029
L.0030
L.UJ31 23
L.0032
L.UJ33
L.0024
L.0533
L.3036
L.i)037
L.UO3J
L.0039
L.0040
L.0041
L.0042 30
L.0063
L.JU44
L.0045
L.0040
L.JJ47
L.304d 32
L.0049
L.305S 35
L.J05l 40
L.>il52 45
L.dOb3 5u
L.3054
L.3055
L.5056 100
L.3057
L.5058 110
L.0054 120
L.0060
Led061
L.SO62
L-SJt3
L.0064 201)
L.0065 44(3./4. I
L.0066
L.0067
L.0068
L.0069 290
L.0070 300
L.0071
L.0072 350
L.0073
L.0074 6 0
L.0075
L.0076 400
L.5077 7 0
L.0078
L.0079 6 0
L.00eO S I O P
L.0081 END

L.5001 S T U d M SEWER C € S I G N
L.0002 P I P E LENGTh C I A C R A O E DSFLC/S STORM S AREA
~.0003 I 100. -576 .0020 -244 1016. 20006.
L.OO0'4 2 150. -514 - 0 0 4 C .155 Y11. 20000.
L.JJU> 3 200. - 6 4 3 -0040 .162 206d. 40000.
L.0306 4 100. -415 .0020 -102 772. 10000+
~.0007 5 100. .574 .0040 .342 20od. 40000.
~.u006 6 200- .613 .0040 .u17 2068. 10000.
L.00U'i 7 2CO. - 2 5 3 .0020 .o9b 2068. 40000.
1.i) J 10 @ 100. - 5 0 5 .0050 1.287 2068. 20000.
LaOJ11 DATA @ .0751440..0010 2563 301) 60
152

After drawing out a plan of the catchment with each pipe, the
longest leg possible is marked, starting from the outfall, then success-
ively shorter legs on first the longest, then successively shorter pipes.

Now the pipes are numbered in the reverse over, starting at the top
of the shortest leg etc. Proceed down each leg w i t h the numbering until
a junction is reached. Never proceed past a branch which has not been
tabulated previously. In this way all pipes leading into a pipe will
have had their diameters calculated before the next lower pipe is
designed.

Sample Input

The d a t a a r e i n metres a n d a r e taken from F i g . 8.2

8 .075 1440 .001 3 300 60 9.8


100 .002 20000 .4 .005 .01 -1

150 .004 20000 .4 .003 .Ol -1

200 .004 40000 .4 .003 .Ol 1

100 .002 10000 .3 .005 .02 -1

100 .004 40000 .4 .003 .Ol -1

200 .004 10000 .5 .005 .01 4

200 .002 ,40000 .4 .002 .01 0

100 .005 20000 .4 .003 .01 3

TRAPEZOIDAL CHANNELS

F i g . 8.3 Trapezoidal channel geometry


153

For trapezoidal channels the h y d r a u l i c equations become

A = Y (b + y/S1 + Y/SI)
P = b + y 1 J(1 + 1/S12) + i ( l + 1/S2’))l
In particular for a vertical sided rectangular channel of limited width
b, employing the M a n n i n g equation.

A yb,
P = b + 2 y

Q - ayb yb )23
(b+2y
= ~ i ( y b ) ~/ (’ b~+ 2 y I Z 3

The analysis of flow in channels must generally be done numer-


ically. The channel is divided into reaches and a suitable time step
selected to simulate flow and depth variations. The continuity and
friction equations are applied conjunctively to c a l c u l a t e increase i n water
depth and flow rate respectively. The method can be employed for
catchment channel flow simulations. Many natural channels can be
approximated by a trapezoid, or else a number of trapezoids. A channel
plus flood plane can be represented by two trapezoids at different bed
levels, the flood plane being at the top of the banks of the channel.
The roughness, and hydraulic radius, a n d consequently the v e l o c i t y will
d i f f e r from channel to overbank a n d t h i s can be accounted f o r .

COMPAR I SON OF K I NEMAT I C AND T I ME-SH I FT ROUT I N G I N CONDU I TS

Whereas overland flow time lag may be predicted quite differently


using kinematic or time lag methods, in the case of conduits, time lag
often provides a sufficiently a c c u r a t e assessment of flow. That is, owing

to the confined cross section of a conduit, flow is more inclined to


emerge at the same rate that it enters a conduit, and travel time
approximates r e a c t i o n time s u f f i c i e n t l y w e l l .
In stormwater drainage, runoff hydrographs from overland flow
c o n s t i t u t e the essential input to h y d r a u l i c c o n d u i t s ; e.g. pipes, channels,
culverts etc. The overland flow hydrographs are attenuated further as
they travel through the conduits. In a stormwater drainage network,
where c o n d u i t s a n d manholes a r e i n t e r l i n k e d to c a r r y water from d i f f e r e n t
subcatchments onto a major outlet, hydrograph attenuation through the
conduits is very important. Hydrographs from conduits leading to the
same manhole have to be summated for designing hydraulic structures
or conduits downstream or for studying the behaviour of an existing

network under c e r t a i n conditions. T h e magni tude of the h y d r o g r a p h peaks


154

as well as their relative time positions are important for the accurate

assessment of design flows.


Various methods exist for routing runoff hydrographs t h r o u g h closed
conduits. The most commonly used are time shift methods. A time shift
method s h i f t s the e n t i r e h y d r o g r a p h in time without a n y storage consider-
ations for attenuation. The time shift or lag time is calculated by
dividing the length of the conduit by the velocity of the water in the
conduit. This velocity is usually taken to be the velocity of water in
the c o n d u i t when the c o n d u i t i s almost f u l l under steady c o n d i t i o n s .
Storage balance methods a r e also used f o r routing. They apply mass
b a l a n c e equations across the c o n d u i t . Such equations a r e solved b y e i t h e r

explicit or implicit schemes. Both time s h i f t a n d storage r o u t i n g methods


ignore non-uniform flow a n d dynamic effects in the system. Other methods
for hydrograph routing include routing through conduits using the
kinemat i t equations o r even the dynamic equations of flow.
The use of the kinemat ic equations for r o u t i n g r e q u i r e s comparat i v e l y

large computational effort in comparison with time s h i f t as the equations


have to be solved a t close g r i d p o i n t s a l o n g the c o n d u i t over short time
increments. Most existing drainage models use time shift methods and
since the solution of the kinematic equation is tedious it may in some
cases be u n w a r r a n t e d .

Section Geometry and E q u a t i o n s for C o n d u i t s

Two section configurations are studied here, one a circular section


and the other a trapezoid. Both sections are assumed to be partly full
as dynamic effects of the system are not studied. For the pipe this
implies that the depth of flow is always less than the pipe diameter
w h i l e for the trapezoid i t s sides a r e assumed to be h i g h enough to a l l o w
a n y depth of water.
For partially filled closed conduits, i.e. where no lateral inflow
e x i s t s a l o n g the c o n d u i t , the k i n e m a t i c c o n t i n u i t y equation i s :

( 8 .1 1 )

where q is discharge (m3/s), a is cross sectional area of flow (m’),


x is distance along the conduit from the inlet (m) and t is time (5).

In kinematic theory discharge can be assumed to be a function of


flow depth as the f r i c t i o n slope i s assumed to equal the bed slope. This
enables the use of u n i f o r m flow equations expressed i n terms of bed slope
instead of friction slope. Such equations are usually described in the

fol lowing form:


155

(a) Pipe

I
(b) Trapezoid

Fig. 8.4 C o n d u i t Sections

m-1
q = a a R (8.12)

where c( and m are friction flow coefficients depending, on the uniform


flow equation used, R is the hydraulic radius of the section, i.e. a/p
( m ) a n d p i s the wetted p e r i m e t e r of the section ( m ) .

0 = 1
n
5 112 a n d m = 5/3 (8.13)

where n = M a n n i n g ' s roughness c o e f f i c i e n t a n d S = b e d slope.


I n s e r t i n g the v a l u e s of a a n d m from 8.13 i n e q u a t i o n 8.12 yields:

1
q = - s
4 .5/3
(8.14)
p2-/'3
The geometry of the c o n d u i t s i s d e s c r i b e d b y e q u a t i o n s 8.15 - 8.18
I56

(8.15)

(8.16)

A = by + y2 t a n (90 - 0 (8.17)
Trapezoid
P = b + 2 y sec (90 - 0) (8.18)

The equations 8.11 and 8.14 were reduced to a dimensionless form


by Constantinides (1983) with the choice of suitable variables. The
dimensionless equations are then solved for different conduit sections
and input hydrographs. The kinematic equations are solved in their
dimensionless form to facilitate generalization of results in terms of
constant parameters that are functions of the input parameters. The
use of the dimensionless equations reduces computational effort as the
number o f cases t o s t u d y r e d u c e s g r e a t l y .
The variables q, a, x and t are reduced to the dimensionless
variables Q, A, X and T by dividing them b y appropriate variables with
i d e n t i c a l u n i t s a s f o l lows:
F o r the p i p e ,
Q = q/qm (8.19)
A = a/d2 (8.20)
P = p/d (8.21)
Y = y/d (8.22)

For the t r a p e z o i d ,
Q = q/qc (8.23)
A = a/b2 (8.24)
P = p/b (8.25)
Y = y/b (8.26)

a n d f o r both sections
x = x/L (8.27)
T = t/t (8.28)
k

where q i s t h e maximum f l o w c a p a c i t y of t h e p i p e ( m 3 / s ) , 7
0 . 335285~h d 8 / 3
m
qc is a discharge variable, being a function of
f r i c t i o n coefficients
0 , m a n d bottom w i d t h of t r a p e z o i d , b (m’/s) i.e. qc = 5 112 b 8 1 31
n tk ’
i s a t i m e c o n s t a n t ( 5 ) a n d L i s t h e l e n g t h of t h e c o n d u i t (rn).
To define the discharge and time constants appropriately the

dimensionless kinematic equations are obtained by substituting the


dimensionless variables in the continuity equation i.e. for the pipe,

(8.29)
157

Rearranging yields:

(8.30)

Furthermore by defining the time constant as in equation 8.31


reduces equation 8.30 to the dimensionless equation 8.33. Similarly for
the t r a p e z o i d the time constant i s d e f i n e d i n e q u a t i o n 8.32.
For the pipe:

t = -Ld2
(8.31)
qm
For t h e t r a p e z o i d :

t = -LbZ
(8.32)
qc
where the d i m e n s i o n a l c o n t i n u i t y equation i s :

-
aQ aA -
a x + aT
- - (8.33)

Similarly, t h e u n i f o r m f l o w e q u a t i o n 8.14 can be r e d u c e d to i t s dimension-


less form, i.e.

f o r the p i p e :

(8.34)

w h e r e t h e m a x i m u m c a r r y i n g c a p a c i t y of a p i p e c a n b e s h o w n to b e

qm = 0.335282 Sf d8/3 (8.35)

S u b s t i t u t i n g i n e q u a t i o n 8.34 and rearranging yields:


A5/3
1 - (8.36)
Q =
0.335282 p2/3

F o r t h e t r a p e z o i d t h e u n i f o r m f l o w e q u a t i o n r e d u c e s to:
Q q c = -1S z - ( A b ‘ ) 5 / 3 (8.37)
(Pd)2/3
Defining q as in equation 8.38 reduced equation 8.37 to the dimension-
C
l e s s f l o w e q u a t i o n f o r t h e t r a p e z o i d g i v e n i n e q u a t i o n 8.39:

‘c-
- tn ,t b8/3 (8.38)

For the t r a p e z o i d :
A 5/3
Q = - (8.39)
2
,3
/

Equations for t for both sections can be obtained by substituting


k
equation 8.35 and 8.38 into equation 8.31 and 8.32. Similarly, for
obtaining the dimensionless area and perimeter variables (A and P)
equations 8.20, 8.21 and 8.22 are substituted i n e q u a t i o n s 8.15 t o 8.18.
The r e s u l t i n g e x p r e s s i o n s a r e summarised below
158

Pipe
1
A = - cos
4
-1
(1-2Y) - (-1 - Y )
2
. (‘f - Y2)1/2 (8.40)
-1
P = cos (1-2Y) (8.41)
L
t = (8.42)

Channel
A = Y + Y2 tan (90-0) (8.43)
P = 1 + 2Y sec (90-0) (8.44)

(8.45)

Two shapes of inflow hydrographs are routed through the conduits,

one a uniform and the other a triangular time distribution. These two
time distributions were chosen as they represent extreme cases, i.e. a
natural runoff hydrograph, from overland flow, would have a shape
between these two extremes depending on the rainfall and catchment
c h a r a c t e r i st i cs.
In addition to the shapes the hydrographs were assumed to have
a variety of durations and intensities. Fig. 8.5 illustrates the inflow
I
h y d r o g r a p h s i n t h e i r dimensionless form.

QIM ‘IM

QIM is the maximum discharge factor or inflow factor and TD = td/tk,

where t d i s d u r a t i o n .

Fig. 8.5 D i f f e r e n t dimensionless i n f l o w h y d r o g r a p h s


159

The dimensionless e q u a t i o n f o r speed of p r o p a g a t i o n i s

_
dT
- C1
(8.46)

where C 1 = 0.335262 f o r the p i p e


= 1.0 f o r the t r a p e z o i d
C1
where f o r p i p e s :
3A
TP= 2(Y - Y Z ) (8.47)

a n d for t rapezoi ds :

aA
2P
- ’
2
cos (90-0) + Y sin (90-0) (8.48)

Equation 8.46 is a function of the depth coefficient, Y. it was solved


in terms of Y u s i n g a computer model.

2. 4
PIPE

2. 0 -

DEPTH/DIhMETER

F i g . 8.6 Dimensionless propagation speed of a disturbance in partially


f i l l e d pipes
160
d
0
cu
4

(0
4
4

-t
N
4

0
...
.LP/XP

0
0
C
(0
9
N
d

Fig. 8.7 Dimensionless propagation speed of a d i s t u r b a n c e i n trapezoids


Typical results are given i n Fig. 8.6 a n d 8.7. I t c a n b e e n seen f r o m

Fig. 8.6 the maximum dimensionless propagation speed i n a p i p e i s 1.63


a n d occurs when the depth over diameter ratio i s 0.62. Fig. 8.7 shows
that for the trapezoid the dimensionless p r o p a g a t i o n speed increases with
an increase in the depth over bottom w i d t h r a t i o . It i s necessary to know
for both the p i p e and the t r a p e z o i d t h e m a x i m u m d e p t h o v e r d i a m e t e r and
d e p t h o v e r b o t t o m w i d t h r a t i o s r e s p e c t i v e l y in o r d e r t o assess t h e m a x i m u m
propagation speed, (dX/dT)m, during any single simulation. I t should be
noted that for the pipe any simulation, where the depth over diameter
r a t i o exceeds 0.62, will have a (dX/dT) of 1.63.
m
The maximum depth of flow in the c o n d u i t to be encountered during
simulation will be a function of the maximum inflow discharge at the
inlet, as the hydrograph will attenuate as it travels away from the

inlet. The maximum dimensionless depth of flow in the conduit (Y) is


related to the maximum dimensionless inflow discharge, or inflow factor
(Q ), b y e q u a t i o n s 8.36 a n d 8.39.
IM
Equation 8.46 yields ma ximu m p r o p a g a t i o n speeds for different inflow
factors.

