HYDRAULIC

DESIGN

HANDBOOK

Larry W. Mays, Editor in Chief

Department of Civil and Environmental Engineering
Arizona State University
Tempe, Arizona

McGraw-Hill
New York San Francisco Washington, D.C. Auckland Bogota Caracas Lisbon
London Madrid Mexico City Milan Montreal New Delhi San Juan
Singapore Sydney Tokyo Toronto

Library of Congress Cataloging-in-Publication Data

Hydraulic design handbook/Larry W. Mays, editor-in-chief.
p. cm.
Includes bibliographical references and index.
ISBN 0-07-041152-2
1. Hydraulic structuresDesign and costruction Hanbooks,
manuals, etc. I. Mays, Larry W.
TC180.H94 1999
627dc21
99-20240
CIP
McGraw-Hill
^n
<x$
A Division of The McGraw-Hill Companies
Copyright () 1999 by The McGraw-Hill Companies, Inc. All rights reserved. Printed
in the United States of America. Except as permitted under the United States
Copyright Act of 1976, no part of this publication may be reproduced or distributed in
any form or by any means, or stored in a data base or retrieval system, without the
prior written permission of the publisher.
3 4 5 6 7 8 9 0 DOC/DOC

043210

ISBN 0-07-041152-2
The sponsoring editor for this book was Larry Hager and the production supervisor
was Pamela A. Pelton. It was set in Times Roman by Compuvision.
Printed and bound by R. R. Donnelley & Sons Company.
This book was printed on acid-free paper.
McGraw-Hill books are available at special quantity discounts to use as premiums
and sales promotions, or for use in corporate training programs. For more information, please write to the Director of Special Sales, McGraw-Hill, Inc. Two Penn Plaza,
New York, NY 10121-2298. Or contact your local bookstore.
Information contained in this work has been obtained by The
McGraw-Hill Companies, Inc. ("McGraw-Hill") from sources
believed to be reliable. However, neither McGraw-Hill nor its
authors guarantee the accuracy or completeness of any information
published herein, and neither McGraw-Hill nor its authors shall be
responsible for any errors, omissions, or damages arising out of
use of this information. This work is published with the understanding that McGraw-Hill and its authors are supplying information but are not attempting to render engineering or other professional services. If such services are required, the assistance of an
appropriate professional should be sought.

A. Osman Akan Old Dominion University (CHAP 14)

Bayard Bosserman II Boyle Engineering Corporation (CHAP 10)
Herman Bouwer USDA-ARS Water Conservation Laboratory (CHAP 24)
Albert J Clemmens USDA-ARS Water Conservation Laboratory (CHAP 21)
Sharon L. Cole Greeley and Hansen (CHAP 22)
H. Wayne Coleman Harza Engineering Company (CHAP 8 and 17)
George K. Cotton Simons and Associates, Inc. (CHAP 16)
Roy G. Dodson Dodson and Associates, Inc. (CHAP 19)
Richard H. French Desert Research Institute, University and
Community College System of Nevada (CHAP 3 and 5)
Marcelo Garcia University of Illinois at Urbana-Champaign (CHAP 6)
David Hobbs Greeley and Hansen (CHAP 22)
Paul Johnson Arizona State University (CHAP 23)
William L. Judy Greeley and Hansen (CHAP 22)
Bryan W. Karney University of Toronto (CHAP 2)
Mariush W. Kemblowski Utah State University (CHAP 4)
Kevin Lansey University of Arizona (CHAP 9)
James E. Lindell Harza Engineering Company (CHAP 8, 17 and 18)
Federico E. Maisch Greeley and Hansen (CHAP 22)
James L. Martin AScI Corporation (CHAP 5)
C. Samuel Martin Georgia Institute of Technology (CHAP 12)
Larry W. Mays Arizona State University (CHAP 1,9, 15 and 20)
Steve C. McCutcheon U.S. EPA National Exposure Research Laboratory (CHAP 5)
Clifford A. Pugh U.S. Bureau of Reclamation (CHAP 21)
John A. Replogle USDA-ARS Water Conservation Laboratory (CHAP 21)
Stuart M. Stein GKY & Associates (CHAP 13)
Frank J. Tantone Greeley and Hansen (CHAP 22)
I. Kaan Tuncok Stanley Consultants, Inc. (CHAP 15)
Yeou-Koung Tung Hong Kong University of Science and Technology (CHAP 7)
Gilberto E. Urroz Utah State University (CHAP 4)
C. Y. Wei Harza Engineering Company (CHAP 8, 17 and 18)
Ben C. Yen University of Illinois at Urbana-Champaign (CHAP 14)
G. Kenneth Young, Jr. GKY & Associates (CHAP 13)
Mark Ysusi Montgomery Watson (CHAP 11)

Larry W. Mays is professor of civil and environmental engineering at Arizona State

University and former chair of the department. He was formerly director of the Center for
Research in Water Resources at the University of Texas at Austin, where he also held an
Engineering Foundation Endowed Professorship.
A widely published expert on water resources, he wrote Optimal Control of
Hydrosystems (Marcel Dekker) and was editor in chief of the Water Resources Handbook
(McGraw-Hill) and the Water Distribution Systems Handbook (McGraw-Hill). He was coauthor of both Applied Hydrology and Hydrosystems Engineering and Management published by McGraw-Hill and was the editor in chief of Reliability Analysis of Water
Distribution Systems published by ASCE. He has published extensively in the water
resources literature on his research efforts. Dr. Mays is a registered professional engineer
in several states and a registered professional hydrologist. He has served as a consultant to
many organizations.

PREFACE

The Hydraulic Design Handbook, referred to herein as the Handbook, has been an extensive effort to develop a comprehensive reference book on hydraulic design. Over the years
it has become more and more obvious to me the need for such a comprehensive reference
book, particularly for the practicing engineer, who needs a central reference to start to
accumulate knowledge for the hydraulic design process. They need a reference
to find state-of-the-art design procedures, worked out example problems, discussions of
what can go wrong in the design process, up-to-date references for further reading, and
limited discussions of the theoretical aspects as they relate to the design process. Also references and discussions are needed for many different computer codes that are used in
hydraulic design practice. This Handbook is an attempt to fulfill these needs of the practicing engineer.
Also I consider this Handbook to be a valuable reference, if not a text, for use in teaching hydraulic design at both the undergraduate and graduate levels. Basically there are no
published books (or textbooks) that comprehensively cover hydraulic design. Hopefully
the Handbook can be used effectively for such courses because the balance of design concepts and procedures with worked out examples and just enough theoretical coverage
should make this, a valuable book for the hydraulic design classroom.
A large amount of knowledge has accumulated over the years on hydraulic design,
mainly in various government publications and reports, in journals and a limited amount
in textbooks. Unfortunately this knowledge has not been accumulated into one central
location. Over the years many new methods have been developed and many computer
programs have been written for hydraulic design. Engineers need an authoritative
source of guidance as to which methods and computer programs are appropriate for
various given tasks. This Handbook is intended to provide this information in a concise
and accessible form. This Handbook is not, however, an encyclopedia of every known
fact about hydraulic design.
The Handbook is divided into four sections, each consisting of a group of chapters. The
first section on principles for hydraulic design consists of chapters on pressurized flow,
open-channel flow, groundwater flow, environmental hydraulics, sedimentation and erosion hydraulics, and risk-reliability analysis. These are fairly comprehensive chapters that
focus on the concepts of the various types of flow. The chapter on risk-reliability presents
many methodologies for risk-reliability analysis and presents the U.S. Army Corps of
Engineers methodology. The second series of chapters are devoted to hydraulic design for
water supply with the major emphasis on the design of water distribution systems. These
chapters include pump design, system design, pipe design, and hydraulic transients.
A chapter is also presented on the hydraulic design for energy generation. The third series
of chapters are devoted to hydraulic design for water excess management. The chapters
are rather comprehensive ranging from highway drainage, urban drainage, highway crossings, flood control channels, spillways, stilling basins, flow transitions, energy dissipators,
and flood-plain hydraulics. The fourth series of chapters are devoted to environmental considerations with chapters on the hydraulics of water and wastewater treatment systems,

groundwater contamination, the hydraulic design of infiltration systems for artificial

recharge, and the design of flow measuring structures.
There are obviously many other topics that could have been covered making the
Handbook even more comprehensive; however, I had to make choices on coverage. These
choices obviously reflect my vision of what is needed most in the Handbook. The topics
covered are the ones that I feel are the most common among the many in hydraulic design.
The detail of coverage of subject areas may vary among chapters and among topics in
chapters. There is also a reflection of both my perspective of the subject with the constraint that all the material fits within one handbook. So hopefully you will have an understanding and appreciation of what I am trying to accomplish in this Handbook.
As I reflect upon my own experiences in the field of water resources, I am continually reminded of how handbooks have always been a part of my learning. These handbooks
have included the Handbook of Applied Hydraulics by C. V. Davis and K. E. Sorenson,
the Handbook of Hydraulics by E. F. Brater and H. W. King, the Handbook of Applied
Hydrology by V. T. Chow, and the Handbook of Fluid Dynamics by V. L. Streeter. These
have all been valuable resources to me throughout my engineering education, consulting
experiences, and teaching and research. More recently David Maidment developed the
Handbook of Hydrology and I developed the Water Resources Handbook. This new effort,
the Hydraulic Design Handbook, hopefully will complement these previous handbooks to
fulfill a gap not provided by these previous handbooks.
Each book that I have worked on has been a part of my lifelong journey in water
resources. The Handbook certainly is no exception. I have gained more from this experience than can ever be measured in words. From my early days as a boy growing up in
Illinois is near Mark Twain Country, between the Mississippi and Illinois Rivers to now,
I have always had this loving appreciation for water and the desire to continually learn
more. Over the years even my spare time has been focused on water in one form or another, fly fishing, scuba diving, and snow skiing.
I dedicate this Handbook to humanity and human welfare.

Larry W. Mays
Scottsdale, Arizona

Acknowledgments
I must first acknowledge the authors who made this handbook possible. It has been a sincere privilege to have worked with such an excellent group of dedicated people. They are
all experienced professionals who are among the leading experts in their fields.
References to material in this handbook should be attributed to the respective chapter
authors.
During the past twenty-three years of my academic career I have received help and
encouragement from so many people that it is not possible to name them all. These people represent a wide range of universities, research institutions, government agencies, and
professions. To all of you I express my deepest thanks.
I would like to acknowledge Arizona State University, especially the time afforded me to
pursue this handbook.
I sincerely appreciate the advice and encouragement of Larry Hager of McGraw-Hill
throughout this project.
This handbook has been a part of a personal journey that began years ago when I was a
young boy with a love of water. Books are companions along the journey of learning. I
hope that you will be able to use this handbook in your own journey of learning about
water. Have a happy and wonderful journey.
Larry W. Mays

CHAPTER 1
INTRODUCTION
Larry W. Mays
Department of Civil and Environmental Engineering
Arizona State University
Tempe, Arizona

1.1

OVERVIEW

Since the Egyptian's and Mesopotamian's first successful efforts to control the flow of
water thousands of years ago, a rich history of hydraulics has evolved. Sec. 1.2 contains
a brief description of some ancient hydraulic structures that are found around the world.
During the 20th century, many new developments have occurred in both theoretical and
applied hydraulics. A number of handbooks and textbooks on hydraulics have been published, as indicated in Fig. 1.1. From the viewpoint of hydraulic design, however, only
manuals, reports, monographs, and the like have been published, mostly by government
agencies. Unfortunately, many aspects of hydraulic design have never been published as
a compendium. This Hydraulic Design Handbook is the first effort devoted to producing
a comprehensive handbook for hydraulic design. The book covers many aspects of
hydraulic design, with step-by-step procedures outlined and illustrated by sample design
problems.

1.2 ANCIENT HYDRAULIC STRUCTURES

1.2.1 A Time Perspective
Although humans are newcomers to earth, their achievements have been enormous. It was
only during the Holocene epoch (10,000 years ago) that agriculture developed (keep in
mind that the earth and the solar system originated 4,600 million years ago). Humans have
spent most of their history as hunters and food-gatherers. Only in the past 9,000 to 10,000
years have humans discovered how to raise crops and tame animals. Such changes probably occurred first in the hills to the north of present-day Iraq and Syria. The remains of
the prehistoric irrigation works in Mesopotamia and Egypt still exist. Table 1.1 presents a
chronology of knowledge about water.
Figure 1.2 illustrates the chronology and locations of various civilizations ranging
from India to Western Europe. This figure, from O. Neugebaur's book titled The Exact
Sciences in Antiquity, illustrates the Hellenistic period the era of "ancient science,"
during which a form of science developed that spread later from Europe to India.
This ancient science was dominant until the creation of modern science dominant in
Isaac Newton's time.

Abbott's Computational Hydraulics (1980)

Fischer et al., Mixing in Inland and
Coastal Waters (1979)

Freeze and Cherry's Groundwater (1979)

Graf's Hydraulics of Sediment Transport

(1971)

Streeter and Wylies' Hydraulic Transients

(1967)
Hendersons' Open-Channel Flow (1966)
Leliavsky's River and Canal Hydraulics
(1965)
Morris and Wiggert's Applied Hydraulics
in Engineering (1963)

U.S. Geological Survey's Roughness

Characteristics of Natural Channels (1967)
Daily and Harleman's Fluid Dynamics
(1966)
Linsley and Franzini's Elements of
Hydraulic Engineering (1964)

USBR Design of Small Dams (1960)

Chow's Open-Channel Hydraulics (1959)
U.S. Bureau of Reclamation's Hydraulic
Design of Stilling Basin and Energy
Dissipators (1958)
Stoker's Water Waves (1957)
Parmakiams' Waterhammer Analysis
(1955)
King's Handbook of Hydraulics (1954)

Leliavsky's An Introduction to Fluvial

Hydraulics (1955)
Addison's Treastise on Applied
Hydraulics (1954)

U.S. Bureau of Reclamation's Hydraulic

Laboratory Practice (1953)
Rich's Hydraulic Transients (1951)
Rouse's Engineering Hydraulics (1950)
FIGURE 1.1 A selected list of books on hydraulics published between 1900 to 1980.

Allen's Scale Models in Hydraulic

Engineering (1947)
ASCE's Hydraulic Models (1942)

Davis and Sorersen's Handbook of

Applied Hydraulics (1942)

Woodward and Posey's Hydraulics of

Steady Flow in Open Channels (1941)
Rouse's Fluid Mechanics for Hydraulic
Engineers (1938)
Muskat's The Flow of Homogeneous Fluids
Through Porous Media (1937)
Bakhmeteff's Hydraulics of Open Channels
(1932)
Schoder and Dawson's Hydraulics (1927)
Le Conte's Hydraulics (1926)

Hoyt and Grover's River Discharge

(1916)
Hoskins's A Text-Book on Hydraulics
(1911)

Merriman's Treatise on Hydraulics (1904)

FIGURE 1.1 (Continued)

Daugherty's Hydraulics (1937)

Bakhmeteff's The Mechanics of Turbulent
Flow (1936)

TABLE 1.1 Chronology of Knowledge About Water

Prehistorical period
3rd -2nd millennium B.C.
3rd millennium B.C.
3 millennium B.C.
Probably very earlyt
2nd millennium B.C.

Springs
Cisterns
Dams
Wells
Reuse of excrement as fertilizer
Gravity flow supply pipes or channels and drains, pressure
pipes (subsequently forgotten)
8th-6th c. B.C.
Long-distance water supply lines with tunnels and bridges,
as well as intervention in and harnessing of karst water
systems
Public as well as private bathing facilities, consisting of:
6th c. B.C. at the latest
bathtubs or showers, footbaths, washbasins, latrines or
toilets, laundry and dishwashing facilities
Use of definitely two and probably three qualities of water:
6th c. B.C. at the latest
potable, subpotable, and nonpotable, including irrigation
using storm runoff, probably combined with waste waters
6th-3rd c. B.C.
Pressure pipes and siphon systems
*Indicates an element discovered, probably forgotten, and rediscovered later,
indicates an educated guess.
Source: Crouch, 1993.

1.2.2

Irrigation Systems

1.2.2.1 Egypt and Mesopotamia. In ancient Egypt, the construction of canals was a major
endeavor of the Pharaohs beginning in Scorpio's time. Among the first duties of provincial
governors was the digging and repair of canals, which were used to flood large tracts of land
while the Nile was flowing high. The land was checkerboarded with small basins defined by
a system of dikes. Problems associated with the uncertainty of the Nile's flows were recognized. During high flows, the dikes were washed away and villages were flooded, drowning
thousands of people. During low flows, the land was dry and no crops could grow. In areas
where fields were too high to receive water directly from the canals, water was drawn from
the canals or from the Nile by a swape or shaduf(fig. 1.3), which consisted of a bucket on
the end of a cord hung from the long end of a pivoted boom that was counterweighted at the
short end (de Camp, 1963). Canals continued to be built in Egypt throughout the centuries.
The Sumerians in southern Mesopotamia built city walls and temples and dug canals
that were the world's first engineering works. It also is of interest that these people, fought
over water rights from the beginning of recorded history. Irrigation was vital to
Mesopotamia, Greek for "the land between the (Tigris and Euphrates) rivers." An ancient
Babylonian curse was, "May your canal be filled with sand" (de Camp, 1963), and even
their ancient laws dealt with canals and water rights. The following quotation from
approximately the sixth century B.C., illustrates such a law (de Camp, 1963):
"The gentleman who opened his wall for irrigation purposes, but did not make
his dyke strong and hence caused a flood and inundated a field adjoining his,
shall give grain to the owner of the field on the basis of those adjoining"
Because the Tigris and Euphrates carried several times more silt per unit volume of water
than the Nile did, flooding problems were more serious in Mesopotamia than in Egypt. As
a result the rivers in Mesopotamia rose faster and changed course more often.

India

Iron fAesopotam. Syria

Writing

Elarn

Esypt Asia minor Greece

Italy

Spain

Writing

Sumer

Indus
Valley

V.cUe
P.ncd

Hyksos

Linear B
Hittite*

Etruscans

Roman Empire
Byzantium

FIGURE 1.2 Chronology and location of different civilizations ranging from India to Western Europe.
(Neugebauer, 1993)
The irrigation systems in both Mesopotamia and the Egyptian Delta were of the basin
type, opened by digging a gap in the embankment and closed by placing mud back into
the gap. (See Fig. 1.4 for a comparison of the irrigation works in Upper Egypt and in
Mesopotamia.) Water was hoisted using the swape, Mesopotamian laws required farmers
to keep their basins and feeder canals in repair; they also required everyone else to wield
hoes and shovels when the rivers flooded or when new canals were required or old ones
needed repair (de Camp, 1963). Some canals may have been used for 1,000 years before

FIGURE 1.3 Shadufs of the Amarna period, from the tomb of Nefer-Hotep at Thebes.
Note irrigation of date palms and other orchard trees and the apparent tank or pool (lower
right). The water pattern in the lowest margin suggests lifting out of an irrigation canal.
(Davies, 1933, pis. 46 and 47). Figure as presented in Butzer (1976).
they were abandoned and others were built. Even today, 4,000 to 5,000 years later, the
embankments of the abandoned canals remain. In fact, these canal systems supported a
larger population than lives there today. Over the centuries, Mesopotamian agriculture
began to decline because of the salty alluvial soil. In 1258, the Mongols conquered
Mesopotamia and destroyed its irrigation systems.

Comparative Irrigation Patterns in Upper Egypt and Mesopotamia

Mesopotamia
A.D. MOO

Upper Egypt
A.O. 1*50
CinalS (.hf-mol
C9tn<itf*t with tf.tct)
SOHAG
Dikes

FIGURE 1.4 Comparative irrigation networks in Upper Egypt and Mesopotamia. A. Example of linear,
basin irrigation in Sohag province, ca. AD 1850. B. Example of radial canalization system in the lower
Nasharawan region southeast of Baghdad, Abbasid (A.D. 883-1150). Modified from R. M. Adams (1965,
(Fig. 9) Same scale as Egyptian counterpart) C. Detail of field canal layout in B. (Simplified from R. M.
Adams, 1965, Fig. 10). Figure as presented in Butzer (1976).
The Assyrians also developed extensive pubic works. When Sargon II invaded Armenia
in 714 B.C., he discovered the qandt (Arabic) or kariz (Persian), a system of tunnels used
to bring water from an underground source in the hills down to the foothills (Fig. 1.5).
Sargon destroyed the system in Armenia but brought the concept back to Assyria. Over the
centuries, this method of irrigation spread across the Near East into North Africa and is

Mountain range

Masonry
lining

(b) Cross section of a qanat

(a) A typical qanat layout for water supply

Water intakd
About area
150'
Vertical air
shafts
Farmland

Infiltration
to qanat

Groundwater level
Aquifer (alluvial silt)
Rock

Shale (impervious)

(c) Longitudinal section of qanat

FIGURE 1.5 Details of the qanat system. (Biswas, 1970).

still used. Sargon's son Sennacherib also developed waterworks by damming the Tebitu
River and using a canal to bring water to Nineveh, where the water could be used for irrigation without the need for hoisting devices. During high water in the spring, overflows
were handled by a municipal canebrake that was built to develop marshes used as game
preserves for deer, wild boar, and birds. When this system was outgrown, a new canal 30
mi long was built, with an aqueduct that had a layer of concrete or mortar under the upper
layer of stone to prevent leakage.
1.2.2.2 Prehistoric Mexico. During the earliest years of canal irrigation in Mexico,
the technology changed little (Fig. 1.6) and the method of flooding tended to be haphazard. The technological achievements were relatively primitive until about 600 or
500 B.C., and few of the early systems remain. Whereas the earlier systems were constructed of loosely piled rocks, the later ones consisted of storage dams constructed of
blocks that were mortared together. Some spillways were improved, and floodgates
were used in some spillways. (Some dams could be classified as arch dams.) The
canals were modified to an extent during this time: Different cross-sectional areas were

CULTURE PERIOD
Year
VfeMey of Mexico
Oaxaca

TECHNOLOGICAL DEVELOPMENTS

-1600
LATE
POSTCLASSlC
-1400 * Relocation of rivers, Large masonry aqueducts.
POSTCLASSIC
-1200
EARLY
POSTCLASSIC -1000
~ 800 * Advanced relocation of ephemeral streams.
CLASSIC

-600

CLASSIC

LATE
FORMATIVE

MICHXE
FORMATiVE

EARLY
FORMATIVE

TERMINAL
FORMATIVE

LATE
FORMATIVE
MIDDLE
FORMATIVE
EARLY
FORMATIVE

- 400 Rock diversion dams, Treliised aqueducts.

Relocation of ephemeral streams.
-200
-If-

-200 Use of valley bottoms, Advanced channelization, Furrows,

r Use of perennial streams, Earth & brush diversion dams,
Z Head gates, Small earthen aqueducts.
-400 ^- Use of permanent springs, Sluice gates.
- 600

Masonry storage dams (arch) with floodgates,

Chiseled canal, Fieid borders,

ttrw_ * Earthen dams.

-oOO
* Incipient channelization, Weirs, Water spreaders.
-1000

r Use of ephemeral streams, Site drainage, Small canals,

-1ZOQ m R00J1 storage dams (gravity) with spillway.

FIGURE 1.6 Regional chronology and dates of developments in various aspects of canal irrigation
technology in Mexico. (Doolittle, 1990)
used, some were lined with stone slabs, and the water for irrigation of crops was more
carefully controlled.
Between 550 and 200 B.C., the irrigation-related features and the entire canal systems
were significantly improved. The channelization of stream beds, the excavation of canals,
and the construction of dams were probably the most significant improvements. However,
the technology stopped improving after 200 B.C., and no significant developments
occurred for approximately 500 years. Around 300 A.D., a few new improvements were
initiated, but the technology remained essentially the same through the classic period
(A.D., 200 - 800/1000) and early postclassical period (A.D. 800/1000-1300). Figure 1.7 is
a map of fossilized canals in the Tehuacan Valley in Mexico.

Tehuacan

San Juan
Ajalpan

San Gabriel
Chilac

San
Sebastian
Zmacatepec

FIGURE 1.7 Map of fossilized canals on the Llano de Ia Taza in the Tehuacan Valley.
(Woodbury and Neely, 1972, as presented in Doolittle, 1990)

1.2.2.3 North America. The canal irrigation systems in the Hohokam and Chaco
regions stand out as two major prehistoric developments in the American Southwest
(Crown and Judge, 1991). The two systems expanded over broad geographic areas of similar size (the Hohokam in Arizona and the Chacoans in New Mexico). Although they were
developed at similar times, they apparently functioned independently. Because the two
systems evolved in different environments, their infrastructures also differed considerably.
The Hohokam Indians inhabited the lower Salt and GiIa River valleys near Phoenix,
Arizona. Although the Indians of Arizona began limited farming nearly 3000 years ago,
construction of the Hohokam irrigation systems probably did not begin until the first few
centuries A.D. Who originated the idea of irrigation in Arizona, whether the technology
was developed locally or it was introduced from Mexico, is unknown. Figure 1.8 illustrates the extensive system in the Phoenix area, and Fig. 1.9 provides a schematic of the
details of its major components.
In approximately 1450 A.D., the Hohokam culture declined, possibly for a combination
of reasons: flooding in the 108Os, hydrologic degradation in the early UOOs, and the
recruitment of laborers by surrounding populations. The major flood in 1358 ultimately
destroyed the canal networks, resulting in movement of the people. Among the Pima
Indians, who were the successors of the Hohokam Indians, use of canals was either
limited or absent. Although the prehistoric people who lived outside the area of Hohokam

Fifth Edition. Copyright 1929, all rights reserved

ii P#HISrOHKCMAiS.flScale of miles. ^
PAVtO KOADS I9Z5
.;'' PREHISTORIC BUILDINGS. MAJ> OF
TOWNSHIPLMfS
PREHISTORIC
IRRIGATION CANALS
DR. OMAR A. TUKKEY T.RG.S.
ARIZONA
The largestorsinSouth
gle bodyPHOENIX
of land irrigated
in prehistoric
times in North
America
andPrehi
perhaps
in the world.
Thi
s
map
accompani
e
s
a
report
on
Dr,Txurney,
published by Major Geo. H.Kelly, State Historistori
an.cCapiIrrigation
tol Bld .,Pbyhoeni
,Arizona
g

FIGURE 1.8 Canal building in the Salt River Valley with a stone hoe held in the hand without a handle. These were the original engineers, the true pioneers who built, used,
and abandoned a canal system when London and Paris were a cluster of wild huts. Turney (1922) (Courtesy of Salt River Project, Phoenix, Arizona)

Diversion weir
Conol intoke oreo
Moin conol
Primory
heodgofe (with
stone poving)
secondory )
(some with stone poving)
Loteroi conol heodgote(topon)
Lateral canal network
Field oreo
Field house
Field turn-out
Junction pool

FIGURE 1.9 Schematic representation of the major components of a Hohokam

irrigation system in the Phoenix Basin. (Masse, 1991)

culture also constructed irrigation systems, none approached the grand scale of the
Hohokam systems.
In the ninth century, the Anasazi people of northwestern New Mexico developed a
cultural phenomenon, the remains of which currently consist of more than 2400
archaeological sites and nine towns, each containing hundreds of rooms, along a 9-mi
stretch. The Chacoan irrigation system is situated in the San Juan Basin in northwestern New Mexico. The basin has limited surface water, most of it discharges from
ephemeral washes and arroyos. Figure 1.10 illustrates the method of collecting and
diverting runoff throughout Chaco Canyon. The water collected from the side canyon
that drained from the top of the upper mesa was diverted into a canal by either an earthen or a masonry dam near the mouth of the side canyon (Vivian, 1990). These canals
averaged 4.5 m in width and 1.4 m in depth; some were lined with stone slabs and others were bordered by masonry walls. The canals ended at a masonry head gate, where
water was then diverted to the fields in small ditches or to overflow ponds and small
reservoirs.

Mesa Top

MULTIPLE HEAOGATE

FIGURE 1.10 Hypothetical reconstruction of the Rincon-4 North water

control system in Chaco Canyon. Similar systems were located at the
mouths of all northern side conyons in the lower 15 m of Chaco Canyon.
(Adapted by Ron Beckwith from Vivian, 1974, Fig. 9.4)

1.2.3

Dams

The Sadd-el-Kafara dam in Egypt, situated on the eastern bank of the Nile near Heluan
approximately 30 km south of Cairo, in the Wadi Garawi, has been referred to as the world's
oldest large dam (Garbrecht, 1985). The explorer and geographer George Schweinfurth
rediscovered this dam in 1885, and it has been described in a number of publications since
that time (see Garbrecht, 1985). It was built between 2950 and 2690 B.C. Although the Jass
drinking-water reservoir in Jordon and the diversion dams on the Kasakh River in Russia
are probably older, they are much smaller than the Sadd-el-Kafara (Dam of the Pagans).
It is unlikely that the Sadd-el-Kafara dam was built to supply water for drinking or irrigation because the dam lies too far from the alabaster quarries situated upstream to have
supplied the labor force with drinking water. Furthermore, there is a vast supply of water
and fertile land in the nearby Nile valley. The apparent purpose of the dam was to protect
installations in the lower wadi and the Nile valley from frequent, sudden floods. The dam
was destroyed during construction by a flood; consequently, it was never completed. To
date, the dam's abutments still exist.

The dam had an impervious core consisting of rubble, gravel, and weathered material. On both the upstream and downstream sides, the core was bordered by sections of
rockfill that supported and protected the core. The diameter of the stones ranged from 0.1
to 0.6 m. One remarkable construction feature is the facing of the section of rockfill
where parts of the facing on the upstream side are still well preserved. The dam had an
approximate crest length of 348 ft and a base length of 265 ft and was built straight
across the wadi at a suitably narrow point, with a maximum height of 32 ft above the
valley bed. See Smith (1971) and Upton (1975) for more on dams.
Dam building in the Americas began in the pre-Colombian period in the civilizations
of Central and South America: the Aztecs in Mexico, the Mayans in Guatemala and
Yucatan, and the Incas in Peru. Where as old-world civilizations developed in the valleys
of the big rivers, the Nile River, the Euphrates and the Tigris Rivers, the Indus River, and
the Yellow River, most of the early civilizations in the New World were not river civilizations. In South America, the civilizations appeared in the semiarid highlands and the
arid coastal valleys traversed by small rivers. In Central America, the Mayans, the Aztecs,
and the predecessors of the Aztecs were not river civilizations.
The Mayans did not practice irrigation; however, they did provide efficient water
supplies to several of their large cities. They developed the artificial well (cenote\ the underground cistern (chultun\ and the large open reservoir (aguado). The Mayans' failure to
develop irrigation may have accelerated their decline. In the Yucatan, the aguados are still
found in some places, but the cenote was the major source of water for drinking and bathing.
1.2.4

Urban Water Supply and Drainage Systems

Knossos, approximately 5 km from Herakleion, the modern capital of Crete, was among
the most ancient and unique cities of the Aegean and Europe. The city was first inhabited
shortly after 6000 B.C. and, within 3000 years, it had became the largest Neolithic
Settlement in the Aegean (Neolithic age, circa 5700-2800 B.C). During the Bronze age
(circa 2800-1100 B.C.), the Minoan civilization developed and reached its culmination as
the first Greek cultural miracle of the Aegean world.
The Minoan civilization has been subdivided into four periods: the prepalatial period
(2800-1900 B.C.), the protopalatial period (1900-1200 B.C.), the neopalatial period
(1700-1400 B.C.), and the postpalatial period (1400-1100 B.C.). During the prepalatial
period, a settlement at Knossos; was leveled to erect a palace. Little is known about the
old palace because it was destroyed in approximately 1700 B.C. A new palace was constructed on leveled fill from the old palace. During the neopalatial period, Knossos was at
the height of its splendor. The city covered an area of 75,000 to 125,000 m2 and had a population estimated to be on the order of tens of thousands.
The irrigation and drainage systems at Knossos were most interesting. An aqueduct
supplied water through tubular conduits from the Kounavoi and Archanes regions and
branched out into the city and the palace. Figure 1.11 shows the type of pressure conduits
used within the palace for water distribution. The drainage system consisted of two separate conduits: one to collect the sewage and the other to collect rain water (Fig. 1.12).
Unfortunately, the Mycenean palace was destroyed by an earthquake and fire in approximately 1450 B.C., as were all the palatial cities of Crete.
Anatolia, also called Asia Minor, which is part of the Republic of Turkey, has been the
crossroads of many civilizations during the past 10,000 years. During the last 4000 years,
going back to the Hittite period (2000-200 B.C.) many remains of ancient urban watersupply systems have been found, including pipes, canals, tunnels, inverted siphons, aqueducts, reservoirs, cisterns, and dams, (see Ozis, 1987 and Ozis and Harmancioglu, 1979).

FIGURE 1.11 Water distribution pipe in Knossos, Crete. (Photograph by L.W. Mays)

FIGURE 1.12 Urban drainage system in Knossos, Crete. (Photograph by L.W. Mays)

An example of one such city is Ephesus, which was founded during the 10th century
B.C. as an Ionian city out of the Temple of Artemis. In the sixth century B.C., the city settled near the temple, and subsequently was reestablished at its present site, where it developed further during the Roman period. Water was supplied to Ephesus from springs at different sites. Cisterns also supplied well water to the city. Water for the great fountain, built
between 4 and 14 A.D., was diverted by a small dam at Marnss and was conveyed to the
city by a system 6 km long consisting of one large and two small clay pipe lines. Figure
1.13 shows the type of clay pipes used at Ephesus to distribute water.