Computer S i m u I a t ion

A computer model was developed for solving the dimensionless kine-


matic equations for closed conduits. The model r o u t e s dimensionless inflow
hydrographs through the c o n d u i t s to produce dimensionless outflow hydro-
graphs at the o u t let. The dimensionless hydrographs were then studied
to evaluate the effects that a section of fixed geometry and length has
in a t t e n t u a t ing inflow hydrographs of varying discharge and duration.
For every inflow factor and inflow hydrograph distribution different
dimensionless storm durations were assumed. The dimensionless storm
durations were assumed. The dimensionless storm durations are defined
as the storm d u r a t i o n o v e r the time constant ratio, i.e.
‘d
TD = - (8.49)
t
k
V a l u e s o f TD varied from 0.2 to 10 according to the inflow time
d i s t r i b u t i o n and s e c t i o n t y p e . The f o l l o w i n g observations a r e made :
a) Simulations indicated the lag time of the outflow relative to the
inflow hydrograph decreases with an increase of inflow hydrograph
duration (for a constant inflow factor). The reason for this is that
longer duration inflows imply higher inflow volumes. Hydrographs with
lower volumes tend to spread more within the conduit r e s u l t i n g in lower

water depths which in turn result in lower flow velocities and p r o p a g a -


I62

tion of disturbance speeds. This inevitably increases their lag time.

The same argument e x p l a i n s the second o b s e r v a t i o n , i.e.

b) The ratio of peak at the outlet over peak at the inlet increases
with increasing storm duration ( f o r a constant inflow factor) or i n other
words inflow hydrographs of smal l e r storm duration undergo higher
discharge attenuation than hydrographs of longer duration. The reason
for this is the same as in a), i.e. lower volumes spread more than
bigger volumes resulting in lower depths of flow and thus lower dis-
charges.

C) The lag time for an inflow hydrograph of fixed duration decreases


with higher inflow factors. The reason for this is identical to a) as

h i g h e r i n f l o w f a c t o r s i m p l y h i g h e r volumes of water.

d) Peak flow attenuation is higher for small inflow factors (for a

constant inflow duration) than for high inflow factors, the reason b e i n g
the same as f o r o b s e r v a t i o n b ) .

Further deductions from the results can be made by representing

the printout results in the form of graphs. This i s done in subsequent


sect ions.

C r i t e r i a for choosing between Time S h i f t and Kinematic R o u t i n g

One of the main objectives of this study was to develop a method


for assessing whether time shift methods can be used without having
to resort to routing methods. The main assumption behind time shift
methods i s the preservation of the hydrograph ordinates without any
attenuation. To accept time shift methods, therefore, the hydrograph
attenuation that would happen in a real life situation must be small.

One must therefore decide what are acceptable limits of attenuation.


As hydrograph attenuation differs throughout the hydrograph duration
one usually refers to peak attenuation. In this study a peak attenua-
tion of 10% i s taken to be the maximum peak attenuation that can be
ignored. This value, although arbitrarily defined, is based on the fact
that more accurate determination of runoff is not justified due to the
corresponding inaccuracies in input determination. Furthermore, in a

drainage system consisting of various conduits interlinked in a network,


tolerating a b i g g e r peak a t t e n u a t i o n can result in a gross overestimation
of the outflow peak. This occurs since a small peak attenuation is
propagated downstream through various conduits and doing that increases
i n magni tude.
163

Inflow h y d r o g r a p h d u r i t i o n / t c

F i g . 8.8 Diagram i n d i c a t i n g when time shift routing can be used


w i t h p a r t i a l l y f i l l e d pipes.

Fig. 8.9 Diagram i n d i c a t i n g when time shift r o u t i n g can b e used


w i t h trapezoids a t a n g l e o f side to h o r i z o n t a l of 30°
164

g/
8
9
' Uniform lnpui

/'

6-

KI nemal I c
r u u t 1 n g R",t
be used TlrneSt11fi ,
- '

1 LNTEKMEUlAlt AREA /routing ndy b e u s e d


/'

F i g . 8.10 Diagram i n d i c a t i n g when time s h i f t r o u t i n g can be used


w i t h trapezoids a t a n g l e of s i d e to h o r i z o n t a l of 90'

1.2
\\ PIPE

1.1

1 .o

0.9

.
I

-," 0.8

0.7

0.6

I I 1 I
0.5
0 0.2 0,Q 0.5 0.8 0
iinnffllooww ppeeaakk dd >> ss cc hh aa rr gg ee // qq m
m

F i g . 8.11 Time l a g for hydrographs routed through partially filled


pipes
165

Having decided on an acceptable peak attenuation to be neglected


it is assumed that kinematic routing describes accurately routing in a

real life situation. The results obtained by kinematic routing are


employed to assess the conditions under which time shift methods are
acceptable, i.e. in this case the conditions under which the peak
attenuation is lower than 10%. To do this the results were used to
obtain a dimensionless inflow duration for a 0.9 outflow to inflow peak
ratio for every type of section and inflow factor. The dimensionless
inflow duration was obtained either by I inear interpolation or whenever
thought necessary by plotting dimensionless duration against outflow to

inflow peak ratio and obtaining the dimensionless duration for a peak
ratio of 0.9, the peak ratio of 0.9 corresponding to a 10% peak
attenuation. The r e s u l t s a r e summarised i n Figs. 8.8 - 8.10.

Lag Time for R o u t i n g H y d r o g r a p h s U s i n g T i m e S h i f t Methods

Using a similar method to the above the dimensionless time lag of


hydrographs with a peak a t t e n u a t i o n of 10% was obtained. The dimension-
less lag times are surnmarised in Figs. 8.11 to 8.13 for pipes and
selected trapezoids.
A dotted l i n e represents l a g times as o b t a i n e d b y time shift methods
for comparison purposes.

Comparison of Methods for E v a l u a t i n g L a g T i m e

Two assumptions are currently popular for calculating the time lag
of a hydrograph to be routed by time shift methods. The time lag is

either assumed to be the l e n g t h of the conduit divided by the velocity


of the water when the conduit is discharging at full capacity or it is
assumed that the time lag is the length of the conduit divided by the

velocity of water in the conduit corresponding to the maximum d i s c h a r g e


of the i n f l o w h y d r o g r a p h .

Method 1 (TLp = L/(qm/am).) The time constant (tk) for the pipe is

g i v e n b y e q u a t i o n 8.31. The dimensionless time l a g i s thus

tLP - a rn
- -
t dZ
k
where a /d2 i s the dimensionless flow a r e a for a pipe discharging at
rn
maximum c a p a c i t y . S u b s t i t u t i n g f o r arn/dz

‘LP
- = 0.7653 (8.50)
t
k
166

This gives the dimensionless lag time for a pipe and plots i n Fig.

8.11 as a straight line. As can be seen the lag time c a l c u l a t e d b y this

method over-estimates the true value for h i g h inflow factor values ( b i g g e r


than 0.6) and grossly underestimates it for low inflow factor values
(lower than 0.25). For intermediate inflow factor values this method
yields time l a g s which l i e between the r a n g e set up by the two d i f f e r e n t
input distributions.

Method 2 (tLp = L/(qim/qm).) The following relationship holds for the


dimensionless time l a g :

(8.51)

where A. is the dimensionless flow area corresponding to the maximum


I
d i s c h a r g e of the i n f l o w h y d r o g r a p h (4. 1.
im
Equation 8.51 was solved in the following way to express (t /t )
LP k
in terms of the inflow factor (qim/qm). The dimensionless water depth
(Yi) corresponding to the flow depth Ai is solved knowing the inflow
factor and a Newton-Raphson iterative scheme, using equation 8.36.
Y. is used to solve for A. and consequently for tLp/tk. The c a l c u -
lated values of tLp/tk are plotted in Fig. 8.11 for comparison. It can
I
be seen that this present method yields time l a g s closely resembling the
results obtained from kinemat ic theory using uniform input hydrographs.
This occurs as uniform input hydrographs (which do not attenuate
significantly - 10% only) maintain an approximately constant depth
through their travel through the conduit, thus having a speed of flow
similar to that calculated by the existing method. The fact that time

lag as developed by kinematic routing is slightly less than that using


the present method is because some attenuation (10%) occurs during
r o u t i n g for p r o d u c i n g the r e s u l t s .

Time Lag for Trapezoids

Method (1) as outlined above is not applicable for trapezoids in


this study as they are assumed to be deep enough to accommodate

incoming hydrographs of any discharge. As their depth i s not restricted


one cannot talk of maximum discharge through trapezoids. Method (Z),
however, can b e . used to express the dimensionless lag time (tLp/tic) in
terms of the inflow ( 4 . / q c ) to compare time l a g s w i t h
factor the present
im
method w i t h the r e s u l t s shown i n F i g s . 8.12 a n d 8.13.
The time constant t i s given by equation 8.32 for the trapezoid.
k
T h i s y i e l d s the dimensionsless time l a g .
167

2.5

Time s h i f t method

i.c

_' i .5
Unlforrn I n p u t

1.0

i
_I
0 7 4 6 8

Inflow peak d i r c h a r g e / q c

Fig. 8.12 Time l a g f o r h y d r o g r a p h s routed through trapezoids with


a n g l e of side t o h o r i z o n t a l of 30°
168

TRAPEZOID
A N G L E - 90'

\\\

--------

/
Triangular Input

I I I
6
I n f l o w peak d>scharge/q,

Fig. 8.13 Time l a g f o r h y d r o g r a p h s roubed through trapezoids with


a n g l e of side t o h o r i z o n t a l of 90
169

t
LP Ai
~-
tk -0 (8.52)

This equation is solved to yield the r a t i o t /t for different values


Lp k
of the inflow factor. Note that the relationship will differ for different
angles for the trapezoid as the dimensionless flow area is a function
of the side angle. The r e s u l t s a r e p l o t t e d i n F i g s . 8.12 a n d 8.13 together
with the kinematic routing results for comparison purposes. As can be
seen ( t h e dashed l i n e s ) the time l a g s from the present method a r e s l i g h t l y

higher but closely resemble the ones from kinematic routing using a

uniform input. Note that t h i s was also the case f o r the p i p e . The reasons
for their resemblance a r e s i m i l a r to those for the p i p e a n d a r e discussed
i n the p r e v i o u s section.
It can be seen from Figs. 8.8 to 8.10 that the dimensionless inflow
duration is much more critical than the dimensionless inflow peak
discharge for determining whether time shift methods can be used. This
is more apparent in the case of trapezoids where the 10% peak attenua-
tion curves appear almost vertical for dimensionless inflow peak dis-
charge values g r e a t e r than 2.0.
Furthermore, it can be seen that the dimensionless infow duration
decreases with increasing inflow factor. This is expected as inflow
hydrographs with a similar inflow factor need bigger durations than
ones with a higher inflow factor for both inflow hydrographs to have
similar volumes. As was discussed earlier, higher inflow volumes will
imply smal l e r peak attenuation, other parameters being constant, one
exception to this observation being the pipe for inflow factors higher
than 0.8. It can be seen from Fig. 8.8 that as the inflow factor
approaches unity the dimensionless duration (causing a 10% peak
a t t e n u a t i o n to the i n f l o w h y d r o g r a p h ) increases.
This is probably due to the fact that a pipe discharges more when
not f l o w i n g f u l I as a l r e a d y discussed.
It will also be noted that for trapezoids and discharge inflow

factors of less than 2.0 the dimensionless inflow duration increases


sharply as the dimensionless discharge decreases. This is probably due
to the fact that at low depths of flow side friction effects cause a
stabi I i t y effect on the flow highly attenuating peak discharges. This
in turn implies higher inflow durations for maintaining a peak
a t t e n u a t i o n of 10%.

For a constant inflow factor, the inflow duration (implying a 10%


peak attenuation of the routed hydrograph) is bigger for the triangular

distribution than for the uniform one. This is to be expected as a


170

triangular distribution has a lower inflow volume than a uniform one,


both d i s t r i b u t i o n s h a v i n g the same d u r a t i o n a n d i n f l o w factors.

The triangular distribution would therefore need a greater duration


(for a constant inflow factor) or a greater inflow factor ( f o r a constant
duration) to yield a similar r e s u l t to the u n i f o r m d i s t r i b u t i o n . Note t h a t
a constant volume will not imply identical results between the two
distributions as the shape also plays an important role in the routing;
for examp I e:
From Fig. 8.8, for a pipe and an inflow factor of 0.6, the corres-

ponding dimensionless durations resulting in a 10% peak attenuation of


the inflow hydrograph, are found to be 0.18 for a uniform input distri-
bution and 0.82 for a triangular distribution. This implies that the
triangular distribution has a bigger inflow volume than the u n i f o r m one
i n the r a t i o o f :

This r a t i o (triangular to uniform volume) varies depending on the


inflow factor and type of section but is always found to be more than
unity. This implies f u r t h e r that a u n i f o r m time d i s t r i b u t i o n i s attenuated
less than a triangular one even if both have the same volume when
r o u t e d t h r o u g h a closed c o n d u i t .

A further comparison of the effects the inflow distribution has on


the results i s shown in Figs. 8.11 to 8.13. The u n i f o r m h y d r o g r a p h takes
more time to travel along the conduit (it has a bigger lag time) than

the triangular one (both hydrographs attenuated at their peak by 10%).


The reason for this is that for a constant inflow factor a n d a constant

peak attenuation the triangular distribution has a much b i g g e r d u r a t i o n


than the uniform one. Furthermore, i n the case of the triangular distri-
bution, the peak discharge in the outflow hydrograph corresponds to
the peak of the inflow hydrograph which lies, in time, in the middle
of its duration. In the case of the uniform distribution however, the
outflow hydrograph peak will correspond to the inflow peak at a much
earlier stage of the distribution, i.e. at the beginning of the inflow
hydrographs. This implies a later entry time (in the conduit) for the
peak of the u n i f o r m d i s t r i b u t i o n r e s u l t i n s i n a longer l a g time.
The engineer faced with the problems of routing a runoff hydrograph
through a pipe or a channel will find the results presented here of
direct use. The runoff hydrograph could be the result of overland flow
or the outflow from another conduit. Figs. 8.8 to 8.10 can be used to

establish the necessity of routing while Figs. 8.11 to 8.13 can be used
171

to calculate a lag time for the cases for which time shift routing is

shown to b e ade a u a te .

REFERENCES

A r g a m a n , Y., S h a m i r , U. and S p i v a k , E . 1973. D e s i g n o f o p t i m a l s e w e r a g e


s y s t e m s , P r o c . ASCE, ( 9 9 ) , EE5, Oct., p 703-716.
B a r n e s , A.H., 1967. C o m p a r i s o n o f c o m p u t e d and o b s e r v e d f l o o d r o u t i n g i n
a c i r c u l a r cross section. I n t l . H y d r o l . Sympos. C o l o r a d o S t a t e U n i v . ,
F o r t C o l l i n s , p p 121-128.
C o n s t a n t i n i d e s , C.A., 1983. C o m p a r i s o n o f k i n e m a t i c and t i m e s h i f t r o u t i n g
in c l o s e d c o n d u i t s . R e p o r t 3/1983. Water Systems Research Programme,
U n i v e r s i t y of the Witwatersrand.
G r e e n , I .R.A., 1984. WITWAT s t o r m w a t e r d r a i n a g e p r o g r a m . W a t e r S y s t e m s
R e s e a r c h P r o g r a m m e , R e p o r t 1/1984. U n i v e r s i t y o f t h e W i t w a t e r s r a n d . 6 7 p
Martin, C. and K i n g , D., 1981. A n a l y s i s o f s t o r m s e w e r s u n d e r s u r -
c h a r g e . P r o c . C o n f . U r b a n S t o r m w a t e r , I I I i n o i s . pp 74-183.
M e r r i t t , L.B. and B o g a n , R.H., 1973. C o m p u t e r b a s e d o p t i m a l d e s i g n of
s e w e r s y s t e m s . P r o c . ASCE, ( 9 9 ) , EE1, F e b . pp 35-53.
S t e p h e n s o n , D., 1980. D i r e c t d e s i g n a l g o r i t h m f o r s t o r m d r a i n n e t w o r k s .
P r o c . I n t . Conf. U r b a n S t o r m D r a i n a g e , U n i v . K e n t u c k y , L e x i n g t o n .
Stephenson, D., 1981. S t o r m w a t e r H y d r o l o g y and D r a i n a g e , E l s e v i e r , 276
PP.
Y e n , B.C. and S e v u k , A.S., 1975. D e s i g n o f s t o r m s e w e r n e t w o r k s . P r o c .
ASCE, 101, EE4, A u g . 535-553.
175

Case 1 , 2 and 3

Fig. 9.2 Simple catchment a n a l y z e d

From F i g . 3.6 ( f o r U = 0.40) r e a d e q u i l i b r i u m te > 4 h ( o f f t h e graph) b u t


the peak runoff factor for this F i s QF = 0.23 which corresponds to a
storm d u r a t i o n of t d = 2.2h. The p e a k r u n o f f r a t e i s
Qp = 0.23Baa5"/10 = 0.23x1000x1x70 5'3/105 = 2.74m3/s

The total precipitation rate over t h e catchment of area A for the


same storm d u r a t i o n i s
70 x 1000 x 2000 = 17.6m3/s
Ai =
( 0 . 2 1 + + 2 . 2 ) - * ~x 3600 x 1000

so the r a t i o n a l c o e f f i c i e n t C = 2.74/17.6 = 0.16.


Note however that the full catchment is not contributing at the
time of peak runoff for the design storm, so C does not o n l y represent
the reduction in runoff due to losses, it also accounts for o n l y p a r t of
the catchment contributing. The runoff for the full catchment would be
less as the storm duration would be longer than 2.2 h so t h e i n t e n s i t y
w o u l d be less a n d the losses r e l a t i v e l y h i g h e r .

ii) Reduction in Infiltration

If the infiltration and i n i t i a l abstractions a r e reduced b y urbaniz-


ation, the p e a k r u n o f f increases. The c o n s t r u c t i o n of b u i l d i n g s and roads
could reduce infiltration rate to 7 mm/h and initial abstraction to 14
mm. For F = 7/70 = 0.1 and U = 14/70 = 0.20 (Fig. 3.5) then for LF
= 3.27 as for case (i), t h e time t o e q u i l i b r i u m i s o f f t h e c h a r t b u t the
critical storm has a duration of 2.2 hours and the corresponding peak
flow is
176

= 0.44 x 1000 x 1 .O x 70 /lo5 = 5.24m3/s


QP
The corresponding r u n o f f coefficient C works out to be 0.30

Case 4

F i g . 9.3 Catchment w i t h channel

iii) Effect of Reduced Roughness due to Paving

With the c o n s t r u c t i o n of roads, pavements a n d b u i l d i n g the n a t u r a l


retardation of the surface runoff is el iminated and concentrat ion time
reduces. That is, the system response i s faster a n d as a r e s u l t s h o r t e r ,
sharper showers are the worst from the point of view of runoff peak.
For the sample catchment the effective Manning roughness could quite
easily be reduced to 0.03. Then a = 3.33 and L F = 0.98. The time to
equilibrium would therefore be 3h but the peak intensity storm has a
duration of 2.2h as before. In this case extent of the storm over the
catchment i s g r e a t e r however, a n d the peak r u n o f f i s
Q = 0.23 x 1000 x 3.33 x 705’3/10 = 9.12m3/s
P
The corresponding increase in C is from 0.16 to 0.52 an appreciable
increase if it is borne in mind this i s only due to reduced roughness
and does not account for reduced infiltration. It will be noted t h a t the

effect of reducing roughness i s even greater t h a n decreasing i n f i I tration


for t h i s case. The same effect i s m a g n i f i e d i n the f o l l o w i n g example.

iv) Effect of Canalization

The effect of a stream down the centre of the catchment is illus-


trated in the following example. The same s u r f a c e roughness ( n = 0.1)
177

and permeability (f = 10 rnm/h, u = 30 mm) a s f o r c a s e ( i ) a r e assumed.


The overland flow cross slope is t a k e n a s 0.04 a n d 0.01 for a 8 m wide
channel down the catchment. The dimensionless hydrographs in Chapter
six a r e used a g a i n .
2L u.6 ba u.6
The s t r e a m c a t c h m e n t r a t i o G = (2
) 2
ba s 2Lo
- ( 2 x 2000 o.6 8 x 2 o.6 = o.50
8 x 1 2 x 500
By t r i a l , guess storm duration resulting in peak runoff of 1.5h, then

i = __
(o.24 70 1 . 5 ) . m- 10 = 42.7-10 = 32.7 mm/h
e
td - t = 1.5 - 30/42.7 = 0.80h
ted =

F = 10/32.7 = 0.31

Lo )l/m = 500 3/5


tCO = i m-1 2 x ( 32.7/3600000 )3'2
2860s = 0.80h
011

TD = (5/3)ted/tco = (5/3)0.8/0.8 = 1.67

Therefore t d = t + t = 0.8 + 30/42.7 = 1.50 h w h i c h agrees w i t h guess


ed u
Interpolating Figs. 6.10 a n d 6.11 t h e p e a k f a c t o r Q = 0.85

Peak f l o w Q = QAie = 0 . 8 5 ~ 2 ~ 6
1x32.7/3.6x106
0 = 15.4m3/s

Rational coefficient C = 15.4/(42.7x2/3.6) = 0.65

v) Combined r e d u c e d roughness and r e d u c e d losses

If roughness is reduced by paving to 0.03 then a = 3.33 a n d L F


= 0.98 as for case ( i i i ) . The reduced loss factors become F = 0.1 and
U = 0.2 as f o r case ( i i ) . F r o m F i g . 3.5 t = 1.7 h a n d the corresponding
P F = 0.43.
Hence the peak flow Q = 0.43 x 1000 x 3.33 x 70 5 3 = 17.0m3/s. The
r a i n f a l l r a t e f o r a s t o r m o f t h i s d u r a t i o n i-s
70 x 1000 x 2000
(0.24 + 1 . 7 ) 0 8 9 x 36000 x 1000 = 21.6 m 3 / s so C = 0.79.

The r e l a t i v e e f f e c t of each v a r i a b l e on p e a k r u n o f f c a n b e compared


with the aid of Table 9.1. The effect of reducing infiltration 30% a n d
initial abstraction 40% i s to double the peak runoff. The critical storm
duration was not affected but the effective area contributing increased
slightly. The effect of reducing surface roughness is even more
remarkable however. Even maintaining the same losses (both initial and
abstraction and infiltration) as for the n a t u r a l catchment the runoff peak
178

increased by a factor of 4. The area contributing increased noteably

although the c r i t i c a l storm duration was not affected. Reducing roughness


even more would not necessarily increase runoff much as p r a c t i c a l l y the
e n t i r e catchment contributes for case (iii) whereas the a r e a c o n t r i b u t i n g
in case ( i ) was much less. Only for case (v) with reduced roughness
and losses is the concentration time equal to the c r i t i c a l storm duration.

TABLE 9.1 Showing effect of d i f f e r e n t s u r f a c e c o n f i g u r a t i o n s on peak


runoff from a 2000m long by lOOOm wide catchment.
0.89
So = 0.01, i = 70 rnm/h/(0.24h + t )
d

CASE n f mm/h u mm t h td h i mm/h Q m'/s C


__ - ~ - P ~
i) Virgin 0.1 10 30 5 2.2 36.7 2.74 0.16
catchment

ii) Reduced 0.1 7 14 4 2.2 36.7 5.24 0.30


losses

iii) Reduced 0.03 10 30 3 2.2 36.7 9.12 0.52


roughness

iv) Canaliz- 0.1 10 30 0.8 1.5 42.7 15.4 0.65


ation
(stream
w i d t h 3m)

V) Reduced 0.03 7 14 1.7 1.7 38.8 17.0 0.79


losses a n d
roughness

The effect of canalization i s somewhat similar to r e d u c i n g roughness


- water velocities, and concentration rates, are faster. This is due to
the greater depth in channels (Q = 6 Jsy"/n). Consequently a greater

a r e a c o n t r i b u t e s to the peak.
Not much sense can be made out of comparing the r e s u l t i n g r a t i o n a l
coefficients (ratio of peak runoff rate to rainfall rate times catchment
area). That is because the time of concentration for each case is
different due to differing roughness, rainfall rate etc. In any case it
is irrelevant when it comes to critical storm duration which is shorter
t h a n the time to e q u i l i b r i u m .
179

D E T E N T I O N STORAGE

Although the kinemat i c equations as presented previously cannot

accommodate reservoir storage they may be rearranged to i I lustrate the


storage components in them. The St. Venant equations w h i c h i n c l u d e terms
for storage when water s u r f a c e i s not p a r a l l e l to the bed, are

aA
-- - aa (9.1 1
at ax

(9.2)

The f i r s t equation is the continuity equation and the second the


so-called dynamic equation. The first equation does not give the total
storage in the reach, i t represents the r a t e of change i n cross sectional
area of flow as a function of inflow and outflow. The second equation
contains more about the distribution of storage. The last two terms
represent the wedge component of storage, which are absent in the

kinematic equations. The kinematic equations therefore treat storage as


a prism, with storage i n blocks and no allowance f o r difference i n slope
between bed and water surface is made. Since the second equation is
replaced by a friction equation and So = Sf in the k i n e m a t i c equations,
only the first equation in the case of the kinematic equations can be
used to c a l c u l a t e storage changes.
The c o n t i n u i t y equation may be w r i t t e n as
Az-Ai
-0-1
+-
Ax
=o At
(9.3)

where 0 i s outflow, I is inflow over a r e a c h of length A x, and A t and


A2 are the cross sectional areas before and after A t respectively.

If 0 = (0 + 02)/2 and I = (I, + 12)/2 a n d AAx i s replaced b y S, the


1
storage which is a function of Al and Ao, which in turn a r e functions
of flowrate, e.g. S = XI + (l-X)O, then equation (9.3) becomes the one
frequently used f o r open channel r o u t i n g ,

0 2 = c, I I+ c 21 2 + caO1 (9.4)
where cI, c and cg a r e functions of A x and A t. The l a t t e r equation
is referred to as Muskingum's equation used in routing floods along
channels. if X = 0 the routing equation corresponds to level pool or

reservoir routing. The more general equation with X = 1/2 represents


a 4-point numerical solution of the continuity equation as employed in
kinemat i c models ( B r a k e n s i e k , 1967).
180

CHANNEL STORAGE

Channel storage performs a similar function to pond storage in


retarding flow, and there are many analogies which can be drawn
between the two. Channel storage is a function of friction resistance
a n d channel shape a n d can be c o n t r o l l e d i n v a r i o u s ways.
The form of f r i c t i o n equation, as w e l l as the f r i c t i o n factor, affect
the r e a c t i o n speed of a catchment a n d the volume stored on the catchment.
The excess rain stored on the catchment, whether in channels or on
planes, is a form of detention storage, and as such, affects the con-
centration time and consequently the peak r a t e of runoff. Some friction
formulae used i n stormwater d r a i n a g e p r a c t i c e a r e l i s t e d below.

S.I. units English units

Darcy-Weisbach Q = (8/f)1bA(RSg)1fi Q = ( 8 / f ) l R A(RSg)'n 9.5)

Chezy Q = 0.55CA(RS)'" Q = CA(RS)'/* 9.6)


Manning Q = AR2/3S1/2/n Q = 1 .486AR 2'3S v2/n 9.7)
Strickler Q = 7.7A(R/k) (RSg)V2Q= 7.7A(R/k)v6 (RSg)Ih 9.8)

R is the hydraulic radius A/P where A is the area of flow and P the
wetted perimeter. R can be approximated by depth y f o r wide r e c t a n g u l a r
channels. S is the energy gradient, f is the friction factor and k is
a l i n e a r measure of roughness analogous to the N i k u r a d s e roughness.
Both the roughness. coefficient CY and the exponent m of R or y in
the general flow equation (9.11) affect the peak flow off a catchment.
This is largely due to the attenuating effect of friction resulting in a
larger time to equi l i b r i u m . A rainfal I excess intensity-duration relation-
ship i s required to e v a l u a t e the effect of each coefficient on peak runoff
r a t e and maximum catchment storage. The f o l l o w i n g expression f o r excess
r a i n f a l I intensity i s assumed:

(9.9)

I n t h i s equation it i s customary to express i and a i n mm/h or inches


per hour and b and td i n h o u r s where t d i s the storm d u r a t i o n assumed
equal to time of concentration tc for maximum peak runoff of a simple
ca tchrnent.
S t a r t i n g w i t h the kinemat i c equation f o r c o n t i n u i t y

aa t v . 3ax = 'e
(9.10)
181

a n d a general flow resistance e q u a t i o n

q = aym (9.11)
then i t may be shown that tc = (L/aiem-')'/m where q i s the r u n o f f r a t e
per unit width of the catchment and y is the flow depth. The rising

limb of the h y d r o g r a p h i s g i v e n b y t h e equation


m
q = CY ( i t ) (9.12)

0 p . , , ,
1
, , , , . . , , , ,

2 '
,

m
Fig. 9.4 Hydrograph shapes for different values of m in q = ay

a n d another expression may be d e r i v e d from the f a l l i n g limb (see Chapter


2). In Fig. 9.4 are plotted dimensionless hydrographs to illustrate the
effect of m on the shape, of the hydrograph. The graphs are rendered
dimensionless by plotting Q = q/ieL against T = t/tc. m is used as a
parameter. Thus m = 1/2 represents closed conduit or orifice flow, m
= 1 represents a deep vertical s i d e d channel, m = 3/2 represents a wide
rectangular channel according to Darcy or a rectangular weir, m = 5/3
represents a wide rectangular channel i f M a n n i n g ' s equation i s employed,
and rn = 5/2 represents a triangular. weir. The graphs immediately
indicate the effect of m on catchment detention storage since the area
u n d e r the g r a p h represents storage.
The smaller m, the greater storage. Thus provided storage is
economical by throttling outflow one may increase storage and increase

concentration time thereby reducing discharge rate (which is not immed-


iately apparent from these graphs as they a r e p l o t t e d r e l a t i v e to excess
rainfal I intensity). In practice the concentration time increases the
greater the storage so that the lower intensity storms become the design
storms. This has a compound effect in reducing flow rates since total

volume of losses increases and it i s possible that the e n t i r e catchment


w i l l not c o n t r i b u t e a t the peak flow time.
182

A general solution of peak flow and storage in terms of intensity-

duration relationships is derived below. Solving (9.9) with td = tC f o r


maximum rate of runoff per unit area and generalizing by dividing by

a.