FIGURE 1.13 Water distribution pipe in Ephesus, Turkey. (Photograph by

L. W. Mays)

The latrine, or public toilet shown in Fig. 1.14, was built in the first century A.D. at
Ephesus. The toilets were placed side by side with no partitions. In the middle was a
square pond, and the floors were paved with mosaics.
The Great Theatre at Ephesus, the city's largest and most impressive building, had a
seating capacity for 24,000 people. Built in the Hellenistic period, the theatre was not only
a monumental masterpiece but during the early days of Christianity, one major confrontation between Artemis and Christ took place there. Of notable interest from a waterresources viewpoint is the theatre's intricate drainage system. Figure 1.15 shows a
drainage channel in the floor of the theatre.
Public baths also were a unique feature in ancient cities: for example, the Skolactica
baths in Ephesus had a salon and central heating; a hot bath (caldariuni), a warm bath
(tepidarium), and a cold bath (frigidarium)', and a dressing room (apodyterium). In the
second century A.D., the first building had three floors. In the fourth century, a woman
named Skolacticia modified the baths, making them accessible to hundreds of people.
There were public rooms and private rooms, and people who wished to could stay for
many days. Hot water was provided by a furnace and a large boiler.
Perge is another ancient city in Anatolia that had a unique urban water infrastructure.
The photographs in Fig. 1.16 illustrate the Majestic Fountain (nymphaion), which consisted of a wide basin and a richly decorated architectural facade. Because of its architecture and statues, the fountain was one of Perge's most magnificent edifices. A water channel ran along the middle, dividing each street and bringing life and coolness to the city.
The baths of Perge were magnificent. The first photograph in Fig. 1.17 shows one of the
baths of Perge; the second photograph illustrates the storage of water under the floor to
keep the water warm. Like the baths in other ancient cities in Anatolia, the baths of Perge
had a caldarium, a tepidarium, and a frigidarium.

FIGURE 1.14 A latrine, or public toilet, built at Ephesus, Turkey, in the first century B.C.
(Photograph by L. W. Mays)

FIGURE 1.15 A drainage channel on the floor of the Great Theater at Ephesus,
Turkey. (Photograph by L. W. Mays)

The early Romans devoted much of their time to useful public works projects, including roads, harbor works, aqueducts, temples, forums, town halls, arenas, baths, and sewers. The prosperous early Roman bourgeois typically had a 12-room house, with a square
hole in the roof to let rain in and a cistern beneath the roof to store the water. Although
the Romans built many aqueducts, they were not the first to do so. King Sennacherio built
aqueducts, as did the Phoenicians and the Helenes. The Romans and Helenes needed
extensive aqueduct systems for their fountains, baths, and gardens. They also realized that

FIGURE 1.16 Two views of the Majestic Fountain (nymphaiori) in Perge, Anatolia, Turkey.
(Photographs by L. W. Mays)
water transported from springs was better for their health than river water and that spring
water did not need to be lifted to street level as did river water. Roman aqueducts were
built on elevated structures to provide the needed slope for water flow. Knowledge of pipe

FIGURE 1.17A View of the baths at Perge, Anatolia, Turkey. (Photographs by

L.W. Mays)
making-using bronze, lead, wood, tile, and concrete-was in its infancy, and the difficulty of making strong large pipes was a hinderance. Most Roman piping was made of lead,
and even the Romans recognized that water transported by lead pipes was a health hazard.
The source of water for a typical Roman water supply system was a spring or a dug
well, which usually was equipped with a bucket elevator to raise the water. If the well

FIGURE 1.17B View of the baths at Perge, Anatolia, Turkey. (Photographs by L.W. Mays)

water was clear and of sufficient quantity, it was conveyed to the city by aqueduct. Also,
water from several sources was collected in a reservoir, then conveyed by an aqueduct or
a pressure conduit to a distributing reservoir (castellum). Three pipes conveyed the water:
one to pools and fountains, one to the public baths for public revenue, and one to private
houses for revenue to maintain the aqueducts (Rouse and Ince, 1957). Figures 1.18 and
1.19 illustrate the layout of the major aqueducts of ancient Rome. Figure 1.20 shows the
Roman aqueduct in Segovia, Spain, which is probably among the most interesting of
Roman remains in the world. This aqueduct, built during the second half of the first century A.D. or the early years of the second century, has a maximum height of 78.9 m. See
Van Deman (1934) for more details on Roman aqueducts.
Irrigation was not a major concern because of the terrain and the intermittent rivers.
However, the Romans did, drain marshes to obtain more farmland and to eliminate the bad
air, or "harmful spirits," rising from the marshes because they believed it caused disease
(de Camp, 1963). The disease-carrying mechanism was not the air, (or spirits) but the
malaria-carrying mosquito. Empedocles, the leading statesman of Acragas in Sicily during the Persian War in the fifth century B.C., drained the local marshes of Selinus to
improve the people's health (de Camp, 1963).
The fall of the Roman Empire extended over a 1000-year period of transition called
the Dark Ages during which the concepts of science related to water resources probably
retrogressed. After the fall of the Roman Empire, clean water, sanitation, and public health
declined in Europe. Historical accounts tell of incredibly unsanitary conditions: polluted
water, human and animal wastes in the streets, and water thrown out of windows onto
passersby. As a result, various epidemics ravaged Europe. During the same period, the
Islamic cultures on the periphery of Europe religiously mandated high levels of personal
hygiene, highly developed water supplies, and adequate sanitation systems. For furthen
reading see Needham (1959) Payne (1959), Reynolds (1970) Robbins (1946), Sarton
(1952-59) and Wittfogel (1956).

Aqueducts
in
ANCIENT ROME

REGIONES
Porta Capena
IIHII Caelimontium
IsisetSerams
IVV Templum
Esquiliae Pads
VI ViaLata
AltaSemita
.VU
Forum Maximus
Romanum
IXVIII Circus
XII
Piscina
Publica
XIH
Aventinus
XVI Transtiberim
FIGURE 1.18 Termini of the major aqueducts in ancient Rome. (Evans, 1993)

Claudia et
Anio Novus

Marcia
Tepula
Julia

SPES VETUS

Arcus
Caelimontani

Marcia
Tepula
Julia

FIGURE 1.19 The area of Spes Vetus showing the courses of the major aqueducts entering the
city above ground. (From R. Lanciani, Forma Urbis Romae), as presented in Evans (1993).

FIGURE 1.20 Roman aqueduct in Segovia, Spain. (Photograph by L.W. Mays)

1.3 DEVELOPMENT OF HYDRAULICS

The historical development of hydraulics as a modern science has been described by
Biswas (1970), Rouse (1976), and Rouse and Ince (1963). More recently, the book titled,
The Science of Water (Levi, 1995) presents an excellent history of the foundation of
modern hydraulics. The reader is referred to these excellent books for details on the development of hydraulics.

1.4

FEDERAL POLICIES AFFECTING HYDRAULIC DESIGN

Federal legislation contains policies that can affect the design of various types of
hydraulic structures. These policies are listed in Appendix l.A, where they are categorized
into the following sections: environment, health, historic and archeological preservation,
and land and water usage. The appendix also lists the abbreviations used in the policies,
(adapted from AASHTO, 1991).

1.5

CONVENTIONAL HYDRAULIC DESIGN PROCESS

Conventional procedures for hydraulic design are basically iterative trial-and-error procedures. The effectiveness of conventional procedures depends on an engineer's intuition,
experience, skill, and knowledge of hydraulic systems. Therefore, conventional procedures
are closely related to the human element, a factor that could lead to inefficient results for the
design and analysis of complex systems. Conventional procedures are typically based on
using simulation models in a process of trial and error to arrive at an optimal solution. Figure
Next Page

CHAPTER 2
HYDRAULICS OF
PRESSURIZED FLOW
Bryan W. Karney
Department of Civil Engineering
University of Toronto,
Toronto, Ontario,
Canada

2.1

INTRODUCTION

The need to provide water to satisfy basic physical and domestic needs; use of maritime and fluvial routes for transportation and travel, crop irrigation, flood protection, development of stream power; all have forced humanity to face water from the
beginning of time. It has not been an easy rapport. City dwellers who day after day
see water flowing from faucets, docile to their needs, have no idea of its idiosyncrasy. They cannot imagine how much patience and cleverness are needed to handle our great friend-enemy; how much insight must be gained in understanding its
arrogant nature in order to tame and subjugate it; how water must be "enticed" to
agree to our will, respecting its own at the same time. That is why a hydraulician
must first be something like a water psychologist, thoroughly knowledgeable of its
nature. (Enzo Levi, The Science of Water: The Foundations of Modern Hydraulics,
ASCE, 1995, p. xiii.)
Understanding the hydraulics of pipeline systems is essential to the rational design,
analysis, implementation, and operation of many water resource projects. This chapter
considers the physical and computational bases of hydraulic calculations in pressurized
pipelines, whether the pipelines are applied to hydroelectric, water supply, or wastewater
systems. The term pressurized pipeline means a pipe system in which a free water surface
is almost never found within the conduit itself. Making this definition more precise is difficult because even in a pressurized pipe system, free surfaces are present within reservoirs and tanks and sometimesfor short intervals of time during transient (i.e.,
unsteady) eventscan occur within the pipeline itself. However, in a pressurized pipeline
system, in contrast to the open-channel systems discussed in Chapter 3, the pressures
within the conveyance system are usually well above atmospheric.
Of central importance to a pressurized pipeline system is its hydraulic capacity: that is,
its ability to pass a design flow. A related issue is the problem of flow control: how design
flows are established, modified, or adjusted. To deal adequately with these
two topics, this chapter considers head-loss calculations in some detail and introduces the
topics of pumping, flow in networks, and unsteady flows. Many of these subjects are treated in greater detail in later chapters, or in references such as Chaudhry and Yevjevich (1981).
Rather than simply providing the key equations and long tabulations of standard
values, this chapter seeks to provide a context and a basis for hydraulic design. In addition to the relations discussed, such issues as why certain relations rather than others are
used, what various equations assume, and what can go wrong if a relation is used incor-

rectly also are considered. Although derivations are not provided, some emphasis is placed
on understanding both the strengths and weaknesses of various approaches. Given the
virtually infinite combinations and arrangements of pipe systems, such information is essential for the pipeline professional.

2.2

IMPORTANCEOFPIPELINESYSTEMS

Over the past several decades, pressurized pipeline systems have become remarkably
competitive as a means of transporting many materials, including water and wastewater.
In fact, pipelines can now be found throughout the world transporting fluids through every
conceivable environment and over every possible terrain.
There are numerous reasons for this increased use. Advances in construction
techniques and manufacturing processes have reduced the cost of pipelines relative to
other alternatives. In addition, increases in both population and population density have
tended to favor the economies of scale that are often associated with pipeline systems. The
need for greater conservation of resources and, in particular, the need to limit losses
caused by evaporation and seepage have often made pipelines attractive relative to openchannel conveyance systems. Moreover, an improved understanding of fluid behavior has
increased the reliability and enhanced the performance of pipeline systems. For all these
reasons, it is now common for long pipelines of large capacity to be built, many of which
carry fluid under high pressure. Some of these systems are relatively simple, composed
only of series-connected pipes; in other systems, the pipes are joined to form complex
networks having thousands of branched and interconnected lines.
Pipelines often form vital links in the process chain, and high penalties may be associated with both the direct costs of failure (pipeline repair, cost of lost fluid, damages
associated with rupture, and so forth) and the interruption of service. This is especially
evident in industrial applications, such as paper mills, mines, and power plants. Yet, even
in municipal systems, a pipe failure can cause considerable property damage. In addition,
the failure may lead indirectly to other kinds of problems. For example, a mainline break
could flood a roadway and cause a traffic accident or might make it difficult to fight a
major fire.
Although pipelines appear to promise an economical and continuous supply of fluid,
they pose critical problems of design, analysis, maintenance, and operation. A successful
design requires the cooperation of hydraulic, structural, construction, survey, geotechnical, and mechanical engineers. In addition, designers and planners often must consider the
social, environmental, and legal implications of pipeline development. This chapter focuses on the hydraulic considerations, but one should remember that these considerations are
not the only, nor necessarily the most critical, issues facing the pipeline engineer. To be
successful, a pipeline must be economically and environmentally viable as well as technically sound. Yet, because technical competence is a necessary requirement for any successful pipeline project, this aspect is the primary focus.

2.3

NUMERICAL MODELS: BASIS FOR PIPELINE ANALYSIS

The designer of a hydraulic system faces many questions. How big should each pipe be
to carry the required flow? How strong must a segment of pipe be to avoid breaking? Are
reservoirs, pumps, or other devices required? If so, how big should they be and where
should they be situated?

There are at least two general ways of resolving this kind of issue. The first way is to
build the pipe system on the basis of our "best guess" design and learn about the system's
performance as we go along. Then, if the original system "as built" is inadequate, successive adjustments can be made to it until a satisfactory solution is found. Historically, a
number of large pipe systems have been built in more or less this way. For example, the
Romans built many impressive water supply systems with little formal knowledge of fluid
mechanics. Even today, many small pipeline systems are still constructed with little or no
analysis. The emphasis in this kind of approach should be to design a system that is both
flexible and robust.
However, there is a second approach. Rather than constructing and experimenting with
the real system, a replacement or model of the system is developed first. This model can
take many forms: from a scaled-down version of the original to a set of mathematical
equations. In fact, currently the most common approach is to construct an abstract numerical representation of the original that is encoded in a computer. Once this model is
"operational," experiments are conducted on it to predict the behavior of the real or proposed system. If the design is inadequate in any predictable way, the parameters of the
model are changed and the system is retested until design conditions are satisfied. Only
once the modeller is reasonably satisfied would the construction of the complete system
be undertaken.
In fact, most modern pipelines systems are modeled quite extensively before they are
built. One reason for this is perhaps surprisingexperiments performed on a model are
sometimes better than those done on the prototype. However, we must be careful here,
because better is a relative word. On the plus side, modeling the behavior of a pipeline
system has a number of intrinsic advantages:
Cost. Constructing and experimenting on the model is often much less expensive than
testing the prototype.
Time. The response rate of the model pipe system may be more rapid and convenient
than the prototype. For example, it may take only a fraction of a second for a
computer program to predict the response of a pipe system after decades of projected
growth in the demand for water.
Safety. Experiments on a real system may be dangerous or risky whereas testing the
model generally involves little or no risk.
Ease of modification: Improvements, adjustments, or modifications in design or
operating rules can be incorporated more easily in a model, usually by simply editing
an input file.
Aid to communication. Models can facilitate communication between individuals and
groups, thereby identifying points of agreement, disagreement, misunderstanding, or
issues requiring clarification. Even simple sketches, such as Fig. 2.1, can aid discussion.
These advantages are often seen as so overwhelming that the fact that alternative
approaches are available is sometimes forgotten. In particular, we must always remember
that the model is not reality. In fact, what makes the model useful is precisely its simplicityit is not as complex or expensive as the original. Stated more forcibly, the model is
useful because it is wrong. Clearly, the model must be sufficiently accurate for its intended purpose or its predictions will be useless. However, the fact that predictions are imperfect should be no surprise.
As a general rule, systems that are large, expensive, complex, and important justify
more complex and expensive models. Similarly, as the sophistication of the pipeline system increases, so do the benefits and advantages of the modeling approach because this

FIGURE 2.1 Energy relations in a simple pipe system.

strategy allows us to consider the consequences of certain possibilities (decisions, actions,
inactions, events, and so on before they occur and to control conditions in ways that may
be impossible in practice (e.g., weather characteristics, interest rates, future demands,
control system failures). Models often help to improve our understanding of cause
and effect and to isolate particular features of interest or concern and are our primary tool
of prediction.
To be more specific, two kinds of computer models are frequently constructed for
pipeline systems: planning models and operational models.
Planning models. These models are used to assess performance, quantity or economic impacts of proposed pipe systems, changes in operating procedures, role of devices,
control valves, storage tanks, and so forth. The emphasis is often on selection, sizing, or
modification of devices.
Operational models. These models are used to forecast behavior, adjust pressures or
flows, modify fluid levels, train operators, and so on over relatively short periods (hours,
days, months). The goal is to aid operational decisions.
The basis of both kinds of models is discussed in this chapter. However, before you
believe the numbers or graphs produced by a computer program, or before you work
through the remainder of this chapter, bear in mind that every model is in some sense a
fake it is a replacement, a stand-in, a surrogate, or a deputy for something else. Models
are always more or less wrong. Yet it is their simultaneous possession of the characteristics of both simplicity and accuracy that makes them powerful.

2.4

MODELING APPROACH

If we accept that we are going to construct computer models to predict the performance
of pipeline systems, then how should this be done? What aspects of the prototype can and
should be emphasized in the model? What is the basis of the approximations, and what
principles constrain the approach? These topics are discussed in this section.
Perhaps surprisingly, if we wish to model the behavior of any physical system, a
remarkably small number of fundamental relations are available (or required). In essence,
we seek to answer three simple questions: where?, what? and how? The following sections
provide elaboration.

The first question is resolved most easily. Because flow in a pipe system can almost
always be assumed to be one-dimensional, the question of where is resolved by assuming
a direction of flow in each link of pipe. This assumed direction gives a unique orientation
to the specification of distance, discharge, and velocity. Positive values of these variables
indicate flow in the assumed direction, whereas negative values indicate reverse flows.
The issues of what and how require more careful development.
2.4.1 Properties of Matter (What?)
The question of 'What?' directs our attention to the matter within the control system. In
the case of a hydraulic system, this is the material that makes up the pipe walls, or fills
the interior of a pipe or reservoir, or that flows through a pump. Eventually a modeller
must account for all these issues, but we start with the matter that flows, typically consisting mostly of water with various degrees of impurities.
In fact, water is so much a part of our lives that we seldom question its role. Yet water
possesses a unique combination of chemical, physical, and thermal properties that makes
it ideally suited for many purposes. In addition, although important regional shortages
may exist, water is found in large quantities on the surface of the earth. For both these reasons, water plays a central role in both human activity and natural processes.
One surprising feature of the water molecule is its simplicity, formed as it as from two
diatomic gases, hydrogen (H2) and oxygen (O2). Yet the range and variety of water's
properties are remarkable (Table 2.1 provides a partial list). Some property values in the
TABLE 2.1 Selected Properties of Liquid Water
Physical Properties
1. High densitypliq < 1 000 kg/m3
2. Density maximum at 40Ci.e., above freezing!
3. High viscosity (but a Newtonian fluid)JJL ~ 10~3 N s/m2
4. High surface tensionor = 73 N/m
5. High bulk modulus (usually assumed incompressible)K 2.07 GPa
Thermal Properties
1. Specific Heathighest except for NH3-C 4.187 kJ/(kg-C)
2. High heat of vaporizationcv 2.45 MJ/kg
3. High heat of fusioncf 0.36 MJ/kg
4. Expands on freezingin almost all other compounds, psolid > /9liq
5. High boiling pointc.f., H2 (20 K), O2 (90 K) and H2O (373 K)
6 Good conductor of heat relative to other liquids and nonmetal solids.
Chemical and Other Properties
1. Slightly ionizedwater is a good solvent for electrolytes and nonelectrolytes
2. Transparent to visible light; opaque to near infrared
3. High dielectric constantresponds to microwaves and electromagnetic fields
Note: The values are approximate. All the properties listed are functions of temperature, pressure, water purity, and
other factors that should be known if more exact values are to be assigned. For example, surface tension is greatly
influenced by the presence of soap films, and the boiling point depends on water purity and confining pressure. The
values are generally indicative of conditions near 1O0C and one atmosphere of pressure.

table especially density and viscosity valuesare used regularly by pipeline engineers.
Other properties, such as compressibility and thermal values, are used indirectly, primarily to justify modeling assumptions, such as the flow being isothermal and incompressible. Many properties of water depend on intermolecular forces that create powerful
attractions (cohesion) between water molecules. That is, although a water molecule is
electrically neutral, the two hydrogen atoms are positioned to create a tetrahedral charge
distribution on the water molecule, allowing water molecules to be held strongly together with the aid of electrostatic attractions. These strong internal forcestechnically called
'hydrogen bonds'arise directly from the non-symmetrical distribution of charge.
The chemical behavior of water also is unusual. Water molecules are slightly ionized,
making water an excellent solvent for both electrolytes and nonelectrolytes. In fact, water
is nearly a universal solvent, able to wear away mountains, transport solutes, and support
the biochemistry of life. But the same properties that create so many benefits also create
problems, many of which must be faced by the pipeline engineer. Toxic chemicals, disinfection byproducts, aggressive and corrosive compounds, and many other substances can
be carried by water in a pipeline, possibly causing damage to the pipe and placing consumers at risk.
Other challenges also arise. Water's almost unique property of expanding on freezing
can easily burst pipes. As a result, the pipeline engineer either may have to bury a line or
may need to supply expensive heat-tracing systems on lines exposed to freezing weather,
particularly if there is a risk that standing water may sometimes occur. Water's high viscosity is a direct cause of large friction losses and high energy costs whereas its vapor
properties can create cavitation problems in pumps, valves, and pipes. Furthermore, the
combination of its high density and small compressibility creates potentially dramatic
transient conditions. We return to these important issues after considering how pipeline
flows respond to various physical constraints and influences in the next section.
2.4.2 Laws of Conservation (How?)
Although the implications of the characteristics of water are enormous, no mere list of its
properties will describe a physical problem completely. Whether we are concerned with
water quality in a reservoir or with transient conditions in a pipe, natural phenomena also
obey a set of physical laws that contributes to the character and nature of a system's
response. If engineers are to make quantitative predictions, they must first understand the
physical problem and the mathematical laws that model its behavior.
Basic physical laws must be understood and be applied to a wide variety of applications and in a great many different environments: from flow through a pump to transient
conditions in a channel or pipeline. The derivations of these equations are not provided,
however, because they are widely available and take considerable time and effort to do
properly. Instead, the laws are presented, summarized, and discussed in the pipeline context. More precisely, a quantitative description of fluid behavior requires the application
of three essential relations: (1) a kinematic relation obtained from the law of mass conservation in a control volume, (2) equations of motion provided by both Newton's second
law and the energy equation, and (3) an equation of state adapted from compressibility
considerations, leading to a wavespeed relation in transient flow and justifying the
assumption of an incompressible fluid in most steady flow applications.
A few key facts about mass conservation and Newton's second law are reviewed
briefly in the next section. Consideration of the energy equation is deferred until steady
flow is discussed in more detail, whereas further details about the equation of state are
introduced along with considerations of unsteady flow.

2.4.3 Conservation of Mass

One of a pipeline engineer's most basic, but also most powerful, tools is introduced in this
section. The central concept is that of conservation of mass and its key expression is the
continuity or mass conservation equation.
One remarkable fact about changes in a physical system is that not everything changes.
In fact, most physical laws are conservation laws: They are generalized statements about
regularities that occur in the midst of change. As Ford (1973) said:
A conservation law is a statement of constancy in naturein particular, constancy
during change. If for an isolated system a quantity can be defined that remains
precisely constant, regardless of what changes may take place within the system,
the quantity is said to be absolutely conserved.
A number of physical quantities have been found that are conserved in the sense of
Ford's quotation. Examples include energy (if mass is accounted for), momentum, charge,
and angular momentum. One especially important generalization of the law of mass conservation includes both nuclear and chemical reactions (Hatsopoulos and Keenan, 1965).
2.4.3.1 Law of Conservation of Chemical Species "Molecular species are conserved in
the absence of chemical reactions and atomic species are conserved in the absence of
nuclear reactions". In essence, the statement is nothing more a principle of accounting,
stating that the number of atoms or molecules that existed before a given change is equal
to the number that exists after the change. More powerfully, the principle can be transformed into a statement of revenue and expenditure of some commodity over a definite
period of time. Because both hydraulics and hydrology are concerned with tracking the
distribution and movement of the Earth's water, which is nothing more than a particular
molecular species, it is not surprising that formalized statements of this law are used frequently. These formalized statements are often called water budgets, typically if they
apply to an area of land, or continuity relations, if they apply in a well-defined region of
flow (the region is well-defined; the flow need not be).
The principle of a budget or continuity equation is applied every time we balance a
checkbook. The account balance at the end of any period is equal to the initial balance
plus all the deposits minus all the withdrawals. In equation form, this can be written as
follows:
(balance)^ = (balance), + ^ deposits - ^ withdrawals
Before an analogous procedure can be applied to water, the system under consideration must be clearly defined. If we return to the checking-account analogy, this requirement simply says that the deposits and withdrawals included in the equation apply to one
account or to a well-defined set of accounts. In hydraulics and hydrology, the equivalent
requirement is to define a control volumea region that is fixed in space, completely surrounded by a "control surface," through which matter can pass freely. Only when the
region has been precisely defined can the inputs (deposits) and outputs (withdrawals) be
identified unambiguously.
If changes or adjustments in the water balance (AS) are the concern, the budget concept can be expressed as
AS = Sf- S1 = (balance), - (balance). = V1 - V0

(2.1)

where V1 represents the sum of all the water entering an area, and V0 indicates the total
volume of water leaving the same region. More commonly, however, a budget relation
such as Eq. 2.1 is written as a rate equation. Dividing the "balance" equation by At and
taking the limit as At goes to zero produces

& = JjL = I-O

(2.2)
at
where the derivative term S' is the time rate of change in storage, S is the water stored in
the control volume, I is the rate of which water enters the system (inflow), and O is the
rate of outflow. This equation can be applied in any consistent volumetric units (e.g., m3/s,
ftVs, L/s, ML/day, etc.)
When the concept of conservation of mass is applied to a system with flow, such as a
pipeline, it requires that the net amount of fluid flowing into the pipe must be accounted
for as fluid storage within the pipe. Any mass imbalance (or, in other words, net mass
exchange) will result in large pressure changes in the conduit because of compressibility
effects.
2.4.3.2 Steady Flow Assuming, in addition, that the flow is steady, Eq. 2.2 can be
reduced further to inflow = outflow or 7 = O. Since the inflow and outflow may occur at
several points, this is sometimes re-written as
E V A = S VA
inflow 1 1 outflow

(2-3)

Equation (2.3) states that the rate of flow into a control volume is equal to the rate of
outflow. This result is intuitively satisfying since no accumulation of mass or volume
should occur in any control volume under steady conditions. If the control volume were
taken to be the junction of a number of pipes, this law would take the form of Kirchhoff's
current lawthe sum of the mass flow in all pipes entering the junction equals the sum
of the mass flow of the fluid leaving the junction. For example, in Fig. 2.2, continuity for
the control volume of the junction states that
G1 + Q2 = C3 + Q4

2.4.4

(2-4)

Newton's Second Law

When mass rates of flow are concerned, the focus is on a single component of chemical
species. However, when we introduce a physical law, such as Newton's law of motion, we
obtain something even more profound: a relationship between the apparently unrelated
quantities of force and acceleration.
More specifically, Newton's second law relates the changes in motion of a fluid or solid
to the forces that cause the change. Thus, the statement that the resultant of all external
forces, including body forces, acting on a system is equal to the rate of change of momentum of this system with respect to time. Mathematically, this is expressed as
x^
d(mv)
ix=v

(2 5)

where t is the time and Fext represents the external forces acting on a body of mass m
moving with velocity i). If the mass of the body is constant, Eq. (2.5) becomes

FIGURE 2.2 Continuity at a pipe junction Q1 + Q2 = Q3 + Q4

2Fex, = f = mfl

(2.6)

where a is the acceleration of the system (the time rate of change of velocity).
In closed conduits, the primary forces of concern are the result of hydrostatic pressure,
fluid weight, and friction. These forces act at each section of the pipe to produce the net
acceleration. If these forces and the fluid motion are modeled mathematically, the result
is a "dynamic relation" describing the transient response of the pipeline.
For a control volume, if flow properties at a given position are unchanging with time,
the steady form of the moment equation can be written as
2X*,= | PV(V n)dA
./CS

(2.7)

where the force term is the net external force acting on the control volume and the right
hand term gives the net flux of momentum through the control surface. The integral is
taken over the entire surface of the control volume, and the integrand is the incremental
amount of momentum leaving the control volume.
The control surface usually can be oriented to be perpendicular to the flow, and one
can assume that the flow is incompressible and uniform. With this assumption, the
momentum equation can be simplified further as follows:
2> = (P<4w)out - (pAvv)in = PG(vout - vin)

(2.8)

where Q is the volumetric rate of flow.

Example: Forces at an Elbow. One direct application of the momentum relation is
shown in Fig. 2.3, which indicates the flows and forces at an elbow. The elbow is assumed
to be mounted in a horizontal plane so that the weight is balanced by vertical forces (not
shown).

FIGURE 2.3 Force and momentum fluxes at an elbow.

The reaction forces shown in the diagram are required for equilibrium if the elbow is
to remain stationary. Specifically, the force Fx must resist both the pressure force and must
account for the momentum-flux term. That is, taking x as positive to the right, direct application of the momentum equation gives
(PA)1 - Fx= -P(Su1

(2.9)

Fx = (PA)1 + PGu1

(2.10)

Thus,
In a similar manner, but taking y as positive upward, direct application of the momentum equation gives
(PA)2 - Fy = pQ( -V2)

(2.11)

(here the outflow gives a positive sign, but the velocity is in the negative direction). Thus,
Fy = (PA)2 + PGu2

(2.12)

In both cases, the reaction forces are increased above what they would be in the static case because the associated momentum must either be established or be eliminated
in the direction shown. Application of this kind of analysis is routine in designing thrust
blocks, which are a kind of anchor used at elbows or bends to restrain the movement of
pipelines.

2.5 SYSTEM CAPACITY: PROBLEMS IN TIME

AND SPACE
A water transmission or supply pipeline is not just an enclosed tube it is an entire system that transports water, either by using gravity or with the aid of pumping, from its
source to the general vicinity of the demand. It typically consists of pipes or channels with
their associated control works, pumps, valves, and other components. A transmission sys-

tern is usually composed of a single-series line, as opposed to a distribution system that

often consists of a complex network of interconnected pipes.
As we have mentioned, there are many practical questions facing the designer of such
a system. Do the pipes, reservoirs and pumps have a great enough hydraulic capacity? Can
the flow be controlled to achieve the desired hydraulic conditions? Can the system be
operated economically? Are the pipes and connections strong enough to withstand both
unsteady and steady pressures?
Interestingly, different classes of models are used to answer them, depending on the
nature of the flow and the approximations that are justified. More specifically, issues of
hydraulic capacity are usually answered by projecting demands (water requirements) and
analyzing the system under steady flow conditions. Here, one uses the best available estimates of future demands to size and select the primary pipes in the system. It is the
hydraulic capacity of the system, largely determined by the effective diameter of the
pipeline, that links the supply to the demand.
Questions about the operation and sizing of pumps and reservoirs are answered by
considering the gradual variation of demand over relatively short periods, such as over an
average day or a maximum day. In such cases, the acceleration of the fluid is often negligible and analysts use a quasi-steady approach: that is, they calculate forces and energy
balances on the basis of steady flow, but the unsteady form is used for the continuity equation so that flows can be accumulated and stored.
Finally, the issue of required strength, such as the pressure rating of pipes and fittings,
is answered by considering transient conditions. Thus, the strength of a pipeline is determined at least in part by the pressures generated by a rapid transition between flow states.
In this stage, short-term and rapid motions must be taken into account, because large
forces and dangerous pressures can sometimes be generated. Here, forces are balanced
with accelerations, mass flow rates with pressure changes. These transient conditions are
discussed in more detail in section 2.8 and in chapter 10.
A large number of different flow conditions are encountered in pipeline systems. To
facilitate analysis, these conditions are often classified according to several criteria. Flow
classification can be based on channel geometry, material properties, dynamic considerations (both kinematic and kinetic), or some other characteristic feature of the flow. For
example, on the basis of fluid type and channel geometry, the flow can be classified as
open-channel, pressure, or gas flow. Probably the most important distinctions are based
on the dynamics of flow (i.e., hydraulics). In this way, flow is classified as steady or
unsteady, turbulent or laminar, uniform or nommiform, compressible or incompressible,
or single phase or multiphase. All these distinctions are vitally important to the analyst:
collectively, they determine which physical laws and material properties are dominant in
any application.
Steady flow: A flow is said to be steady if conditions at a point do not change with
time. Otherwise a flow is unsteady or transient. By this definition, all turbulent flows, and
hence most flows of engineering importance, are technically unsteady. For this reason, a
more restrictive definition is usually applied: A flow is considered steady if the temporal
mean velocity does not change over brief periods. Although the assumption is not formally required, pipeline flows are usually considered to be steady; thus, transient conditions represent an 'abnormal', or nonequilibrium, transition from one steady-state flow to
another. Unless otherwise stated, the initial conditions in transient problems are usually
assumed to be steady.
Steady or equilibrium conditions in a pipe system imply a balance between the physical laws. Equilibrium is typified by steady uniform flow in both open channels and closed
conduits. In these applications, the rate of fluid inflow to each segment equals the rate of

outflow, the external forces acting on the flow are balanced by the changes in momentum,
and the external work is compensated for by losses of mechanical energy. As a result, the
fluid generally moves down an energy gradient, often visualized as flow in the direction
of decreasing hydraulic grade-line elevations (e.g., Fig. 2.1).
Quasi-steady flow. When the flow becomes unsteady, the resulting model that must
be used depends on how fast the changes occur. When the rate of change is particularly slow, typically over a period of hours or days, the rate of the fluids acceleration is
negligible. However, fluid will accumulate or be depleted at reservoirs, and rates of
demand for water may slowly adjust. This allows the use of a quasi-steady or extended-duration simulation model.
Compressible and Incompressible. If the density of the fluid p is constantboth in
time and throughout the flow fielda flow is said to be incompressible. Thus, p is not a
function of position or time in an incompressible flow. If changes in density are permitted or required, the flow is compressible.
Surge. When the rate of change in flow is moderate, typically occurring over a period
of minutes, a surge model is often used. In North America, the term surge indicates an
analysis of unsteady flow conditions in pipelines when the following assumptions are
made: the fluid is incompressible (thus, its density is constant) and the pipe walls are rigid
and do not deform. These two assumptions imply that fluid velocities are not a function
of position along a pipe of constant cross-section and the flow is uniform. In other words,
no additional fluid is stored in a length of pipe as the pressure changes; because velocities
are uniform, the rate at which fluid enters a pipe is always equal to the rate of discharge.
However, the acceleration of the fluid and its accumulation and depletion from reservoirs
are accounted for in a surge model.
Waterhammer. When rapid unsteady flow occurs in a closed conduit system, the transient condition is sometimes marked by a pinging or hammering noise, appropriately
called waterhammer. However, it is common to refer to all rapidly changing flow conditions by this term, even if no audible shock waves are produced. In waterhammer models,
it is usually assumed that the fluid is slightly compressible, and the pipe walls deform with
changes in the internal pressure. Waterhammer waves propagate with a finite speed equal
to the velocity of sound in the pipeline.
The speed at which a disturbance is assumed to propagate is the primary distinction
between a surge and a waterhammer model. Because the wavespeed parameter a is related to fluid storage, the wavespeed is infinite in surge or quasi-steady models. Thus, in
effect, disturbances are assumed to propagate instantly throughout the pipeline system. Of
course they do no such thing, because the wavespeed is a finite physical property of a pipe
system, much like its diameter, wall thickness, or pipeline material. The implication of
using the surge or quasi-steady approximation is that the unsteady behavior of the pipe
system is controlled or limited by the rate at which the hydraulic boundary conditions
(e.g., pumps, valves, reservoirs) at the ends of the pipe respond to the flow and that the
time required for the pipeline itself to react is negligible by comparison.
Although unsteady or transient analysis is invariably more involved than is steadystate modeling, neglecting these effects in a pipeline can be troublesome for one of two
reasons: the pipeline may not perform as expected, possibly causing large remedial
expenses, or the line may be overdesigned with respect to transient conditions, possibly
causing unnecessarily large capital costs. Thus, it is essential for engineers to have a clear
physical grasp of transient behavior and an ability to use the computer's power to maximum advantage.
One interesting point is that as long as one is prepared to assume the flow is compressible, the importance of compressibility does not need to be known a priori. In fact,

all the incompressible, quasi-steady, and steady equations are special cases of the full transient equations. Thus, if the importance of compressibility or acceleration effects is
unknown, the simulation can correctly assume compressible flow behavior and allow the
analysis to verify or contradict this assumption.
Redistribution of water, whatever model or physical devices are used, requires control
of the fluid and its forces, and control requires an understanding not only of physical law
but also of material properties and their implications. Thus, an attempt to be more specific and quantitative about these matters will be made as this chapter progresses.
In steady flow, the fluid generally moves in the direction of decreasing hydraulic
grade-line elevations. Specific devices, such as valves and transitions, cause local pressure drops and dissipate mechanical energy; operating pumps do work on the fluid and
increase downstream pressures while friction creates head losses more or less uniformly
along the pipe length. Be warned, howeverin transient applications, this orderly situation rarely exists. Instead, large and sudden variations of both discharge and pressure can
occur and propagate in the system, greatly complicating analysis.