-_ 1
1 L/a(a/3600000)m-1 I
'-'lm
(9.13)
I c + I P
3600 ( ie/a )

m-1
The term L/ua is r e f e r r e d to as the length factor. The constants are
introduced for a in mm/h, and time of concentration i n hour units. The
maximum peak flow factor ie/a is plotted against length factor in Fig.
9.5, since i t i s not easy to solve (9.13) d i r e c t l y f o r i /a

i e / a and s/a
max
2 c=o.9
b=0.25h 1

,/f )

s * .. . I I I , * ] I
00
n 1 I 1 I

L 1.i a
.
rn- 1 . . jC
I0

Fig. 9.5 Peak flow a n d storage versus l e n g t h factor

An expression for the corresponding catchment storage is derived

below. At equilibrium the flow per unit width at a d i s t a n c e x down the


catchment is
183

q = i x
e
m
= ay

therefore y = (iex/a)lIm
I n t e g r a t i n g y w i t h respect to x y i e l d s the t o t a l volume on the catchment

or i n terms of the average depth of storage s = V/L

1 /m
(
L 1/m
) - - -
, (9.14)
m+l a a ( a/3600000) m-' 3600

where s i s i n mm, and i and a a r e i n mm/h. s/a i s also p l o t t e d a g a i n s t


length f a c t o r in Fig. 9.5. It will be obzerved t h a t a v e r a g e storage depth
does not increase in proportion to L / a a m - ' . I n fact the r a t e of increase
reduces beyond L/aam-' = 50, and the r a t e of reduction i n peak flow

ie/a also decreases beyond the figure, indicating reducing advantage


in increasing channel length or roughness ( 01 = K1 J(S) /n). Since t o t a l
channel cost is a direct function of storage capacity it would appear
to be an optimum at some intermediate value of L/aam-' if there is a
cost associated with peak d i s c h a r g e e.g. c u l v e r t s or flooding downstream
(see Fig. 9.6).

minimum

cost
$

Fig. 9.6 Optimum catchment storage volume.

Note that infiltration after the rainfall stops, is neglected in the


above analysis. Inclusion of that effect would lower the ie/a and s/a
l i n e s to the right, implying a l a r g e r L/aam-' i s best. The model p r o v i d e s

an indication of total storage in the system. The location ( a n d volume)


184

of storage could be f u r t h e r optimized using dynamic programming methods


or by detailed modelling. It should be found generally that it is most
economical to provide pond storage (m = 1/2) at the outlet, whereas
channel or catchment storage (m = 5/3) is most economical at the head
of the system.

K I NEMATIC EQUAT I ONS FOR CLOSED CONDU I T SYSTEMS

If the open channel kinematic equations are applied to closed


conduit flow the problem becomes a steady state flow one since flow
rates become independent of cross section. This i s provided the conduits
remain ful I and there are no storage ponds at nodes j o i n i n g conduits.
If one permits storage variation at nodes one has the reservoir-pipe
situation encountered in water supply which i s often analyzed employing
pseudo-steady flow equations.

Fig. 9.7 Input-output node storage

T h e c o n t i n u i t y e q u a t i o n becomes ( s e e F i g . 9.7)

(Qi+,-Qi)- qi + Ai-
dhi
=O (9.15)
dt
where the reservoir surface area A. replaces B dx in the open channel
continuity equation where B is the catchment width. q is the reservoir

i n f l o w here. The d y n a m i c e q u a t i o n i s r e p l a c e d b y

Q. = aAm (9.16a)

where A is the (constant) conduit cross sectional area. Since the kine-
matic equations omit the dependency of Q o n h e a d d i f f e r e n c e h, t h e l a t t e r
equation assumes the head gradient along the pipe equals the pipe
gradient, i.e. free-surface just full flow. Since A is a constant it is

r e l a t i v e l y easy to r e p l a c e the l a s t e q u a t i o n b y o n e of t h e f o r m
185

Q. = 01 Ah.m (9.16b)
This equation is applicable to free discharge from an orifice o r over a

weir.One more a p p l i c a b l e to c o n d u i t flow would be


m
Q = aA(hi-l-hi) (9.16~)
Any one of the above three equations c o u l d be a p p l i c a b l e i n storm-

water drainage. For channel or overland flow (9.16a) applies, for


complete storage control (9.16b) applies and for closed conduit control
(9.16~) is applicable. The latter form of equation has in fact been
employed i n water r e t i c u l a t i o n p i p e network a n a l y s i s . I t can b e a p p l i e d in
storm drainage to closed systems (not of great interest in stormwater
management p r a c t i c e ) or to p i p e - r e s e r v o i r problems. Surface detention and
artificial detention storage ponds can be handled in an overall flow
balance employing the closed conduit kinematic method. It should be
noted that the numerical instabi I i t y problems associated with s o l u t i o n of

the open channel k i n e m a t i c equations a r e absent. Time steps can be much


larger than for open channel kinemat ic model I ing. Storage fluctuations
may be computed i n steps and the effect of changes i n pond water levels
on flows i n c o n d u i t s can be accounted f o r .
One p o s s i b l e a p p l i c a t i o n of such a program i s to an inter-connected
pond system with reversible flows in conduits. Overload from one pond
can be forced back to another pond. Such situations can readily arise

from s p a t i a l l y v a r i a b l e storms a n d p o s s i b l y for t r a v e l l i n g storms.


Off-channel storage can also be accounted for. Such ponds have
the a d v a n t a g e that water l e v e l v a r i a t i o n s a r e not as marked as the head
variations in the drain pipes (which may in fact be surcharged). This
i s due to the r e v e r s i b l e head loss between the m a i n c o n d u i t a n d the pond.

3
1.4m 3/s

1. Om

Fig. 9.8 Conduit a n d storage storm d r a i n network.


The simplified layout in Fig. 9.8 was analyzed employing the

accompanying k inematic closed conduit continuous simulation program.


Input a n d output a r e appended to i l l u s t r a t e the s i m p l i c i t y i n t h i s t y p e of

anlysis. Flow r e v e r s a l , p o n d level v a r i a t i o n s a n d the l a r g e a t t e n u a t i o n in


peak flow will be observed due to the ponds (from 5.6m3/s down to
1.5m3/s). By adjusting individual pond areas and conduit sizes an
optimum design could be achieved for any design storm input. A
sensitivity analysis for alternative storms such as different storm
d u r a t i o n s o r ones w i t h s p a t i a l v a r i a b i l i t y would then be performed.

COMPUTER PROGRAM TO SIMULATE RESERVOIR LEVEL V A R I A T I O N S I N A PIPE


NETWORK

Closed c o n d u i t d r a i n a g e networks can as e x p l a i n e d above, be used to


ameliorate peak flows by directing water into storage. Flow can be in
either direction a n d depends on the difference i n water levels a t the two
ends of the conduit, not on the conduit gradient as for open channels.
Apart from t h i s , the p r i n c i p l e s a r e the same as f o r open channel k i n e m a t i c
flow. That i s steady state conditions ( h e a d loss/flow equations) are used
together with the c o n t i n u i t y equations. The accompanying computer program
written in HP 85 'BASIC' will simulate the variations in water level in
r e s e r v o i r s i n a d d i t i o n to p e r f o r m i n g a network flow balance.
The program is based on the l i n e a r node method (Stephenson, 1984)
network analysis with an additional variable, a r e a of reservoir for each
'fixed head' or, in this case, 'reservoir type' node. If the simulation
duration T 4 i n h o u r s a n d time increment T5 a r e i n p u t , f o r example 24 a n d
1, then the heads a t each node a n d water level i n each r e s e r v o i r w i l l be
p r i n t e d out every hour. The actual network i t e r a t i o n s each time interval
after the f i r s t should be m i n i m a l s i n c e the network flows a r e b a l a n c e d i n
the first iteration and only u n b a l a n c e d due to reservoir level changes
which will have to be corrected at subsequent time intervals. Although
drawoffs are time-fixed in the present program, they c o u l d be a l t e r e d a t
pauses i n the r u n n i n g o r inserted i n e q u a t i o n form.

The output, namely level variations, could be used to estimate


r e q u i r e d r e s e r v o i r depths ( u s i n g t r i a l r e s e r v o i r surface a r e a s ) a n d i n f a c t
to see at which reservoir locations the storage i s most required. Data

requirements are similar to the analysis program with the following


additions.
187

In the first data line after the name, the simulation duration and
increment in hours is added at the end of the line. In the p i p e d a t a ,
the first pipes should be from the various reservoirs with the surface
areas of the up-stream reservoirs in square metres given at the end of
the pipe data lines. In order to display the reservoir levels in the
biggest reservoir it is necessary to have a supply pipe from a pseudo
fixed head, very large, reservoir to represent a pumped supply feeding
i n t o the a c t u a l b i g g e s t level r e s e r v o i r i n the d i s t r i b u t i o n system.
The selection of 'upper' and 'lower' nodes f o r any pipe, numbered

1 and 2 is somewhat arbitrary. If the incorrect flow direction is


assumed, a negative flow number will appear in the answers thus indi-
c a t i n g the flow d i r e c t i o n i s from node 2 to node 1 as specified.
When data is put in, the order of pipes is to a limited extent
arbitrary, but the 'node 1' of any pipe should have been defined as
a 'node 2' i n some p r e v i o u s p i p e . T h i s does not apply to the f i r s t pipe
which will originate at a reservoir. The order of pipes enables data
on successive nodes, i.e. initial estimates of heads and flows, to proceed
down the system from p r e v i o u s l y d e f i n e d nodes.
Node numbering i s also open to the user except the r e s e r v o i r - t y p e
nodes (with specified initial water levels) should be numbered first,
from 1 to J3.
There is scopc for setting all Darcy friction factors the same to
minimize data requirements, or to vary each factor. Note if old data
in files is used those friction factors, not the 'common' factor, i s used

even if a common factor is fed in. To print out 'old data' in file, it
is necessary to go into revision mode (2) of pipe data input. To get
out of r e v i s i o n mode, type 0 f o r p i p e number to be revised.
Part of the data i s read interactively on the keyboard. The first
lines of data (name, duration, no. of nodes, reservoir data and pump
data) is typed for each run. The pipe and node data can b e typed in
or r e t r i e v e d from a f i l e o r ammended i n a f i l e .
The time increment between iterations for simulatiot mode must
be small enough to avoid large variations in water levels n reservoirs
between iterations. The reservoir surface area and flows will control
this.
Additional pipes can be added in edit ( 2 ) mode a n d dill then be
stored in the data file. Pipes can only be removed by limiting the
number of pipes in the initial lines to e l i m i n a t e those not required at

the end. The other way is to put a very small diameter for a pipe to
be removed from the network. New nodes or reservoirs can be added b y

r e t y p i n g i n data.
188

When reading in initial data however, no more than the number

of pipes in the data file should be specified, The number will auto-
m a t i c a l l y be increased when more d a t a l i n e s a r e added.
The l a s t s p e c i f i c a t i o n of any drawoff i s retained if a node happens
to be s p e c i f i e d more than once in input. One should also make sure each
node i s s p e c i f i e d ( a s a N2) at least once to define i t s drawoff.

Data Input

Each line may contain more than one unit of data separated by
commas.

Line 1 Name of network ( a n d r u n no.)


Line 2 Analysis ( 0 ) or simulation (1) - type 0 o r 1
Line 3 Drawoff duration in minutes, thus if drawoff is over
8 hours, type 480. Simulation duration mins. If 24
hours, type 1440, Time increment DT, mins. Suggest
30 - 120.
Line 4 Constant ( 0 ) or various ( 1 ) Darcy f ’ s - type 0 or 1
Line 5 No. p i p e s ,
No. nodes ( t o t a l including reservoirs)
No. r e s e r v o i r type nodes.
Lines 6 . . . (one for each reservoir node in successive order)
I n i t i a l water level, rn

Surface a r e a of r e s e r v o i r , mz
Line 7 O l d ( 0 ) o r new ( 1 ) o r revised pipe data ( 2 ) ;
type 0 , l or 2.
L i n e s 8.. . (one f o r each p i p e i n new p i p e d a t a )
Node 1 no.
Node 2 no.
Pipe l e n g t h rn
Pipe i n s i d e d i a . , rn
Drawoff a t node 2 , m3/s
(Darcy f r i c t i o n f a c t o r i f line 4 is 1 )
Line 9 I f line 7 i s 2, w i l l ask p i p e no. for revision.
L i n e 10 Pipe data for new p i p e s as f o r Line 8 including Darcy
f r i c t i o n factor.
Line 11 No. of pumps or pressure reducing valves (one per

pipe).
L i n e s 12... Pipe no. in which pump or PRV is installed, pumping
head or PRV head loss ( - ) in rn.
189

L i s t of Symbols i n P r o g r a m

1 = analysis, 2 = simulation
0 = constant f, 1 = v a r y i n g Darcy f.
0 = o l d data, 1 = new d a t a , 2 = revise old data
0 = no data listing required, 1 = required

h e a d Ios s/Q I Q 1
ZH f o r e a c h SOR
CAF
p i p e diameter (m)
old v a l u e of H ( I )
Darcy friction f a c t o r e.g. 0.012 l a r g e dia. clean p i p e
0.03 small tuberculated pipe

common D a r c y f a c t o r
head a t node or j u n c t i o n I
node counter
n u m b e r of n o d e s
u p p e r node number of p i p e
lower node number on p i p e
n u m b e r of r e s e r v o i r t y p e nodes
i tera t ion
pipe counter
node counter
pipe counter
number of connecting p i p e s
M 2 ( L , M l ( L ) )p i p e n u m b e r c o n n e c t i n g
N$ a l p h a n c r m e r i c n a m e of s y s t e m , up t o 12 c h a r a c t e r s
NO maximum number m a i n i t e r a t i o n s p e r m i t t e d e.g. 4: t 5

N1 m a x i m u m n u m b e r SOR ( s u c c e s s i v e o v e r - r e l a x a t i o n o f
simu l taneous e q u a t i o n s ) i t e r a t i o n s e.g. 4T t 10
N2 counter for main iterations
N3 c o u n t e r f o r SOR i t e r a t i o n s
P n u m b e r of p i p e s
P1 n u m e r o f p i p e s and P R V ' s ( 1 p e r p i p e m a x i m u m )

Q(K) flow in p i p e
Q1 d r a w o f f m3/ s
Q2( I ) d r a w o f f m3 /s
R(k) pump head in m, (or pressure reducing valve head in m
if n e g a t i v e )

S g n2 / 8
190

S ( 2 )I CKij
s3 CHj
S4( I CKijHj
55 o l d Q(K) f o r averaging
T3 drawoff duration, m i n s e.g. 8 h x 60 = 480
T4 simulation duration, m i n s e.g. 24 x 60 = 1440
T5 time increment i n simulation, m i n s e.g. 60
TO tolerance on head in m e.g. 0.0001
T1 t o l e r a n c e o n SOR i n m e . g . 0.01
W-SOR f a c t o r e.g. 1.3 (1-2)
X(K) pipe length m

REFERENCES

Brakensiek, D.L., 1967. K i n e m a t i c f l o o d r o u t i n g . T r a n s Am. SOC. A g r i c .


E n g r s . lO(3) p 340-343.
Colyer, P.J., 1982. T h e v a r i a t i o n o f r a i n f a l l o v e r an urban c a t c h m e n t .
P r o c . 2 n d I n t I. C o n g . U r b a n S t o r m D r a i n a g e . U n i v e r s i t y o f I I l i n o i s .
H u f f , F.A. and C h a n g n o n , S.A., 1972. CI i m a t o l o g i c a l a s s e s s m e n t o f urban
effects on precipitation at St. Louis. J. Appl. Meteorology, 11,
p 823-842.
Stephenson, D . , 1984. K i n e m a t i c a n a l y s i s o f d e t e n t i o n s t o r a g e . P r o c . S t o r m
W a t e r M a n a g e m e n t and Qua1 i t y u s e r s G r o u p M e e t i n g , USEPA, D e t r o i t .
S t e p h e n s d n , D . , 1984. P i p e f l o w A n a l y s i s , E l s e v i e r , A m s t e r d a m , 204 p p .
Sutherland, F.R., 1983. A n i m p r o v e d r a i n f a l I i n t e n s i t y d i s t r i b u t i o n f o r
h y d r o g r a p h s y n t h e s i s . W a t e r S y s t e m s R e s e a r c h P r o g r a m m e , R e p o r t 1/1983,
U n i v e r s i t y of the Witwatersrand.
191

P r o g r a m Listing 546 122 \ J C ( E r' I =O 1


556 PRINT# 1,K i JI(K>,JE(K>,X(K
16 ! NETSIFI KINEMHTJC~CONTIN SI ),D<K),Q2<J2~K?),F(K>
MULN OF NETWORKS WITH STORAG 566 NEXT K
E
578 GOTO 748
26 ASSIGN# 1.~0 "DF~TNET*!
TE"DATNE1 166988
CREA 586 FOR Kl=l TO 168
J
598 DISP "REVISE PIPE NO.";
38 DIN C(5b,,Q<?6~,H<5b).Q2(56) 668 INPUT K
,S2(56),S4(56>,F(56> 616 IF K=b THEN 666
46 DIM J1<5b), JE(56>,D(58), X(56 614 IF K<=P THEN 626
), R(96) Ill (58) M2<58,5)
J J 616 P=K
58 DISP "NAPlE OF NETYORK'; 628 DISP "NODEl.HODE2,Lm,Dn,DRA~
66 INPUT NS DFF~NIJ/s,DARCY~"~
76 DISP "ANALYSIS OR SIPlULATION 636 INPUT Jl(K),J2(K),X(K),D(K>,
( 112)" i Q2( J2(K) ), F < K )
8 8 INPUT ~i 646 PRINT# 1,K ;'Jl<K),JE(K>,X<K
56 IF' A1=2 THEN 146 ),D(K),QZ(J2(K)),F<K>
l 6 b T3=1 653 NEXT F l
116 T4=1 666 DISP DATALIST REQD ( b / ) " i
126 T5=1 67b INPUT A 5
136 GOTO 166 686 IF R5=! THEN 748
146 DISP 'DRAWOFF DURATIONnin,SI 696 PRINT NODE1 N 2 Xra Dm Qm3
M DURNm in,DTm in" i /s f
158 INPUT T3,T4,T5 788 FOR ~ = iTO P
16b Q2<1>=6 718 PRINT USING 736 i Jl(K> J2(K
176 T3=T3$6b ),X<K),D<K),Q2<J2<Kj>,F K)
186 T4=T4X60 728 NEXT K
196 T5=T5*6b 77c* V Y n r r nnr.n nnnn nnnnn n nnn

268 DISP 'VARYING fs<B/l)";


216 INPUT A2
226 IF R2=8 THEN .- . . - --
2C;n 758 Q(K)=3.14159*D(K>*2/4
238 DISP-~NPIPES,NODES, NRESS"; 766 R<K)=6
246 INPUT P,J,J3 776 C<K)=S*D(K>*S/F(KZ/X(K) ! I/
258 GOTO 286 K
786 IF J2<K)<=J3 THEN 866
798 H ( J 2 ( K > ) = H ( J l < K ) ) - l / C < K ) * Q ( K
276 INPUTUP,J, J3,Fl ) *2
286-DISP INITL WATER LEVELm,SUR 866 NEXT K
FACE AREA m2"; 816 DISP "NO.PUtlPS/PRVs'i
256 FOR L=l TO J3 826 INPUT P1
386 DISP L; 83b FOR P;=l TO P1
316 INPUT H(L),A(L) 846 DISP PIPEN,+HEADm Nl-N2"iP2
326 NEXT L
336 6 = 5 . 8 856 INPUT K,R(K)
346 S=3.14155*2tG/S 866 NEXT P 2
356 DISP "OLD OR NEW OR REVISE P 876 FOR L=l TO J
IPEOATA(6/1/2)- " i 886 M1 (L)=6
366 INPUT A4 896 FOR H=l TO P
~~

376 IF A4=1 THEN 436 NElJ DATA


f 566 IF Jl<M)=L THEN 526
388 FOR K=l TO P ! OLD DATA 918 IF JE<M)<>L THEN 546
356 READ# 1,K Jl(K),JP(K),XCK>
J 926 Hl(L)=Ml(L>+l
,D(K>,Q2(JZ<K)>,Ftk) 9 3 8 HE(L,Hl<L)>=H
468 NEXT K
416 IF H4=2 THEN 586
426 GOTO 746
436 IF H 2 = 6 THEN 466
446 DISP "NODE1,NODE2,Lm,Dm,DRAW
OFFEm3/s,DARCYf " j
458 GOTO 476 946 NEXT M
466 DISP " N ~ D E 1 , N O l l E 2 ~ L m , D m ~ D R A W 9 5 6 NEXT L
0F2m 3 / s'' ; '166 W = l . 3 ! SOR FHCTOR
47B FOR K=l TO P ! P I P E Dt3TH 376 T6=.6B61 ! TOLERANCE n
Y8B Tl=.Bl ! S O R TUL rn
996 NB=SQH<J)+5 ! ITNS PIPES
lB6O Nl=SQR(Jj+lB ! ITNS SOR
1618 N2=6
1626 N3=6
1638 PRINT "PIPEtiET" NZ I

i64B FOR T6=T5 TO T4 STEP T5


1056 I F Td<=T3 THEN la%@
1068 FOE L=i Tit J
1678 122cL:i=@
1888 NEXT L
le98 FOR 1 = 1 Ti? t.rS
192

1110
1128
1138 NEXT K
1 1 4 0 FOR K = l TO P
1150 S2(Jl<K))=S2<Jl(K))-C(K)/hB
StQCK))
1168 S 2 ( J 2 ( K ) ) = S 2 < J 2 < K ) > - C < K ) / A B
S(Q(K))
1170 NEXT K
1188 FOR K = l TO N 1
1190 C 2 = 8
1280 S 3 = 0
1210 N3=N3+1
1220 I F J3+1>J THEN 1380
1238 FOR L=J3+1 TO J
1 2 4 8 S 4 <L)=0
1250 FOR fl3=l TO M l ( L )
1260 Fl=H2<L,M3)
1270
1288 /hB
1298
1388 /AB
1310 NEXT M 3
1320 OE=H<L>
1338
1348
1350
1360 NEXT L
1370 I F C2/S3<=T1 THEN 1390
1388 NEXT K
1398 FOR K = l TO P ! NEW FLOtJS
1400
1410