2.6

STEADYFLOW

The design of steady flow in pipeline systems has two primary objectives. First, the
hydraulic objective is to secure the desired pressure and flow rate at specific locations in
the system. Second, the economic objective is to meet the hydraulic requirements with the
minimum expense.
When a fluid flows in a closed conduit or open channel, it often experiences a complex interchange of various forms of mechanical energy. In particular, the work that is
associated with moving the fluid through pressure differences is related to changes in both
gravitational potential energy and kinetic energy. In addition, the flow may lose mechanical energy as a result of friction, a loss that is usually accounted for by extremely small
increases in the temperature of the flowing fluid (that is, the mechanical energy is converted to thermal form).
More specifically, these energy exchanges are often accounted for by using an
extended version of Bernoulli's famous relationship. If energy losses resulting from friction are negligible, the Bernoulli equation takes the following form:

H+*-?+!+*
where P1 and/?2 are the pressures at the end points, y is the specific weight of the fluid, V1
and V2 are the average velocities at the end points, and Z1 and Z2 are the elevations of the
end points with respect to an arbitrary vertical datum. Because of their direct graphical
representation, various combinations of terms in this relationship are given special labels,
historically called heads because of their association with vertical distances. Thus,
Head

Definition

Associated with

Pressure head
Elevation head
Velocity head
Piezometric head
Total head

p/j
z
v2/2g
p/j + z
p/j + z + v2/2g

Flow work
Gravitational potential energy
Kinetic energy
Pressure H- elevation head
Pressure + elevation + velocity head

Next Page

CHAPTER 3
HYDRAULICS OF OPEN
CHANNEL FLOW

Richard H. French
Desert Research Institute,
University and Community College System of Nevada
Reno, Nevada

3.7

INTRODUCTION

By definition, an open channel is a flow conduit having a free surface: that is, a boundary
exposed to the atmosphere. The free surface is essentially an interface between two fluids
of different density. Open-channel flows are almost always turbulent, unaffected by surface tension, and the pressure distribution within the fluid is hydrostatic. Open channels
include flows ranging from rivulets flowing across a field to gutters along residential
streets and highways to partially filled closed conduits conveying waste water to irrigation and water supply canals to vital rivers.
In this chapter, the basic principles of open channel hydraulics are presented as an
introduction to subsequent chapters dealing with design. By necessity, the material presented in this chapter is abbreviatedan abstract of the fundamental concepts and
approachesfor a more detailed treatment, the reader is referred to any standard references or texts dealing with the subject: for example, Chow (1959), French (1985),
Henderson (1966), or Chaudhry (1993)
As with any other endeavor, it is important that a common vocabulary be established
and used:
Critical slope (Sc): A longitudinal slope such that uniform flow occurs in a critical
state.
Flow area (A): The flow area is the cross-sectional area of the flow taken normal to
the direction of flow (Table 3.1).
Froude number (Fr): The Froude number is the dimensionless ratio of the inertial and
gravitational forces or
Fr = =
(3.1)
VgD
where V = average velocity of flow, g = gravitational acceleration, and D = hydraulic
depth. When Fr = 1, the flow is in a critical state with the inertial and gravitational forces
in equilibrium; when Fr < 1, the flow is in a subcritical state and the gravitational forces
are dominant; and when Fr > 1, the flow is in a supercritical state and the inertial forces
are dominant. From a practical perspective, sub - and supercritical flow can be differentiated simply by throwing a rock or other object into the flow. If ripples from the rock

TABLE 3.1 Channel Section Geometric Properties

Channel Definition

Rectangle

Trapezoid
with
equal side
slopes
Trapezoid
with
unequal
side slopes
Triangle
with equal
side slopes
Triangle
with
unequal
side slopes
Circular

Area

Wetted Perimeter

Hydraulic Radius

Top Width

Hydraulic Depth

progress upstream of the point of impact, the flow is subcritical; however, if ripples from
the rock do not progress upstream but are swept downstream, the flow is supercritical.
Hydraulic depth (D). The hydraulic depth is the ratio of the flow area (A) to the top
width (T)OtD= A/T (Table 3.1).
Hydraulic radius (R). The hydraulic radius is the ratio of the flow area (A) to the wetted perimeter (P) or R = A/P (Table 3.1).
Kinetic energy correction factor (a). Since no real open-channel flow is one-dimensional, the true kinetic energy at a cross section is not necessarily equal to the spatially
averaged energy. To account for this, the kinetic energy correction factor is introduced,
or

K^) "-"^*

and solving for a,

-^
<
When the flow is uniform, a = 1 and values oc of for various situations are summarized
in Table 3.2.
Momentum correction coefficient (ft): Analogous to the kinetic energy correction factor, the momentum correction factor is given by
PpgV = IJpV2JA
RP = Jlv2JA
-^4-

n
^
(33)

When the flow is uniform, P = I and values of (3 for various situations are summarized
in Table 3.2
Prismatic channel. A prismatic channel has both a constant cross-sectional shape and
bottom slope (S0). Channels not meeting these criteria are termed nonprismatic.
Specific energy (E). The specific energy of an open-channel flow is

E = y + a^

(3.4)

where y = depth of flow and the units of specific energy are length in meters or feet.
TABLE 3.2 Typical Values of a and p for Various Situations
Situation

Min.
Regular channels,
flumes,
spillways
Natural streams and torrents
Rivers under ice cover
River valleys, overflooded
Source: After Chow (1959).

Value of oc
Avg. Max.

Min.

Value of $
Avg. Max.

1.10

1.15

1.20

1.03

1.05

1.07

1.15
1.20
1.50

1.30
1.50
1.75

1.50
2.00
2.00

1.05
1.07
1.17

1.10
1.17
1.25

1.17
1.33
1.33

Specific momentum (M). By definition, the specific momentum of an open-channel

flow is

M = ^- + ~zA
gA

(3.5)

Stage: The stage of a flow is the elevation of the water surface relative to a datum. If
the lowest point of a channel section is taken as the datum, then the stage and depth of
flow (y) are equal if the longitudinal slope (S0) is not steep or cos (0) ~ 1, where 0 is the
longitudinal slope angle. If 0 < 10 or S0 ^ 0.18, where S0 is the longitudinal slope of the
channel, then the slope of the channel can be assumed to be small.
Steady. The depth (y) and velocity of flow (v) at a location do not vary with time; that
is, (dy/dt = O) and (3v/9r = O). In unsteady flow, the depth and velocity of flow at a location vary with time: that is, (3y/9/ ^ O) and (3v/9r ^ O).
Top width (T). The top width of a channel is the width of the channel section at the
water surface (Table 3.1).
Uniform flow. The depth (y). flow area (A), and velocity (V) at every cross section are
constant, and the energy grade line (Sy), water surface, and channel bottom slopes (S0) are
all parallel.
Superelevation (Ay). The rise in the elevation of the water surface at the outer channel boundary above the mean depth of flow in an equivalent straight channel, because of
centrifugal force in a curving channel.
Wetted perimeter (P). The wetted perimeter is the length of the line that is the interface between the fluid and the channel boundary (Table 3.1).

3.2

ENERGYPRINCIPLE

3.2.1 Definition of Specific Energy

Central to any treatment of open-channel flow is that of conservation of energy. The
total energy of a particle of water traveling on a streamline is given by the Bernoulli
equation or

p
V2
H = Z + - + oc
Y
2g
where H = total energy, z = elevation of the streamline above a datum, p = pressure, y
= fluid specific weight, (p/y) = pressure head, V1IZg = velocity head, and g = acceleration of gravity. H defines the elevation of the energy grade line, and the sum [z + (p//)]
defines the elevation of the hydraulic grade line. In most uniform and gradually varied
flows, the pressure distribution is hydrostatic (divergence and curvature of the streamlines
is negligible) and the sum [z + (p/y)] is constant and equal to the depth of flow y if the
datum is taken at the bottom of the channel. The specific energy of an open-channel flow
relative to the channel bottom is

E=

y+ a^ = y + a^v

(3 6)

where the average velocity of flow is given by

V=|

(3.7)

where Q = flow rate and A = flow area.

The assumption inherent in Eq. (3.6) is that the slope of the channel is small, or cos(0)
a 1. If 9 < 10 or S0 < 0.18, where S0 is the longitudinal slope of the channel, Eq. (3.6)
is valid. If 6 is not small, then the pressure distribution is not hydrostatic since the vertical depth of flow is different from the depth measured perpendicular to the bed of the
channel.
3.2.2 Critical Depth
If y in Eq. (3.6) is plotted as a function of E for a specified flow rate Q, a curve with two
branches results. One branch represents negative values of both E and y and has no physical meaning; but the other branch has meaning (Fig. 3.1). With regard to Fig. 3.1, the following observations are pertinent: 1) the portion designated AB approaches the line y =
E asymptotically, 2) the portion AC approaches the E axis asymptotically, 3) the curve has
a minimum at point A, and 4) there are two possible depths of flowthe alternate
depthsfor all points on the E axis to the right of point A. The location of point A, the
minimum depth of flow for a specified flow rate, can be found by taking the first derivative of Eq. (3.6) and setting the result equal to zero, or

Depth of Flow
y

f*-
It can be shown that dA = (T = dy) or (dA/dy = T) (French, 1985). Substituting this
result, using the definition of hydraulic depth and rearranging, Eq. (3.8) becomes

Subcritlcal
Decreasing Increasing
Supercritical

Specific Energy
Specific Momentum
FIGURE 3.1 Specific energy and momentum as a function of depth when the channel geometry and flow rate are specified.

!_
CLf^
_-eLZ: _-^ o
1
gA^dy = 1l gA*A =l 1 gD= U
or

V2
D
Tg = T

<3-9>

and

-= = Fr=l
VgD

(3.10)

which is the definition of critical flow. Therefore, minimum specific energy occurs at the
critical hydraulic depth and is the minimum energy required to pass the flow Q. With this
information, the portion of the curve AC in Fig. 3.1 is interpreted as representing supercritical flows, where as AB represents subcritical flows.
With regard to Fig. 3.1 and Eq. (3.6), the following observations are pertinent. First,
for channels with a steep slope and a * 1, it can be shown that
Fr=

V
IgD COs(Q)
V
a

(3.11)

Second, E - y curves for flow rates greater than Q lie to the right of the plotted curve,
and curves for flow rates less than Q lie to the left of the plotted curve. Third, in a rectangular channel of width b, y = D and the flow per unit width is given by
-

(3-12)

and

V=-

(3.13)

Then, where the subscript c indicates variable values at the critical point,
HT)"

(3 14)

f=y

(3 15)

yc = \Ee

(3.16)

and

In nonrectangular channels when the dimensions of the channel and flow rate are specified, critical depth is calculated either by the trial and error solution of Eqs. (3.8), (3.9),
and (3.10) or by use of the semiempirical equations in Table 3.3.
3.2.3 Variation of Depth with Distance
At any cross section, the total energy is
# = ^ = y+z

(3.17)

TABLE 3.3 Semiempirical Equations for the Estimation of yc

Channel Definition
(1)

Equation for yc
in terms of
V = a Q*/g
(2)

^ t a n g l e f y ] 0.33
(V)
T r a P e Z O l d

27
b
0 8 U-*
l 1 f[ ^0.75*
fcl.25]
J - WZ

Triangle
(WV2Q
U2J
Circle
(MLV*
UH
Source: From Straub (1982).

where y = depth of flow, z = elevation of the channel bottom above a datum, and it is
assumed that a and cos(6) are both equal to 1. Differentiating Eq. (3.17) with respect to
longitudinal distance,
dH _ \2g)
~d^~ dx

dy
~dx

dz
^x

n 1R ,
(J 18)

'

dH
dz
where j = the change of energy with longitudinal distance (Sf), = the channel botx
tom slope (S0), and, for a specified flow rate,
d

\2g) _ Q23 dA dy _ _(FTd

= _(Fr)2^
dx
#A dy dx
gA3 dx
dx

Substituting these results in Eq. (3.18) and rearranging,

= ^A2
dx
1 - Fr

(3.19)

which describes the variation of the depth of flow with longitudinal distance in a channel
of arbitrary shape.

Section

FIGURE 3.2 Channel with a compound section.

3.2.4 Compound Section Channels

In channels of compound section (Fig. 3.2), the specific energy correction factor oc is not
equal to 1 and can be estimated by
N / \
Yl* 3 ' 1
^Uv
a =

(3.20)
A2

where K1 and A1. as follows the conveyance and area of the /th channel subsection, respectively, K and A are conveyance are as follows:
N
K=^K1
i =1
and
N
A = EAI=1
N = number of subsections, and conveyance (K) is defined by Eq. (3.48) in Sec. 3.4.
Equation (3.20) is based on two assumptions: (1) the channel can be divided into subsections by appropriately placed vertical lines (Fig. 3.2) that are lines of zero shear and do
not contribute to the wetted perimeter of the subsection, and (2) the contribution of the
nonuniformity of the velocity within each subsection is negligible in comparison with the
variation in the average velocity among the subsections.

3.3

MOMENTUM

3.3.1 Definition of Specific Momentum

The one-dimensional momentum equation in an open channel of arbitrary shape and a
control volume located between Sections 1 and 2 is

YZ1A1 - 7 Z2A2 - Pj = 1 Q (V2 - V1)

O

(3.21)

where y = specific weight of water, A1 = flow area at sections 1 and 2; V. = average

velocity of flow at sections 1 and 2, Pf = horizontal component of unknown force acting
between Sections 1 and 2 and z,= distances to the centroids of the flow areas 1 and 2 from
the free surface. Substitution of the flow rate divided by the area for the velocities and
rearrangement of Eq. (3.21) yields
j = (K

~^)-(K

^)

or

-^ = M1-M2

(3.22)

where
M1 = ^- + Z1A1
(3.23)
&Aand M is known as the specific momentum or force function. In Fig. 3.1, specific momentum is plotted with specific energy for a specified flow rate and channel section as a function of the depth of flow. Note that the point of minimum specific momentum corresponds
to the critical depth of the flow.
The classic application of Eq. (3.22) occurs when Pf = O and the application of the resulting equation to the estimation of the sequent depths of a hydraulic jump. Hydraulic jumps
result when there is a conflict between the upstream and downstream controls that influence
the same reach of channel. For example, if the upstream control causes supercritical flow
while the downstream control dictates subcritical flow, there is a contradiction that can be
resolved only if there is some means to pass the flow from one flow regime to the other.
When hydraulic structures, such as weirs, chute blocks, dentated or solid sills, baffle piers,
and the like, are used to force or control a hydraulic jump, Pf in Eq. (3.22) is not equal to zero.
Finally, the hydraulic jump occurs at the point where Eq. (3.22) is satisfied (French, 1985).
3.3.2 Hydraulic Jumps in Rectangular Channels
In the case of a rectangular channel of width b and Pf = O, it can be shown (French,
1985) that
^- - V0.5 [1 + 8(Fr1)* - 1]
y\

(3.24)

^- = 0.5 [Vl + 8(Fr 2 )Z-I]

(3.25)

or

^= 2(Fr2)2 - 4(Fr2Y + 16(Fr2)* - ...

V
2
Equations (3.24) and (3.25) each contain three independent variables, and two must be
known before the third can be found. It must be emphasized that the downstream depth of
flow (y2) is not the result of upstream conditions but is the result of a downstream controlthat is, if the downstream control produces the depth y2 then a hydraulic jump will
form. The second form of Eq. (3.25) should be used when (Fr2)2 < 0.05 (French, 1985).

3.3.3

Hydraulic Jumps in Nonrectangular Channels

In analyzing the occurrence of hydraulic jumps in nonrectangular but prismatic channels, we

see that no equations are analogous to Eqs. (3.24) and (3.25). In such cases, Eq. (3.22) could
be solved by trial and error or by use of semiempirical equations. For example, in circular
sections, Straub (1978) noted that the upstream Froude number (Fr1) can be approximated by
*,-(*)
and the sequent depth can be approximated by
Fr1 < LTy2 = ^

(3.27)

yl.8
Fr1 > 1.Iy2 = 73

(3.28)

For horizontal triangular and parabolic prismatic channel sections, Silvester (1964,
1965) presented the following equations.
For triangular channels:
(^J-5- 1 = 1.5 W [l -(U)

(3.29)

For parabolic channels with the perimeter defined by y = aP-12, where a is a

coefficient:
|J5-l = 1.67W[l-|J5]

Value of Fr1
FIGURE 3.3 Analytic curves for estimating sequent depths in a trapezoidal channel
(From Silvester, 1964)

(3.30)

In the case of trapezoidal channels, Silvester (1964) presented a method for graphical
solution in terms of the parameter
*=

(3.31)

In Fig. 3.3, the ratio of Cv2^1) is plotted as a function of Fr1 and k.

3.4

UNIFORMFLOW

3.4.1 Manning and Chezy Equations

For computational purposes, the average velocity of a uniform flow can be estimated by
any one of a number of semiempirical equations that have the general form
V= CRxSy

(3.32)

where C = a resistance coefficient, R = hydraulic radius, S = channel longitudinal slope,

and x and y are exponents. At some point in the period 1768-1775 (Levi, 1995), Antoine
Chezy, designing an improvement for the water system in Paris, France, derived an equation relating the uniform velocity of flow to the hydraulic radius and the longitudinal slope
of the channel, or
V=C VRS

(3.33)

where C is the Chezy resistance coefficient. It can be easily shown that Eq (3.33) is similar in form to the Darcy pipe flow equation. In 1889, Robert Manning, a professor at the
Royal College of Dublin (Levi, 1995) proposed what has become known as Manning's
equation, or
y = i R^Vs

(3.34)

where n is Manning's resistance coefficient and <|> = 1 if SI units are used and <|) = 1.49
if English units are used. The relationship among C, n, and the Darcy-Weisbach friction
factor (D is
C

= jK = y*

(3.35)

At this point, it is pertinent to observe that n is a function of not only boundary roughness and the Reynolds number but also the hydraulic radius, an observation that was made
by Professor Manning (Levi, 1995).
3.4.2 Estimation of Manning's Resistance Coefficient
Of the two equations for estimating the velocity of a uniform flow, Manning's equation is
the more popular one. A number of approaches to estimating the value of n for a channel
are discussed in French (1985) and in other standard references, such as Barnes (1967),
Urquhart (1975), and Arcement and Schneider (1989). Appendix 3.A lists typical values
of n for many types of common channel linings.

In an unvegetated alluvial channel, the total roughness consists of two parts: grain or
skin roughness resulting from the size of the sediment particles and form roughness
because of the existence of bed forms. The total coefficient n can be expressed as
/i = /i' + n"

(3.36)

where n' = portion of Manning's coefficient caused by grain roughness and n" = portion
of Manning's coefficient caused by form roughness. The value of n' is proportional to the
diameter of the sediment particles to the sixth power. For example, Lane and Carlson
(1953) from field experiments in canals paved with cobbles with d15 in inches, developed
/T = 0.026^f

(3.37)

and Meyer-Peter and Muller (1948) for mixtures of bed material with a significant proportion of coarse-grained sizes with J90 in meters developed
n' = 0.038J91O6

(3.38)

/"= Form Roughness

Darcy-Welsboch friction factor/

In both equations, dxx is the sediment size such that xx percent of the material is smaller
by weight.
Although there is no reliable method of estimating n", an example of the variation of
/for the 0.19 mm sand data collected by Guy et al. (1966) is shown in Fig. 3.4. The n values commonly found for different bed forms are summarized in Table 3.4. The inability
to estimate or determine the variation of form roughness poses a major problem in the
study of alluvial hydraulics (Yang, 1996).
Use of Manning's equation to estimate the velocity of flow in channels where the primary component of resistance is from drag rather than bed roughness has been questioned
(Fischenich, 1996). However, the use of Manning's equation has persisted among engineers
because of its familiarity and the lack of a practical alternative. Jarrett (1984) recognized that

Grain Roughness

Measured unit stream power VS[(ft-lb/lb)/s]

FIGURE 3.4 Variation of the Darcy-Weisbach friction factor as a function of unit stream power.

guidelines for estimating resistance coefficients for high-gradient streams with stable beds
composed of large cobbles and boulders and minimally vegetated banks (S0 > 0.002) were
based on limited data. Jarrett (1984) examined 21 high-gradient streams in the Rocky
Mountains and developed the following empirical equation relating n to S0 and R (in feet):
= ^^
flO.16

(3.39)

Jarrett (1984) stated the following limitations on the use of Eq. (3.39): First, the equation is applicable to natural main channels with stable bed and bank materials (gravels,
cobbles, boulders) with no backwater. Second, the equation can be used for 0.002 < S0 ^
0.04 and 0.15 < / ? < 2.1 m (0.5 < R < 7.0 ft). Results of the regression analysis indicated that for R > 2.1 m ( 7.0 ft), n did vary significantly with depth; therefore, as long as
the bed and bank material remain stable, extrapolation to larger flows should not result in
significant error. Third, the hydraulic radius does not include the wetted perimeter of the
bed particles. Fourth, the streams used in the analysis had relatively small amounts of suspended sediment.
Vegetated channels present unique challenges from the viewpoint of estimating roughness. In grass-lined channels, the traditional approach assumed that n was a function of
vegetal retardance and VR (CoyIe, 1975). However, there are approaches more firmly
based on the principles of fluid mechanics and the mechanics of materials (Kouwen, 1988;
Kouwen and Li, 1980.) Data also exist that suggest that in such channels flow duration is
not a factor as long as the vegetal elements are not destroyed or removed. Further, inundation times, and/or hydraulic stresses, or both that are sufficient to damage vegetation
have been found, as might be expected, to reduce the resistance to flow (Temple, 1991).
Petryk and Bosmajian (1975) presented a relation for Manning's n in vegetated channels based on a balance of the drag and gravitational forces, or
n = ^[C^]'2

(3.40)

where Cd a coefficient accounting for the drag characteristics of the vegetation and (Veg)rf
the vegetation density. Flippin-Dudley (1997) has developed a rapid and objective
procedure using a horizontal point frame to measure (Veg)d . Equation (3.40) is limited
because there is limited information regarding Cd for vegetation (Flippin-Dudley
et aL, 1997).
3.4.3 Equivalent Roughness Parameter k
In some cases, an equivalent roughness parameter k is used to estimate n. Equivalent
roughness, sometimes called "roughness height," is a measure of the linear dimension
of roughness elements but is not necessarily equal to the actual or even the average height
of these elements. The advantage of using k instead of Manning's n is that k accounts for
changes in the friction factor due to stage, whereas the Manning's n does not. The relationship between n and k for hydraulically rough channels is
*=

6/?1/6
7
TH
YlogJ 12.2K
V
J

where F = 32.6 for English units and 18.0 for SI units.

<3-41)

TABLE 3.4 Equivalent Roughness Values of Various Bed Materials

Material

k
(ft)
(2)

(1)

Brass, copper, lead, glass

0.0001-0.0030
Wrought iron, steel
0.0002-0.0080
Asphalted cast iron
0.0004-0.0070
Galvanized iron
0.0005-0.0150
Cast iron
0.0008-0.0180
Wood stave
0.0006-0.0030
Cement
0.0013-0.0040
Concrete
0.0015-0.0100
Untreated gunite
0.01-0.033
Drain tile
0.0020-0.0100
Riveted steel
0.0030-0.0300
Rubble masonry
0.02
Straight, uniform earth
0.01
channels
Natural streambed
0.1000-3.0000
Sources: From Ackers C (1958), Chow (1959), and Zegzhda (1938).

k
M
(3)
0.00003048-0.0009
0.0001-0.0024
0.0001-0.0021
0.0002-0.0046
0.0002-0.0055
0.0002-0.0009
0.0004-0.0012
0.0005-0.0030
0.0030-0.0101
0.0006-0.0030
0.0009-0.0091
0.0061
0.0030
0.0305-0.9144

With regard to Eq. (3.41), it is pertinent to observe that as R increases (equivalent to

an increase in the depth of flow), n increases. Approximate values of k for selected
materials are summarized in Table 3.4. For sand-bed channels, the following sediment
sizes have been suggested by various investigators for estimating the value of k: k = d65
(Einstein, 1950), k = J90 (Meyer-Peter and Muller, 1948), and k = J85 (Simons and
Richardson, 1966).

3.4.4

Resistance in Compound Channels

In many designed channels and most natural channels, roughness varies along the perimeter of the channel, and it is necessary to estimate an equivalent value of n for the entire
perimeter. In such cases, the channel is divided into TV parts, each with an associated wetted perimeter (P1), hydraulic radius (R1), and roughness coefficient (n.), and the equivalent
roughness coefficient (ne) is estimated by one of the following methods. Note that the wetted perimeter does not include the imaginary boundaries between the subsections.
1. Horton (1933) and Einstein and Banks (1950) developed methods of estimating ne
assuming that the average velocity in each of the subdivisions is the same as the
average velocity of the total section. Then

ne =

N ,
\ ~|2/3
sM
i^Lj

(3.42)

2. Assuming that the total force resisting motion is equal to the sum of the subsection
resisting forces,

N
~|l/2
E
w>
ne = i^^Jt

(3.43)

3. Assuming that the total discharge of the section is equal to the sum of the subsection discharges,
PR
ne =
(3.44)
Y ^L!
~i "i
4. Weighting of resistance by area (Cox, 1973),
AT
E A
ne =-^~

(3.45)

5. The Colebatch method (Cox, 1973).

/^
V3
/EM 3 M

"^V'V-}

(3 46)

3.4.5 Solution of Manning's Equation

The uniform flow rate is the product of the velocity of flow and the flow area, or
Q = VA = -AR^VS
(3.47)
n
In Eq. (3.47), ARm is termed the section factor and, by definition, the conveyance of the
channel is
6
K = -- AR
(3.48)
Before the advent of computers, the solution of Eq. (3.34) or Eq. (3.47) to estimate the
depth of flow for specified values of V (or Q\ n, and S was accomplished in one of two
ways: by trial and error or by the use of a graph of AR2/3 versus v. In the age of the
desktop computer, software is used to solve the equations of uniform flow. Trial and error
and graphical approaches to the solution of the uniform flow equations can be found in
any standard reference or text (e.g., French, 1985).
3.4.6 Special Cases of Uniform Flow
3.4.6.1 Normal and critical slopes. If Q, n, and yN (normal depth of flow) and the channel section are defined, then Eq. (3.47) can be solved for the slope that allows the flow to
occur as specified; by definition, this is a normal slope. If the slope is varied while the discharge and roughness are held constant, then a value of the slope such that normal flow

occurs in a critical state can be found: that is, a slope such that normal flow occurs with
Fr = 1. The slope obtained is the critical slope, but it also is a normal slope. The smallest
critical slope, for a specified channel shape, roughness, and discharge is termed the limiting critical slope. The critical slope for a given normal depth is
N
S*c = 8n2
pRjp

(3
49)
(J-W)

where the subscript Af indicates the normal depth value of a variable and, for a wide channel,
gn2
(3 50)
*=^
3.4.6.2 Sheet/low. A special but noteworthy uniform flow condition is that of sheetflow. From the viewpoint of hydraulic engineering, a necessary condition for sheetflow is
that the flow width must be sufficiently wide so that the hydraulic radius approaches the
depth of flow. With this stipulation, the Manning's equation, Eq. (3.48), for a rectangular
channel becomes
Q = ^Ty^ VS
(3.51)
where T = sheetflow width and yN = normal depth of flow. Then, for a specified flow rate
and sheetflow width, Eq. (3.51) can be solved for the depth of flow, or

( nO >/s
(3 52)
^ = UFW/
The condition that the value of the hydraulic radius approaches the depth of flow is not
a sufficient condition. That is, this condition specifies no limit on the depth of flow, and
there is general agreement that sheetflow has a shallow depth of flow. Appendix 3.A summarizes Manning's n values for overland and sheetflow.
3.4.6.3 Superelevation. When a body of water moves along a curved path at constant
velocity, it is acted for a force directed toward the center of the curvature of the path.
When the radius of the curve is much larger than the top width of the water surface, it can
be shown that the rise in the water surface at the outer channel boundary above the mean
depth of flow in a straight channel (or superelevation) is
V2T
A, = _

(3.53)

where r = the radius of the curve (Linsley and Franzini, 1979). It is pertinent to note that
if the effects of the velocity distribution and variations in curvature across the channel are
considered, the superelevation may be as much as 20 percent more than that estimated by
Eq. (3.53) (Linsley and Franzini, 1979). Additional information regarding superelevation
is available in Nagami et al., (1982) and U.S. Army Corps of Engineers (USAGE, 1970).

3.5

GRADUALLY AND SPATIALLY VARIED FLOW

3.5.1 Introduction
The gradual variation in the depth of flow with longitudinal distance in an open channel
is given by Eq. (3.19), or

dy _ S0 - Sf
dx
1 - Fr2
and two cases warrant discussion. In the first case, because the distance over which the
change in depth is short it is appropriate to assume that boundary friction losses are small,
or Sf = O. When this is the case, important design questions involve abrupt steps in the
bottom of the channel (Fig. 3.5) and rapid expansions or contractions of the channel
(Fig. 3.6). The second case occurs when Sf# O.

3.5.2 Gradually Varied Flow with Sf = O

When Sf = O and the channel is rectangular in shape and has a constant width, Eq. (3.19)
reduces to
(1

(3 54)
- Fr2)l+ Tx =
'
and the following observations are pertinent (the observations also apply to channels of
arbitrary shape):

1. If dz/dx > O (upward step) and Fr < 1, then dy/dx must be less than zerodepth
of flow decreases as x increases.
2. If dz/dx > O (upward step) and Fr > 1, then dy/dx must be greater than zerodepth
of flow increases as x increases.
Profile

Profile

FIGURE 3.5 Definition of variables for gradually varied flow over positive and negative steps.

Plan

FIGURE 3.6 Definition of variables for gradually varied flow through contracting
and expanding channel sections.

3. If dz/dx < O (downward step) and Fr < 1, then dy/dx must be greater than zero
depth of flow increases as x increases.
4. If dz/dx < O (downward step) and Fr > 1, then dy/dx must be less than zerodepth
of flow decreases as x increases.
In the case of a channel of constant width with a positive or negative step, the relation
between the specific energy upstream of the step and the specific energy downstream of
the step is
E1 = E2 + Az

(3.55)

In the case dz/dx = O, if the channel is rectangular in shape but the width of the channel changes, it can be shown (French, 1985) that the governing equation is
(1 - Fr^ - Fr*J^ = O
dx
b dx

(3.56)

The following observations also apply to channels of arbitrary shape:

1. If db/dx > O (width increases) and Fr < 1, then dy/dx must be greater than
zero-depth of flow increases as x increases.
2. If db/dx > O (width increases) and Fr > 1, then dy/dx must be less than zerodepth
of flow decreases as x increases.

3. If db/dx < O (width decreases) and Fr < 1, then dyldx must be less than zero
depth of flow decreases as x increases.
4. If db/dx < O (width decreases) and Fr > 1, then dyldx must be greater than zero
depth of flow increases as x increases.
In this case, the relation between the specific energy upstream of the contraction
(expansion) and the specific energy downstream of the step contraction (expansion) is
E1 = E2

(3.57)

It is pertinent to note that in the case of supercritical flow, channel expansions and contractions may result in the formation of waves.
Additional information regarding steps, expansions, and contractions can be found in any
standard reference or text on open-channel hydraulics (e.g., French, 1985).
3.5.3 Gradually Varied Flow with Sf O
In the case where Sf cannot be neglected, the water surface profile must estimated. For a
channel of arbitrary shape, Eq. (3.19) becomes
dy

_S0-Sf

_tfQ2p*3
1

~gA3"

^ gA3

For a specified value of Q, Fr and Sf are functions of the depth of flow y. For illustrative purposes, assume a wide channel; in such a channel, Fr and Sf will vary in much the
same way with y since P-T and both Sf and Fr have a strong inverse dependence on the
flow area. In addition, as y increases, both Sf and Fr decrease. By definition, Sf = S0 when
y = yN. Given the foregoing, the following set of inequalities must apply:
Sf > S0

for

y < yN

Fr > 1

for

y < yc

Sf<S0

for

y>yN

Fr < 1

for

y > yc

and

These inequalities divide the channel into three zones in the vertical dimension. By convention, these zones are labeled 1 to 3 starting at the top. Gradually varied flow profiles
are labeled according to the scheme defined in Table 3.5.
For a channel of arbitrary shape, the standard step methodology of calculating the gradually varied flow profile is commonly used: for example, HEC-2 (USAGE, 1990) or HECRAS (USAGE, 1997). The use of this methodology is subject to the following assumptions: (1) steady flow, (2) gradually varied flow, (3) one-dimensional flow with correction
for the horizontal velocity distribution, (4) small channel slope, (5) friction slope (averaged) constant between two adjacent cross sections, and (6) rigid boundary conditions.
The application of the energy equation between the two stations shown in Fig. 3.7 yields

TABLE 3.5 Classifications of Gradually Varied Flow Profiles

Profile Designation
Channel
Slope

Zone
1

Zone
2

Zone
3

(1)

(2)

(3)

(4)

Mild
O < S0 < Sc

Ml

y>yN>yc
Ml

yN>y>yc
M3

Critical
S0 = SC>0

Cl

yN>yc>y
y>yc = yN

C2

y = yN = yc
C3

Steep
S0 > Sc > O

Relation ofy
to yN and
yc
(5)

Sl

yc = yN>y
y>yc>yN

52

yc>y>yN

Type
of
Curve
(6)
Backwater
(dy/dx > O)
Drawdown
(dy/dx < O)
Backwater
(dy/dx > O)
Backwater
(dy/dx > O)
Parallel to
channel bottom
(dy/dx = O)
Backwater
(dy/dx > O)
Backwater
(dy/dx > O)
Drawdown
(dy/dx < O)

Type
of
Flow
(7)
Subcritical
Subcritical
Supercritical
Subcritical
Uniform
critical
Supercritical
Subcritical
Supercritical

TABLE 3.5:

(Continued)

Channel
Slope

Zone
1

Zone
2

Zone
3

(1)

(2)

(3)

Horizontal

(4)

Relation ofy
to yN and
yc
(5)

Type
of
Curve
(6)

Type
of
Flow
(7)

S3

yc>yN>y

Backwater
(dy/dx > O)

Supercritical

yN>y>yc

Drawdown
(dy/dx < O)
Backwater
(dy/dx > O)

Subcritical

Drawdown
(dy/dx < O)
Backwater
(dy/dx > O)

Subcritical

None

So =

m
m
Adverse
S0 < O

yN>yc>y

Supercritical

None
A2

yN>y>yc
A3

)w > X: > y

Supercritical

Datum

Section 1

Section 2

FIGURE 3.7 Energy relationship between two channel sections.

yi
V2
Z1 + Gi1^- = z2 + a2^ + hf+he

(3.61)

where Z1 and Z2 = elevation of the water surface above a datum at Stations 1 and 2, respectively, he = eddy and other losses incurred in the reach, and hf = reach friction loss.
The friction loss can be obtained by multiplying a representative friction slope, Sf, by
the length of the reach, L. Four equations can be used to approximate the friction loss
between two cross sections:
Sf

( Q. + Q2 >
~v
v~ (average conveyance)
[K1 + K2 )

(3.62)

S +S
Sf = ^ (average friction slope)

(3.63)

2S S
Sf = ^ ,f (harmonic mean friction slope)
Sn + Sj2

(3.64)

~Sf = VSy1 Sfl (geometric mean friction slope)

(3.65)

and

The selection of a method to estimate the friction slope in a reach is an important decision and has been discussed in the literature. Laurenson (1986) suggested that the "true"
friction slope for an irregular cross section can be approximated by a third-degree poly-

nomial. He concluded that the average friction slope method produces the smallest maximum error, but not always the smallest error, and recommended its general use along with
the systematic location of cross sections. Another investigation based on the analysis of
98 sets of natural channel data showed that there could be significant differences in the
results when different methods of estimating the friction slope were used (USAGE, 1986).
This study also showed that spacing cross sections 15Om (500 ft) a part eliminated the differences.
The eddy loss takes into account cross section contractions and expansions by multiplying the absolute difference in velocity heads between the two sections by a contraction
or expansion coefficient, or
h =c

*^Ts~^Tg

(3 66)

'

There is little generalized information regarding the value of the expansion (C6) or the
contraction coefficient (Cc). When the change in the channel cross section is small, the
coefficients C6 and Cc are typically on the order of 0.3 and 0.1, respectively (USAGE,
1990). However, when the change in the channel cross section is abrupt, such as at
bridges, C6 and Cc may be as high as 1.0 and 0.6, respectively (USAGE, 1990).
With these comments in mind,
H1= Z1+ U1^

(3.67)

H2 = z2 + Oc2 ^J

(3.68)

and

With these definitions, Eq. (3.61) becomes

H1 = H2 + hf + he

(3.69)

Eq. (3.69) is solved by trial and error: that is, assuming H2 is known and given a longitudinal distance, a water surface elevation at Station 1 is assumed, which allows the
computation of H1 by Eq. (3.67). Then, hf and he are computed and H1 is estimated by
Eq. (3.67). If the two values of H1 agree, then the assumed water surface elevation at
Station 1 is correct.
Gradually varied water surface profiles are often used in conjunction with the peak
flood flows to delineate areas of inundation. The underlying assumption of using a steady
flow approach in an unsteady situation is that flood waves rise and fall gradually. This
assumption is of course not valid in areas subject to flash flooding such as the arid and
semiarid Southwestern United States (French, 1987).
In summary, the following principles regarding gradually varied flow profiles can be
stated:
1. The sign of dy/dx can be determined from Table 3.6.
2. When the water surface profile approaches normal depth, it does so asymptotically.
3. When the water surface profile approaches critical depth, it crosses this depth at a
large but finite angle.
4. If the flow is subcritical upstream but passes through critical depth, then the feature
that produces critical depth determines and locates the complete water surface profile. If the upstream flow is supercritical, then the control cannot come from the
downstream.

5. Every gradually varied flow profile exemplifies the principle that subcritical flows
are controlled from the downstream while supercritical flows are controlled from
upstream. Gradually varied flow profiles would not exist if it were not for the
upstream and downstream controls.
6. In channels with horizontal and adverse slopes, the term "normal depth of flow"
has no meaning because the normal depth of flow is either negative or imaginary.
However, in these cases, the numerator of Eq. (3.60) is negative and the shape of
the profile can be deduced.
Any method of solving a gradually varied flow situation requires that cross sections be
defined. Hoggan (1989) provided the following guidelines regarding the location of cross
sections:
1. They are needed where there is a significant change in flow area, roughness, or longitudinal slope.
2. They should be located normal to the flow.
3. They should be located in detailupstream, within the structure, and downstreamat structures such as bridges and culverts. They are needed at all control structures.
4. They are needed at the beginning and end of reaches with levees.
5. They should be located immediately below a confluence on a main stem and immediately above the confluence on a tributary.
6. More cross sections are needed to define energy losses in urban areas, channels
with steep slopes, and small streams than needed in other situations.
7. In the case of HEC-2, reach lengths should be limited to a maximum distance of
0.5 mi for wide floodplains and for slopes less than 38,550 m (1800 ft) for slopes
equal to or less than 0.00057, and 370 m (1200 ft) for slopes greater than 0.00057
(Beaseley, 1973).

3.6

GRADUALLY AND RAPIDLY VARIED UNSTEADY FLOW

3.6.1 Gradually Varied Unsteady Flow

Many important open-channel flow phenomena involve flows that are unsteady.
Although a limited number of gradually varied unsteady flow problems can be solved
analytically, most problems in this category require a numerical solution of the governing equations. Examples of gradually varied unsteady flows include flood waves, tidal
flows, and waves generated by the slow operation of control structures, such as sluice
gates and navigational locks.
The mathematical models available to treat gradually varied unsteady flow problems
are generally divided into two categories: models that solve the complete Saint Venant
equations and models that solve various approximations of the Saint Venant equations.
Among the simplified models of unsteady flow are the kinematic wave, and the diffusion
analogy. The complete solution of the Saint Venant equations requires that the equations
be solved by either finite difference or finite element approximations.
The one dimensional Saint Venant equations consist of the equation of continuity
* + * + * = <>
dt
dx
dx

(3.7Oa)

and the conservation of momentum equation

^v
*

3v
^v
Tx + 8irx-8(S-s')

+v

=0

(3 71a)

An alternate form of the continuity and momentum equations is

r|y +

dr

a^) = 0
dJC

and

I*+vav
g 3f
g 3*

*
dx

By rearranging terms, Eq. (3.7Ib) can be written to indicate the significance of each
term for a particular type of flow, or
\
-^
^fS=S ^-"steady
~\ - f ^ l ^
~J lJ steady, nonuniform - f -p1
/)f Iunsteady, nonuniform

(372)
\^''^)

Equations (3.70) and (3.71) compose a group of gradually varied unsteadyflowmodels that are termed complete dynamic models. Being complete, this group of models can
provide accurate results; however, in many applications, simplifying assumptions regarding the relative importance of various terms in the conservation of momentum equation
(Eq. 3.71) leads to other equations, such the kinematic and diffusive wave models (Ponce,
1989).
The governing equation for the kinematic wave model is
f

^f = 0

(3.73)

where e = a coefficient whose value depends on the frictional resistance equation used (e
= 5/3 when Manning's equation is used). The kinematic wave model is based on the equation of continuity and results in a wave being translated downstream. The kinematic wave
approximation is valid when

t SV
-^- > 85

(3.74)

where tR = time of rise of the inflow hydrograph (Ponce, 1989).

The governing equation for the diffusive wave model is
3(2
dQ ( Q Id 2 G
+
v
4 >=(*k}i

(3 75)

where the left side of the equation is the kinematic wave model and the right side accounts
for the physical diffusion in a natural channel. The diffusion wave approximation is valid
when (Ponce, 1989),
**So (y) 0 ' 5 * !5

(3.76)

If the foregoing dimensionless inequalities ( Eq. 3.74 and 3.76) are not satisfied, then the
complete dynamic wave model must be used. A number of numerical methods can be used
to solve these equations (Chaudhry, 1987; French; 1985, Henderson, 1966; Ponce, 1989).

3.6.2 Rapidly Varied Unsteady Flow

The terminology "rapidly varied unsteady flow" refers to flows in which the curvature of
the wave profile is large, the change of the depth of flow with time is rapid, the vertical
acceleration of the water particles is significant relative to the total acceleration, and the
effect of boundary friction can be ignored. Examples of rapidly varied unsteady flow include
the catastrophic failure of dams, tidal bores, and surges that result from the quick operation
of control structures such as sluice gates. A surge producing an increase in depth is termed
a positive surge, and one that causes a decrease in depth is termed a negative surge.
Furthermore, surges can go either upstream or downstream, thus giving rise to four basic
types (Fig. 3.8). Positive surges generally have steep fronts, often with rollers, and are stable. In contrast, negative surges are unstable, and their form changes with the advance of the
wave.
Consider the case of a positive surge (or wave) traveling at a constant velocity (wave
celerity) c up a horizontal channel of arbitrary shape (Fig. 3.8b). Such a situation can
result from the rapid closure of a downstream sluice gate. This unsteady situation is converted to a steady situation by applying a velocity c to all sections; that is, the coordinate
system is moving at the velocity of the wave. Applying the continuity equation between
Sections 1 and 2
(V1 + C)A1 = (V2 + c)A2

(3.77)

Since there are unknown losses associated with the wave, the momentum equation
rather than the energy equation is applied between Sections 1 and 2 or

FIGURE 3.8 Definition of variables for simple surges moving in

an open channel.

FIGURE 3.8 Definition of variables for simple surges moving in an open channel.

YA1Z1 - 7A2Z2 = 1 ^1(V1 + C)(V2 + c - V . - c )

(3.78)
O
where boundary friction has been ignored. Eliminating V2 in Eq. (3.78) by manipulation
ofEq. (3.77) yields
" ( A2 }
~|-5
8\-fA (A 1 Z 1 -A 2 Z 2 )
V i )
V1 + c =
Y^A
L
i A2

(3.79)

In the case of a rectangular channel, Eq. (3.79) reduces to

Vl + c _ V 5 - [ i ( 1 + a ) ]

When the slope of a channel becomes very steep, the resulting supercritical flow at
normal depth may develop into a series of shallow water waves known as roll waves. As
these waves progress downstream, they eventually break and form hydraulic bores or
shock waves. When this type of flow occurs, the increased depth of flow requires
increased freeboard, and the concentrated mass of the wavefronts may require additional
structural factors of safety.
Escoffier (1950) and Escoffier and Boyd (1962) considered the theoretical conditions
under which a uniform flow must be considered unstable. Whether roll waves form or not
is a function of the Vedernikov number (Ve), the Montuori number (Mo), and the concentration of sediment in the flow. When the Manning equation is used, the Ve is

TABLE 3.6 Shape Factor for Common Channel Sections

Channel Definition

Rectangle
Trapezoid
with
unequal side
slopes

Circle

Ve = |rFr

(3.8Ia)

Ve = IfFr

(3.8Ib)

and if the Chezy equation is used

Fr should be computed using Eq. (3.11) and F = a channel shape factor (Table 3.6) or
F = 1 -R^
dA

(3.82)

When Mo 1, flow instabilities should expected. The Montuori number is given by

Mo = ^

(3.83)

It is appropriate to note that in some publications (e.g., Aisenbrey et al., 1978) Mo is

the inverse of Eq. (3.83). Figure 3.9 provides a basis for deciding whether roll waves will
form in a given situation. In the figure, data from Niepelt and Locher (1989) for a slurry
flow are also plotted. The Niepelt and Locher data suggest that flow stability also is a
function of the concentration of sediment.

3.7

CONCLUSION

The foregoing sections provide the basic principles on which the following chapters on
design are based. Two observations are pertinent. First, open-channel hydraulics is incrementally progressing. That is, over the past several decades, there have been incremental
advances that primarily have added details, often important details, but no major new
advances. Second, open-channel hydraulics remains a one-dimensional analytic approach.
However, the assumption of a one-dimensional approach may not be valid in many situations: for example, nonprismatic channels, flow downstream of a partially breached dam,
or lateral flow over a spillway. In some of these cases, the one-dimensional approach may
provide an approximation that is suitable for design. In other cases, however, a two- or
three- dimensional approach should be used. Additional information regarding two- and
three- dimensional aDDroaches can be found in Chaudhrv (1993).

Flows with roll-waves

Flows without roll-waves
Clear Water

Slurry Curve - 43%

Concentration by Weight

Montuor! Number = *&FIGURE 3.9 Flow stability as a function of the Vedernikov and Montuori numbers for clear water and slurryflow.(Based on data from Montuori, 1963;
Niepelt and Locher, 1989

REFERENCES
Ackers, P., "Resistance to Fluids flowing in Channels and Pipes," Hydraulic Research Paper No. 1,
Her Majety's Stationery Office, London, 1958.
Aisenbrey, A. J., Jr., R. B., Hayes, H. J., Warren, D. L., Winsett, and R. B. Young, Design of Small
Canal Structures, U.S. Department of Interior, Bureau of Reclamation, Washington, DC 1978.
Arcement, G. J., and V. R. Schneider, "Guide for Selecting Manning's Roughness Coefficients for
Natural Channels and Flood Plains," Water Supply Paper 2339, U.S. Geological Survey,
Washington, DC, 1989.
Barnes, H. H., "Roughness Characteristics of Natural Channels," U.S. Geological Survey Water
Supply Paper No. 1849, U.S. Geological Survey, Washington, DC. 1967.
Beasley, J. G., An Investigation of the Data Requirements of Ohio for the HEC-2 Water Surface
Profiles Model, Master's thesis, Ohio State University, Columbus, 1973.
Chaudhry, M. H., Open-Channel Flow, Prentice-Hall, New York 1993.
Chaudhry, M. H., Applied Hydraulic Transients, Van Nostrand Reinhold, New York, 1987.
Chow, V. T., Open-Channel Hydraulics, McGraw-Hill, New York, 1959.
Cox, R. G., "Effective Hydraulic Roughness for Channels Having Bed Roughness Different from
Bank Roughness," Miscellaneous Paper H-73-2, U.S. Army Engineers Waterways Experiment
Station, Vicksburg, MS, 1973.
Coyle, J. J. "Grassed Waterways and Outlets," Engineering Field Manual, U.S. Soil Conservation
Service, Washington, DC, April, 1975, pp. 7-1-7-43.
Einstein, H. A., "The Bed Load Function for Sediment Transport in Open Channel Flows.
Technical Bulletin No. 1026, U.S. Department of Agriculture, Washington, DC, 1950.
Einstein, H. A., and R. B. Banks, "Fluid Resistance of Composite Roughness," Transactions of the
American Geophysical Union, 31(4): 603-610, 1950.
Escoffier, F. F., "A Graphical Method for Investigating the Stability of Flow in Open Channels
or in Closed Conduits Flowing Full," Transactions of the American Geophysical Union, 31(4),
1950.
Escoffier, F. F, and M. B. Boyd, "Stability Aspects of Flow in Open Channels," Journal of the
Hydraulics Division, American Society of Civil Engineers, 88(HY6): 145-166, 1962.
Fischenich, J. C., "Hydraulic Impacts of Riparian Vegetation: Computation of Resistance," EIRP
Technical Report EL-96-XX, U.S. Army Engineers Waterways Experiment Station, Vicksburg,
MS, August 1996.
Flippin-Dudley, S. J., "Vegetation Measurements for Estimating Flow Resistance," Doctoral dissertation, Colorado State University, Fort Collins, 1997.
Flippin-Dudley, S. J., S. R. Abt, C. D. Bonham, C. C. Watson, and J. C. Fischenich, "A Point
Quadrant Method of Vegetation Measurement for Estimating Flow Resistance," Technical Report
No. EL-97-XX, U.S. Army Engineers Waterways Experiment Station, Vicksburg, MS, 1997.
French, R. H., Hydraulic Processes on Alluvial Fans. Elsevier, Amsterdam, 1987.
French, R. H., Open-Channel Hydraulics, McGraw-Hill, New York, 1985.
Guy, H. P., D. B. Simons, and E. V. Richardson, "Summary of Alluvial Channel Data from Flume
Experiments, 1956-61," Professional Paper No. 462-1, U.S. Geological Survey, Washington, DC,
1966.
Henderson, F. M., Open Channel Flow, Macmillan, New York, 1966.
Hoggan, D. H., Computer-Assisted Floodplain Hydrology & Hydraulics, McGraw-Hill, New York,
1989.
Horton, R. E., "Separate Roughness Coefficients for Channel Bottom and Sides," Engineering
News Record, 3(22): 652-653, 1933.

Jarrett, R. D., "Hydraulics of High-Gradient Streams," Journal of Hydraulic Engineering,

American Society of Civil Engineers, 110(11): 1519-1539, 1984.
Kouwen, N., "Field Estimation of the Biomechanical Properties of Grass," Journal of Hydraulic
Research, International Association and Hydraulic Research, 26(5): 559-568, 1988.
Kouwen, N., and R. Li, "Biomechanics of Vegetative Channel Linings," Journal of the Hydraulics
Division, American Society of Civil Engineers, 106(HY6): 1085-1103, 1980.
Lane, E. W., and E. J. Carlson, "Some Factors Affecting the Stability of Canals Constructed in
Coarse Granular Materials," Proceedings of the Minnesota International Hydraulics Convention,
September 1953.
Laurenson, E. M., "Friction Slope Averaging in Backwater Calculations," Journal of Hydraulic
Engineering, American Society of Civil Engineers 112(12),1151-1163 1986.
Levi, E., The Science of Water: The Foundation of Modern Hydraulics, Translated from the Spanish
by D. E. Medina, ASCE Press, New York, 1995.
Linsley, R. K. and J. B. Franzini, Water Resources Engineering, 3rd ed., Mc-Graw-Hill, New York,
1979.
Meyer-Peter, P. E., and R. Muller, "Formulas for Bed Load Transport," Proceedings of the 3rd
International Association for Hydraulic Research, Stockholm, 1948, pp. 39-64.
Montuori, C., Discussion of "Stability Aspects of Flow in Open Channels," Journal of the
Hydraulics Division, American Society of Civil Engineers 89(HY4): 264-273, 1963.
Nagami, M., R. Scavarda, G. Pederson, G. Drogin, D. Chenoweth, C. Chow, and M. Villa, Design
Manual: Hydraulic, Design Division, Los Angeles County Flood Control District, Los Angeles,
CA, 1982.
Niepelt, W. A., and F. A. Locher, "Instability in High Velocity Slurry Flows, Mining Engineering,
Society for Mining, Metallurgy and Exploration, 1989, pp. 1204-1209.
Petryk, S., and G. Bosmajian, "Analysis of Flow Through Vegetation," Journal of the Hydraulics
Division, American Society of Civil Engineers, 101(HY7): 871-884, 1975.
Ponce, VM., Engineering Hydrology: Principles and Practices, Prentice-Hall, Englewood Cliffs,
NJ, 1989.
Richardson, E. V, D. B.Simons, and P. Y. Mien, Highways in the River Environment, U.S.
Department of Transportation, Federal Highway Administration, Washington, DC, 1987.
Silvester, R., "Theory and Experiment on the Hydraulic Jump," Proceedings of the 2nd
Australasian Conference on Hydraulics and Fluid Mechanics, 1965, pp. A25-A39.
Silvester, R., "Hydraulic Jump in All Shapes of Horizontal Channels," Journal of the Hydraulics
Division, American Society of Civil Engineers, 90(HYl): 23-55, 1964.
Simons, D. B., and E. V. Richardson, "Resistance to Flow in Alluvial Channels," Professional
Paper 422-J, U.S. Geological Survey, Washington, DC, 1966.
Simons, Li & Associates, SLA Engineering Analysis of Fluvial Systems, Fort Collins, CO, 1982.
Straub, W. O. "A Quick and Easy Way to Calculate Critical and Conjugate Depths in Circular Open
Channels," Civil Engineering, 70-71, December 1978.
Straub, W. O., Personal Communication, Civil Engineering Associate, Department of Water and
Power, City of Los Angeles, January 13, 1982.
Temple, D. M., "Changes in Vegetal Flow Resistance During Long-Duration Flows," Transactions
of the ASAE, 34: 1769-1774, 1991.
Urquhart, W. J. "Hydraulics," in Engineering Field Manual, U.S. Department of Agriculture, Soil
Conservation Service, Washington, DC, 1975.
U.S Army Corps Engineers, HEC-RAS River Analysis System, User's Manual, U.S. Army Corps of
Engineers Hydrologic Engineering Center, Davis, CA, 1997.
U.S Army Corps Engineers, "HEC-2, Water Surface Profiles, Useris Manual," U.S. Army Corps of
Engineers Hydrologic Engineering Center, Davis, CA, 1990.

U.S Army Corps Engineers, "Accuracy of Computed Water Surface Profiles," U.S. Army Corps of
Engineers Hydrologic Engineering Center, Davis, CA, 1986.
U.S Army Corps Engineers, "Hydraulic Design of Flood Control Channels," EM 1110-2-1601.
U.S. Army Corps of Engineers, Washington, DC, 1970.
Yang, C. T., Sediment Transport: Theory and Practice, McGraw-Hill, New York, 1996.
Zegzhda, A.P., Theroiia Podobija Metodika Rascheta Gidrotekhnickeskikh Modele (Theory of
Similarity and Methods of Design of Models for Hydraulic Engineering), Gosstroiizdat,
Leningrad, 1938.

APPENDIX 3.A
VALUES OF THE
ROUGHNESS COEFFICIENT n

Values of the Roughness Coefficient n*

Type of Channel

Minimum

A Closed Conduits flowing partly full

A-I Metal
a. Brass, smooth
0.009
b. Steel
1 . Lockbar and welded
0.010
2. Riveted and spiral
0.013
c. Cast iron
1. Coated
0.010
2. Uncoated
0.011
d. Wrought iron
!.Black
0.012
2. Galvanized
0.013
e. Corrugated metal
1. Subdrain
0.017
2. Storm drain
0.021
A-2 Non-metal
a. Lucite
0.008
b. Glass
0.009
c. Cement
1. Neat, surface
0.010
2. Mortar
0.011
d. Concrete
!.Culvert, straight
and free of debris
0.010
2. Culvert, with bends, connections,
and some debris
0.011
3. Finished
0.011
4. Sewer and manholes, inlet, etc,
straight
0.013
5. Unfinished, steel form
0.012
6. Unfinished, smooth wood form 0.01 2
7. Unfinished, rough wood
form
0.015
e. Wood
1. Stave
0.010
2. Laminated, treated
0.015

Normal

Maximum

0.010

0.013

0.012
0.016

0.014
0.017

0.013
0.014

0.014
0.016

0.014
0.016

0.015
0.017

0.019
0.024

0.030
0.030

0.009
0.010

0.010
0.013

0.011
0.013

0.013
0.015

0.011

0.013

0.013
0.012

0.014
0.014

0.015
0.013
0.014

0.017
0.014
0.016

0.017

0.020

0.012
0.017

0.014
0.020

Type of Channel
f. Clay
1 . Common drainage tile
2. Vitrified sewer
3. Vitrified sewer with
manholes, inlet, etc.
4. Vitrified subdrain with
open joint
g. Brickwork
1. Glazed
2. Lined with cement mortar
h. Sanitary sewers coated with
sewage slimes with bends and
connections
i. Paved invert, sewer, smooth
bottom
j. Rubble masonry, cemented
k.Polyethylene pipe
1. Polyvinyl chloride
B. Lined or Built-up Channels
B-I Metal
a. Smooth steel surface
1. Unpainted
2. Painted
b. Corrugated
B-2 Nonmetal
a. Cement
1. Neat, surface
2. Mortar
b. Wood
1 . Planed, untreated
2. Planed, creosoted
3. Unplaned
4. Plank with battens
5. Lined with roofing paper
c. Concrete
1. Trowel finish
2. Float finish
3. Finished, with gravel
on bottom
4. Unfinished
5. Gunite, good section
6. Gunite, wavy section
7. On good excavated rock
8. On irregular excavated rock

Minimum

Normal

Maximum

0.011
0.011

0.013
0.014

0.017
0.017

0.013

0.015

0.017

0.014

0.016

0.018

0.011
0.012

0.013
0.015

0.015
0.017

0.012

0.013

0.016

0.016
0.018
0.009
0.010

0.019
0.025

0.020
0.030

0.011
0.012
0.021

0.012
0.013
0.025

0.014
0.017
0.030

0.010
0.011

0.011
0.013

0.013
0.015

0.010
0.011
0.011
0.012
0.010

0.012
0.012
0.013
0.015
0.014

0.014
0.014
0.015
0.018
0.017

0.011
0.013

0.013
0.015

0.015
0.016

0.015
0.014
0.016
0.018
0.017
0.022

0.017
0.017
0.019
0.022
0.020
0.027

0.020
0.020
0.023
0.025

Type of Channel

Minimum

Normal

Maximum

0.015
0.017

0.017
0.020

0.020
0.024

0.016
0.020
0.020

0.020
0.025
0.030

0.024
0.030
0.035

0.017
0.020
0.023

0.020
0.023
0.033

0.025
0.026
0.036

0.011
0.012

0.013
0.015

0.015
0.018

0.017
0.023
0.013

0.025
0.032
0.015

0.030
0.035
0.017

0.013
0.016
0.030

0.013
0.016

0.016
0.018
0.022
0.022

0.018
0.022
0.025
0.027

0.020
0.025
0.030
0.033

0.023
0.025

0.025
0.030

0.030
0.033

0.030

0.035

0.040

0.028

0.030

0.035

0.025

0.035

0.040

0.030

0.040

0.050

0.025
0.035

0.028
0.050

0.033
0.060

d. Concrete bottom float with

sides of
1 . Dressed stone in mortar
2. Random stone in mortar
3. Cement, rubble masonry,
plastered
4. Cement rubble masonry
5. Dry rubble or riprap
e. Gravel bottom with sides of
1. Formed concrete
2. Random stone in mortar
3. Dry rubble or riprap
f. Brick
1. Glazed
2. In cement mortar
g. Masonry
1. Cemented rubble
2. Dry rubble
h. Dressed ashlar
i. Asphalt
1. Smooth
2. Rough
J. Vegetal cover
C-I Excavated or Dredged
C.I General
a. Earth, straight and uniform
1. Clean and recently completed
2. Clean, after weathering
3. Gravel, uniform section, clean
4. With short grass, few weeds
b. Earth, winding and sluggish
1 . No vegetation
2. Grass, some weeds
3. Dense weeds or aquatic plants
in deep channels
4. Earth bottom and rubble
sides
5. Stony bottom and weedy
banks
6. Cobble bottom and clean
sides
c. Dragline-excavated or dredged
1 . No vegetation
2. Light brush on banks

0.500

Type of Channel

Minimum

d. Rock cuts
1. Smooth and uniform
0.025
2. Jagged and irregular
0.035
e. Channels not maintained, weeds
and brush uncut
1. Dense weeds, high as
flow depth
0.050
2. Clean bottom, brush on sides 0.040
3. Same, highest stage of flow 0.045
4. Dense brush, high stage
0.080
C.2 Channels with maintained vegetation
and velocities of 2 and 6 ft/s
a. Depth of flow up to 0.7 ft
1. Bermuda grass, Kentucky bluegrass,
buffalo grass
Mowed to 2 in
0.07
Length 4 to 6 in
0.09
2. Good stand, any grass
Length approx. 12 in
0.18
Length approx. 24 in
0.30
3. Fair stand, any grass
Length approx. 12 in
0.014
Length approx. 24 in
0.25
b. Depth of flow up to 0.7-1.5 ft
1. Bermuda grass, Kentucky bluegrass,
buffalo grass
Mowed to 2 in
0.05
Length 4-6 in
0.06
2. Good stand, any grass
Length approx. 12 in
0.12
3. Length approx. 24 in
0.20
Fair stand, any grass
Length approx. 12 in
0.10
Length approx. 24 in
0.17
D Natural streams
D-I Minor streams (top width at
flood stage < 100 ft)
a. Streams on plain
1. Clean, straight, full stage
no rifts or deep pools
0.025
2. Same as above, but with more
stones and weeds
0.030
3. Clean, winding, some pools and
shoals
0.033
4. Same as above, but with some
weeds and stones
0.035

Normal

Maximum

0.035
0.040

0.040
0.050

0.080
0.050
0.070
0.100

0.120
0.080
0.110
0.14

0.045
0.05
0.09
0.15
0.08
0.13

0.035
0.04
0.07
0.10
0.16
0.09

0.030

0.033

0.035

0.040

0.040

0.045

0.045

0.050

Type of Channel

Minimum

Normal

5. Same as above, lower stages

more ineffective slopes and
sections
0.040
6. Same as no. 4, more stones
0.045
7. Sluggish reaches, weedy, deep
pools
0.050
8. Very weedy, reaches, deep pools
or floodways with heavy stand of
timber and underbrush
0.075
b. Mountain streams, no vegetation in
channel, banks usually steep, trees
and brush along banks submerged
at high stages
1. Bottom: gravels, cobbles and
few boulders
0.030
2. Bottom: cobbles with large
boulders
0.040
D-2 Floodplains
a. Pasture, no brush
!.Short grass
0.025
2. High grass
0.030
b. Cultivated areas
1. No crop
0.020
2. Mature row crops
0.025
3. Mature field crops
0.030
c. Brush
1. Scattered brush,
heavy weeds
0.035
2. Light brush and trees
in winter
0.035
3. Light brush and trees
in summer
0.040
4. Medium to dense brush
in winter
0.045
5. Medium to dense brush in
summer
0.070
d. Trees
1. Dense willows, summer,
straight
0.110
2. Cleared land with tree stumps,
no sprouts
0.030
3. Same as above but with a heavy
growth of sprouts
0.050
4. Heavy stand of timber, a few down
trees, little undergrowth, flood stage
below branches
0.080
5. Same as above, but with flood stage
reaching branches
0.100

Maximum

0.048
0.050

0.055
0.060

0.070

0.080

0.100

0.150

0.040

0.050

0.050

0.070

0.030
0.035

0.035
0.050

0.030
0.035
0.040

0.040
0.045
0.050

0.050

0.070

0.050

0.060

0.070

0.110

0.070

0.110

0.100

0.160

0.150

0.200

0.040

0.050

0.060

0.080

0.100

0.120

0.120

0.160

Type of Channel

Minimum

Normal

Maximum

D-3 Major streams (top width at flood stage

> 100 ft); the n value is less that for
minor streams of similar description
because banks offer less effective resistance
a. Regular section with no boulders
or brush
0.025
b. Irregular and rough section
0.035

0.060
0.100

0.020
0.030
0.040
0.025

0.015
0.020

0.17
0.17
0.20
0.20
0.10
0.30
0.05

0.80
0.48
0.40
0.30
0.20
0.40
0.13

0.09
0.05

0.34
0.25

0.008
0.06
0.06
0.30
0.04
0.07
0.17

0.012
0.22
0.16
0.50
0.10
0.17
0.47

0.10

0.20

0.10
0.08
0.04
0.02

0.15
0.12
0.10
0.05

D-4 Alluvial sandbed channels (no vegetation

and data is limited to sand channels with
D50< 1.0mm
a. Tranquil flow, Fr < 1
!.Plane bed
0.014
2. Ripples
0.018
3. Dunes
0.020
4. Washed out dunes or transition 0.014
b. Rapid flow, Fr > 1
1. Standing waves
0.010
2.Antidunes
0.012
E. Overland Flow (Sheetflow)
E-I Vegetated areas
a. Dense turf
b. Bermuda and dense grass
c. Average grass cover
d. Poor grass cover on rough surface
e. Short prairie grass
f. Shrubs and forest litter, pasture
g. Sparse vegetation
h. Sparse rangeland with debris
1.0% cover
2.20% cover
E-2 Plowed or tilled fields
a. Fallowno residue
b. Conventional tillage
c. Chisel plow
d. Fall disking
e. No tillno residue
f. No till (20-40% residue cover)
g. No till (100% residue cover)
E-3 Other surfaces
a. Open ground with debris
b. Shallow flow on asphalt or
concrete
c. Fallow fields
d. Open ground, no debris
f. Asphalt or concrete

Source: From Chow (1959), Richardson et al. (1987), Simons, Li, & Associates (SLA),
1982, and others. * The values in bold are recommended for design

CHAPTER 4
SUBSURFACE FLOW
AND TRANSPORT

Mariush W. Kemblowki and Gilberto E. Urroz

Utah Water Research Laboratory
Utah State University
Logan, Utah

4.7

INTRODUCTION

This chapter begins with the mathematical description of the constitutive relationships for
flow and transport in porous media. Following this, simple analytical solutions are presented for a variety of subsurface flow and transport problems. The principles of flow and
transport are outlined, and solutions are provided for practical problems of flow and transport in both the saturated and the unsaturated zones. The latter includes problems of
transport in the vapor phase. The major focus is on the processes that are relevant to
subsurface mitigation.