1428
1430
1440
1450
1468
1470

1488 NEXT K
1450 I F C3/P<=T0 THEN 1510
1588 NEXT I
1510
~~~~ FOR L = l TO ~3 ! RES LEVEL.s
1520 H ( L ) = H ( L ) - Q 2 < L ) * T S /~-
H(L~
1530 FOR M 3 = 1 TO M l < L >
1540 M=M2<L,M3)
1550 IF J l ( M ) < > L THEN 1578
1560 H(L)=H(L)-Q(M>*T5/A(L)
1570 I F J2<M)<>L THEN 1590
1580 H(L)=H(L)+Q(M)*T5/A<L)
1558 NEXT M 3
1680 NEXT L
1610 P R I N T USING "K,DDDDOD,X,K
DDDD . O D " i " T s = " , T6, " H i = "
(1)
1 6 2 8 P R I N T "NODE1 NZ Xm Om 12
m3/s H2m ''
1630 F O R K=I TO r
1 6 4 8 P R I N T CLSING 1668 j J l ( K ? , J 2
iK),X(K),D(K>,Q(K),H~J~~~))
1658 NEXT k
1 6 6 B IMAGE O D D . ~ D D O ~ O D D D D , O D . D D D ,
O D D . DDD, O D D D O . n
1678 NEXT TE.
1688 STOP
ASSIGN# 1 TO X
END
193

lOOm
-
1

1000mx0.15
0.09

Fig. 9.9 Pipe network analyzed

F'l PENET TESTSIm 1


Tc=i.4.413O h1= 1138.813
tr0OEl t12 XDI Dm Grm3,s H?m
1 5 1E:OB .3WB .143 74.8
5 t. 868 .i06 .137€. 5c.l
5 8 550 ,258 .6i9 ee.3
d 7 1088 . 1 5 @ -.El14 54.3
8 7 S88 ,288 -654 54.3
5 4 788 .156 .818 72.5
4 3 988 .158 .018 78.7
8 3 708 .268 -.835 78.7
3 2 688 .288 -.B55 78.8
Ts=28888 H 1 = 10B.88
t4.IODEl N2 X m Dm Grm3/s H2m
1 5 1808 . 3 6 8 .146 73.1
5 6 888 . 2 8 Q .876 49.8
28 8

7
958
7 1888
8Bb
.ZSB .678
.l5G
.2@8
-.614
.854
E5. 1
53.2
53.2
5 4 708 -158 .816 71.4
4 3 988 .1SO .81b 69.2
8 3 788 .ZBB -.634 65.2
3 2 688 .2BU -.853 76.1
T s = 4 3 2 8 6 H1= 188.88
NODE1 t42 Xn Drn QmS/s H2m
1 5 1588 .JBB -.810 5B.2
5 6 888 .280 .689 97.9
5 8 958 .258 .837 95.9
.i' - 77 1088
Y 808
.158
.2B8 -.888
96.'
96.3
5 4 76P .is(? ,017 53.9
4 3 98B ,150 817 8:3. 3
8 3 768 -288 Ei46 88.3
3 2 688 .288 ,863 78.4
Ts=576813 H1= 1 8 0 . 88
H-lnDEl N 2 X m Om U N I ~ / C H2m
8 , 3 , 7 6 8 ,. 2 , . 0 3 1 5 1888 -.817 95.4
9 'i 5 6 860 .288 ,885 59.6
3,2,668, . 2 , 8 5 8 958 .2513 .,836 97.2
N O . PUMPS/PRVs? rj 7 I8Sb .I58 97.5
1 8 7 888 .280 -.Be8 97.5
PIPEN,+HEADm Nl-t42 1 ? 5 4 7QB .150 ,816 95.3
1>1 4 3 588 .150 .81G 50.6
8 3 ;fit3 .200, .845 re.@
3 T' .2QB ,861 S8.G
194

CHAPTER 10

K I NEMAT I C MODELL I NG

INTRODUCTION

Kinematic flow holds for those cases when a unique relationship


exists between the depth of flow and the volumetric flowrate. Model
equations are derived through simp1 i f i c a t i o n s to the f u l I equations
governing gradual ly-varied, unsteady overland and open channel flow.
When applied to one-dimensional overland flow, if the rainfall rate is
steady and the watershed geometry is a regular geometry such as a
plane, analytical solutions can be o b t a i n e d f o r the e q u i v a l e n t character-
istic form of the g o v e r n i n g equations. Otherwise, one must use numerical

solution techniques. Currently, kinematic models best apply to highly


impervious (urban) and/or smal I watersheds. However, research is on-
going in several countries to extend the applicability to large and
mu1 t i p l e l a n d use watersheds.
The study of kinematic hydrology and modelling must begin with

the derivation of the f u l l equations governing overland a n d open channel


flow, followed w i t h an examination of model simplifications and when they
can be invoked, development of the c h a r a c t e r i s t i c roots, and then proceed

with analytical and numerical solutions and example applications. In


this chapter, a discussion is given of general modelling concepts and
definitions to provide i n s i g h t s a n d u n d e r s t a n d i n g of the r o l e of k i n e m a t i c
model l i n g as one approach to h y d r o l o g i c model I i n g .

STORMWATER MODELL I NG

Kinematic model I i n g fa1 Is under the umbrel l a of stormwater model I i n g .


Stormwater is defined as the direct watershed response to rainfall

(Overton and Meadows, 1976). It is the runoff which enters a ditch,


stream or storm sewer which does not have a significant base flow
component. This definition does not assume that a1 I stormwater reaches
an open channel by the overland flow route, although in urban areas

the direct response is mostly through overland flow due to the high
degree of imperviousness. I n contrast, in rural watersheds, an overland
flow component may be nonexistant and direct storm response may be
o n l y near the stream a n d occur as shallow subsurface flow.

As defined, stormwater i s associated with small u p l a n d or headwater


watersheds where base flow is not a significant portion of the total
195

streamflow during periods of r a i n f a l I. Therefore, the emphasis of

stormwater m o d e l l i n g i s on the storm hydrograph and not the streamflow


hy d r o g r a p h .

MATHEMAT I CAL MODELS

A mathematical model is simply a quantitative expression of a


process or phenomenon one i s observing, analyzing, or predicting. Since
no process can be completely observed, any mathematical expression of
a process will involve some element of stochasticism, i.e. uncertainty.
Hence, any mathematical model formulated to represent a process or
phenomenon will be conceptual to some extent and the reliability of the

model will be based upon the extent to which it can be or has been
verified. Model verification is a function of the data available to test

the model scientifically and the resources available (time, manpower,


a n d money) to perform the tests. Since time, manpower, a n d money a l w a y s
have f i n i t e limits, decisions must be made as to the degree of complexity
the model i s to have, and the extensiveness of the v e r i f i c a t i o n tests t h a t
a r e to be performed.
The initial task of the modeler then i s to make decisions as to w h i c h
model to use or to build, how to verify it, and how to determine its
statistical reliability in application, e.g., feasibility, planning, design,
. .
or management. This decision-making process IS initiated by clearly
formulating the objective of the m o d e l l i n g endeavour and placing it in
the context of a v a i l a b l e resources.
If the initial model form does not achieve the intended objective,
then it simply becomes a m a t t e r of revising the model and repeating the
experimental verifications until the project objective is met. Hence,
mathematical modelling is by its nature heuristic and iterative. The
choice of model revisions as well as the initial model structure will also
be h e a v i l y affected by the r a n g e of choice of modelling,concepts a v a i l a b l e
to the modeler, and by the skill which the modeler has o r can develop
i n a p p l y i n g them.
Figure 10.1 is a schemat i c representat ion of the model I i n g process.

The modelling process is not new but is nothing more than a modern
expression of the classical scientific thought processes involved in the
design of an experiment. What i s new i s that today a very l a r g e number
of concepts can be evaluated efficiently in a very s m a l l amount of time
at a relatively s m a l l expense u s i n g computers a n d the body of a n a l y t i c a l

techniques termed systems a n a l y s i s .


OBJECTIVE-
OEFl NED
CONCEPT
OF
HYDROLOGIC
PROCESSES
MODEL
- FORMULLTION .
EXPERIMENTAL
VERl FlCATlON -
08JECTIVE

L
t
FEEDBACK
FEEDBACK

Fig. 10.1 T h e rnodeiling p r o c e s s


197

SYSTEM DEF I N I T I ON

Dooge (1976) has developed a good working definition of a system


as being any structure, device, scheme, or procedure that interrelates
an i n p u t t o an o u t p u t in a g i v e n t i m e r e f e r e n c e .
The k e y concepts of a system a r e :

1. A system consists of parts connected together in accordance with


some s o r t o f p l a n , i.e. i t i s an o r d e r e d a r r a n g e m e n t .

2. A system h a s a time frame


3. A system h a s a cause-effect relation.
4. A system has the main function to interrelate an input and o u t p u t ,

e.g., s t o r m r a i n f a l l and s t o r m r u n o f f .

In the strictest definition, the systems approach is an overall one


and does not concern itself with details which may or may not be
important and which, in any case, may not be known. This seemingly
limits the systems approach to an attempt to get around the complex
geometry and physics of the hydrologic system. If we were solely
concerned with problems of identification (defined by Doodge as the
recognition of the overall nature of a system's operation, but not any
details of the nature of the system itself), t h i s attitude of ignoring the
details of the system would be a r e a s o n a b l e one. However, when we are
going to simulate a hydrologic system and its response, the elements
of physicaz hydrology become important. For instance, if we build or
use a model that is in c o n f l i c t with the physical realities, then we c a n

hardly expect to obtain good results from such a model, or even to be


able to calibrate the model to achieve good results. Thus, the systems
a p p r o a c h to stormwater modeling must consider the a s s i m i l a t i o n of process

models i n t o an o v e r a l l representation of the hydrologic cycle, or portions


thereof. How we1 I the model components must represent the different

processes depends on the purpose for the model and how m u c h data a r e
a v a i l a b l e w i t h w h i c h to v e r i f y the model.
In conclusion, the essence of systems analysis as applied to storm-
water modelling i s to interrelate rainfall (input) to stormwater (output)
w i t h a r e l i a b l e model in a c o m p u t a t i o n a l l y e f f i c i e n t m a n n e r .
TERM I NOLOGY AND DEF I N I T I O N S

There has been an e v o l u t i o n of systems j a r g o n , and i t i s important


to review the main p a r t s to b e t t e r u n d e r s t a n d h y d r o l o g i c m o d e l l i n g .
A variable has no f i x e d v a l u e ( e . g . , d i s c h a r g e ) whereas a parameter
i s a constant whose value v a r i e s w i t h the circumstances of i t s application
(e.g., Manning n - v a l u e ) .
The d i s t i n c t i o n between Linear a n d nonlinear systems i s of paramount
importance in understanding the mechanism of h y d r o l o g i c a l model I i n g .
A linear system is defined mathematically by a linear differential
equation, the principle of superposition applies and system response is

only a function of the system itself. An example of a linear system


representation is the unit hydrograph model. A nonlinear system is
represented by a nonl inear differential equation and system response
depends upon the system itself and the input intensity. An example of
a nonlinear system representation is kinematic overland flow. It i s well
known that real world systems are highly nonlinear, but linear
representations have often been made because the system is not under-

stood we1 I o r because of the pressures exerted b y resource c o n s t r a i n t s .

The state of a system is defined as the values of the v a r i a b l e s of


the system'at an instant in time. Hence, if we know exactly where a l l

of the stormwater i s and i t s flowrate in a basin, then we know the s t a t e


of the system. The s t a t e of a stormwater system i s determined e i t h e r from
historical d a t a o r b y assumption.
System memory i s the l e n g t h of time i n the p a s t over which the i n p u t
affects the present state. if stormwater from a basin today is affected
by the stormwater flow yesterday, the system ( w a t e r s h e d ) i s s a i d to h a v e
a f i n i t e memory. If it i s not a f f e c t e d a t all, the system has no memory;

and, if it is affected by storm flows since the b e g i n n i n g of the world,


the system i s said to h a v e i n f i n i t e memory. Memory of s u r f a c e water flow
systems i s mostly a f u n c t i o n of antecedent moisture c o n d i t i o n s .
A time-invariant system is one in which the input-output relation
is not dependent upon the time at which the input is applied to the
system. To i Ilustrate, unit hydrograph models represent the catchment

as a time-invariant system because the same unit hydrograph (response


function) is maintained throughout the storm regardless of variations
in watershed conditions. Usual l y , time-variance is considered among
storm events, seasons of the year, etc., and not within individual

storms. Time-invariance indicates constant land use, ground cover, and


drainage system configuration and capacity, and ignores soil moisture
v a r i a t i o n s a n d the effects of erosion.
199

A lwnped variable or parameter system is one in which the

variations i n space e i t h e r do not e x i s t or h a v e been ignored. Conversely,


a distributed parameter system recognizes spatial variations. The input
is said to be lumped if rainfall into a system is considered to be
spatial ly uniform. Lumped systems are represented by ordinary differen-
tial equations and distributed systems a r e represented by partial differ-
ent i a l equations.
A system i s said to be stochastic if for a given input there i s an
element of chance or probabi I i t y associated with obtaining a certain
output. A deterministic system has no element of chance in it, hence
for a given input a completely predictable output results for given
initial and/or boundary values. A purely random process has no deter-
ministic component and output is completely given to chance. A para-
metric or conceptual model does have an element of chance built into

i t since there a l w a y will be e r r o r s in verifying i t on r e a l data. I t does


therefore have a stochastic component. A alack boz model relates input
to output by an arbitrary function, and has no inherent physical
significance.
Model Optimization i s the o b j e c t i v e determination of the "best" values
for the model parameters u s i n g h y d r o l o g i c d a t a f o r the type of watersheds
and range of hydrologic conditions for which the model has been

designed. This function is I imited to parametric stormwater models, and


is a p p l ied in the r e g i o n a l izat ion process. To regionalize a model means
to develop a scientific basis for predicting the model parameters on
ungauged watersheds from hydrologic and physiographic characteristics
of that watershed. Regionalization can be accomplished o n l y if there a r e
enough benchmark watersheds with adequate periods of record that a
statistical inference can be drawn, i.e., statistically significant
parameter p r e d i c t i o n equations can be developed.
Model catibration basically is the fine-tuning of model parameter
values to achieve the best fit between observed and predicted runoff
hydrographs. To verify a model is to compare model predictions with
observed runoff values without adjusting parameter values to confirm
the model is doing a reasonable job in simulating the true watershed

response to known i n p u t .
Two concepts that are frequently confused (misused) are analysis
and simuhtion. The confusion with analysis stems from what it i s being
used to describe. As it relates to stormwater models, analysis is the
procedure used to calibrate a model to the data. It i s an attempt to

improve the state-of-the-art and i s fundamental l y a research a n d develop-


200

ment tool. Simulation, by contrast, utilizes the results of previous

analyses (and r e g i o n a l i z a t ion methods) to synthesize ( p r e d i c t ) stormwater


runoff from either design or real time rainfall on ungauged watersheds.
Simulations also can be performed at gauged watersheds to generate
runoff data for design events or events not contained in the available
record. Analysis i s often applied, for example, i n the context of s t u d y i n g
the probable performance of a storm sewer system during design storm
events. We are prone to say that we have analyzed the system and
found that it should work! Actually, what we are doing is using
simulation results to predict the probable performance characteristics
of the storm sewer system.

MODELL I NG APPROACHES

There a r e two conceptual approaches that h a v e been used i n develop-


ing stormwater models. An approach often employed in urban planning
has been termed deterministic modeling o r system s i m u l a t i o n . These models
have a theoretical structure based upon physical laws and measures of
initial and boundary conditions. When conditions are adequately
specified, the output from such a model should be known with a high
degree of certainty. In reality, however, because of the complexity of
the stormwater flow process, the number of physical measures required
would make a complete model intractable. Simp1 i f i c a t i o n s and approx-
imations must therefore be made. Since there are always a number of
unknown model coefficients and parameters that cannot be directly or

easily measured, it is required that the model be verified. This means


that the results from usable deterministic models must be verified by
being checked against real watershed data wherever such a model is
to be a p p l i e d .
The second conceptual stormwater modeling a p p r o a c h has been termed

parametric modellinq. In this case, the models are somewhat less


rigorously developed a n d g e n e r a l l y simpler i n approach. Model parameters
are not necessarily defined as measurable p h y s i c a l entities although they
are generally rational. Parameters for these models are determined by

fitting the model to hydrologic data with an optimization technique.


Application of parametric models to ungauged watersheds i s possible only
if regionalized parameter prediction equations are available and are
based on data from watersheds within the same geographical area and

with simi l a r geomorphic and land use characteristics as the watershed

being considered. As with deterministic models, user confidence stems

from v e r i f i c a t i o n studies u s i n g local d a t a .


20 1

EXAMPLES OF PARAMETRIC AND DETERMINIST I C MODELS

An excellent example of a parametric stormwater model is the TVA


S tormwat er Model (Betson, et al., 1980). T he model i s an e v e n t simulation

model formulated with a variation of the SCS c u r v e n u m b e r r u n o f f model


for determining rainfall excess and a unit hydrograph model. An event