4.2

CONSTITUTIVERELATIONSHIPS

This section presents the basic concepts and laws used to describe flow and transport in the
subsurface. In particular, the constitutive relationships defining the fluid flow in fully and
partially saturated media are given as well as the relationships that describe diffusive and dispersive mass fluxes in porous media. Finally, we show the relations used to describe partitioning of chemicals in the subsurface environment.
4.2.1 Darcy's Law
Consider the flow of a fluid through a pipe filled with a granular material, as shown in
Fig. 4.1. In the figure, Z1 and Z2 represent the elevations of the pipe centerline above a reference level at Sections 1 and 2, respectively, whereas p/y and p2/j represent the water
pressure head at Sections 1 and 2, respectively. We define the piezometric head at any
location in the porous media as
h = z+p/J

(4.1)

where y = specific weight (weight per unit volume) of water, typically, y = 9810 N/m3 or
62.4 lb/ft3. Let q be the average water velocity in the cross section of the pipe: i.e.,

q = Q/A

(4.2)

Arbitrary datum
FIGURE 4.1 Porous media flow.
where Q = volumetric discharge (volume per unit time) and A = total cross-sectional
area of the pipe (including the soil matrix). French hydrologist Henry Darcy discovered
that the average flow velocity could be estimated from
q = KQi1 -H2)IL

(4.3)

where L is the distance, measured along the pipe, between cross sections 1 and 2, and K
is a parameter that depends on the nature of the porous media as well as on the properties
of the transported fluid. For water, K is known as the hydraulic conductivity or the coefficient of permeability. Typical values of K are given in many references (e.g., Bureau of
Reclamation, 1985) see tables 4.1 ard 4.2. Eq. (4.3) is known as Darcy1s law and is commonly used to model the flow of fluids in porous media. Notice that the velocity V is not
the fluid velocity in the soil pores, it is an average velocity calculated over the entire area
of the flow cross section. The average pore velocity is calculated as
v = -j(4.4)
vw
where Qw is the volumetric water moisture content. Note that for saturated flow, Qw = n,
where n is porosity.
Darcy's law (i.e., Eq. 4.3) also can be written more concisely as

q = KI

(4.5)

where / is the hydraulic gradient defined as

/ = ^)

4.6)

Hydraulic conductivity K is a function of aquifer and fluid properties -specifically of

the intrinsic soil permeability k, fluid viscosity ^, and fluid density p-and is given by
K = kpg/\L

(4.7)

For saturated flow of a constant density fluid in isotopic porous media, the Darcy law
can be written as

*=-*!)
For anisotropic media, the Darcy law is written as

()

(4 9)
* = -*(f|)
where A^ is the conductivity tensor. For saturated flow of a fluid of variable density in
anisotropic porous media, we have

*-fe(t+^)
where ksoilij is the intrinsic permeability tenser. Finally, for unsaturated flow of variabledensity fluid in anisotropic porous media, we have
+
aX
-HrHr
Pfluid \.OXj *-!)
j)
where kr(Q) = relative permeability of the porous media. Relative permeability is a function of soil saturation, which in turn is a function of the capillary pressure. These relationships for partially saturated flow are discussed in the next section.

4.2.2 Unsaturated Flow-Constitutive Relationship

In unsaturated flow, the concern is water movement in the zone above the water table. In
this case, the water saturation Sw is a function of the difference between air and water
pressures because the water is resulting from held by capillary forces resulting from surface tension. This difference is known as the capillary pressure and is defined as
haw = ha-hw

(4.12)

Typically, in unsaturated flow theory we assume negligible resistance to the gas-phase

flow in porous media; as a result, we also can assume that the gas-phase pressure is uniform and equal to the atmospheric pressure. Hence haw = hw. The aqueous pressure in
the unsaturated zone is lower than the atmospheric pressure; thus, the capillary pressure
is positive. The negative pressure head hw also is known as the soil matrix suction *F.
Thus, the total head in the aqueous phase is h = * + z.
To remove water from the pore space i.e., to reduce the water content we have to apply
more negative pressure to the aqueous phase, i.e., increase the capillary pressure haw. This
relationship is typically called the soil-water retention curve, and can be expressed by the
following commonly used parametric models: the Brooks-Corey (BC) model and the van
Genuchten (VG) model The BC model is
9- 0

(m>

"-^-Rj
for > hb and is otherwise (capillary fringe zone),
e. = i

(413)

(4.14)

where n = porosity, G = volumetric moisture content (equal to n SJ, Qe = effective volumetric moisture content, 9r = residual (irreducible) moisture content, X = Brooks-Corey
parameter, and hb = capillary fringe height. The van Genuchten model is

ee = e - -^ - er = [i + (<x)Tw

(4.15)

where: = curve fitting parameter that depends on the type of soil, a = soil property
index, a l/hb, and m = 1 1/rc.
Note that parameter n depends on pore-size distribution. For a well-graded soil (wide
pore-size distribution, which results in a flatter moisture content curve 6(1F)), n is small,
whereas for poorly graded soils (narrow pore-size distribution, which results in a steeper
moisture content curve G(1F), n is large: typically values of n higher than 2.5.
Also note that although the BC and VG models are the most commonly used models
in analysis, there is no restriction on using different mathematical representations to
describe the characteristics of soil-water retention. For example, a simple exponential
model, such as Qe = exp( P *F), is in some cases, sufficient to describe the physics of the
retention.
As the moisture content in partially saturated media decreases, so does the volume of
pores available to fluid flow. Thus, hydraulic conductivity for the partially saturated
media depends on the water content and, in turn, on the metric suction. To describe this
relationship, we modify the value of intrinsic permeability k by the factor of kr(Q) or krC)t
called relative permeability. Several models for kr are shown below:
BC model of relative permeability:

2-3X
fcr = e e

(4.16)

kr = 6e5(l - (1 - 9/T)2

(4.17)

VG model of relative permeability:

Mualem model of relative permeability:

kr = exp[-a]

(4.18)

4.2.3 Diffussive and Dispersive fluxes

4.2.3.1 Molecular diffusion. Molecular diffusion describes the process by which a contaminant species dissolved in an environmental fluid moves from regions of higher
concentration to regions of lower concentration. When the only mechanism affecting the
diffusion of the contaminant species is the random motion of its molecules, the process is
referred to as molecular diffusion. The mass flux of a solute along a single direction, in a
liquid or gaseous body, is described by Pick's law:
= D (f)

(4 19

- >

In this equation, q = mass flux of solute per unit area per unit time [ML" T"1],
D diffusion coefficient (L2 T -1), C = solute concentration = mass of solute/volume of
solution (M/L3), dC/dx = concentration gradient along the x direction. The minus sign in
Eq. (4.19) indicates that the solute flux will go from regions of larger concentration to
those of lower concentration. Values of the diffusion coefficient, Z), depend on the type of
solute and the type of environmental fluid. For major cations and amnions dissolved in
water, values of D range from 1 X 10~9 to 2 X 10 m2/s (Fetter, 1994).

4.2.3.2 Molecular diffusion in porous media Molecular diffusion in a porous medium

is affected by the nature of the medium. For example, the paths that diffusing molecules
follow in a porous medium are, in general, more complicated than if they were diffusing
in water. Although Eq. (4.19) can still be used to describe the diffusion in a porous medium, the diffusion coefficient must be modified, and Eq. (4.19) is rewritten in terms of an
effective diffusion coefficient. One widely accepted expression for the effective diffusion
coefficient in porous media is the Millington-Quirk equation
93.33
D = D0 sol*ion
(4.20)
where D is the molecular diffusion coefficient of the compound in pure solution fluid,
^solution is me solution fluid-filled porosity of the soil, and n is the total porosity of the soil.
4.2.3.3 Mechanical dispersion and macro-dispersion. Mechanical dispersion refers to
the component of dispersion caused by differences in velocity at the pore level that are a
consequence of the pore geometry. Water will move at different rates as a result of differences in pore sizes and tortuosity. A contaminant dissolved in the water flowing through
a porous medium will be dispersed in both the longitudinal and transverse directions
because of the fluctuations in the water velocity field. A way to incorporate the influence
of pore geometry in the dispersion process is to define longitudinal and transverse dispersivities OCL and ocr. Longitudinal and transverse mechanical dispersion coefficients can
thus be defined in terms of the dispersivities and the average pore velocity. For example,
the longitudinal mechanical dispersion coefficient (DL)mech will be given by (DL)mech OCL
v. This dispersion coefficient is not treated separately from the effective diffusion coefficient defined in (8); instead, they are both combined in a coefficient of hydrodynamic dispersion. Thus, Pick's Law, which describes molecular diffusion in a fluid, can be used to
describe longitudinal and transverse dispersion in a porous medium if the diffusion coefficient D in Eq. (4.19) is replaced by a coefficient of longitudinal (or transverse) hydrodynamic dispersion, D1 or Z)r, given by
DL = CCL v + D*

(4.21)

DT = aT v + D*

(4.22)

or

where OCL, ocr = longitudinal and transverse dispersivities, respectively, and D* = effective porous-media diffusion coefficient.
In addition to the pore-scale dispersion, we also have formation-scale dispersion or,
more accurately, spreading, which is a result of the variability in transport velocity caused
by the heterogeneity of the hydraulic conductivity field. In terms of magnitude, this
microdispersive flux is significantly larger than the one related to mechanical dispersion.
Mathematically, macrodispersive flux is the flux equal to the expected value of the product of Darcian velocity (qf) and of the contaminant concentration (C') fluctuations:
* macrodispersive = < *' C' >

(4.23)

In otherwords, this flux can be described using an expression similar to the Fickian diffusion equation (Eq. 4.19) with a macrodispersivity coefficient. In the most general (threedimensional) case, the equation defining the macrodispersive flux in the flowing fluid is

rlC1
^ = (Ap)SgOXj

(4.24)

where Afj represents the macrodispersivity tensor and the bar indicates averaged quantities. The fluctuations q' and c' result from the heterogeneous nature of the aquifer. This is
expressed in the way the macrodispersivity tensor is estimated. For example, the longitudinal macrodispersivity is estimated using
A11 = ^

(4.25)

where <jf2 = variance of log-conductivity (f= Ln[K]), K1 = correlation scale in the direction of flow, and y is given by
= cx
v = -rT
P\\
KgJ1
Lo J

(4 26)

In summary, the longitudinal and transverse components of the total diffusive and dispersive mass flux (per unit bulk area) in heterogeneous geologic formations is estimated
as follows:
VL= -^FLuiDALvFLUID + D)^

(4.27)

9TJiOR= -QnjunAwVFiMD + D)^

(4-28)

T^C
<7r,vert = -(^FumAj^Fum + W f^

(4-29)

where D = Millington-Quirk effective dispersion coefficient.

4.2.4

Partitioning

Equilibrium partitioning and sorption are the most common chemical processes that
affect reactive transport. These processes are dealt with by equating the total concentration to the sum of the concentrations in each phase multiplied by their respective volumes.
Furthermore, by equating the concentration in each phase to the concentration in a common phase say, the concentration in water the total concentration can be expressed in
terms of the common phases concentration and a retardation coefficient R
CT = ^WATER^XTER(^ ~*~

T^- /\
2*i "gWATER '= ^WATER^-WATERR
/ * WATER

(4.30)

where: K1 = partitioning coefficient between the 1th phase and the common phase K1= C1/
CWATER, ^i = tne volumetric content of the ith phase, and QWATER = the volumetric content
of the common phase. This type of equilibrium partitioning is frequently used to describe
the relationships between concentrations in the following scenarios: (1) vapor and aqueous phases, (2) soil and vapor phases, (3) soil and aqueous phases, (4) partitioning of a
tracer between aqueous and NAPL or DNAPL phases, and (5) partitioning of a tracer
between vapor and NAPL or DNAPL phases.

For partitioning of compounds present in a NAPL or DNAPL mixture and aqueous

phase, we use
CIWATER ~ xi$i

(4.31)

where S1 is the solubility of compound 7 in water and X1 is its mole fraction in the mixture.
Finally, for partitioning of compounds present in a NAPL or DNAPL mixture and vapor
phase, use
"w - '^Tf=raSSWir

<4'32'

where V1 is the vapor concentration of compound /, Pvl is the vapor pressure of compound
/, MWI is its molecular weight, R = gas constant, and T = temperature in 0K.
4.2.5

Degradation

In addition to partitioning, degradation of compounds also may affect the fate and
transport of reactive compounds significantly. Typically, degradation is modeled using
either power-order decay models or growth-process-based models. In environmental
subsurface hydrology, three basic power-order models are used: (1) zero-order decay,
(2) first-order decay, and (3) a combination of the first two models. According to these
models, the total decrease of mass in unit bulk volume caused by degradation is
expressed by

(^C1
-^ = - E e/*<y - S eA/c/
<4-33)
/ = PHASES
I = PHASES
where K01 = zero-order degradation rate of the compound in phase 7, K11 = first-order
degradation rate of the compound in phase 7, and C1 = mass/volume concentration of the
compound in phase 7.
In addition to the zero-order and first-order degradation processes, the Monod kinetics
is frequently used to describe oxygen limited aerobic degradation of organic compounds
in the aqueous phase. According to the Monod model, the degradation rate in terms of the
total concentration is expressed by the following system of equations:
C
dCr = _ Q
WATER } ( WATER ]
M1
3f
WATER^N Kr +. Cr
K J- n

( c + WATER)(KO + OWATER)

^
^ ' '

(4

and
CWATER
D
Jvl if

YM
}hr
( 4WATER
}
^t = - 64TK>C
tH r , rC
n
dt
(Kc + WATER )(Ko + OWATER)

(4 35}
^HOj;

where CWATER is the aqueous concentration of the contaminant, OWATER is the dissolved oxygen concentration, M1 is the total concentration of the active microbial biomass, i is the
maximum rate of organic solute utilization, Kc is the concentration of the organic solute
at which the utilization rate is half the maximum, K0 is the electron acceptor (oxygen)
concentration at which the utilization rate is half the maximum, and % is the substrate utilization ratio.

4.3

FLOW AND TRANSPORT IN SATURATED ZONES

In this section, we present some solutions to saturated flow and transport problems that
are encountered in the practice of subsurface hydrology. We begin with well hydraulics,
an understanding of which is important to the design of pump-and-treat systems, and discuss several models of transport of soluble plumes.
4.3.1 Flow to a Single Well
Darcy's law describes the average water flux through a porous medium when the local
hydraulic gradient is known. To determine discharges we use the law of conservation of
mass for the water, also known as the continuity equation. For the analysis of wells, it is
assumed that the flow toward the well caused by the well pumping is radially symmetric.
In this situation, it is convenient to use the equation of continuity in radial coordinates
(r, 6). For steady-state state flow toward a single well in a confined aquifer without
recharge, the equation reads
Q = 271/ ^)

(4.36)

where K and B = aquifer conductivity and thickness, respectively. The equation of continuity for well flow in an unconfined aquifer is similar to Eq. (4.19): namely,
Q = 2nrKh(^\

(4.37)

with B replaced by the variable flow depth h. Implicit in Eq. (4.36) and (4.37) is the
Dupuit-Forchheimer assumption, the implication of which is that the flow in the aquifer
can be assumed to be practically horizontal.
Equations (4.36) and (4.37) are used to obtain solutions for the steady-state discharge
to a well in confined and unconfined aquifers, respectively, if the piezometric heads (confined aquifers) or water-table elevations (unconfined aquifers) H1 and h2 are known at two
radial distances T1 and r2, respectively:
Q = 2nKB(h2 - H1)IIn(T2Ir1) = 2nT(h2 - H1)IIn(T2Ir1)

(4.38)

where T = KB = aquifer transmissivity, and

Q = nKD(h22 - Hl)IIn(T2Ir1)

(4.39)

The solutions given in Eqs. (4.38) and (4.39) assume that the well penetrates to the
impermeable bottom of the aquifer and that there is no recharge into the aquifer. They also
assume an infinitely large aquifer with no interaction with surface streams or impermeable boundaries. Well solutions are often given in terms of the drawdown s as a function
of the radial distance r. The drawdown is defined as
s = H- h

(4.40)

where H is the elevation of the original piezometric surface before pumping at the well
starts. The distance R for which h = H and s = O is called the radius of influence of the
well. Using the concepts of drawdown and radius of influence, the steady-state flow equation in a confined aquifer can be rewritten as
,-&L

(4.4,,

By combining the solutions for the steady state flow in confined and unconfined
aquifers, one can derive a relationship between the drawdown calculated from confined
conditions (assuming constant in space and time aquifer thickness equal to //), and that
estimated for unconfined conditions (when the change in aquifer thickness caused by
pumping is taken into account):
SUNC = H- V^ - 2sCONFH

(4.42)

Using this formula, known as Jacob's correction, one can initially assume constant
aquifer thickness B = Hm calculations and use "confined" aquifer solutions to calculate
drawdown, then correct the drawdown using Jacob's correction. This approach is particularly useful when dealing with transient flow. For transient flow conditions in an aquifer
with constant flow thickness, the transient drawdown is given by
<">-&l:s!r**-&'"
where
M(r

' = Wt

(4 44)

'

and S = storativity (for confined aquifers) or porosity (for unconfined aquifers) and W(u)
is known in subsurface hydrology as well function and in mathematics as exponential
integral. This function is tabulated in almost every groundwater hydrology textbook. It
also is available in many engineering mathematics software packages, such as
Mathematica, as a library function.
4.3.2 Superposition and Convolution
For a time-variable pumping rate, the principle of convolution can be used to estimate the
transient drawdown. This approach is strictly valid for linear systems: i.e., systems in
which the response (drawdown) is a linear function of the excitement (pumping rate). The
linearity assumption is strictly valid for confined aquifers only; however, as long as the
drawdowns do not exceed 20% of the initial aquifer thickness, it also may be used for
unconfined aquifers. Using the convolution approach, the transient drawdown for a pumping rate changing in a step-wise fashion is given by
n
S
=
^ -^T
^ (Q* + Q* - 1)^^ - 1
<4'45)
*nik = 1
where the drawdown is estimated at time f, tn<t<tn + l, QK is the pumping rate for
t
K - i < *< * K,to = 0> Qo = 0-0, and &tK _ l = t - tK _ r When several wells are present,
the superposition approach is used to estimate the cumulative drawdown by adding the
drawdown contributions from all the wells:
m
*W = 4Sr 2 QLW(u(rL,t))
(4.46)
where rL is the distance between the point of interest (where the drawdown is estimated) and
well L. When several wells are pumping at variable rates, the superposition and convolution
approaches are used simultaneously. The superposition principle also can be used to superimpose the drawdown on the natural (ambient) flow conditions. Using this principle leads to

h(x9 y, t) = H(x, v, t) - s(x, v, t)

(4.47)

where h = transient potentiometric surface that combines ambient conditions and well
impact, H = potentiometric surface under natural (ambient) conditions, and s = transient
drawdown.
4.3.3 Interception Wells
With respect to contaminant transport in the subsurface, interception wells are used to trap
the contaminant plume within the well flow field. It is assumed in this analysis that there
is an ambient steady-state uniform flow through the aquifer. The combination of wellrelated flow and ambient uniform flow satisfies the conditions of two-dimensional potential flow in a horizontal plane, where the discharge described by stream function y is related to potential <|), or the piezometric head h. The extent of the aquifer through which water
travels to the well and is captured by it is called the capture zone. The derivation of the
analytical solution for steady-state flow capture-zone uses the following assumptions:
(1) a homogeneous, isotropic, infinitely large aquifer, (2) uniform flow, (3) no leakage, (4)
aquifer storativity or specific yield neglected, (not relevant for steady-state analysis),
(5) hydrodynamic dispersion neglected, (6) the Dupuit assumption applies, and (7) the
well is fully penetrating and pumping at a constant rate. Three important parameters are
used in delineating the capture zone: namely, the stagnation point, the upgradient maximum width of the capture zone, and the equation for the capture zone boundary.
For a confined aquifer, the distance from the well to the stagnation point (measured in
the direction of the uniform flow) is
^0 = Jf7

(4-48)

where Qw = well discharge, T = aquifer transmissivity = KB (K= aquifer permeability,

B = aquifer depth), and I = natural hydraulic gradient: i.e., the gradient responsible for
the ambient steady-state uniform flow in the aquifer. The upgradient divide, defined by
the maximum width of the capture zone far upgradient of the well, for the confined aquifer
is given by
"W = %

(4-49)

and the equation of the dividing streamline is

jc =

(4.50)
/
27tf7y\
tan [n-
\
&I
The procedure for delineating the capture zone consists of the following steps: (1) estimate the location of the stagnation point (XSTAG, O), (2) estimate the maximum width of the
capture zone wD/v, and (3) vary y between zero and wD/v/2 and use the capture zone boundary to estimate the boundary location (;c,v).
4.3.4 Partially Penetrating Wells
Performance of wells that penetrate only partially through the bearing strata is discussed
in this section. The simplest case consists of a well that is barely penetrating into an semi-

infinite porous medium so that the aquifer flow is three-dimensional and spherically symmetric. In this case, the following relationship applies between the flow into the partially
penetrating well Qp and the flow to a fully penetrating one Q:
t-K
where B is the aquifer thickness, rw is the well radius, and R is the radius of influence of
the partially penetrating well. Because, in general, rw B, then the equation above indicates that the spherical flow to a partially penetrating well is highly inefficient compared
with simple radial flow: i.e., for the same drawdown in the well, it results in a significantly
smaller pumping rate.
In the general case of partial penetration, one may consider the total drawdown sr,
which consists of the drawdown equivalent to that of a fully penetrating well s and additional head loss because of the partial penetration of the well As:

ST = s + As

(4.51)

Additional head loss for a well penetrating from the top (or the bottom) of the aquifer
is estimated as follows:
Aj =

CO^)J 0^1
2nTp
I
rw \

(4.53)

where p = penetration factor; p = hJB; hs = penetration depth; and B = aquifer

thickness (Fig. 4.2). For the well centrally positioned in the aquifer, the following formula is used:
^, = CO^
2nTp JO^b
l 2rw 1I]

(4,4)

Thus, when the pumping rate is defined for a well, we calculate the drawdown correction As and add it to the full penetration drawdown s. However, when the drawdown is
given for a well, we have to recalculate the pumping rate. In this case, the true pumping

FIGURE 4.2 Partially-penetrating well.

rate is given by
c =2

'

TTAJ = 2H^

(455)

where s = drawdown defined at the well, Q = pumping rate estimated using s, and
Qp = actual pumping rate
4.3.5 WellDuplets
Well duplets, each of which consists of one pumping and one recharge well, are frequently used as a means of injecting and removing aquifer mitigation solutes, such as co-solvents, surfactants, or both. Typically, the recharge well is positioned directly upgradient
from the discharge well, and the magnitudes of pumping and injection rates are the same.
In this case, the two wells form a flow circulation cell: i.e., all the injected water is
pumped out by the discharge well. In the case of co-solvent flushing, it is important to
understand what region of the aquifer is subject to the mitigation: i.e., what the boundary
is of the circulation cell. This boundary is defined by the upgradient and downgradient
stagnation points (X STAG , O) and (XSTAG, O) and the cell boundary equation. For the x-axis
parallel to the direction of ambient flow and the origin of the coordinate system located at
the mid-point between the two wells, the two stagnation points, (~XSTAG, O) and (XSTAG, O)
are given as the roots of the quadratic equation
Q0Bd
~27

i ( x/d
_ x/d 2 ] _ Q
^((x/d + l ) 2 (x/d-1)) u

f456 ,
1

where q0 = ambient flux, 2d = distance between the wells, and Qw = pumping/injection

rate. The boundary of the circulation cell is defined by
+ 4H3rfSu]-~-'[<s#hj])-i

<4-57'

The circulation cell is symmetric with respect to the y axis. The cell delineation procedure consists of estimating the locations of the stagnation points and varying x between
zero and XSTAG and using the cell-boundary equation to solve for y. This implicit equation
can be solved by any calculation software, such as Mathematical or MS Excel.
4.3.6 Transport Equations
The following general form of mass transport equation in the saturated zone is derived assuming one-dimensional advective and three-dimensional diffusive-dispersive transport in the
aqueous phase, linear partitioning of a compound between the three phases (water-soil(D)NAPL), and first-order degradation in the aqueous phase. For this conditions, we have
<^H((<""+)t)-t-^
where Cw = aqueous phase concentration, v = pore-water velocity, and 9W = volumetric
moisture content, and oc. = longitudinal, transverse horizontal, and transverse vertical

macrodispersivities, and D = Millington-Quirk dispersion coefficient, and X = first-order

degradation rate in the aqueous phase, and R = retardation factor,
R = I + SOILY$ _|_ ^NAPL^NAPL
w
^w

/^ CQ\

where kSOIL = water-soil partitioning coefficient, kSOIL = S/CW, S = weight/weight concentration of absorbed compound in soil, kNAPL = NAPL - water partitioning coefficient,
^NAPL = CNAPLW> CNAPL = concentration of compound in the NAPL phase, JB = bulk density of dry soil, and QNAPL = volumetric WAPL content.
Assuming that the diffusive fluxes are negligible compared the macrodispersive fluxes, the transport equation can be simplified to yield
3CW
3 (
3CW "1 V dCw
C
-^3r = ^-WVR^3Jt1 ^ l R 3Jt1 Jr *R-^djtj ~ ^RR w

(4-6)

where VR = v/R and KR = UR.

4.3.7 Selected Analytical Solutions
Closed-form solutions are available for a variety of flow, boundary, and initial conditions.
Van Genuchten and Alves (1982) presented a good summary of these solutions. Some of
the most useful solutions are presented below.
4.3.7.1 One-dimensional transport with step change in concentration-no degradation. This simple case has the initial condition C(x,Q) = O for x ^ O, and it is subject to
the following boundary conditions: C(0,f) = C0, t > O and C(,0 = O, t > O. The solution
of the transport equation for these conditions is given by

^-Sfc&MsM^])

<"

4.3.7.2 One-dimensional transport with step change in concentration and first-order

degradation. The initial and boundary conditions are the same as in Sec. 4.7.1.
The solution is given by
<*>-M^H+^n]
/
4X w a r \i/2 ~
* - v J i + -M
t
V
\
R
Erfc
w
[
2(W)1/2
J

(4.62)

where Erfc = complementary error function.

4.3.7.3 Continuous point injection, 2-D dispersive transport, no retardation, and no
degradation. A tracer is continuously injected at a rate Q (per unit depth of the aquifer)

with a concentration C0 into a uniform flow field from a point (x = O, y = O). Let the
uniform velocity be v^. The asymptotic solution, i.e., for t <*>, is given by
CC*y)
(51 -P
P P^kJ ^ [- + ^]I
^ = UrtVS^J
(-2D1Pr) ]_^l4DL(DL D 7 JJ

(4-63)

where K0 = modified Bessel function of the second kind and of Oth order (Bear, 1972).
The time-dependent solution is
C(X

L Wn n V*/feWmP)] - W,P)
U^VDLDrJ 1,2DJ

(4.64)

where
/ v 2 / r2
V2 \
>- VAt& + )

'4^

W (f, ) = leaky well function (see, for example, Hunt, 1983, p. 100).
4.3.7.4 Point slug injection into a uniform flow field3-D transport and retardation.
In this case, a slug of contaminant of the mass M = C0V is injected at point (0,0,0). The
transient distribution of concentration is described by
C(x y Z t}=

' ' '

eJ_(*-V)
L
4axvRt

VC
%(nvRt)m(axa^m
2

_^i___jM
4ayv/?r 4a2v/?rJ

(4.66)

4.3.7.5 Continuous injection from a finite-sized source with retardation and degradation. In this case, consider transport from a rectangular source that is perpendicular
to the direction of flow. The source width is F, and its depth below the water table is Z.
The transient concentration distribution in the presence of retardation and degradation is
given by
. C0 \{ x V 1 fi + 4V*H1
cfc^o^m-ji-HH}]
172
j-, / fjc-v^l
A ^ +4^)
Kv' I
l
2(CW)\

ErfC

\y + Y/2]

F_J

y - YI2\

M^^rM^HI
{E*[^}-E*[^}

(4 67)

'

4.4 FLOW AND TRANSPORT IN UNSATURATED

ZONE AQUEOUS PHASE
In this section, we briefly discuss the flow continuity equations and present some simple
solutions to selected flow problems. That discussion is followed by a presentation of mass
transport in the water phase of unsaturated zone.
4.4.1 Flow in an Unsaturated Zone
The continuity equation for an unsaturated flow system can be written as
U = -V.q

(4.68)

= -gi + & + ^i]

3r
(dx
dy
dz)

(4.69)

or

Combining Darcy's law with the mass continuity equation, we can write the final flow
equations is
^ = V(KmVh)
at

(4.70)

The flow equations can be simplified for horizontal and vertical flow conditions.
a. One-dimensional horizontal flow:
w = a {K?rg\
Bt dx(
dx)

(4.7i)

where 0 = volumetric water content. In this, the contribution of the elevation head, z,
vanishes, since 3z/3jc = O.
b. One-dimensional vertical flow:
W = JL(Ww + 1I
ot oz {
(oz
Jj

(472)

Note that the flow equations are characterized by the presence of two dependent, albeit
related, variables: namely, 0 and 1F. To simplify this situation, we describe the relationship between 0 and 1F by a term called soil diffusivity D(0) as
=m
where C(0) is called specific moisture capacity and is defined as
C(0) = U

>
(4.74)

Using these definitions, the flow equation can be written as follows:

c. One-dimensional horizontal flow:
^ = |W^]
dt dx[
dx)

(4.75)

d. One-dimensional 1 -D vertical flow:

r)ft
_.
~\Q
^ = IOW-5- + ^))

(4-76)

As can be seen, we now have only one dependent variable: namely, 6. The only limitation of
this formulation is that specific moisture capacity C(Q) becomes zero in the capillary fringe
zone, thus making the solution impossible. Therefore, this formulation is valid only in the partially saturated zone (water content less than saturated value), not in the capillary fringe.
Another way to solve this problem, which does not have the limitation discussed
above, is to formulate the flow equations in terms of soil suction \(/. For the vertical flow,
we obtain
C(e)

? = i (^(XF)(lr+ l)j

(4 74)

Exercise. Consider steady-state vertical infiltration from the soil surface to the water table
at depth L. The relative hydraulic conductivity of the soil is described by the following
exponential law:
kr(V) = exp[-aV]

(4.78)

Derive an expression for the vertical distribution of h. After the first integration of the
flow equation, we obtain
~\\TJ
\
+1
a- J =
(4-79)

where q = infiltration rate and A" is a function of capillary pressure y.

Kf) = Kjxp\-d\

(4.80)

where h = z x|/. Substitution of these relationships into the flow equation leads to

or

ext\a(h - z)l^ = -?L

-
KSAT

(4.81)

exp[ah]dh = -/-exp[ax]dx
^-SAr
After the second integration, we obtain

(4.82)

exp(ah] = -J- exp[az] + C1

(4.83)
oc
ASAr
Substituting the boundary condition /z(0) = O and solving for C1 yields the following
expression for the total head h:
h(z) = -ln \J-(exp[az] - 1) + ll
a L^5Ar
J

(4.84)

4.4.2 Transport in an Unsaturated Zone

The mass continuity equation for an unsaturated flow system with advection and diffusion/dispersion in the aquoeus phase, diffusion in the vapor phase, partitioning between
four phases (soil, water, vapor, and (D)NAPL), and first-order degradation in the aqueous
phase can be written as

wW
4=
T?-^fk
+ ^N^l
^ - XC
d/ =3x,
' +6
Qw ) to,J- v &,
W

(4.85)

where Cw = aqueous phase concentration; v = pore-water velocity; 6W = volumetric

moisture content; a, = longitudinal, transverse horizontal, and transverse vertical
macrodispersivities; D = Millington-Quirk dispersion coefficient; Dv = Millington-Quirk
dispersion coefficient in the vapor phase; K = first-order degradation rate in the aquoeus
phase; and R = retardation factor;
R = 1 + ^SOIL^B _^_ ^NAPL^NAPL _j_ ^AIR^AIR
vw
vw
Uw

/^ g^x

where kSOIL = water-soil partitioning coefficient, kSOIL = S/CW, S = weight/weight concentration of absorbed compound in soil, kNAPL = NAPL-water partitioning coefficient,
^VAPL = CNAPI! Cw> CNAPL = concentration of compound in the WAPL phase, kAIR = vaporwater partitioning coefficient, kAIR = V/CW, V = concentration of compound in the vapor
phase, J8 = bulk density of dry soil, QAIR = volumetric vapor content, and QNAPL volumetric NAPL content.
Assuming again that the diffusive fluxes are negligible compared with the macrodispersive and advective fluxes, the transport equation can be simplified to yield
T'l-hfh-f-v.

<">

where VR vIR and KR = K/R. Note that the form of this equation is the same as the form
of the one for transport in the saturated zone; therefore, the analytical solutions presented
for the saturated transport are valid for the unsaturated conditions. This is particularly true
for the one-dimensional (vertical) transport equations, which are of primary interest in the
case of unsaturated fate and transport of compounds.