model simulates the runoff from a one-time rainfall event, whereas a

continuous model simulates a time series of daily flows and hydrographs.


The curve number model was modified somewhat to include a constant
abstraction rate which allows for infiltration during lulls and a f t e r the
cessation of rainfall but before runoff was ceased. This introduced a
new parameter, PHI, which is analogous to the soil saturated hydraulic
conductivity, but which i s determined solely through optimization studies.
The unit hydrograph shape i s described with two triangles, the so-called
double triangle unit hydrograph, and r e q u i r e s f o u r parameters, the peak
flowrate and time to peak of the first triangle, the time base for both
hydrographs, and the time to peak of the second t r i a n g l e . A f i f t h model
parameter, the peak ordinate of the second triangle, is determined from
the constraint that the volume u n d e r the u n i t h y d r o g r a p h equal one b a s i n
i n c h o r mm o f r u n o f f .

The TVA developed regional prediction equations for each of the


model parameters using data from over 500 events on 38 rural, urban
and s u r f a c e m i n e d watersheds in the Tennessee Valley region. Using these
equations, the model can be applied to o t h e r watersheds within t h e same
physiographic regions with reasonable success, as demonstrated in
verification studies by Betson, et al. (1981). However, t h i s model should
not be used outside the limits of its regionalization. This was demon-
strated by Meadows, et al. (1983) in a study of the application of four
unit hydrograph models to watersheds in 14 physiographic provinces
across the U n i t e d States. The r e s u l t s which they obtained for e a c h model
basically were acceptable o n l y w i t h i n t h e r e g i o n s of t h e i r development.
Most deterministic models are formulated with the kinematic runoff
model, of which there are several, including the EPA Storm Water
Management Model (Metcalf and E d d y , et al., 1971). the USGS D i s t r i b u t e d

Rainfall-Runoff R o u t i n g Model (Dawdy, et al., 1978), and WITWAT (Green,


1984), to name a few. These models differ, but each i s formulated with
a 2 or 3 parameter model for infiltration, and kinematic overland and
channel routing. The infiltration model parameters generally can be
estimated from site measures or as typical values in textbooks and
pub1 i s h e d reports. Similarly, the routing parameters, e.g. Manning's
n-value, can be estimated from published sources. Thus, these models
202

are applicable to an ungauged site because model parameters generally


are measurable o r typical values are known. Confidence in model simu-
lation is high, but should be confirmed t h r o u g h v e r i f i c a t i o n s t u d i e s once
local d a t a became a v a i l a b l e .
The best of both worlds is illustrated by the USGS model. It can
be a p p l i e d d i r e c t l y as a deterministic model, or if local calibration data
are a v a i lab1 e, the soi I-moi s t u r e accounting and i n f i I t r a t ion parameters
can be optimized. The USGS terms this version of the model a p a r a m e t r i c -
d e t e r m i n i s t i c r u n o f f model (A1 ley, et a l . , 1980).
Engineers have designed drainage systems for decades using the
we1 I-known Rational Method, and have simulated watershed runoff with
unit hydrograph models, e.g. SCS curvilinear unit hydrograph. Why is
it necessary, or even useful, to work with kinematic stormwater models
now? The answer to this question lies in an examination of what kine-
m a t i c models w i l l do f o r the engineer - a n d perhaps i t also l i e s i n what
the other methods w i l l not do.
First, the role of models in general should be acknowledged.
E n g i n e e r i n g design of drainage systems a n d environmental impact assess-
ment of land use change require informat ion about watershed response
to prescribed "design" events which most often are extreme events.
Since most s m a l l b a s i n s a r e not gauged f o r both r a i n f a l l a n d streamflow,
little hydrologic data is available to quantify the necessary response

characteristics. Further, if a watershed is gauged, it is unlikely that


a suitable "design" event, is contained in the record unless the gauge
has been in operation for many years. Even so, the data are for the
watershed response in i t s current l a n d use c o n d i t i o n and a r e not a true
measure of the watershed response f o l l o w i n g l a n d use change. To p r o p e r l y
quantify the watershed response for "after" development conditions, the
record would h a v e to be extended f o r several y e a r s to i n s u r e the p r o b a b -

ility of an adequate number of acceptable events. But the land use


change must still be planned, the associated drainage system designed,
and impact statements prepared. We do not have the luxury of being
a b l e to wait for the data to be collected, so we must r e s o r t to p r e d i c t i o n
methods.

It is widely accepted that mathematical stormwater models are the


only a v a i l a b l e means of m a k i n g r e l i a b l e prGdictions of watershed response
to design events and of the effects of land use change on stormwater
runoff and q u a l i t y . I t must be s t a t e d e m p h a t i c a l l y , however, that models
are not a substitute for field gathered data or knowledge of the
h y d r o l o g i c / h y d r a u l i c and water qua1 i t y processes on the p a r t of the user.
No model can predit how a natural system behaves as dependably as
direct measurements of the system itself. The principal use of models

is in situations where direct measurements are either impossible or


impractical, such as the "after" development conditions. When a d r a i n a g e
system is under design, for example, a model will l e t the designer look
at many alternative configurations. More importantly, the designer can
answer the "what if" questions, and can do so within a reasonable
framework of time a n d costs. Models also p e r m i t a more a c c u r a t e a n a l y s i s
of complex watershed and drainage systems. The advent of models has
changed the engineer from a cookbook artist who relied heavily on
judgement to a serious a n a l y s t a n d p l a n n e r .
The selection of a model typically i s a statement of user confidence,
which has been defined as "the belief in the reliability or credibility
of the results and exists either consciously or subconsciously in the
minds of the model user o r clientele" (ASCE, 1983). This belief i s derived
from experiences in the use, development, or testing of a model, from

user understanding of watershed hydrologic processes and model repre-


sentation of these processes, and from confidence in authority, e.g.,
textbooks, technical journals, a n d federal agency endorsement. Ultimately,
confidence i s founded on verification studies at the watershed where the
simulations a r e required.
The keyword is "reliable". When using a model, one must remember
the model is merely a mathematical expression of the true system and
cannot account for all the s u b t l e t i e s of the v a r i o u s phenomena (processes)
involved. Reliable r e s u l t s a r e those o n w h i c h the model user can foster
the belief that if such an e'vent occurs, the p r o b a b l y runoff hydrograph
w i l l be v e r y much l i k e the model p r e d i c t i o n s .
So why use kinematic models? Perhaps the best answer i s that they
are deterministic, distributed parameter models that can account for the
spatial watershed and rainfal I variations and the nonlinearity of the
runoff process. In other words, kinematic models are a better model of
the true process. Because kinematic models are based' on the physics
of the runoff process; the model structure is rational, the parameters
are measurable or are available from published studies and textbooks,
a n d the model can be a p p l i e d with a minimum of calibration data. (They

can be applied in the absence of calibration data; user confidence is


supported by the extensive testing and documentat ion of kinemat i t
models.) Though young in evolution, there are now several models
a v a i l a b l e for computer use, even personal computer use. Thus, kinematic
models are as readily used as other models and have the advantages
offered b y d e t e r m i n i s t i c models.
204

Fig. 10.2 Contour p l o t of topography

TWO - D I MENS I ONAL OVERLAND FLOW MODELL I NG

Topography a n d catchment s u r f a c e c h a r a c t e r i s t i c s can not be p r o p e r l y

accounted for in one-dimensional models in all cases. The cone shaped


catchment is a typical example. Also the effect of varying surface
roughness, slope and losses is often two-dimensional. Storm patterns
cannot be accounted for properly and the assumption of a rectangular
hyetograph over the entire catchment is often dangerous. It may be
necessary in the case of complex catchments or rain to resort to two-
dimensional model l i n g .

Two-dirnensiona I k i nernat ic equations

One - d i m e n s i o n a l equations can be extended into two dimensions as


fol lows:
205

Fig. 10.3 Topography i n 3-Dimensions

The c o n t i n u i t y e q u a t i o n becomes

(10.1)

where q is the flow in the x direction (m2/s) and qz is the flow in


the z d i r e c t i o n ( m ’ / s ) .
A proof of this equation can be found in Dronkers ( 1 9 6 4 ) . For two
dimensional flow two motion equations are required. In kinematic theory
these a r e o b t a i n e d b y assuming

sox = sfx (10.2)

soz = Sfz (10.3)

where S i s the bed slope i n the x d i r e c t i o n , So* i s the bed slope in


ox
the z d i r e c t i o n , S f x i s the f r i c t i o n slope i n the x direction a n d 5 fz is
t h e f r i c t i o n slope i n the z d i r e c t i o n .
For the general form of headloss equation one can o b t a i n
1 m
qx = 4 b,Y )z
(10.4)

(10.5)

(10.6)

and a x = f u n c t i o n of 5
ox
a z= f u n c t i o n of S
02

This idea for two dimensional flow was used by Orlob (1972). It
will be noticed that qt is always positive while q and q can be

Positive Or negative as ( a x ) 2 and are functions of sox a n d Soz


respective1 y.
Boundary conditions

There are two boundary conditions that can be used on watersheds.


One can assume that the water depth at the boundary is always zero
and that all the water entering the origin leaves it in the form of a
discharge. This has been assumed in all existing theories. One must
then define
(10.7)

yk = 0.0 (10.8)
I

so qk = i ~ x / 2 (10.9)
I e
One could alternatively assume that the discharge at the origin is
controlled by the depth of water at the origin as assumed for the rest
of the points. For this case we must then use the same equations as
with the other point's. The effect of using the two different boundary

conditions w i l l be shown later.

I n i t ia I conditions

After the first time step it may be assumed that the water depth
at al I points, except at the origin in the case of the first boundary
condition, for the case of a n i n i t i a l l y d r y catchment

(10.10)

The proposed equations may be solved at grid p o i n t s over a plane


provided runoff is adequately described by the kinemat i c equations.
Where there a r e flow concentrations such as an i n l a n d depression, storage
will not be accounted for except with a separate routine to account for
net volume stored. If outflow eventually occurs when the depression is
f i Iled, a g a i n a s e p a r a t e r o u t i n e i s needed to detect this.
The effect of channelization, for example rills and furrows in which
runoff collects, can be accounted for by reducing the effective dx or

dy over which runoff occurs. Where channel side friction is applicable


however condui t equations may be r e q u i r e d .
The effect of spatially varying soil types a n d cover can also affect

losses to a significant extent. l n f i Itration, and possible re-emergence


of interface flow can be accounted for with a two-layer model with per-
meable i n t e r f a c e . A sample of such a model i s discussed l a t e r .

Fig. 10.4 F l o w directions 0 from model


207

2000 00

.o +.P +.o .+.o. -0.0 +1eo.o+lso.o+~eo.o+rso.o+1so.o

f 1900

+reo.o+-a6.13+.0 +o.o +iae.e--1500

+iii.7 +ieo.o +12s.s +ias.o+1213.9--1400

+71.8 - 0.0 +l~E.E+l7l.E+?*8.,0--~300

+70.6 4-7E.e -0.0 +155.4+1EO.E--1200

+?ie.8+107.6+7i.6 +o.o +?se.7--1 100

+i4i.e+iee.e+se.a +o.o +iss.4--1 OD0

a.e +47.7 +o.o +7e.e +7e.e +4.e +o.o +i4?.a+?4,e+s~.~

f ' 70.0
208

REFERENCES

Alley, W.M., Dawdy, D.R. and S c h a a k e , J.C., Jr., 1980. Parametric-


d e t e r m i n i s t i c urban w a t e r s h e d m o d e l . J. H y d r . D i v . ASCE, V o l . 106, No.
HY5, pp. 679-690.
ASCE, 1983. Q u a n t i f i c a t i o n o f l a n d u s e c h a n g e e f f e c t s u p o n h y d r o l o g y , b y
t h e T a s k C o m m i t t e e o n Q u a n t i f y i n g Land Use C h a n g e E f f e c t s , R.P. B e t s o n ,
Chmn., p r e s e n t e d a t t h e J u l y 20-22, 1983. ASCE I r r i g a t i o n and D r a i n a g e
D i v i s i o n S p e c i a l t y Conference, h e l d a t Jackson, Wyoming.
B e t s o n , R.P., B a l e s , J . and P r a t t , H.E., 1980. U s e r s G u i d e t o TVA-
HYSIM, A h y d r o l o g i c p r o g r a m f o r q u a n t i f y i n g l a n d u s e c h a n g e e f f e c t s .
EPA-600/7-80-048, Tennessee Va l ley A u t h o r i t y , W at er Systems Development
B r a n c h , N o r r i s , Tennessee.
B e t s o n , R.P., B a l e s , J. and D e a n e , C.H., 1981. M e t h o d o l o g i e s f o r a s s e s s -
i n g s u r f a c e m i n i n g i m p a c t s . R e p o r t No. WR28-1-550-108, Tennessee V a l l e y
A u t h o r i t y , W a t e r S y s t e m s D e v e l o p m e n t B r a n c h , N o r r i s , Tennessee.
D a w d y , D.R., S c h a a k e , J.C., J r . and A l l e y , W.M., 1978. U s e r ' s g u i d e f o r
d i s t r i b u t e d r o u t i n g r a i n f a l I - r u n o f f m o d e l . U.S. Geological Survey Water
R e s o u r c e s I n v e s t i g a t i o n s 78-90.
DJoge, J.C.I., 1973. L i n e a r t h e o r y o f h y d r o l o g i c s y s t e m s . U.S. Dept. o f
A g r i c u l t u r e , A g r i c u l t u r a l R e s e a r c h S e r v i c e , T e c h . B u l I . No. 1468.
Dronkers, J.J., 1964. T i d a l c o m p u t a t i o n s i n r i v e r s and c o a s t a l w a t e r s .
N o r t h H o l l a n d P u b l i s h i n g Co., A m s t e r d r a m .
Green, I .R.A., 1984. WITWAT stormwater d r a i n a g e program - Theory,
Applications and U s e r ' s M a n u a l . Report No. 1/1984, Water Systems
Research Programme, Dept. of Civi I Engineering, University of the
W i twatersrand, Johannesburg, South A f r i c a .
M e a d o w s , M.E., H o w a r d , K.M. and C h e s t n u t , A . L . , 1983. D e v e l o p m e n t o f
models f o r s i m u l a t i n g stormwater r u n o f f from s u r f a c e coal mined l a n d s :
U n i t h y d r o g r a p h m o d e l s . R e p o r t No. G5115213, V o l . 1 , U.S. D e p t . o f t h e
I n t e r i o r , O f f i c e o f S u r f a c e M i n i n g , D i v i s i o n of R e s e a r c h , W a s h i n g t o n ,
D.C.
O r l o b , G.T., 1972. M a t h e m a t i c a l m o d e l l i n g o f e s t u a r i a l s y s t e m s . Interna-
t i o n a l Symposium o n m a t h e m a t i c a l m o d e l i n g t e c h n i q u e s in w a t e r r e s o u r c e s
systems, E d i t o r Asi t K . Biswa s. Pro ce e d in gs Volume 1 .
M e t c a l f and E d d y , Inc.,. University of F l o r i d a , and W a t e r R e s o u r c e s
Engineers, 1971. Storm w a t e r management model. U.S. Environmental
P r o t e c t i o n A g e n c y , W a s h i n g t o n , D.C.
Overton, D.E. and M e a d o w s , M.E., 1976. S t o r m w a t e r M o d e l i n g , A c a d e m i c
P r e s s , New Y o r k , N.Y.

Fig. 10.5 Water depth v a r i a t i o n a t t = 8 m i n o v e r the catchment


209

CHAPTER 11

APPL I CAT IONS OF K I NEMAT I C MODELL I NG

APPROACHES

This chapter contains examples of kinemat ic stormwater simulation


models and their application to rural and urban watersheds. These

models were selected from the range of available models because they
are simple in concept and structure, have been tested extensively, and
are representative of the approaches taken in developing kinematic
watershed models. For these reasons, they should help the r e a d e r to more
f u l l y u n d e r s t a n d k i n e m a t i c m o d e l s and t h e i r a p p l i c a t i o n s .
The reader is reminded that with any watershed model, approx-
imations and simplifications are made. Previously, we have seen that
kinematic models are simplifications to the dynamic wave models; and
that their solution, whether analytical or numerical, requires approx-

imations to the watershed geometry, drainage layout, rainfal I pattern,


etc. The examples in this chapter illustrate different approaches to
m a k i n g these a p p r o x i m a t ions,

A MODEL FOR URBAN WATERSHEDS

A model that has been successfully applied to urban watersheds is


the U.S. Geological Survey model, DR3M (Dawdy, et al., 1978). T h i s model
combines the soi I moisture accounting and rainfal I excess components

of the model developed by Dawdy and o t h e r s (1972) with the kinematic


wave routing components of the model developed by LeClerc a n d Schaake

(1973). Input to the model includes daily rainfall, storm rainfall, daily
pan evaporation and a physical definition of the drainage basin
discretized into as many as 50 segments, including overland flow,
channel and reservoir segments. D u r i n g storm days, t h e model generates
a simulated discharge hydrograph based on input data from as many
as three rain gauges. Th e model consists of two m a i n sets o f components:

parametric rainfal I excess and deterministic runoff routing components.

Parametric Rainfal I Excess Components

The parametric rainfal I excess components are a soil moisture


accounting component, an infi I t r a t ion component, an impervious area

rainfal I excess component, and an optimization component. A substantial


210