4.5

FLOW AND TRANSPORT IN VAPOR PHASE

This section presents the soil vapor flow equations. This is followed by selected solutions
to vapor flow problems. Finally, we discuss diffusive transport in the vapor phase.
4.5.1 Soil Vapor Flow
The flow of gases in a porous medium can be described by combining the following
equations:
a. The equation of continuity is
V(pv) = -^

(4.88)

with p = vapor density, v = Darcy's flux vector, 9AIR = air-filled porosity, and t = time.
b. The perfect gas law is
P- = -

(4-89)

with m = gas mass, V = gas volume, p = absolute pressure, M = molar mass, T =

absolute temperature 0K, and R = universal gas constant.
c. Darcy's law for vapor flow is
v=

-Av/?

(4.90)

where k = soil intrinsic permeability and JJL = gas viscosity. The resulting governing
equation is 1953
1V.-2S*
(Bruce et al., 1953).
d. The molar flux [moles/unit area-time] is given by
= ^PVp

(4.92)

For one-dimensional flow, the governing equation reduces to

a^ = 29^3p
ftc*
k
^t

'

and for radial flow, the governing equation is

?
+ l z 2y.fr
dr2
r 3r
k
3f
For steady state, exact analytical solutions for gas flow are obtained by using the following transformation (Cho, 1991):
M = (/
(4.95)
where m is referred to as the discharge potential. The governing equation now becomes
Laplace's equation:
V2m = O
(4.96)
For example, the exact solution for the point source in three-dimensional space is given by
m= --J=(4.97)
4nr
where Q = source strength and r = distance from source point.
To estimate the time required to achieve steady-state vapor flow, Johnson et al. (1990)
presented a method based on the solution of radial flow of vapor to a well. Their results
are summarized in Fig. 2 of Johnson et al. (1990 b) for values of k corresponding to sandy
soil. They also presented a method to estimate vapor flow rates, pressure distributions, and
vapor velocities in unsaturated soils based on the steady-state solution to the governing
equation of vapor radial flow: namely,

W = ^l
The pressure p can be expressed in terms of the ambient pressure pAtm and a deviation
/?' from this pressure: p' is equivalent to the vacuum that would be measured in the soil.
If this substitution is used in the flow equation and if we neglect the product p'2 relative
to the product pAtm /?' (linearization), then the resulting equation for radial flow is

k
F^
PATM %- -3-}
r dr( Br )

(4

-99)

The solution to this equation for the following boundary conditions is

p' = O, for r > oo

"^)-SH5

'"*

as given by
P

' = 4K&W

eXp[

-X]dx=4^)WM

(4 101)

'

where Q is the volumetric flow rate to the vapor well. The well function W(u) is tabulated in almost all groundwater textbooks. The behavior of the integral is such that for
(^A/aM^fcp^t) < 0.001, its value is close to the asymptotic steady-state limit.
Exercise. Given the following parametershydraulic conductivity K= 10~2 cm/sec, air
viscosity |HA//? = 0.018 cp, volumetric air content (equal to porosity) 6 AIR = 0.3, and local
pressure gradient dp/dx = 0.01 atm/cmestimate the pore-vapor velocity. From the conversion table (Domenico and Schwartz, 1990), we have
k[darcy] = K\mls\* 1.04*105

(4.102)

Thus, k = 10.4 darcy. The vapor flux is given by

J^gP+
v#* w

<y
&J

Assuming that kr = 1.0, we obtain

Y^^
I?1 - "8[cmAc]
q[
*cm,sec} - ^[centipoise]
Bx N
[ cm J - "0.018

(4.104)

Finally, we estimate the pore-vapor velocity v = ql$Am 19.26 cm/s.

4.5.2 Transport in Vapor Phase
It is usual to assume that whenever advective vapor flow is present, it dominates the transport process and the diffusive/dispersive processes can be neglected. In this case, the fate
and transport equation for a compound that partitions between the four phases (soil, water,
vapor, and (DNAPL) and is subject to first-order degradation in the aqueous phase is
*? = ~v^-^k*-v

(4105)

where: V = vapor phase concentration, VAIR = pore-air velocity, 0W = volumetric moisture content, QAIR = volumetric vapor content, X = first-order degradation rate in the
aqueous phase, kA _ w = air-water partitioning coefficient, kA_w = C^V, and R = retardation factor,

/? = 1 +

~ ^SOIL^B -L A - W^NAPL^NAPL , ^A - W^W

VAIR
^AIR
^AIR

,A ]f\\

where: kSOIL = water-soil partitioning coefficient, kSOIL = SlCw, S = weight/weight concentration of absorbed compound in soil, kNAPL = WAPL-water partitioning coefficient,
^NAPL = CNAPL/CW CNAPL = concentration of compound in the TVAPL phase, kAIR = vaporwater partitioning coefficient, kAIR = Vl Cw, V = concentration of compound in the vapor
phase, Y6 = bulk density of dry soil, QAIR = volumetric vapor content, and QNAPL = volumetric NAPL content.
Division of the transport equation by the retardation factor yields
?--"-**-v

(4 107)

where vAirR = vAIRIR and AT? = kA _ ^^w/C^A//? ^)- Note that the form of this equation is the
same as the form of the one for transport in the saturated zone, except for the absence of
the dispersive term. Therefore, the analytical solutions presented for the saturated transport are valid for the vapor transport conditions.
When there is no advective transport in the vapor phase, the transport equation must
include diffusive fluxes in vapor and aqueous phases to yield
K^V= 3f D vc|V
& dx,( Dx,

avl.x^.*
dx/)
Qm

wv

(4.108)

where D = Millington-Quirk dispersion coefficient for the aqueous phase, and Dv =

Millington-Quirk dispersion coefficient in the vapor phase. Assuming that the dispersion
coefficients do not vary in space leads to the following form of the transport equation:
= /-*V
dt
dx]

(4.109)

D = " + ?*-*
K

(4.110)

where

and

*-&

Again, we note the similarity of this fate and transport equation to the one presented
for saturated transport and conclude that all the analytical solutions presented in Sec. 4.2
can, in principle, be used to analyze vapor phase transport.
Exercise. Given that water saturation Sw = 0.20, porosity n = 0.4, compound concentration in soil vapor at depth L = 2 m C0 = 100 mg/L, compound concentration at the soil
surface Cs = 0.01, and molecular diffusion coefficient of the compound D0 =
0.087cm2/sec, estimate the compound mass flux in the vapor phase at the soil surface. The
effective vapor-phase diffusion coefficient is given by
03.33
D = D0 -4= 0.0026 cmVs
nl
and the mass flux of compound A is estimated as follows:

(4.20)

 = -0 = D^^ = 0.0026[cm%] 100^[If"31 = 1.3Ng/^)] (4.112)

Exercise. Consider advective transport of a compound in vapor phase. The compound
partitions between vapor and aqueous phases according to the relationship
V=KC

(4.113)

where V = concentration in vapor phase, C concentration in aqueous phase, and K =

partitioning coefficient. Given vapor flux q, porosity n, and water saturation Sw9 estimate
the apparent (retarded) velocity of the compound. From the advective transport equation,
we have

--
where
CT = CSwn + V(I - Sw)n = Vn (1 - SJ S"
+ l\
[K(L bw
)

(4.115)

Thus, for nonretarded tracers, we have

T-f

where pore-vapor velocity is

V

= *T^J

<4'117>

whereas for retarded compounds, we have

ff

<>

where the retarded velocity is given by

VR = ^ =
R

j
+1
^w
K(I - Sw)

(4.119)

Exercise. Consider diffusive vertical transport of a compound in vapor phase. The compound is subject to first-order degradation in the aqueous phase at rate X and to partitions
between vapor and aqueous phases according to the following relationship:
V=KC

(4.113)

where V = concentration in vapor phase, C = concentration in aqueous phase, and K =

partitioning coefficient. At the depth of 100 cm below the ground surface, the vapor concentration of the compound was measured to be V0, whereas at the ground surface the concentration was Vs. Given the compound's diffusion coefficient in vapor phase D0, porosity n, and water saturation Sw, estimate the diffusive flux of the compound at the soil surface. The relevant mass transport equation is given by

DJfV-I
SC = Q
dx2 n

(4.120)

where
Q3.33
D - D0-^n2
Substituting C = VIK into the mass transport equation leads to
32V
^JL-X2V=O
dr
where

(4.19)

(4.121)

X" = ^

We solve the modified mass transport equation to obtain

V(Jt) = C1BXpI -Xx] + C2exp[Kx\

(4.123)

where constants C1 and C2 are obtained from the boundary conditions

V(O) - V0, and V(IOO) = V,

(4.124)

The compound's mass flux at the soil surface is estimated from

q = -D^VQ
djt

@x = 10Q

(4.125)

REFERENCES
Bear, J., and Y. Bachmat, Introduction to Modeling of Transport Phenomena in Porous Media,
Kluwer Academic, Dordrecht, The Netherlands, 1990.
Bear, J., Dynamics of Fluids in Porous Media, Dover Publications, New York,k 1972.
Bear, J., Hydraulics of Groundwater, McGraw Hill, New York, 1979.
Bouwer H., Groundwater Hydrology, McGraw Hill, New York, 1978.
Bruce, G. H., D. W. Peaceman, and H. H. Rachford, Jr., 1953. "Calculations of Unsteady-State Gas
Flow Through Porous Media," Petroleum Transactions, AIME, 198: 1953.
Bureau of Reclamation, Ground Water Manual (Reprint), U.S. Department of the Interior
Washington, DC, 1995.
Cedergren H. R., Seepage, Drainage, and Flow Nets, 3rd ed., John Wiley & Sons, Inc., New York,
1989.
Charbeneau, R. J., "Kinematic Models for Soil Moisture and Solute Transport," Water Resources
Research, 20: 699-706, June, 1984.
Chirlin, G. R., "A Critique of the Hvorslev Method for Slug Test Analysis: The Fully Penetrating
Well," Ground Water Monitoring Review, 130-138, 1989.
Cho, J. S., 1991. Forced Air Ventilation for Remediation of Unsaturated Soils Contaminated by
VOC., Publication No. EPA/600/S2-91/016, U.S. Environmental Protection Agency, Washington,
D.C.

Dagam, G., "Solute transport in heterogeneous porous formations," Journal of Fluid Mechanics,
145: 151-177, 1984.
De Josselin Jong, G. "Singularity Distribution for the Analysis of Multiple-Fluid Flow Through
Porous Media," Journal of Geophysical Research, 65: 3739-3758, 1960.
De Marsily G., Quantitative HydrogeologyGroundwater Hydrology for Engineers, Academic
Press, San Diego, CA,
De Smedt R, and P. J. Wirenga, "Solute Transport Through Soil With Nonuniform Water Content,"
Soil Science Society of America Journal, 42. (1): 1978.
Domenico, P. A., and F. W. Schwartz, Physical and Chemical Hydrogeology, John Wiley & Sons,
New York, 1990.
Dullien F. A. L., Porous Media: Fluid Transport and Pore Structure, 2nd ed., Academic Press, San
Diego, CA, 1992.
Edelman J. H., Groundwater Hydraulics of Extensive Aquifers, 2nd., International Institute for
Land Reclamation and Improvement, Bulletin No. 13, The Netherland, 1983.
Fetter, C. W., Contaminant Hydrogeology, Macmillan, New York, 1993.
Fetter, C. W., Applied Hydrogeology, Simon & Schuster, Company Englewood, NJ, 1994.
Freeze, R. A., and J. A. Cherry, Groundwater, Prentice-Hall, Englewood Cliffs, NJ, 1979.
Gelhar, L. W. and C. L. Axness. "Three-Dimensional Stochastic Analysis of Macrodispersion in
Aquifers", Water Resources Research, 19 (1): 161-180, 1983.
Germann, P. R, M. S. Smith, and G. W. Thomas, "Kinematic Wave Approximation to the Transport
of Escherichia coli in the Vadose Zone," Water Resources Research. 23 (7), 1281-1287, 1987.
Girinsky, N. K. Determination of the Coefficient of Permeability, Gosgeolizdat, 1950.
Grubb, S. "Analytical Model for Estimation of Steady-state Capture Zones of Pumping Wells in
Confined and Unconfmed Aquifers," Ground Water, 31(1) 27-32, 1993.
Hantush, M. S., Hydraulics of Wells, "in Advances in Hydroscience," V. T. Chow, ed., Academic
Press, New York, 1964.
Hantush, M. S. "Growth and Decay of Groundwater-mounds in Response to Uniform Percolation."
Water Resources Research, 3 (1): 227-234, 1967.
Harr, M. E., Groundwater and Seepage, Dover Publications, New York, 1990.
Haverkmp, R., M. Vauclin, J. Touma, P. J. Wierenga, and G. Vachaud, A "Comparison of
Numerical Simulation Models for One-Dimensional Infiltration," Soil Science Society America
Journal, 41: 285-294, 1977.
Hinchee, R. E. ed., Air Sparging for Site Remediation, Lewis Publishers, Boca Raton, FL, 1994.
Huisman, L., Groundwater Recovery, Winchester Press, New York, 1972.
Hunt, B. "Seepage to Collection Gallery Near Seacoast," Water Resources Research, 21: 311-316,
1985.
Hunt, B., Mathematical Analysis of Groundwater Resources, Butterworths, London, UK, 1983.
Jaffe, P. R., and R. A. Ferrara, "Desorption Kinetics in Modeling of Toxic Chemicals," Journal of
Environmental Engineering, American Society of Civil Engineers, 109: 859-867, 1983.
Javandel, I., and C. F. Tsang, "Capture-zone Type Curves: A Tool for Aquifer Cleanup." Ground
Water, 24: 616-625, 1985.

L, C. Doughty, and C. F. Tsang, Groundwater Transport: Handbook of Mathematical Models,

American Geophysical Union, Washington, DC, 1987.
Johnson, P. C., M. W. Kemblowski, and J. D. Colthart, "Quantitative Analysis for the Cleanup of
Hydrocarbon-Contaminated Soils by in-situ Soil Venting," Ground Water, 1990.
Johnson, P. C., C. C. Stanley, M. W. Kemblowski, D. L. Byers, and J. D. Colthart. A "Practical
Approach to the Design, Operation, and Monitoring of in situ Soil-Venting Systems." Ground
Water Monitoring Review, Spring, 1990.
Jury, W. A., "Chemical Transport Modeling: Current Approaches and Unresolved Problems,"
Chemical Mobility and Reactivity in Soil Systems, 1983, pp. 49-64.
Jury, W. A., R. Grover, W. F. Spencer, and W. J. Farmer, "Modeling Vapor Losses of Soil
Incorporated Triallate," Soil Science Society of America Journal, 44: 445^-50, 1980.
Jury. W. A., W. F. Spencer, and W. J. Farmer, "Use of Models for Assessing Relative Volatility,
Mobility, and Persistence of Pesticides and other Trace Organics in Soil Systems." Hazard
Assessment of Chemicals: Current Developments, Vol. 2, 1983.
Keely, J. F. and C. F. Tsang, "Velocity Plots and Capture Zones of Pumping Centers for GroundWater Investigations." Ground Water, 21: 701-714, 1983.
Kishi, Y. and Y. Fukuo, "Studies on Salinization of Groundwater," L Journal of Hydrology, 35:
1-29, 1977.
Kool, J. B., J. C. Parker, and M. T. van Genchten, "Parameter Estimation for Unsaturated Flow and
Transport ModlesA Review." Journal of Hydrology, 91: 255-293, 1987.
Kozeny, J., Thorie und Berchnung der Brunnen. Wasserkraft und Wasserwirtschaft, Nos. 8-10,
1933.
Marino, M. A., "Artificial Groundwater Recharge: I. Circular Recharging Area," Journal of
Hydrology, 25: 201-208, 1975.
Marshall, T. J., J. W. Holmes, and C. W. Rose, Soil Physics, 3rd ed., Cambridge University Press,
Cambridge, UK, 1996.
McElwee, C., and M. Kemblowski, "Theory and Application of an Approximate Model of
Saltwater Upconing in Aquifers," Journal of Hydrology, 115: pp 139-163, 1990.
McWhorter, D. B. Steady and Unsteady Flow of Fresh Water in Saline Aquifers, Water
Management Technical Report No. 20, Council of U.S. Universities for Soil and Water
Development in Arid and Sub-Humid Areas, 1972.
Musa, M. and M. W. Kemblowski. "Effective Capture Zone for a Single Well," Submitted to
Ground Water July 1994.
Newsom, J. M., and J. L. Wilson, "Flow of Ground Water to a Well Near a Stream - Effect of
Ambient Ground-water Flow Direction." Ground Water, 25: 703-711, 1988.
Oberlander, P. L. and R. W. Nelson, "An Idealized Ground-Water Flow and Chemical Transport
Model (S-PATHS)," Ground Water, 22: 441^49, 1984.
Ostendorf, D. W, R. R. Noss, and D. O. Lederer, "Landfill Leachate Migration through Shallow
Unconfined Aquifers," Water Resources Research, 20: 291-296, 1984.
Palmer, C. M., Principles of Contaminant Hydrogeology, Lewis Publishing, Chelsea, MI, 1992.
Pankow, J. E, R. L. Johnson, and J. A. Cherry, "Air Sparging in Gate Wells in Cutoff Walls and
Trenches for Control of Plumes of Volatile Organic Compounds (VOCs)," Ground Water, 31:
654-663, 1993.

Parker, J. C., and M. T. van Genuchten, Determining Transport Parameters from Laboratory ad
Field Tracer Experiments, Virginia Agricultural Experiment Station Bulletin No. 84-3, Virginia
Polytechnic Institute and State University, Blacksburg, 1984.
Parker, J. C., and M. T. van Genuchten, "Flux-averaged and Volume-Averaged Concentrations in
Coninuum Approaches to Solute Transport," Water Resources Research, 20: 886-872, 1984.
Parker, J. C., K. UnIu, and M. W. Kemblowski, "A Monte Carlo Model to Assess Effects of Land
Disposed E & P Waste on Groundwater," SPE Annual Technical Conference & Exhibition, 1993.
Philip, J. R. "The theory of infiltration: 1. The infiltration equation and its solution." Soil Science.
83: 345-357, 1957.
Raudkivi, A. J., and R. A. Callander, Analysis of Groundwater Flow, Edward Arnold, London, UK,
1976.
Rosenshein, J., and G. D. Bennet, eds., Groundwater Hydraulics, American Geophysical Union,
Washington, DC, 1984.
Rubin, H., and G. F. Pinder., "Approximate Analysis of Upconing." Advances in Water Resources,
1 (2): 97-101, 1977.
Sallam, A., W. A. Jury, and J. Letey, "Measurement of Gas Diffusion Coefficient Under Relatively
Low Air-Filled Porosity," Soil Science Society of America Journal 48:3-6, 1983.
Schiegg, H. O., "Considerations on Water,. Oil and Air in Porous Media," Water Science
Technology, 17: 467^76. 1984.
Shafer, J. M., "Reverse Pathline Calculation of Time-Related Capture Zones in Nonuniform Flow,"
Ground Water 25: 283-289, 1987.
Sikkema, P. C. and J. C. Van Dam, "Analytical Formulae for the Shape of the Interface in a SemiConfined Aquifer," Journal of Hydrology, 56: 201-220, 1982.
Sposito, G., and W. A. Jury, "Inspectional Analysis in the Theory of Water Flow Through
Unsaturated Soil," Soil Science Society of America Journal, 42 (1): 1985.
Sposito, G., "Chemical Models of Inorganic Pollutants in Soils," CRC Critical Reviews in
Environmental Control, 15 (1): 1-24, undated.
Strausberg, S. I. "Estimating Distances to Hydrologic Boundaries from Discharging Well Data,"
19th Annual Meeting of the Rocky Mountain Section of the Geological Society of America, Las
Vegas, NV, 1966.
Thornton, J. S., and W. L. Wootan, Jr., "Venting for the Removal of Hydrocarbon Vapors from
Gasoline Contaminated Soil," Journal of Environmental Science and Health, All (1), 3144, 1982.
Todd, D. K., Groundwater Hydrology, 2th John Wiley & Sons, New York, 1980.
Todd, D. K., "Salt-Water Intrusion and Its Control." Journal of the American Water Works
Associations, 180-187, 1973.
U. S. Department of Agriculture Agricultural Research Service, Analytical Solutions of the OneDimensional Convective-Dispersive Solute Transport Equation, Technical Bulletin No. 1661,
UnIu, K., M. W. Kemblowski, J. C. Parker, D. Stevens, P. K. Chong, and I. Kamil, "A Screening
Model for Effects of Land-Disposed Wastes on Groundwater Quality" Journal of Contaminant
Hydrology, 11: 27-^9, 1992. Washington, DC, 1982
Van Genuchten M. T. and W. J. Alves, "Analytic Solution of the One-Dimensional Convective
Solute Transport Equation, Technical Bulletin. 1661, U.S. Departament of Agriculture,
Washington D.C., 1982.
Ward, C. H., M. B. Tomson, P. B. Bedient, and M. D. Lee, "Transport and Fate Processes in the
Subsurface," Water Resources Symposium, VoI 13, WARSAG, 1987.

Warrick, A. W. , J. W. Biggar, and D. R. Nielsen, "Simultaneous Solute and Water Transfer for an
Unsaturated Soil," Water Resources Research. 7: 1216-1225, 1971.
Watson, K. K., and M. J. Jones, "Algebraic Equations for Solute Movement During Absorption,"
Water Resources Research, 20: 1131-1136, 1984.
Wilson, J. L., and L. W Gelhar, "Analysis of Longitudinal Dispersion in Unsaturated Flow 1: The
Analytical Method," Water Resources Research, 17 (1): 122-130, 1984.
Wilson, J. L., and P. J. Miller, "Two-Dimensional Plume in Uniform Ground-Water Flow Discussion," Journal of the Hydraulics Division, American Society of Civil Engineers, 103
(HY12): 1567-1570, 1979.
Wilson, J. L., and P. J. Miller., "Two-Dimensional Plume in Uniform Ground-Water Flow," Journal
of the Hydraulics Division, American Society of Civil Engineers, 104 (HY4): 503-514, 1978.
Wirojanagud, P., and R. Charbeneau., "Saltwater Upconing in Unconfined Aquifers," Journal of
Hydraulic Engineering, American Society of Civil Engineers, 111: 417^-34, 1985.
Yeh, G. T., Analytical Transient One-, Two-, and Three-Dimensional Simulation of Waste Transport
in the Aquifer System, Environmental Sciences Division Publication No. 1439, Oak Ridge
National Laboratory, Oak Ridge, TN, 1981.

CHAPTER 5
ENVIRONMENTAL
HYDRAULICS

Richard H. French
Water Resources Center
Desert Research Institute
University and Community College System of Nevada
Reno, Nevada
Steven C. McCutcheon
Ecosystems Research Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Athens, Georgia
James L Martin
AScI Corporation
Athens, Georgia

5.1

INTRODUCTION

The thermal, chemical, and biological quality of water in rivers, lakes, reservoirs, and near
coastal areas is inseparable from a consideration of hydraulic engineering principles;
therefore, the term environmental hydraulics. In this chapter we discuss the basic principles of water and thermal budgets as well as mixing and dispersion.

5.2

WATERANDTHERMALBUDGETS

5.2.1 Water Budget

A water budget is a statement of the law of conservation of mass or
(change in storage) = (input) (output)

(5.1)

and the expressions of the water budget can range from simple to very complex. For
example, consider the lake or reservoir shown in Figure 5.1. For this situation, a generic
water budget could be written as follows
^ = (/, + / + /, + Pr + Rr) ~(Ev + Tr+Gs + Oc+W) (5.2)

Aquatic
vegetation
Water table

Water table

Storage, 5s

FIGURE 5.1 A hypothetical lake illustrating the variables in the water budget.

where Ic = channel inflow rate, I0 = overland inflow rate, Ig = groundwater inflow rate,
Pr = precipitation rate, Rr = return flow rate, Ev = evaporation rate, Tr = transpiration
rate, G5 = groundwater seepage rate, O0 = channel outflow rate, W = consumptive withdrawal, and Ss = lake/reservoir storage rate at time t (volume).
The solution of Eq (5.2) quantifies the terms, and, in many cases, the goal of the modeling effort is to estimate the value of a single term or group of terms: for example, evapotranspiration (Ev + Tr). The reliability of using a water budget is directly related to the
accuracy of the prediction techniques used, the availability and quality of gauged data,
and the time period involved. Among the methods of evaluating the individual terms in
Eq. (5.2) are the following:

Channel inflow and outflow ( I c and Oc)gauging, statistical simulation.

Overland inflow (I0)gauging, rainfall-runoff relationships.
Groundwater inflow and seepage rate (Ig and Gs)seepage equations, gauging.
Precipitation (P1)gauging, statistical simulation (Smith, 1993).
Evaporation and transpiration (E and T)gauging, evaporation/transpiration prediction relationships (Bowie et al. 1985; Shuttleworth, 1993).
Return flow and withdrawal (Rr and W)-gauging.

5.2.2 Thermal Budget

The total thermal budget for a body of water includes atmospheric heat exchange at the
air water interface (usually the dominant process), the effects of inflows (tributaries,
wastewater, and cooling water discharges), heat resulting from chemical-biological reactions, and heat exchange with the stream bed. In the following sections, the primary components of the air-water interface heat budget will be briefly discussed; for further details
the reader is referred to Bowie et al., (1985), McCutcheon (1989), or Shuttleworth
(1993).
Atmospheric heat exchange at the air-water interface is given by
H=Qs-Qsr + Qa- Qar - Qbr ~QeQc

(5.3)

where H = net surface heat flux, Q5 = shortwave radiation incident to the water surface
[30-300 (kcal/m2)/h], Qsr = reflected shortwave radiation [5-25 (kcal-m2)/h], Qa =
incoming longwave radiation from the atmosphere (225-360 kcal/m2/h), Qar = reflected
longwave radiation [5-15 (kcal-m2)/hr], Qbr = longwave back radiation emitted by the
water body [220-345 (kcal-m2)/h], Qe = energy utilized by evaporation [25-900
(kcal-m2)/h], and Q0 = energy converted to or from the body of water (35-50 kcal-m2
/hr). Note that the ranges given are typical for the middle latitudes of the United States
(Bowie et al., 1985).
The equations for estimating the terms of the thermal budget use a mixed set of units,
and appropriate conversions among the different units used are provided in Table 5.1.
5.2.2.1 Net atmospheric shortwave radiation (Qs Qsr). The net shortwave radiation
(Qsn) is that portion of the incident shortwave radiation captured at the water surface, taking into account losses caused by reflection. Although solar radiation can be measured
with specialized meteorological stations equipped with radiometers, these instruments
require painstaking calibration and maintenance. In most cases, measured values of solar
radiation are not available at the location of interest and must be estimated from equations.
Among the formulations for estimating net shortwave solar radiation is
Qsn = QS- Qsr = 0.94fcc(l - 0.65C2C)

(5.4)

where Qsc = clear sky solar radiation [kcal Tn Xh) and C0 = fraction of sky covered by
clouds (Anderson, 1954; Ryan and Harleman, 1973). It is pertinent to note that Eq. (5.4)

TABLE 5.1 Useful Energy Conversions for Energy Budget Calculations

IBtu/ftVday
lwatt/m 2
1 Ly/day
1 (kcal-m2)/hr

=
=
=
=

0.131 W/m2
=
7.61 Btu-ft2)/day =
0.485 W/m2
=
1.16 W/m2
=

0.271 Ly/day
=
2.07 Ly/day
=
3.69 (Btu/-ft2)/day =
2.40 Ly/day
=

0.113 (kcal-m2)/h
0.86 (kcal-m2)/h
0.42 (kcal-m2)/h
8.85 (Btu-ft2)/day

1 kpa
lmb
lmm Hg
I m Hg

=
=
=

10 mb
0.1 kpa
1.3mb
33.0mb

7.69 mm Hg
0.769 mm Hg
0.13 kpa
25.4 mm Hg

0.303 in (Hg)
0.03 in (Hg)
0.039 in (Hg)
3.3 kpa

=
=
=
=

=
=
=
=

Abbreviations Ly = Langleys; mb = millibar; and Btu = British Thermal Unit

assumes average reflectance at the water surface and uses clear sky solar radiation. In
some situations, the effects of atmospheric attenuation are much greater than normal and
more complex equations are required (e.g., 1972). Clear sky radiation (Qsc) can be estimated as a function of calendar month and latitude from Fig. 5.2.
Shortwave solar radiation is absorbed at the water surface and penetrates the water
column, depending on the wavelength of the radiation, the properties of the water, and the
matter suspended in the water. The degree of penetration of shortwave solar radiation
(sunlight) into the water column has a significant effect not only on water temperature but
also on the rate of photosynthesis by aquatic plants and the general clarity, color and aesthetic quality of the water. The penetration of shortwave solar radiation is described by
/ - I0exp (-k e y)

(5.5)

where / = light intensity at depth v, K6 = extinction coefficient, and I0 = light intensity

at the surface (y = O).
Values of the extinction coefficient can be estimated by several methods. For example,
measurement of total light penetration into a water column can be made by using a pyreheliometer positioned at the surface that measures the total incoming solar radiation.
Simultaneously, an underwater photometer is lowered and the radiation is recorded at each
of a series of depths throughout the water column. Then, a value of Ke can be estimated
by linear least-squares regression. An alternative but traditional, simpler, and less accurate method to estimate Ke is to lower a target into the water column until, by eye, the target just disappears. A standardized target (Secchi disk) is commonly used, and a number
of investigators (Beeton, 1958; French et al., 1982; Sverdrup et al, 1942;) have developed
empirical relationships between the Secchi disk depth (ys) and the extinction coefficient
of the form
^ = M^
(5.6)
y,
Finally, the depth (ye) at which 1 percent of the surface radiation still remains (the
euphotic depth) is given from Eq. (5.5) as

4.61
ye = -jp

(5.7)

5.2.2.2 Net atmospheric long-wave radiation (Qa - Qar). Atmospheric radiation is characterized by much longer wavelengths than solar radiation because the major emitting
elements are water vapor, carbon dioxide, and ozone. The approach generally used to
estimate this flux involves the empirical estimation of an overall atmospheric emissivity
and the use of the Stephan-Boltzman law (Ryan and Harleman, 1973). Swinbank (1963)
developed the following equation, which has been used in many water quality models
Q0n = Qa~ Qar = U6 X 1013(1 + 0.11C2J(T0 + 46O)6

(5.8)

where Qan = net long-wave atmospheric radiation (Btu/ft2/day), Cc = fraction of sky covered by clouds, and T0 = dry bulb air temperature (0F).
5.2.2.3 Long-wave back radiation (Qbr). The long-wave back radiation from a water
surface in most cases is the largest of all the fluxes in the heat budget (Ryan
and Harleman, 1973). The emissivity of a water surface is well known; therefore, this
flux can be estimated with a high degree of accuracy as a function of the water surface
temperature:

FIGURE 5.2 Clear sky solar radiation. (From Hamon et al. 1954)

Qbr = 0.9707?

(5.9)

where Qbr = long-wave back radiation (cal/m /s), Ts = surface water temperature (K), and
a = Stefan-Boltzman constant (1.357 X IQ-8 cal-m2/s/K4)
5.2.2.4 Evaporative heat flux (Qe). Evaporative heat loss (kcal/m2/s) occurs as a result
of the change of state of water from a liquid form to vapor and is estimated by
Q. = P^A

(5.10)

where Lw = latent heat of vaporization (= 597 Q.51TS, kcal/kg), Ts = surface water temperature (0C), Ev = evaporation rate (m/s), and p = water density (kg/m3).
A standard expression for evaporation from a natural water surface is
Ev = (a + bW)(es - ea)

(5.11)

where Ev = evaporation rate (m/s), a and b = empirical coefficients, W = wind speed at

some specified distance above the water surface (m/s), es = saturation vapor pressure at
the temperature of the water surface (mb), and ea = vapor pressure of the overlying atmosphere (mb). In many cases, the empirical coefficient a has been taken as zero with 1 X
ICh9 ^ b < 5 X ICh9 (Bowie et al., 1985). The saturated vapor pressure can be estimated
(Thackston, 1974) by
(5 12)

^^H^Tr^o)

'

where es is in inches of Hg, and Ts = water surface temperature ( F). There are a number
of ways of estimating ea, depending on the available data. For example, if the relative
humidity (Rn) is known, then
RH = ^

(5-13)

and then if the wet bulb temperature and atmospheric pressure are known (Brown and
Barnwell, 1987)
ea = e,- 0.000367Pa(7; - T.'Jl + ^32J

(5.14)

where all pressures are in (in Hg), all temperatures are in (0F), P0 = atmospheric pressure,
and Twb = wet bulb temperature. The relationship among the air and wet bulb temperatures (0F) and relative humidity (Thackston, 1974) is
Twb= (0.655+ 036RH)Ta

(5.15)

There are many equations for estimating the rate of evaporation. For example, Jobson
(1980) developed a modified formula that was used in the temperature modeling of the
San Diego Aqueduct and subsequently was modified for use on the Chattahoochee River
in Georgia (Jobson and Keefer, 1979). McCutcheon (1982) noted that, in many models,
the wind speed function is a catchall term that compensates for many factors, such as (1)
numerical dispersion in some models, (2) the effects of wind direction, fetch, channel
width, sinuosity, bank, and tree height, (3) the effects of depth, turbulence, and lateral
velocity distribution; and (4) the stability of air moving over the stream. (Fetch is the distance over which the wind blows or causes shear over the water's surface.) Finally, it is

important to note that evaporation estimators that work well for lakes or reservoirs will
not necessarily provide the same level of performance when used in streams, rivers, or
constructed open channels.
5.2.2.5 Convective heat flux (Qc). Convective heat is transferred between air and water
by conduction and is transported to or from the air-water interface by convection. The
convective heat flux is related to the evaporative heat flux (Qe) by the Bowen ratio (Bowie
et al., 1985), or
RB = ^JT = (6.19 X 1(M)P, eTs~eTa
Qe
s~ a

(5.16)

where all temperatures are in (0C), all pressures are in (mb), and R8 = Bowen ratio.
5.2.2.6 Conclusion. The foregoing is a brief summary of the approaches used most frequently to estimate surface heat exchange in numerical models. The reader is referred to
other publications for a more detailed discussion of the approaches (Bowie et al., 1985)
and meteorological data requirements (Shanahan, 1984). Note that each situation should
be considered carefully from the viewpoint of specific factors that must be taken into
account. For example, in most lakes, estuaries, and deep rivers, the thermal flux through
the bottom is not significant. However, in water bodies with depths less than 3 m (10 ft),
bed conduction of heat can be significant in determining the diurnal variation of temperatures within the body of water (Jobson, 1980, Jobson and Keefer, 1979).

5.3

EFFECTS AND CAUSES OF STRATIFICATION

5.3.1 Effects
The density of water is strongly affected by temperature and the concentrations of dissolved and suspended solids. Regardless of the cause of differences in water density, water
with the greatest density is found at the bottom, whereas water with the least density resides
at the surface. When density gradients are strong, vertical mixing is inhibited. Stratification
is the establishment of distinct layers of water of different densities (Mills et al., 1982).
Stratification is enhanced by quiescent conditions and is destroyed by wind stress, turbulence caused by large inflows, and destabilizing changes in water temperature. In many
bodies of water (rivers, lakes, and reservoirs), stratification is the single most important
phenomena affecting water quality.
When stratification is absent, the water column is mixed vertically and dissolved oxygen (DO) is present in the vertical water column from the top to the bottom: that is, fully
mixed water columns do not have DO deficit problems. For example, when stratification
occurs, in reservoirs and lakes mixing is limited to the epliminion or surface layer. Since
stratification inhibits, vertical mixing is inhibited by stratification, and reaeration of the
bottom layer (the hypoliminion) is inhibited if not eliminated. The thermocline (the layer
of steep thermal gradient between the epiliminion and hypoliminion) limits not only mixing but also photosynthetic activity as well. The hypolimnion has a base oxygen demand
and benthic matter and the settling of particulate matter, from the epiliminion only adds
to this demand. Therefore, while the demands of DO in the hypoliminion increase during
the period of stratification, inhibition of mixing between the epiliminion and the
hypolimnion and the lack of photosynthetic activity deplete the DO concentrations in the

hypolimnion. Finally, a rule of thumb suggests that when water temperature is the predominant cause of differences in water density a temperature gradient of at least lC/m is
required to define the thermocline (Mills et al., 1982).
The density of water can be estimated by
p = pr + Aps

(5.17)

where p = water density (kg/m3), pr = water density as a function of temperature, and

Ap5 = increments in density caused by solids.
5.3.2 Water Density as a Function of Temperature
A number of formulations have been proposed to estimate pT and among these are
pr = 999.8452594 + 6.793952 X 10 -2T6
- 9.095290 X 10-3 77 + 1.001685 X IQ-4T4,3
(5.18)
6
9
- 1.120083 X 10- T/ + 6.536332 X H)- T/
where Te = water temperature in 0C (Gill, 1982).
5.3.3 Water Density as a Function of Dissolved Solids or Salinity and
Suspended Solids
In most cases, data for dissolved solids are in the form of total dissolved solids
(TDS); however, in some cases, salinity may be specified. The density increment for dissolved solids can be estimated by
A

PTDS = CTDS(8.221 X 10"4 - 3.87 X 10'6Te + 4.99 X 10-8 Te2)

(5.19)

(Ford and Johnson, 1983), where CTDS = concentration of TDS (g/m3 or mg/L). If the concentration of TDS is specified in terms of salinity (Gill, 1982)
ApSL = CSL(0.824493 - 4.0899 X 10"3 Te + 7.6438 X 10~5 T2
-8.2467 X 10~7 T/ + 5.3875 X 10~9 Te4)
+ C81L (-5.72466 X 10"3 + 1.0277 X 104 Te
-1.6546 X 10"6 Te2) + 4.8314 X 10"4 CSL~2

(5.20)

where CSL = concentration of salinity (kg/m ). The density increment for suspended
solids is
Ap88 = CJl. --1 X l O 3
(5.21)
I
*GJ
where SG = specific gravity of the suspended sediment (Ford and Johnson, 1983).
The total density increment caused by solids is then
Ap8 - (AoTDS or ApSL) + Apss

(5.22)

5.4

MIXINGANDDISPERSIONINOPENCHANNELS

Turbulent diffusion (mixing) refers to the random scattering of particles in a flow by turbulent motions, whereas dispersion is the scattering of particles by the combined effects
of shear and transverse turbulent diffusion. Shear is the advection of a fluid at different
velocities at different positions within the flow.
When a tracer is injected into a homogeneous channel flow, the mixing process can be
viewed as composed of three stages. In the first stage, the tracer is diluted by the flow in the
channel because of its initial momentum. In the second stage, the tracer is mixed throughout the cross section by turbulent transport processes. In the third stage, longitudinal dispersion tends to erase longitudinal variations in the tracer concentration. In some cases, the
second stage is eliminated because the tracer discharge has a significant amount of initial
momentum associated with it; however, in many cases, the tracer flow is small and the
momentum associated with it is insignificant. In the latter case, the first transport stage is
eliminated. In this section, only the second and third transport stages will be treated, with
the implied assumption that if there is a first stage, it can be treated separately. Section 5.6
details how excessive initial momentum must be analyzed.
The reader is cautioned that, in this chapter, y is the vertical coordinate direction and
z is the transverse coordinate direction.
5.4.1 Vertical Turbulent Diffusion
To develop a quantitative expression for the vertical turbulent diffusion
coefficient,
consider a relatively shallow flow in an infinitely wide rectangular channel. It can be
shown that the vertical transport of momentum in such a flow is given by
^ ^ Ty

where T = shear stress at a distance y above the bottom boundary, p = fluid density, ev =
vertical turbulent diffusion coefficient, and v = longitudinal velocity (French, 1985).
Because the one-dimensional vertical velocity profile and shear distribution are known, it
can be shown that
* = 4 v *()N)

(5 24)

where k = von Karman's turbulence constant (0.41), yd = depth of flow, v* = shear

velocity (= Vgy^S), and S = longitudinal channel slope (French, 1985). The depth-averaged value of ev is
e; = 0.067^v*

(5.25)

When the fluid is stably stratified, mixing in the vertical direction is inhibited, and one
often quoted formula expressing the relationship between the unstratified and stratified
vertical mixing coefficient was provided by Munk and Anderson (1948)
* - = l + 3.MRi)

(5 26)

'

where evs = the stratified vertical mixing coefficient and Ri = the gradient Richardson
number.

5.4.2 Transverse Turbulent Diffusion

In the infinitely wide channel hypothesized to derive Eq. (5.24), there is no transverse
velocity profile; therefore, a quantitative expression for et the transverse turbulent diffusion coefficient, cannot be derived from theory. The following equations to estimate et
derived from experiments by Fischer et al., (1979), and Lau and Krishnappen (1977).
In straight rectangular channels, an approximate average of the results available is
e, - 0.15y^v* 50%

(5.27)

where the 50 percent indicates the error incurred in estimating et. In natural channels, e,
is significantly greater than the value estimated by Eq. (5.27). For channels that can be
classified as slowly meandering with only moderate boundary irregularities
e= 0.6Oy^v* 50%

(5.28)

If the channel has curves of small radii, rapid changes in channel geometry, or severe
bank irregularities, then the value of et will be larger than that estimated by Eq. (5.28).
For example, in the case of meanders, Fischer (1969) estimated that
V 2V3J
^.

e =2

~>

where a slowly meandering channel is one in which

TV
.2

(5 29)

(5.30)

and V = average channel velocity, T = channel topwidth, and Rc = radius of the curve.
A comparison of Eqs. (5.25) and (5.27) shows that the rate of transverse mixing is
roughly 10 times greater than the rate of vertical mixing. Thus, the rate at which a plume
of tracer spreads laterally is an order of magnitude larger than is the rate of spread in the
vertical direction. However, most channels are much wider than they are deep. In a typical case, it will take approximately 90 times as long for a plume to spread completely
across the channel as it will take to mix in the vertical dimension. Therefore, in most
applications, it is appropriate to begin by assuming that the tracer is uniformly distributed
over the vertical.
In a diffusional process in which the tracer is added at a constant mass flow rate (M*)
at the center line of a bounded channel (dC/dz = O at z O and 3C/3z = O at z = 7), the
downstream concentration of tracer is given approximately by

^-vfc.iH-^^^)

+ e5p (_<^^2j]

where

r - M*
VTyd

xet
w?

x=

and
, _ z_
T
A reasonable criterion for the distance required for "complete mixing " (where the concentration is within 5 percent of its mean value everywhere in the cross section) from a
center-line discharge is
O IVT 2
L = ^iLi.
(5.32)

$
If the pollutant is discharged at the side of the channel, the width over which the mixing must take place is twice that for center-line injection, but the boundary conditions are
otherwise identical and Eq. (5.32) applies if T is replaced with 2T.
5.4.3 Longitudinal Dispersion
After a tracer becomes mixed across the cross section, the final stage in the mixing
process is the reduction of longitudinal gradients by dispersion. If a conservative tracer is
discharged at a constant rate into a channel, the flow rate of which also is constant, there
is no need to be concerned about dispersion; however, in the case of an accidental release
(spill) of a tracer into a channel or the release is cyclic, dispersion is important. The onedimensional equation governing longitudinal dispersion is
f+=*f+5

where K = the longitudinal dispersion coefficient and S = sources or sinks of materials.

The initial work in dispersion, beginning with Taylor (1954), assumed a prismatic channel. However, natural streams have bends, sandbars, side pools, in-channel pools, bridge
piers, and other natural and anthropogenic changes, and every irregularity in the channel
contributes to longitudinal dispersion. Some channels may be so irregular that no reasonable approximation of dispersion is possible: for example, a stream consisting of pools
and riffles.
Fischer et al. (1979) presented a number of methods of approximating K in a natural
open channel. Of these, the most practical is
K=- 0 1 1 V ^
(5.34)
yy*
Equation (5.33) depends on a crude estimate of er and does not reflect the existence of
"dead zones" in natural channels. However, it does have the advantage of relying only on
the usually available estimates of depth, velocity, width, and surface slope.
With regard to the solution of the dispersion equation, the following observations are
pertinent:
1. The longitudinal dispersion analysis is not valid until the end of the initial period,
when
* a -4F2
^t

(5.35)

2. In the case of a slug of dispersing material (mass M), the longitudinal length of the
cloud after the initial period can be estimated approximately by

O VT^ TCF
24qji!L-o.07
, (Vf
JJ

(5.36)

and the peak concentration within the dispersing cloud is

c

(5 37)
=~frr
A I/4TrAjC
V~v^
where A = channel cross-sectional area.
Note that the observed value of the peak concentration will generally be less than this
estimate because some of the material is trapped in dead zones and some of the typical
tracers (Martin and McCutcheon, 1999) sorb onto sediment particles.

5.5 MIXING DISPERSION IN LAKES AND RESERVOIRS

Important factors in the hydraulic design, operation, and analysis of spills in reservoirs
and lakes include (1) determining vertical stratification to guide lake monitoring and the
design of withdrawal structures, (2) locating the plunge point or separation point to
determine how inflows mix, (3) computing the dilution and mixing of inflows and the
time required to travel through a reservoir or lake, and (4) determining the quality of
withdrawals or outflows and effects on the quality of reservoir water. The elevation and
flow through withdrawal structures at dams are selected to control flooding and achieve
certain water-quality targets or standards. The stratification, mixing, and travel of
inflows are determined to design water-intake structures at dams or other locations in
lakes, to forecast the habitat and fisheries that a proposed reservoir may support, and to
track chemical spills or flood waters through reservoirs. This section is based on
Chaps. 8 and 9 in Martin and McCutcheon (1999), which provide a number of sample
calculations.
Many lakes and reservoirs stratify for part of the year into an epilimnion, thermocline,
and hypolimnion as illustrated in Fig. 5.3. The depth and thickness of the thermocline or
metalimnion vary with location and time of the year, and even time of the day to a limited extent. The thermocline represents the interface between a well-mixed surface layer, or
epilimnion, and the cooler, deeper hypolimnion. In freshwater lakes, the thermocline is
defined by a minimum temperature gradient of I0CAn. When a distinct interface does not
exist, the thermocline, epilimnion, and hypolimnion may not be defined. Mixing processes also are different in riverine, transition, and lacustrine zones (Fig. 5.3). Mixing in the
riverine zone is dominated by advection and bottom shear, and turbulence is generally dissipated under the same conditions. Seiche, wind mixing, boundary shear, boundary intrusion, withdrawal shear, internal waves, and dissipation of turbulence generated elsewhere
cause mixing in the lacustrine zone. Buoyancy resulting from stable stratification stabilizes or prevents mixing. In the transition zone, ending at the plunge point or separation
point, buoyancy begins to balance the advective force of the inflow. There are three
sources of energy for mixing: (1) inflows from tributaries, overland runoff, and discharges, (2) withdrawal at dams, discharges at control structures, and natural outflows,
and (3) wind shear, solar heating and cooling, heat conduction and evaporation, and other
meteorological forces.

TRANSITION

LACUSTRINE

RIVERINE

PLUNGE UNE

HYPOLIMNION
active mixing
patch

oss Ii step
structure

boundary
mixing
withdrawal
layer

FIGURE 5.3 Mixing processes in zones of lakes and reservoirs. (Modified from Fischer, et all979)

SEPARATION POINT

Shallow lakes and reservoirs that do not stratify are normally analyzed in the same
fashion as rivers or as a completely mixed body of water. For a completely mixed system,
the residence time (T in seconds or more typically years) or time for an inflow to travel
through the body of water is simply tr = (|)/g, where ty is the volume of the lake (m3) and
Q is the sum of the inflows or the average reservoir discharge (m3/s).
Freshwater lakes tend to stratify when the mean depth exceeds 10 m and the residence
time exceeds 20 days (Ford and Johnson, 1986). The densimetric or internal Froude number Prd (Norton et al., 1968) provides a better indication of the stratification potential of a
reservoir where
Fr, =

^4 = /.y
= Frp
Mp]
/JAH
^ V gp
V gp y

(5.38)

LL = the length of the reservoir (m), yavg = the its mean depth (m), g = gravitational acceleration (m/s2), Ap = the difference in density over the depth for the internal Fr or between
the inflow and surface waters of the lake or reservoir at the plunge point or separation
point (kg /m3), p = average density of the lake for the internal Fr or density of the inflow
(Turner, 1973) at plunge or separation points (kg/m3), V0 = the average velocity of the
inflow (m/s), and y0 = the hydraulic depth or cross-sectional area divided by the top width
of the inflow (m). The Fr at the plunge point Frp, also defined in Eq. (5.38), will be used
in the next section. For design projections, the dimensionless density gradient Ap/(pyavg)
normally is taken to be 10'6 m-1 (Norton et al., 1968). IfFr I/TC, the reservoir is expected to be well mixed. If Fr 1/Ti9 the reservoir is expected to be strongly stratified, and
when Fr 1/Ti9 the reservoir is expected to be weakly or intermittently stratified.
Using the length or depth scales of Sundaram (1973) and Ford and Johnson (1986) the
depth to which wind can mix and destroy the stratification of a lake or reservoir for a particular surface heat flux is
.^A = 5.9 X 109^
(5.39)
^gHn
Hn
where w* = the shear velocity of the wind (m/s), K = an empirical coefficient approximately equal to the von Karman constant of 0.4, a = the volumetric coefficient of thermal
expansion for water (1.8 X 10^/0C), Hn = surface heat flux (W/m2), p = the density of
water ( 1000 kg/m3), and Cp = the specific heat of water (4186J/kgC).
The wind shear velocity is
DI =

w* = y Pflp^w 1.27 X 10-X,

(5.40)

where uw = the wind speed (m/s), pa = the density of air (kg/m3), and Cd = the drag coefficient, which usually is taken to be 1.3 X 10~3. (See Martin and McCutcheon (1999) to
estimate the net thermal energy flux.)
5.5.1 Annual Stratification Cycle
In freshwater lakes, stratification results when the sun heats the water faster than wind
shear can mix the heat over the depth. In saline lakes, differences in both temperature and
dissolved solids cause stratification. Stratification involving salinity may persist yearround in deeper saline lakes. The onset of stratification in freshwater lakes occurs in late

spring or early summer and persists into the fall or early winter, depending on latitude.
The surface heats rapidly, becoming less dense than deeper layers and forming stable differences in vertical density that inhibit vertical mixing until the fall overturn. As stratification develops, wind and currents mix the upper layers and tend to deepen the thermocline to form the well-mixed epilimnion. Although storms in late spring and summer
episodically lower the thermocline, the thermocline generally rises as solar heating
increases until midsummer. After late summer cooling begins, the thermocline deepens
until the fall overturn occurs. The decreased difference in temperature in the fall with the
hypolimnion allows more mixing that deepens the epilimnion and thermocline. The variable depth of the thermocline at any time is controlled by seasonal climate, the occurrence
of storms, water temperature, water depth, lake bathymetry, the strength of inflow and
outflow current, and other factors covered in more detail by Chapra and Reckhow (1983),
Ford and Johnson (1986), Hutchinson (1957), and Wetzel (1983).
The onset of cooler fall conditions causes the epilimnion to lose heat to the atmosphere. As heat is lost, mixing tends to become more dominant. The overturning or complete mixing of the reservoir or lake dominates as the epilimnion and hypolimnion
approach the same temperature. During winter, lakes and reservoirs remain unstratified
except in the higher latitudes where the hypolimnion approaches 40C and the surface
approaches O0C. The slight winter stratification of these colder water bodies is the result
of the usual decrease in water density as temperature decreases from 4 to O0C. Ice cover
maximizes and prevents wind mixing and erosion of the mild differences in density.
Stratification is so mild that a distinct thermocline does not form and the epilimnion and
hypolimnion are not well defined. Winter stratification persists until spring warming melts
the ice and heats the surface layer to the temperature of the hypolimnion (usually 40C)
when the spring overturn occurs.
The arrival of spring begins the cycle of heating and stratification anew. A difference
in temperature of just a few degrees results in a difference in density sufficient to inhibit
or prevent most vertical mixing in lakes and reservoirs. Vertical mixing is inhibited almost
completely during summer heating because wind and inflows and outflows do not have
sufficient energy to erode the differences in density that arise. The wind and energy available from wind and currents cannot overcome the potential energy differences that tend to
prevent mixing of the denser hypolimnion and lighter epilimnion. Fresh water flows into
a saline lake cause salinity gradients that have the same damping effect. Density stratification also is caused by suspended sediments, primarily resulting in sediment-laden
underflows. Martin and McCutcheon (1999) have illustrated the stratification cycle for
warm-water lakes and reservoirs.
Run-of-the-river reservoirs and shallow lakes that are weakly stratified because of high
flows or wind mixing follow only the general stratification trend. Complete mixing may
occur during the summer stratification period as a result of wind or runoff events, and the
thermocline may be difficult to define. Fall overturn occurs earlier in these bodies of water
than it does in deeper lakes.
5.5.2 Plunge and Separation Point-End of the Transition
Between Riverine and Lacustrine Conditions
The plunge point or separation point marks the downstream end of the transition zone
defined where buoyancy begins to exceed advective forces. These points move seasonally and, to a limited degree, during the day. Usually distinguished by a line of foam or floating debris across the reservoir or lake, the plunge point occurs when a denser inflow dives
below the lake surface and continues to flow along the bottom as a density current. The
separation point occurs when an underflow has the same density as the lake water at a

given depth, and separates from the bottom to flow into a discrete layer of the lake as an
interflow. Some underflows may be dense enough to flow to the lowest point in a lake or
to the dam that forms a reservoir. If the inflow is less dense than water at the lake surface,
an overflow occurs. Fig. 5.3 illustrates these three types of inflows.
At the plunge or separation point, the internal Fr of the stratified lake Prd is equal to
the Fr of the inflow at that point (Frp), as noted in Eq. (5.38). If the difference in density
between the lake surface and the inflow Ap is positive, an overflow occurs, and if Ap is
negative, an underflow occurs. If the slope of the reservoir bottom, or valley, is mild (S8
< 0.007), then the hydraulic depth (y0) is the normal depth of flow. For steep slopes (SB
> 0.007), the hydraulic depth is the critical depth (Akiyama and Stefan, 1984).
For tributary or river channels that are approximately rectangular or triangular, the
hydraulic depth and location of the plunge point or separation point can be calculated. For
a rectangular cross section of constant width, the hydraulic depth is
1/3
22
q2
=
yo =
kpT
2 |AP| 2
F
^nr*J
l^nr1

1/3

<5-41>

where Q = the riverine inflow rate (mVs) equal to VA, A = the cross-sectional flow in
area of the river (m2), B = the conveyance width (m), and q = the flow per unit width
(m2/s). Similar expressions were proposed by Akiyama and Stefan (1984), Jain (1981),
Singh and Shah (1971), and Wunderlich and Elder (1973), among and others. Savage and
Brimberg (1973) developed an independent expression for the Froude number at the
plunge point or point of separation (Frp) based on the conservation of energy and the theory of two-layered flow in stratified water bodies, which can be expressed as
\).478
f\
h)

(5.42)

where fb = the dimensionless bed friction factor and f. = dimensionless interfacial friction. Martin and McCutcheon (1999) have illustrated the calculations and summarized the
validation of these equations by an example derived from Ford and Johnson (1981, 1983).
For a triangular cross section with an angle 29 between the channel or valley walls, the
hydraulic depth is one-half the total depth. The area of the cross section (m2) is A = y\
tan(9), which, when substituted into the expression for the normal densimetric number Frn
and solved for the hydraulic depth V0 (m), is
2 2
T
Q
11^5
;^r
V0 = 0.5 / g2Ap
\
2
(Fr n^tan (e)J

(5.43)

where the bottom depth (distance between water surface and apex of the triangular cross
section) is twice the hydraulic depth for a triangular cross section. Hebbert et al. (1979)
derived an expression for the downstream densimetric Froude number at the plunge point
or separation point Frp for normal flow (S8 < 0.007) in a triangular cross-section, related
to the reservoir characteristics as

FrJ ~ sin(9)tan(SB) [1 - 0.85 C^sin (6)]

(5.44)
CD
where CD = the dimensionless bottom drag coefficient [C0 = (/J + fb)/4].
Equations (5.42) and (5.44) are based on characteristics of the reservoir or lake. (See
Martin and McCutcheon (1999) and Gu et al. (1996) for an example of the calculations.)
5.5.3 Speed, Thickness, and Width of Overflows
Martin and McCutcheon (1999) have noted that the speed of an overflow (vof with dimensions m/s) can be estimated from the celerity of a wave in a frictionless flow, but this
consistently overestimates the rate of spread. Instead, Koh (1976) developed a more
practical semiempirical expression based on uniform flow which reduces to (Ford and
Johnson, 1983)
^/= L 0 4 V^v
where the thickness of the overflow yof (m) can be estimated from (Kao, 1976) as
[ 2 ]w
1 24

^= -

(5.46)

[w
Lnr-i

In natural settings, overflows are usually dissipated by mixing caused by wind and solar
heating before traveling too far.
Horizontal spreading of an overflow is estimated using the inflow Fr defined by Eq.
(5.38). Safaie (1979, cited in Ford and Johnson, 1983) found that for Fr^ < 3, the flow is
an unsteady, buoyancy-driven spread and can be assumed to be completely mixed laterally except for abrupt changes in the entrance geometry. Typically, reservoirs widen gradually where major tributaries enter, but lakes may have an abrupt widening at the mouth of
tributaries. For Fr^ > 3, the inflow acts like a jet that expands proportionally with distance
B(x) = BQ + ex where B(x) = the overflow width (m) at distance x measured from the
separation point (m), B0 = the width of the riverine or tributary flow at the separation
point (m), and c = a dimensionless empirical constant (Ford and Johnson, 1983). From
laboratory experiments with plane jets, the value of c has been determined to be approximately 0.16 (Fischer et al. 1979; Ford and Johnson, 1983).
5.5.4 Underflow or Density Current Mixing
Underflows are dominated by two mixing processes. First, significant mixing occurs during the plunge beneath the surface. Second, shear at the interface with ambient lake or
reservoir water will result in mixing and entrainment as the underflow moves downward.
The initial turbulent mixing of the plunging flow will increase the total flow rate of the
underflow and reduce the density and concentration gradients. The fraction entrainment
^ caused by plunging is (Qp Q)IQ, where Qp is the flow rate at the plunge point (mVs)
and Q is the river flow-rate (mVs). For mild slopes SB < 0.007, ^ is on the order of 0.15
(Akiyama and Stefan, 1984). The depth of the underflow is the normal depth of flow. For
steep slopes S8 > 0.007, is on the order of 1.18 and the density current depth is the crit-

ical depth (Akiyama and Stefan, 1984). However, the entrained fraction ^ is highly variable. The dilution of concentrations or temperatures resulting from mixing in plunging
flows follows from a simple mass or heat balance
Cp = ^^
(5.47)
1+5
where C is the inflow concentration (g/m3 or mg/L3) or temperature (0C), C0 is the ambient concentration (g/m3 or mg/L3) or temperature (0C) of the lake, and Cp is the concentration (g/m3 or mg/L3) or temperature (0C) of the plunging flow after initial mixing.
The mixing after plunging results from bottom shear as well as shear at the interface
of the underflow with ambient lake water. For a triangular cross section, the entrainment
coefficient is (Imberger and Patterson, 1981)
E = I^CfFr2

(5.48)

where laboratory experiments indicate that Ck is approximately 3.2 (Hebbert et al. 1979),
C0 = the dimensionless bottom drag coefficient defined following Eq (5.42), and Frb the
internal Froude number
Fr, = -^=
VeA

(5.49)

where ub = underflow velocity, hb = underflow depth, and efc = relative density difference. The entrainment coefficient E is a constant for a specific body of water. The depth
or thickness of the underflow (m) is a linear function of the entrainment coefficient
(Hebbert et al. 1979; Imberger and Patterson, 1981)
yuf = (6/5)Ex + y0
where x is the distance downstream from the plunge point (m) and y0 is the initial thickness of the underflow (m) that is approximately equal to the depth at the plunge point. If
entrainment is limited, the depth of the underflow remains approximately constant as long
as the bottom slope remains constant. The increase in flow rate because of entrainment for
an underflow in a triangular cross section is solved iteratively as
IY >/3
1
GCO = G 1LW
HH ""MJ
(5.50)
where Q1 = the discharge (m3/s) and V1 = the depth (m) from the previous calculation
step. For the initial iteration, Q1 = the discharge at the plunge point Qp (m3/s) and V1 =
the plunge point depth y0 (m).
Because of more significant differences in density and less internal mixing contrasted
with the epilimnion, underflows tend to remain more coherent than overflows. Sedimentladen underflows, especially, tend to travel to the lake outlet or dam.
5.5.5 Interflow Mixing
After experiencing approximately 15 percent entrainment at the plunge point (for mild
slopes) and mixing as an underflow, an interflow intrudes into a lake at the depth at which
neutral buoyancy is achieved. The turbulence generated by bottom shear is dissipated
quickly, and entrainment into the interflow is dominated by interfacial shear with ambient
lake water above and below the intrusion layer.

When the momentum of inflow is small, an interflow is analogous to a withdrawal

from a dam discussed in Sec. (5.5.6). Interflows are governed chiefly by three conditions
based on the dimensionless number R = Fr.Gr1/3 where Fr1 is the internal Fr defined in Eq.
(5.51) and Gr is the Grashof number (Gr), both of which are computed at the depth of
intrusion. The internal Froude Number computed at the intrusion depth is
ft,= * = _A_
(5.51)
NL]
B1NL]
where ql = the interflow rate per unit width following entrainment at the intrusion point
(m2/s), L1 = the length of the reservoir at the level of intrusion (m), Q1 = the interflow rate
(m2/s), B1 = the intrusion width (m), and N = the buoyancy frequency (s-1) expressed as
/Mp7
^=Vw
<5-52>
where Ap1 = density difference between the layers into which the flow is intruding
(kg/m3), p7 = density of the intrusion (kg/m3), and yl = the thickness of the intrusion (m).
The dimensionless Grashof number Gr is the square of the ratio of the dissipation time to
the internal wave period or

N2L4
Gr = ^
^V

(5.53)

where ev = the vertically averaged diffusivity (m2/s). Generally, if Gr > 1, then an internal wave field will decay slowly, but if Gr < 1 then viscous dissipation damps waves
quickly (Fischer et al. 1979). Imberger and Patterson (1981) also introduced a dimensionless time variable
tN
1
~ Gr1/6
where t = time(s), which, along with the Prandtl number Pr = ev/e,, where etis vertically averaged diffusivity of heat (mVs), is used to define three interflow conditions:
1. If R > 1, the intrusion is governed by a balance of the inertial and buoyancy forces
so that the actual intrusion length L1. is proportional to time, as given by (Ford and
Johnson, 1983; Imberger et al., 1976)
L1 = 0.44L1. VRf

= 0.44 V^7M

(5.54)

If the speed of the intrusion is constant or uniform, the velocity V7 is L/f, so that
( g AD
A Vr V2
,
7
V1 = 0.44 V^V = 0.194 ^p
where pw = the density of the intrusion. The difference in density in the computation
of the buoyancy frequency is that occurring over the thickness of the intrusion hm,
which, along with the relationship um = qjhm, can be substituted into the above equation to yield an alternative formulation for the speed of intrusion. The thickness of the
interflow can be solved by assuming uniform flow (Ford and Johnson, 1983)
fcm = 2.99 Ml;]"3
2
I8P1 J

(5.55)

where hm is generally distributed equally above and below the center line of the
intrusion.
2. If R < t*R < P2/3, then the flow regime is dominated by the balance between viscous and buoyancy forces and the intrusion length becomes
L1. = 0.51 L Rm f516

(5.56)

The thickness of the interflow is

hm = 5.5LmG->

(5.57)

In this regime, the flow is generally distributed so that 64 percent lies above the center line of the intrusion (Imberger, 1980); thus, the half-thickness (HnJ of the interflow
above the center line is given by
hma = 3.5LJG-^

(5.58)

and the half-thickness below the center line is given by

hmb = 2-OLmG-6

(5.59)

3. If P2/3< f < R'1 then the flow regime is dominated by viscosity and diffusion and
the intrusion length becomes
L1 = C6LK3" t3/4

(5.60)

where C6 is a coefficient, that generally is unknown (Fischer et al. 1979).

Ford and Johnson (1986) indicated that unless dissolved solids dominate the density profile (i.e., Pr is high), intrusions into most reservoirs have R > 1, where inertia and
buoyancy dominate. Because the difference in density varies with the location of the
limits of the interflow zone above and below its center line, the solution proceeds by
estimating a value of hm and then by computing the difference in density, which is then
used to compute a revised estimate of hm. This process is repeated until convergence
occurs.
The equations for intrusion require information on both the morphometry of the reservoir and the temperature distribution. The widths used in the formulations should represent the conveyance width (Ford and Johnson, 1983). Because the time for the intrusion
to pass through a lake can be relatively long, the flow rates used in the calculations
should represent an average value over the period of intrusion. To estimate the time scale
in their analysis of intrusions in DeGray Lake in Arkansas, Ford and Johnson (1983) used
the length of the lake and Ap7 and hm across the thermocline. For DeGray Lake, the intrusion time scale ranged from 4 to 6 days. Changes in outflow during the period of the
intrusion also can affect the movement through the lake. Interflows may stall and collapse if the inflow or outflow ends. Interflows also may be diverted or mixed because of
changes in meteorological conditions that influence epilimnion mixing and thermocline
depth.
The temperature or density of the interflow will remain constant. However, the interflow will spread laterally and the thickness will increase caused by entrainment of ambient water. The resulting concentrations can be computed from a mass balance vinCBhm =
constant, where vin is the velocity of the interflow; C = the concentration or temperature;
B = the reservoir width, which may vary with distance from the separation or detachment
point; and hm = the thickness of the interflow.