part of the rainfa I excess components was adopted from a model


developed by Dawdy et al. (1972). T h i s component i s used d u r i n g model
calibration to estab i s h opt irnal parameter values for site i n f i I tration
a n d soi I moisture storage functions.

Soil Moisture Accounting

The soil moisture accounting component determines the effect of ante-


cedent conditions on infiltration. Soil moisture is modelled as a two
layered system, one representing the antecedent base moisture storage
(EMS), and the other, the upper wetted part caused b y infiltration into
a s a t u r a t e d moisture storage (SMS).
During rainfall days, moisture i s added to SMS based on the P h i l i p
i n f i l t r a t i o n equation (Philip, 1954). On o t h e r d a y s , a specified proportion
of daily rainfall (RR) infiltrates into the soil. Irrigation ( f o r example,
lawn watering) can be included in the daily water balance. This is
achieved through user supplied irrigation rates for each month. If a
daily precipitation is less than the daily irrigation rate, the daily

precipitation i s set equal to the i r r i g a t i o n r a t e .


Evapotranspiration takes place from SMS, based on availability,
otherwise from BMS, with the r a t e determined from p a n e v a p o r a t i o n m u l t i -
plied by a pan coefficient (EVC). Moisture in SMS drains i n t o BMS w i t h
a controlling parameter (DRN) determining the rate. Storage in BMS has
a maximum v a l u e (BMSN) e q u i v a l e n t to the f i e l d c a p a c i t y moisture storage
of an active zone. Zero storage in BMS is assumed to correspond to
wilting point conditions in the active soil zone. When storage in BMS
exceeds BMSN, the excess i s s p i l l e d to deeper storage. These s p i l l s c o u l d
be the b a s i s for routing interflow and baseflow components, if desired.
However, t h i s option i s not included in the present version of the model.
A schematic flow c h a r t of the s o i l moisture a c c o u n t i n g component i s shown
in Figure 11.1.

I n f i l t r a t i o n Component

Infiltration is computed with the Philip equation (Philip, 1954),


which is merely a variation to the Green and Ampt equation. One form
of the Green a n d Ampt equation is

(11.1)

where F is the accumulated infiltration depth, K is the effective


21 1

RAINFALL
I NPUT

I EVAPO-
TRANSPIRATION
I COMPUTE:
INFILTRATION

1
, DRAINAGE TO.- USE BMS
WITH RGF
BMS TO COMPUTE
t
PS 1

SPILL TO:
DEEPER
STORAGE

Fig. 11.1 S c h e m a t i c of DR3M s o i l m o i s t u r e a c c o u n t i n g c o m p o n e n t


21 2

hydraulic conductivity, H is the depth of water ponded on the soil


surface, P i s the w e t t i n g f r o n t section, a n d Z i s the depth to the w e t t i n g
front. Using the r e l a t i o n s h i p

z = -0 F
- 0.
(11.2)
S I

i n which BS is the volumetric soil moisture content at saturation and


0 . is the initial (unsaturated) moisture content, Eq. 11.1 is transformed
i n t o the Phi I i p equation.

H + P(3s-0i)
-
dF
dt
= K[l +
F
1 (11.3)

Since the wetting front suction is generally several orders of magnitude


greater than the depth of ponded water, the H term may be ignored.
The mnemonic i d e n t i f i e r s used to designate the resulting infiltration are

FR = KSAT ( 1 + SMSz) (11.4)

i n which FR=dF/dt, KSAT=K, PS=P(@- @ . ) , a n d SMS=F.


S I
The wetting front suction i s not constant, but varies with the soil
moisture c o n d i t i o n . The e f f e c t i v e v a l u e of PS i s assumed to v a r y linearly
between a wilting point and field capacity, and is computed with the
relationship
PS = P S P ~ R G F - (RGF - 1 b$$ (11.5)

i n which BMS is the initial moisture storage in the soil column; BMSN
is the moisture storage in the soil column at field capacity; PSP i s the
effective value of PS at field capacity; and RGF is the ratio of PS a t
wilting point to that at field capacity. This relationship is shown in
Figure 11.2.
Point potential infiltration (FR) computed by the Philip equation
is converted to effective infiltration over the basin using the scheme
of Crawford and Linsley (1966). Letting SR represent the supply rate
of rainfal I for infiltration and OR represent the rate of generation of
r a i n f a l I excess, the equations a r e

OR = -’SR . i f SR < FR (11.6a)


2F R
FR .
OR = SR - -; I f SR > FR ( 1 1 .6b)
2

A schematic of these r e l a t i o n s h i p s i s shown i n F i g u r e 11.3. The rainfall


excess rate, OR, is represented by the a r e a between the dashed SR line
and the linear infiltration capacity curve. The parameters for soil

moisture accounting and infiltration are enumerated in the following:


21 3

LL
0 RGF X PSP
w
3
B
w
>
5w PSP
LL
LL WILTING FIELD
w
POINT CAPACITY
(BMS = 0 1 (BMS=BMSN 1
SOIL- MOISTURE CONTENT

Fig. 11.2 Relationship determining effective value of soil-moisture


p o t e n t i a l (PS)

/tF
t I I 1

w, 'R

v, RAINFALL
INFILTRATION

-
z
a
a
0 25 50 75 100

PERCENTAGE OF AREA WITH INFILTRATION CAPACITY


EQUAL TO OR LESS THAN INDICATED VALUE

Fig. 11.3 Relationship determining rainfall excess (OR) as function


of maximum infiltration capacity (FR) a n d s u p p l y r a t e of
r a i n f a l l (SR)
214

1. Soil Moisture Accounting. The parameters consist of: (a) DRN -


A constant drainage r a t e for redistribution of soil moisture between SMS
and BMS, in inches per day; ( b ) EVC - A pan coefficient for converting
measured pan evaporation to potential evapotranspiration; (c) RR - The
average proportion of daily rainfall that infiltrates into the soil for
the period of simulation excluding storm days; and (d) BMSN - Soil
moisture storage a t f i e l d c a p a c i t y , i n inches.
2. Infiltration. The parameters consist o f : ( a ) KSAT - The h y d r a u l i c
conductivity at natural saturation, in inches per hour; ( b ) RGF - Ratio
of suction at the wetting front for soil moisture a t wilting point to that
at f i e l d c a p a c i t y .

I m p e r v i o u s A r e a R a i n f a l I Excess Component

Two types of impervious surfaces a r e considered by the model. The


first type, effective impervious surfaces, a r e those impervious areas that
are directly connected to the channel drainage system. Roofs that drain
onto driveways, streets and paved parking lots that drain onto streets
are examples of effective impervious surfaces. The second type, non-
effective ihpervious surfaces, are those impervious areas that drain to
pervious areas. An example of a noneffective impervious area is a roof
that d r a i n s onto a lawn.
The only abstraction from rainfall on effective impervious areas is
impervious retention. Thi's retention, which is user specified, must be
filled before runoff from effective impervious areas can occur. Evapo-
ration occurs from impervious retention during periods of no rainfal I.
Rain falling on noneffective impervious areas is assumed to runoff
onto the surrounding pervious area. The model assumes this occurs
instantaneously and that the volume of runoff is uniformly distributed
over the contributing pervious area. This volume is added to the rain
fa1 l i n g on the pervious areas prior to computation of pervious area
r a i n f a l I excess.

O p t i m i z a t i o n Component

An option is jncluded in the model to calibrate the soil moisture


and infi Itration parameters for d r a i n a g e b a s i n s h a v i n g measured r a i n f a l I
runoff data. The method of determining optimum parameter values is
based on an opt imizat ion technique devised b y Rosenbrock (1960).
21 5

Impervious area is not included as a parameter to be optimized,


but i s a parameter to w h i c h simulated r u n o f f volumes a r e v e r y sensitive.
Therefore, values of imperviousness should be determined accurately
before u s i n g the o p t i m i z a t i o n option. I f i n i t i a l estimates of imperviousness
are grossly in error, resulting volumes and peaks will be grossly in
error. In that case, estimates of imperviousness must be adjusted by
the modeler. This adjustment may include revising estimates of the
effective and noneffective impervious areas, perhaps by trial and e r r o r .

Determini st i c Runoff R o u t i n g Components

After determining "opt imum" parameter values and computing the


time series of rainfall excess, control i n the model i s transferred to the
runoff routing component. The mathemat ical representation of an urban
basin requires discretization of the total drainage area into a set of
segments. There a r e three b a s i c types of segments d e f i n e d f o r the model:
channel segments, overland flow segments and reservoir segments. There
is wide flexibility to the approach one can take in dividing a basin
into segments for r u n o f f computations. Guide1 ines for basin segmentation
are presented by Alley and Veenhuis (1979) and Dawdy, et al., (1978).

Channel and O v e r l a n d Flow Segments

A channel segment is permitted to receive upstream inflow from as


many as three other segments, including other channel segments and
reservoir segments. It also may receive lateral inflow from o v e r l a n d flow
segments. T h e o v e r l a n d flow segments receive u n i f o r m l y d i s t r i b u t e d l a t e r a l
inflow from rainfal I excess. A schemat i c illustrating the relationships
between channel a n d o v e r l a n d flow segments i s shown in F i g u r e 11.4 - 5.
Kinematic wave theory is applied in the . r a i n f a l l runoff model to
both overland flow anci channel routing. The necessary equations to be
solved f o r each channel a n d o v e r l a n d flow segment a r e
aa
- aA
(11.7)
ax + at = q;
b
Q = aA (11.8)

in which the terms a r e as p r e v i o u s l y defined.


Finite difference approximations are used to solve Eqs. 11.7 and
11.8. To avoid the convergence and stability problems that can occur
with particular numerical grid spacings (i.e. the relative sizes of A t
and Ax), two finite d i f f e r e n c e methods of solution a r e used to solve f o r
216

JOVERIAND FLOW[
[oEFNDi
I

PHYSICAL CHARACTERISTICS OF COMPONENTS


IN THE SCHEMATIC REPRESENTATION

Fig. 11.5 D i s c r e t i z a t i o n of u r b a n catchment i n t o segments


21 7

Q and A at the unknown grid points. The chosen solution procedure is

made in the model program and depends upon the ratio, G , of the
k i n e m a t i c w a v e s p e e d t o Ax/At.

(11.9)

in which Q3 is the discharge at node point 3 in the finite difference


grid as shown in Figure 11.6. If 0 is greater than or equal to unity,
the equations used a r e

This involves only mesh p o i n t s 1, 2 and 4. If 0 is less than unity, the


equations used a r e
At
A 4 = A + q t + - (11.12)
3 Ax ('1 - '3)
b
Q4 = aA ( 1 1 .13)
4

Ax a n d At values are chosen to ensure about 10 nonzero ordinates


under the rising limb of an equilibrium hydrograph and to keep
computat ional errors within acceptable bounds. The U.S.G.S. recommends
t h a t A t be selected as
At = 0.1 (t
eo +
1 (11.14)

where t and tec a r e t h e k i n e m a t i c o v e r l a n d a n d c h a n n e l times of


eo
equilibrium, respectively. Computational error should be minimized if
Ax and At are selected so that the characteristic passing through point
1 also passes through point 4. A c c o r d i n g l y , it is recommended that ux

be selected a s
LO
ax = - At (11 . 1 5 a )
eo
Lc a t
Ax = - ( 1 1 .15b)
C
ec
for the overland and channel segments, respectively. When Eqs. 11.14
and 11.15 result in non-integer values, the user must round to the
nearest integer.

Reservoir Segments

Provision is made in the model for reservoir routing based o n the


continuity equation. Either of two routing methods can be used. One

method i s l i n e a r storage r o u t i n g
21 8

Fig. 11.6 Four p o i n t f i n i t e d i f f e r e n c e g r i d

5 = co ( 11 .16)

in which S is the storage; Q is the outflow; and C is a constant.

Alternatively, the m o d i f i e d P u l s r o u t i n g method can be used

- +At o2 = I) + l2 + --
At
( 1 1 .17)

i n which I is the inflow to the reservoir and the subscripts 1 and 2


refer to the beginning and end of the time interval, At, respectively.
The modified Puls method utilizes a t a b l e of storage outflow values as
s u p p l i e d b y the user.

There are many ways of accounting for storage with the k i n e m a t i c

method, all of which are, due to the limitation of the k i n e m a t i c method,


approximations. That is, due to the f a c t that the dynamic e q u a t i o n omits
acceleration and deceleration of the water in time and space, wedge
storage i s omitted. Storage can o n l y be i n c l u d e d as channel t y p e storage
(see Chapter 9) if the discharge relationship can be described i n terms
of a kinematic t y p e of d i s c h a r g e - depth equation.
The continuity equation in the kinematic equations can however
be used to account for the lag effect of storage in one of two ways.
Inflow to a r e a c h can be spread over the f u l l s u r f a c e a r e a of the reach
as demonstrated in the example later i n this chapter, or the stream (or
overland flow) surface area can be replaced b y a storage area at the
junctions of reaches with conduit reaches as indicated in the program
i n Chapter 8.

Example Application

The model was applied to the Sand Creek Tributary watershed near
Denver, Colorado. This drainage basin is a 183 acre area of predom-
inantly single family residential land use with some multifamily land
use, a church, a r e c r e a t i o n a l center, a f i r e station, a n d two s m a l l p a r k s .
The basin has some storm sewers in i t s upper end but r e l i e s mostly on

street gutters and concrete lined open ditches for flow conveyance.
Detailed records of rainfal I and streamflow are collected at 5 minute
intervals. A stage discharge relation was developed using flow profile
a n a l y s i s a n d d i s c h a r g e measurements made d u r i n g storm r u n o f f .
Two sample runs are discussed. The first run was an optimization
run to calibrate the model on an antecedent period of record. In the
second run, the soi I moisture accounting and i n f i l t r a t i o n parameters were
set to their final values from the first run and ten storm events were
simulated.
Before any simulations were performed, the watershed was delineated
i n t o sub-basins (overland flow segments) and a drainage network
(channel segments). A schematic showing how the watershed was
approximated with the overland and channel segments is given in
F i g u r e 11.7.

The rationale behind the basin segmentation is as follows: starting


at the basin o u t f a l I, it was first noted that the major drainage system
of the basin consisted of concrete lined ditches which were located in

the positions marked by channel segments CH20, CH21 and CH22. In


analyzing the r e a c h of concrete l i n e d d i t c h comprised o f CH20 a n d CH21,

it was noted that a street, which drained 14 acres of l a n d Q(F03, i n t e r -


sected this reach. Therefore, t h i s r e a c h of concrete l i n e d d i t c h was sub-
divided into channel segments CH20 a n d CH21, and the intersecting street
was designated as channel segment CH23. Overland flow segments QIFOl,
220

,c)Fo2. OVERLAND FLOW SEGMENT AND NUMBER


----
-
CH23 CHANNEL SEGMENT
JTOl JUNCTION AND NUMBER
GENERAL DIRECTION OF OVERLAND FLOW

Fig. 11.7 Schematic representation of Sand Creek tributary watershed


22 1

OFO2, OF03, OF04 and OF09 were then delineated based on this channel
segmentation. It should be noted that overland flow segment F04 does
not h a v e b a l a n c e d lengths of o v e r l a n d flow to CH21. To f u r t h e r s u b d i v i d e
this overland flow segment would also require that segments CH21 and
OF09 be f u r t h e r s u b d i v i d e d .
The unallocated concrete lined ditch was then assigned as channel
segment CH22. O v e r l a n d flow segments OF06 and OF05 were then delineated.
To a v o i d the need to s u b d i v i d e channel segment CH20 which would r e q u i r e
subdividing overland flow segments OF01 a n d OF02, channel segment CH25
was used to bypass channel segment CH20. A junction segment, JTOl,
was required to sum the flow from the two channel segments a t the o u t l e t
of the basin. Finally, the remaining part of the basin was drained by

a street which was assigned as channel segment CH24.


Once the basin was segmented, the sub-basin boundaries were field
checked and representative channel cross sections determined. Channel
slopes were determined from the d r a i n a g e maps, and overland flow slopes
were estimated from the U.S. Geological Survey topographic map for the
area and the street corner elevations shown on the City of Denver

drainage maps. Sub-basin areas were planimetered and lengths of over-


land flow were computed by dividing the area of each sub-basin, in
square feet, by the length, in feet of the channel segment into which
i t contributes lateral inflow.

TABLE 11.1 Model Simulation Results f o r Sand Creek T r i b u t a r y Watershed

Runoff Runoff Volume Peak Flow


Event . .
Date i n inches in cfs
Number Measured Simulated Measured Simulated

1 7-7 2-73 0.08 0.08 32 23


2 7-1 9-73 0.16 0.19 68 74
3 7-22-73 0.055 0.052 22 14
4 7-24-73 0.33 0.28 104 97
5 7-30-73 0.063 0.082 32 28
6 8-07-73 0.70 0.76 2 36 2 80
7 9-1 1-73 0.073 0.14 48 58
8 9-1 1-73 0.23 0.16 143 68
9 7-22-74 0.20 0.32 98 117
10 7-30-74 0.53 0.47 251 2 16
222

The p e r i o d of record to be simulated was J u l y 12, 1973 to J u l y , 30,


1974. To establish initial moisture conditions for the beginning data,
an optimization run was conducted for the period May 1, 1973 to July
12, 1973. Using input values for rainfall daily pan evaporation and
recorded runoff the model was "calibrated" by determining optimal values
for the infiltration and soil moisture accounting parameters. A second

r u n was then conducted to s i m u l a t e watershed r u n o f f (hydrographs) during


the specified s i m u l a t i o n p e r i o d . Results a r e shown i n Table 11.1.

A MODEL FOR RURAL WATERSHEDS

The model described for u r b a n stormwater s i m u l a t i o n c o u l d be a p p l i e d


easily to rural watershe