UPPER LIMIT
Withdrawal
Zone

VELOCITY
PROFILE

FIGURE 5.4 Reservoir withdrawal. (Adapted from Martin and McCutcheon, 1999)

DENSITY
PROFILE

5.5.6 Outflow Mixing

The withdrawal velocity profile is used in models CE-QUAL-Rl (Environmental
Laboratory, 1985) and CE-QUAL-W2 (Cole and Buchak, 1993) and in calculations to
predict the effects of withdrawals on reservoir and tail race water quality. The extent of a
withdrawal zone (Fig. 5.4) strongly depends on the ambient lake stratification and release
rate, location of the withdrawal, and reservoir bathymetry. For a given outflow rate and
location, the withdrawal zone thins as the density gradient increases. Depending on the
degree of stratification, withdrawal rate and location, and other factors related to the
design of the dam and the bathymetry of the reservoir, the withdrawal zone may be thin
or may extend to the reservoir bottom or water surface. Within the withdrawal zone, the
velocity distribution will vary from a maximum velocity to zero at the limits of the zone,
depending on the shape of the density profile. The maximum velocity is not necessarily
centered on the withdrawal port.
A number of methods predict the extent of withdrawal zones and the resulting velocity distributions. Fischer et al. (1979) described methods of computing withdrawal patterns
similar to those used in the analysis of interflows in the previous section. The Box
Exchange Transport, Temperature, and Ecology of Reservois (BETTER) model and the
SELECT model based on the original work of Bohan and Grace (1973) are the more practical approaches. The BETTER model, applied to a number of Tennessee Valley Authority
reservoirs, computes the thickness of the withdrawal zone above and below the outlet elevation from Ay = cw Qout, where Qout = the total outflow rate and cw is a thickness coefficient. The model assumes a triangular or Gaussian flow distribution to distribute flows
within the withdrawal zone (Bender et al. 1990).
The SELECT model (Davis et al. 1985) computes the in-pool vertical distribution of
outflow and concentrations of water quality constituents, the outlet configuration and
depth, and the discharge rate (Stefan et al. 1989). The SELECT code also is applied as
subroutines in generalized reservoir models, such as CE-QUAL-Rl (Environmental
Laboratory, 1985). The model is based on the following equations.
The theoretical limits of withdrawal (Bohan and Grace, 1973) were modified by Smith
et al. (1985) to include the withdrawal angle as
IS=I

where Z = distance from the port center line to the upper or lower withdrawal limit; 6 =
the withdrawal angle (radians); and TV = the buoyancy frequency [gAp/(pZ)]1/2, in which
Ap = the difference in density between that at the upper or lower withdrawal limit and at
the port centerline; and p = the density (kg/m3) at the port center line. The convention is
that Ap is positive for stably stratified flows such that Ap = p (upper limit) - p (withdrawal port) or Ap = p (withdrawal port) - p (lower limit). The elevation of the water
surface, the bottom, of the reservoir, and the withdrawal port and the density profile must
be known. The equation must be solved iteratively since both the distance from the port
center line Z and the density as a function of Z are unknown. A typical solution procedure
where the upper and lower withdrawal zones can form freely within the reservoir without
interference at the surface or bottom is as follows:
1. Rearrange the equation as Qout Z3M)/7i = O.
2. Check to see if interference exists by, first, using Z equal to the distance from the
port center line to the surface. Estimate the density at the center line of the withdrawal port and the water surface and substitute the values into the rearranged
equation. If the solution is non-zero and is positive, surface interference exists.

truncated due to
surface interference

interference limit

VELOCITY

free

PROFILE

lower
limit

FIGURE 5.5 Definition of withdrawal characteristics. (From Martin and McCutcheon, 1999)

Similarly, substitute the distance from the port center line to the bottom, along with
the density at the bottom of the reservoir, and determine if a bottom interference
exists.
3. If both of the evaluations from Step 2 are negative, the withdrawal zone forms
freely in the reservoir. The limit of the surface withdrawal zone above the port
can be determined by using iterative estimates of values for Z and the density at
the height above the center line until the equation approaches zero to within some
tolerance. The lower limit of withdrawal below the port center line can be determined in a similar manner.
4. If surface or bottom interference exists, a theoretical withdrawal limit can be
determined using values of Z computed using elevations above the water surface
for surface interference or below the reservoir bottom for bottom interference.
However, this solution requires an estimate of density for regions outside the limits of the reservoir. Davis et al. (1985) estimated these densities by linear interpolation using the density at the port center line and the density at the surface or
bottom of the reservoir.
For the case where one withdrawal limit intersects a boundary and the other does not,
the freely forming withdrawal limit cannot be estimated precisely using the rearranged
equation. Smith et al. (1985) proposed an extension to estimate the limit of the freely
forming layer similar to that described above
O81 = 0.125(D-JPeL
n
N
[

I
n

sin

I D^d } + D--d\
( D
)
D J

where d = the distance from the port center line to the boundary of interference (m) and
D = the distance between the free withdrawal limit and the boundary of interference (m)
shown in Fig. 5.5. The length scale in the buoyancy frequency Af is D in place of Z, and
Ap is the difference in the density between that at the surface for withdrawals that extend
to the surface and between the lower free limit or density at the bottom for withdrawals
that extend to the bottom and upper free limit. For consistency with the definition of stable stratification as positive, the convention is that Ap = p(surface layer) - p(free limit)
or Ap = p(upper free limit) p(bottom layer).
Once the limits of withdrawal are established, the distribution of withdrawal velocity is estimated by dividing the reservoir into layers, the density of which is determined
at the center line of each layer. The computation of the vertical velocity distribution is
based on the location of the maximum velocity, which can be estimated from (Bohan
and Grace, 1973)

(
Y
YL = HsirAL57^\
(5.63)
H
I
)
where Y1 = the distance from the lower limit to the elevation of maximum velocity (m),
H = the vertical distance between the upper and lower withdrawal limits (m), and ZL =
the vertical distance between the outlet center line and the lower withdrawal limit (m)
shown in Fig. 5.4. If the withdrawal intersects a physical boundary, the theoretical withdrawal limit is used, which may be above the water surface or below the reservoir bottom.
Once the location of the maximum velocity Vmax (m/s) is determined, the normalized
velocity VN(I) = V(I)/Vmax in each layer / is estimated for withdrawal zones that intersect
a boundary as (Bohan and Grace, 1973)

-'-(l
or for a withdrawal that does not intersect a boundary
^ = (,-*M>J

(5.65)

where V(T) = the velocity in layer 7 (m/s), y(/) = the vertical distance from the elevation
of maximum velocity to the center line of layer / (m), Y1 = the vertical distance from the
elevation of maximum velocity to the upper or lower withdrawal limit (m) determined by
whether the centerline of layer / is above or below the point of maximum velocity, Ap(T)
= the density difference between the elevation of maximum velocity and the center line
of layer /, and Apmax = the difference in density between the point of maximum velocity
and the upper or lower withdrawal limit.
If the withdrawal intersects the surface or the bottom, velocities are calculated for locations either above the water surface or below the reservoir bottom and the distribution is
truncated at the reservoir boundaries to produce the final velocity distribution. The flow
rate in each layer / is
w = J^

Q~

<5-66)

Ew
i= i
where Qout = the total release rate and m = the number of layers. The quality of the release
can be determined from a simple flow-weighted average or mass balance as

q(I) C(I)
C* = -^2

(5-67)

E 4
/= i
where CR = the concentration or temperature of water quality constituent C in the release
and C(I) = the concentration or temperature in each layer.
For discharge over a weir, the withdrawal limit Z and average velocity in the withdrawal zone Vweir is derived from the densimetric Froude number [Eq. 5.38] as (Grace,
1971, Martin and McCutcheon, 1999),

Q=y^- C *^ + g ^ + <W^ + g ->

(5-68)

where Ap = the difference in density between the weir crest and the lower withdrawal
limit, p = the density at the weir crest elevation, Hw = head above the weir crest elevation, Z = distance between the crest elevation and the lower withdrawal limit, and C1 and
C2 are constants, which have values of

C1 = 0.54 and C2 = O forZ + H > 2.0

Hw
and

(5.69)

C1 = 0.78 and C2 = 0.70 forZ +H < 2.0

Hw

5.5.7 Mixing Caused by Meteorological Forces

Wind-generated waves and convective cooling cause significant mixing at the water surface. Wind shear causes waves at the surface and at each density interface within a lake
or reservoir, such as the thermocline, and larger scale surface mixing by Langmuir circulation results from sustained wind. Wind setup, seiche, and upwelling are caused by meteorological events that generate mixing over much larger areas. Internal waves are caused
by shearing currents set up by both wind and other currents and, although not as obvious
as surface waves, these can be larger and more effective in causing mixing. The intensity
of wave mixing and turbulence is a direct result of wind energy or the energy in other
shearing currents.
The basic characteristics of waves are amplitude or height between trough and crest
and the length between crests. The wave period is the time required for successive waves
to pass a given point. Progressive waves move with respect to a fixed point, whereas
standing waves remain stationary while water and air currents move past. The height and
period of wind waves are related to wind speed, duration, and fetch. Fetch is the distance
over which the wind blows or causes shear over the water surface. As fetch increases, the
wavelength increases; long wavelengths are only produced in the presence of a long fetch.
The shortest wavelengths require only limited contact between wind and water. Waves
with a wavelength less than 2n cm (6.28 cm) are capillary waves, which are not important in the modeling of lakes and reservoirs. The more important gravity waves have
wavelengths longer than 2n cm. The two types of gravity waves are short waves and long
waves, distinguished by the interaction with the benthic boundary. The wavelength of
short waves seen by eye on lakes and reservoirs is much less than the water depth, and
they are not affected by bottom shear. Long waves, such as lake seiche, are influenced by
bottom friction. Seiches are periodic oscillations of the water surface and density interfaces resulting from a displacement.
Shortwave motion is circular in a vertical plane, making a complete revolution as each
successive wave passes. The orbital motion mixes surface layers or layers at an interface.
With no net advection of water, the overall effect is dispersive. Thus, the mixing terms in
transport and water quality models are generally increased to account for wave mixing,
especially in the epilimnion. In a few cases, specific mixing-length formulas (Kent and
Pritchard, 1957, Rossby and Montgomery, 1935) were derived for wave mixing, but these
formulas have not been applied in current models of water quality. No appreciable orbital
motion occurs below a depth of approximately one-half the wavelength in unstratified
flow, a depth referred to as the wind mixed depth. The wind mixed depth increases with
fetch because the wave height and wavelength increase with increasing fetch. This is illustrated by a simple relationship discovered by Lerman (1978) relating fetch to the depth of
the summer thermocline for a wide variety of lakes of different sizes and shapes.
As wavelength becomes longer in relation to the depth, or as water becomes shallower, wave orbits become increasingly flatter or elliptical. As the orbits flatten, the motion
of the water essentially becomes horizontal oscillation (Smith, 1975) so that the motion
of the water caused by, long waves is more advective rather than dispersive. For long
waves, the wave speed or celerity is c = (gY)05.
As short waves enter shallow water, the bottom affects orbital motion. From this point
inland to the line where wave breaking occurs, the depth is less than one-half the wave
period. In this shore zone, wave velocity decreases with the square root of the depth,
which results in a corresponding increase in wave height. Waves distort as water at the
crest moves faster than the wave, creating an instability. These unstable waves may eventually collapse, forming breakers or whitecaps, depending on the wave steepness of the
waves, the wind speed and direction, the direction of the waves, and the shape and rough-

ness of the bottom. A spilling breaker tends to form over a gradually shoaling bottom and
tends to break over long distances, with the wave collapsing downward in front of the
wave. Plunging breakers occur when the bottom shoals rapidly or when the direction of
the wind opposes the wave. The plunging breaker begins to curl and then collapses before
the curl is complete. A plunging or surging breaker does not break or collapse but forms
a steep peak as the wave moves up the beach. The type of breaking wave and the associated energy controls beach erosion, aquatic plant growth, surf-zone mixing, and the
exchange of contaminants between surface and ground waters.
After breaking, waves continue to move up a gradually sloping beach until the force
of gravity forces the water back. The extent to which the water runs up the beach is called
the swash zone. The movement of the swash up the beach may result in the deposition of
particles and debris, causing swash marks at the highest point of the zone. Wave run-up
in the swash zone also sets up an imbalance of momentum along the porous beach face
that pumps contaminants into and out of the beach (McCutcheon, 1989).
In large lakes and reservoirs with an extremely long fetch, parallel pairs of large vertical vortices or circulatory cells known as Langmuir circulation develop at an angle of
15 clockwise with the general direction of a sustained wind, when wave and current conditions are favorable. The depth of the vortices depends on stratification and may interact
with internal waves formed on the thermocline, deepening over the troughs of internal
waves. Where the counterrotating Langmuir cells converge, visible streaks or bands form
on the surface that tend to accumulate floating debris. In the convergence zone, downward
velocities of 2-6 cm/s carry surface waters toward the thermocline. These downward currents move in a circular fashion and turn upward into a divergence zone midway between
the Langmuir=streaks. Water near the thermocline moves to a zone near the surface at a
velocity of about 1 to 2 cm/s over a larger area. As first proposed by Langmuir (1938),
this type of large-scale circulation also contributes to the vertical mixing of the epilimnion. Like smaller-scale orbital wave mixing, the effect of Langmuir circulation is
lumped into values selected for the eddy viscosities and eddy diffusivities of the epilimnion.
Because of the smaller differences in density across density interfaces within a body of
water, internal waves travel more slowly than do surface waves, but they achieve greater
wave heights. Internal waves include standing waves, such as seiches (Mortimer, 1974) and
internal hydraulic jumps (French, 1985), but most are progressive waves that radiate energy
from the point at which the waves were generated (Ford and Johnson, 1986). Wind shear,
water withdrawals, hydropower releases, and thermal discharges as well as local disturbances produce internal waves. The most significant mixing between stratified layers occurs
when internal waves break (Turner, 1973). Before breaking, internal waves mix the water
adjacent to the interface and sharpen the density interface to increase the likelihood of breaking. When wave breaking does occur, the entrained water is mixed through the adjacent
layer.
Among the most important internal waves is the seiche. As defined above, seiches are
periodic oscillations of the water surface and density interfaces resulting from a displacement. Displacements are typically caused by large scale wind events or large withdrawals.
Sustained wind across a lake surface increases the elevation of the water surface at the
downwind boundary of the lake, causing wind setup. As the wind subsides, the water surface tilt or displacement results in a sloshing motion, or seiche, of the lake surface and in
thermocline if the lake is stratified. If hydropower operations or reservoir releases change
the net flow toward the dam, the water piles up at the dam and forms a seiche, often resulting in noticeable differences in thermocline depths between periods of operation and nonoperation, such as between weekdays and weekends. More rarely, a seiche may result

from earthquakes or other geologic events. During the rocking or sloshing, potential energy is converted to kinetic energy and is dissipated by bottom friction.
Wind setup in Lake Erie may exceed 2 m during severe storms (Wetzel, 1975), but for
a moderate storm blowing over the long axis of Green Bay, Wisconsin the wind setup has
reached approximately 12 cm (Martin and McCutcheon, 1999). An estimate of wind setup
can be obtained from the onedimensional equation of motion assuming constant depth,
negligible bottom stress, and steady-state conditions in an unstratified lake, or
ai = p^ = l
(5/70)
fa
spy
gy
where = the deviation of the water surface (m), x = the horizontal distance (m), pa =
the density of air (kg/m3), p = the density of water (kg/m3), C0 = the dimensionless drag
coefficient, uw = the wind speed (m/s), y = the water depth (m), and v* = the friction
velocity in water (m/s) or (Ts/p)05, in which T8 is the surface shear stress (kg/m- s2).
The term dfydx is positive in the direction of the wind. The divergence between wind and
shear force is negligible in shallow lakes and reservoirs but not in deep oceans.

5.6

PLUME AND JET HYDRAULICS

A jet is the discharge of a fluid from an opening into a large body of the same or similar fluid
that is driven by momentum. A plume is a flow that, while resembling a jet, is the result of
an energy source providing the fluid with positive or negative buoyancy rather than momentum relative to its surroundings. Many discharges into the environment are discussed in
terms of negatively or positively buoyant jets, implying that they derive from sources that
provide both momentum and buoyancy. In such cases, the initial flow is driven primarily by
the momentum of the fluid exiting the opening; however, if the exiting fluid is less or more
dense than the surrounding fluid, it is subsequently acted on by buoyancy forces.
Jets and plumes can be classified as either laminar or turbulent, with the difference
between the two being described by a Reynolds number, as with pipe flow. Near the
source of the flow, the flow of a jet or plume is controlled entirely by the primary initial
conditions that include the mean velocity of the jet's exit, the geometry of the exit, and
the initial difference in density between the discharge and the surrounding, or ambient,
fluid. Secondary initial conditions include the intensity of the exiting turbulence and the
distribution of the velocity. Following Fischer et al. (1979), the factors of prime importance to jet dynamics can be defined as follows:
1. Mass flux the mass of fluid passing a jet cross section per unit time
mass flux =

(pu)dA
(5.71)
JA
where A is the cross-sectional area of the jet and u the time-averaged velocity of the
jet in the axial direction.
2. Momentum flux amount of momentum passing a jet cross section per unit time
(pu2)dA
(5.72)
JA
3. Buoyancy flux buoyant or submerged weight of the fluid passing a jet cross section per unit of time
momentum flux =

buoyancy flux = J (gApw)JA

(5.73)

where Ap = the difference in density between the surrounding fluid and the fluid in the
jet. It is convenient to define g(Ap)/p = g' as the effective gravitational acceleration.

5.6.1 Simple Jets

The two dimensional or plane jet issuing from a slot and the round jet issuing from a nozzle into a quiescent ambient fluid are among the simplest cases of jets that can be considered. These jets have been studied extensively, and there is a reasonable understanding of
how they behave. The boundary between the ambient and jet fluids is sharp at any instant,
and if a tracer were present in the jet fluid, time-averaged measurements would show a
Gaussian distribution of tracer concentration (Q across the jet or
^exp-kM
(5.74)
c
L U
where the subscript m = the value of C on the jet axis, x = the distance along the jet axis,
kj = experimental coefficients, and y = the transverse (or radial) distance from the jet
axis. The Gaussian distribution also is valid for the time-averaged velocity profile across
the jet provided that the measurement is taken downstream of the zone of established flow.
In the case of a circular jet, the length of the zone of established flow is approximately 10
orifice diameters downstream.
Downstream of the zone of established flow, the jet continues to expand and the mean
velocity and tracer concentrations decrease. Within the zone of established flow, the
velocity and concentration profiles are self-similar and can be described in terms of a
maximum value (measured at the jet's center line) and a measure of the width or, in the
case of the velocity, distribution
t-'fe)
where vm = the value of v on the jet's center line, y = a coordinate transverse to the jet's
axis, and bw = the value of x at which v is some specified fraction of vm (often taken as
either 0.5 or 0.37; Fischer et al. 1979). The functional form of/in Eq, (5.75) is most often
taken as Gaussian.
Almost all the properties of turbulent jets that are important to engineers can be
deduced from dimensional analysis combined with empirical data (Fischer et al. 1979).
These results are summarized in Table 5.2.
5.6.2 Simple Plumes
Because the simple plume has no initial volume or momentum flux (e.g., smoke rising
from a fire), all variables must be a function of only the buoyancy flux (E), the vertical
distance from the origin (y), and the viscosity of the fluid where
B = g\^\Q = g'0Q

(5.76)

and Ap0 = difference in density between the plume fluid and the ambient fluid and g\ =
apparent gravitational acceleration.
Results similar to those for jets are summarized in Table 5.3, and the numerical constants given are from Chen and Rodi (1976).

TABLE 5.2 Summary of the Properties of Turbulent Jets

Parameter

Round Jet

Plane Jet

Initial volume flow rate

Q0

b0y0 V0

Initial momentum flux

M0

-4

^oVo^o

Characteristic length scale

1Q

VM

Maximum time-averaged
velocity
ym

Q
^\
vm =- = (7.0 O. Ij -
M
yj

Q
(j\
vw1 -^ = (2.41 0.04) -2"M
\yj

Maximum time-averaged
tracer concentration
Cm

^
fa\

-f
(5.6
O.I)U
C

V)

^
(^ \

-f
=
(2.38
0.04)
-^
C

V)

Mean dilution

-^- = (0.25 0.01) -f

^o
^cJ

(2\G0 = (0.50 0.02) -^

'G J

Ratio
C1-/^

, o

1.4 * 0.1

-^
M

1.2 0.1

Source: After Fischer et al. 1979.

TABLE 5.3 Summary of Plume Properties

Parameter

Round Plume

Plane Plume

Maximum time-averaged
velocity
vm

(4.7 0.2)B1/3 y1'3

1.66B1'3

Maximum time-average
tracer concentration
C1n

(9.1 0.5)M B1'3 v-5/3

2.38M B1'3 B1

Volume flux
Q

(0.15 0.015)1/3 y5/3

0.34 B^y

1.4 0.2

0.81 0.1

Ratio
Cn/Cavg
Source: After Fischer et al. 1979.

REFERENCES
Akiyama, J. and H. Stefan, "Onset of Underflow in Slightly Diverging Channels," Journal of the
Hydraulics Division, American Society of Civil Engineers 113(HY7), 1987, pp. 825-844.
Akiyama, J., and H. Stefan, "Plunging Flow into a Reservoir, Theory," Journal of the Hydraulics
Division, American Society of Civil Enginners, 110(HY4): 484^99, 1984.
Anderson, E. R. "Energy Budget Studies, Part of Water Loss Investigations Lake Hefner
Studies.", U.S. Geological Survey, Professional Paper No. 269, Washington, DC, 1954.
Beeton, A. M., "Relationship Between Secchi Disk Readings and Light Penetration in Lake
Huron," American Fisheries Society Transactions, 87:73-79, 1958.
Bender, M. D., G. E. Hauser, M. C. Shiao and W. D. Proctor, "BETTER: A Two-Dimensional
Reservoir Water Quality Model, Technical Reference Manual and User's Guide," Report No.
WR28-2-590-152, Tennessee Valley Authority, Engineering Laboratory, Norris, TN, 1990.
Bohan, J. P., and J. L. Grace, Jr., "Selective Withdrawal from Man-Made Lakes: Hydraulics
Laboratory Investigation," Technical Report No. H-73-4, U.S. Army Waterways Experiment
Station, Vicksburg, MS, 1973.
Bowie, G. L., W. B. Mills, D. B. Porcella, C. L. Campbell, J. R. Pagenkopf, G. L. Rupp, K. L.
Johnson, W. H. Chan, S. A. Gherini, and C. E. Chamberlin, "Rates, Constants, and Kinetics
Formulations in Surface Water Quality Modeling", 2nd ed. EPA/600/3-85/040, U.S.
Environmental Protection Agency, Environmental Research Laboratory, Athens, GA 1985.
Brown, L. C. and T. O. Barnwel, The Enhanced Stream Water Quality Models QUAL2E and
QUAL2E-UNCAS: Documentation and User Manual,. EPA/600/3-87/007, U.S. Environmental
Protection Agency, 1987.
Chapra, S. C., and K. H. Reckhow, Engineering Approaches for Lake Management, Butterworth,
Boston, 1983.
Chen, C. J., and W. Rodi, A Review of Experimental Data of Vertical Turbulent Buoyant Jets,
Hydraulic Research Report No. 193, Iowa Institute of Hydraulics, Iowa City, IA, 1976.
Cole, T. M., and E. M. Buchak, CE-QUAL-W2: A Two Dimensional, Laterally Averaged,
Hydrodynamic and Water Quality Model, Version 2.0, User Manual, Instruction Report No.
EL-95-1, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS, 1995.
Davis, J. E., J. P. Holland, M. L.Schneider, and S. C. Wilhelms, SELECT, A Numerical, One
Dimensional Model for Selective Withdrawal, Technical Report, U.S. Army Engineer Waterways
Experiment Station, Vicksburg, MS, 1985.
Environmental Laboratory, CE-QUAL-Rl: A Numerical One-Dimensional Model of Reservoir
Water Quality, User's Manual, Instruction Report No. E-82-1, U.S. Army Engineer Waterways
Experiment Station, Vicksburg, MS, 1985.
Fischer, H. B., "The Effects of Bends on Dispersion in Streams," Water Resources Research, 5
496-506, 1969.
Fischer, H. B., E. J. List, R. C. Y. Koh, J. Imberger, and N. H. Brooks, Mixing in Inland and
Coastal Waters, Academic Press, New York, 1979.
Ford, D. E., and M. C. Johnson, An Assessment of Reservoir Mixing Processes, Technical Report
No E-86-7, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS, 1986.
Ford, D. E., and M. C. Johnson, An Assessment of Reservoir Density Currents and Inflow
Processes, Technical Report No. E-83-7, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, MS, 1983.
Ford, D. E., and M. C. Johnson, "Field Observations of Density Currents in Impoundments," in H.
G. Stefan, ed. Proceedings Symposium on Surface Water Impoundments, ASCE, New York, 1981.
French, R. H., Open-Channel Hydraulics, McGraw-Hill, New York, 1985.
French, R. H., J. J. Cooper, and S. Vigg, "Secchi Disc Relationships," AWRA, Water Resources
Bulletin, 18 (1):121-123, 1982.
Gill, A. E., "Appendix 3, Properties of Seawater," in Atmosphere-Ocean Dynamics, Academic
Press, New York, 1982, pp. 599-600.

Gu, R., S. C. McCutcheon, and P. F. Wang, "Modeling Reservoir Density Underflow and Interflow
from a Chemical Spill in a River," Water Resources Research, 32:695-705, 1996.
Grace, J. L., Jr., "Selective Withdrawal Characteristics of Weirs: Hydraulics Laboratory
Investigation," Technical Report H-71-4, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, MS, 1971.
Hamon, R. W, L. L. Weiss, and W. T. Wilson, "Insulation as an Empirical Function of Daily
Sunshine Duration," Monthly Weather Review, 82(6), 1954.
Hebbert, J., J. Imberger, I. Loh, and J. Patterson, "Collie River Flow into Wellington Reservoir,",
Journal of the Hydraulics Division, American Society of Civil Engineers 105(HY(5): 533-545,
1979.
Hutchinson, G. E., A Treatise on Limnology: Vol. 1. Geography, Physics and Chemistry, John
Wiley Sons, New York, 1957.
Imberger, J., "Selective Withdrawal: A Review," in T, Carstens and T. McClimans, eds.,
Proceedings of the 2nd International Symposium on Stratified Flow, International Association for
Hydraulic Research, Tapir, Trondheim, Norway, 1980.
Imberger, J. and J. C. Patterson, "Dynamic Reservoir Simulation Model DYRESM:5," in H.B.
Fischer, ed., Transport Models for Inland and Coastal Waters, Academic Press, Orlando, FL,
1981, pp. 310-561.
Imberger, J., R. T. Thompson, and C. Fandry, "Selective Withdrawal from a Finite Rectangular
Tank," Journal of Fluid Mechanics, 78:489-512, 1976.
Jain, S. C., "Plunging Phenomena in Reservoirs," in H. G. Stefan, ed., Proceedings of the
Symposium on Surface Water Impoundments, American Society of Civil Engineers, new York,
1981.
Jobson, H. E., "Thermal Modeling of Flow in the San Diego Aqueduct, California, and Its Relation
to Evaporation," Professional Paper No. 1122, U.S. Geological Survey, Washington, DC, 1980.
Jobson, H. E., and T. N. Keefer, "Modeling Hight Transient Flow, Mass, and Heat Transport in the
Chattahoochee River Near Atlanta, Georgia," Professional Paper No. 1136, U.S. Geological
Survey, Washington, DC, 1979.
Kent, R. E. and D. W. Pritchard, "A Test of Mixing Length Theories in a Coastal Plain Estuary,"
Journal of Marine Research, 1:456-466, 1957.
Kao, T. W, "Principal State of Wake Collapse in a Stratified Fluid, Two-Dimensional Theory,"
Physics of Fluids, 19:1071-1074, 1976.
Langmuir, L, "Surface Motion of Water Induced by Wind," Science, 87:119-123, 1938.
Lau, Y. L., and B. G. Krishnappen, "Transverse Dispersion in Rectangular Channels," ASCE,
Journal of the Hydraulics Division, American Society of Civil Engineers 103(HYlO): 1173-1189,
1977.
Lerman, A., ed, Lakes: Chemistry, Geology, Physics, Springer-Verlag, New York, 1978.
Martin, J. L. and S. C.McCutcheon, Hydrodynamics and Transport for Water Quality Modeling,
Lewis Publishes-Boca Raton, FL, 1999.
McCutcheon, S. C., Water Quality Modeling: Vol. I. Transport and Surface Exchange in Rivers,
CRC Press, Boca Raton, FL, 1989.
McCutcheon, S. C., "Discussion with Harvey Jobson on Windspeed Coefficients for Stream
Temperature Modeling," Memorandum, U.S. Geological Survey, NSTL Station, MS, March 29,
1982.
Mills, W. B., J. D. Dean, D. B. Porcella, S. A. Gherini, R. J. M. Hudson, W. E. Frick, G. L. Rupp,
and G. L. Bowie, "Water Quality Assessment: A Screening Procedure for Toxic and Conventional
Pollutants," EPA-600/6-82-004b, U.S. Environmental Protection Agency, Athens, GA, 1982.
Mortimer, C. H., "Lake Hydrodynamics," International Association of Applied Limnology
Mitteilungen, 20:124-197, 1974.
Munk, W., and E. R. Anderson, "Notes on a Theory of the Thermocline," Journal of Marine
Research, 7:276-295, 1948.

Norton, W. R., L. A. Roesner, and G. T. Orlob, Mathematical Models for Predicting Thermal
Changes in Impoundments, EPA Water Pollution Control Research Series, U.S. Environmental
Protection Agency, Washington, DC, 1968.
Rossby, G. G. and R. Montgomery, "The Layers of Frictional Influence in Wind and Ocean
Currents," Papers Phy. Oc. Meth., 111(3), 1935.
Ryan, P. J. and D. R. F. Harleman, An Analytical and Experimental Study of Transient Cooling
Pond Behavior. Technical Report No. 161, R.M. Parsons Laboratory, Massachusetts Institute of
Technology, 1973.
Safaie, B."Mixing of Buoyant Surface Jet over Sloping Bottom," ASCE, Journal Waterway,
Port, Coastal and Ocean Engineering Division, 105(WW4): 357-373, 1979.
Savage, S. B. and J. Brimberg, "Analysis of Plunging Phenomenon of Density Currents in
Reservoirs," IAHR, QQ, 13(2): 187-204, 1973.
Shanahan, P., "Water Temperature Modeling: A Practical Guide," in Proceedings of the U.S.
Environmental Protection Agency Stormwater and Water Quality Users Group Meeting, LAMR,
jounal, Hydraulic, Research, , April 1984.
Shuttleworth, W. J., "Evaporation," in D. R. Maidment, ed., Handbook of Hydrology, McGrawHill, New York, 1993.
Singh, B., and C. R. Shah, "Plunging Phenomenon of Density Currents in Reservoirs," La Houille
Blanche, 26(1): 59-64, 1971.
Smith, I. R., Turbulence in Lakes and Rivers, Scientific Publication No. 29, Freshwater Biological
Association, Ambleside, Cumbria, UK, 1975.
Smith, J. A., "Precipitation," in D. R. Maidment, ed., Handbook of Hydrology, McGraw-Hill, New
York, 1993.
Smith, D. R., S. C. Wilhelms, J. P. Holland, M. S. Dortch, and J. E. Davis, Improved Description of
Selective Withdrawal Through Point Sinks, Technical Report No, E-87-2, U.S. Army Engineer
Waterways Experiment Station, Vicksburg, MS, 1985.
Stefan, H. G., R. B. Ambrose, Jr., and M. S. Dortch, "Formulation of Water Quality Models for
Streams, Lakes and Reservoirs: Modeler's Perspective," Miscellaneous Paper E-89-1, .S. Army
Engineer Waterways Experiment Station, Vicksburg, MS., 1989.
Sundaram, T. R., "A Theoretical Model for Seasonal Thermocline Cycle of Deep Temperate
Lakes," Proceedings of the 16th Conference of Great Lakes Research, 1973, pp. 1009-1025.
Sverdrup, H. U., M. W. Johnson, and R. H. Fleming, The Oceans, Prentice-Hall, Englewood Cliffs,
NJ, 1942.
Swinbank, W. C., "Longwave Radiation from Clear Skies." Quarterly Journal of the Royal
Meteorological Society, 89: 339-348, 1963.
Taylor, G. L, " The Dispersion of Matter in a Turbulent Flow Through a Pipe," Proceedings of the
Royal Society of London, Series A, 223: 446-468, 1954.
Thackston, E. L., Effect of Geographical Variation on Performance of Recirculating Cooling
Ponds, EPA-660/2-74-085, U.S. Environmental Protection Agency, Corvallis, OR, 1974.
Tennessee Valley Authority,(TVA) "Heat and Mass Transfer Between a Water Surface and the
Atmosphere." TN Report No. 14, TVA Water Resources Research Engineering Laboratory,
Norris, TN, 1972.
Turner, J. S., Buoyancy Effects in Fluids, Cambridge University Press, Cambridge, UK, 1973.
Wetzel, R. G., Limnology, Saunders College Publishing, Philadelphia, PA, 1975, and 1983.
Wunderlich, W. O. and R. A. Elder, "Mechanics of Flow Through Man-Made Lakes," in ManMade Lakes: Their Problems and Environmental Effects, W. C. Ackerman, G. F. White, and E. B.
Worthington, eds., American Geophysical Union, Washington, DC, 1973.

CHAPTER 6
SEDIMENTATION AND
EROSION HYDRAULICS

Marcelo H. Garcia
Department of Civil and Environmental Engineering
University of Illinois at Urbana-Champaign
Urbana, IL

6.1

INTRODUCTION

Since the beginning of mankind, sedimentation processes have affected water supplies,
irrigation, agricultural practices, flood control, river migration, hydroelectric projects,
navigation, fisheries, and aquatic habitat. In the last few years, sediment also has been
found to play an important role in the transport and fate of pollutants; thus, sedimentation
control has become an important issue in water quality management. Toxic chemicals can
become attached to, or adsorbed by, sediment particles and then be transported to and
deposited in other areas. By studying the quantity, quality, and characteristics of sediment
in rivers and streams, scientists and engineers can determine the sources of the sediment
and evaluate the impact of pollutants on the aquatic environment. In the United States,
sedimentation control is a multibillion-dollar issue. For example, approximately $500
million are spent every year to dredge waterways and harbors for navigation purposes.
Most of the dredged sediment is the result of substantial soil erosion in watersheds.
Estimates by the U.S. Department of Agriculture indicate that annual offside costs of sediment derived from Copland erosion are on the order of $2 billion to $6 billion, with an
additional $1 billion arising from loss in compared productivity.
The sediment cycle starts with the process of erosion, where by particles or fragments
are weathered from rock material. Action by water, wind, glaciers, and plant and animal
activities all contribute to the erosion of the earth's surface. Fluvial sediment is the term
used to describe the case where water is the key agent for erosion. Natural, or geologic,
erosion takes place slowly, over centuries or millennia. Erosion that occurs as a result of
human activity may take place much faster. It is important to understand the role of each
cause when studying sediment transport.
Any material that can be dislodged is ready to be transported. The transportation
process is initiated on the land surface when raindrops result in sheet erosion. Rills, gullies, streams, and rivers then act as conduits for the movement of sediment. The greater
the discharge, or rate of flow, the higher the capacity for sediment transport.
The final process in the cycle is deposition. When there is not enough energy to transport the sediment, it comes to rest. Sinks, or depositional areas, can be visible as newly
deposited material on a floodplain, on bars and islands in a channel, and on deltas.
Considerable deposition occurs that may not be apparent, as on lake and river beds. A
knowledge of sediment dynamics is an integral part of understanding the aquatic ecosystem.
This chapter presents fundamental aspects of the erosion, transport, and deposition of
sediment in the environment. The emphasis is on the hydraulics of bedload and suspend-

ed load transport in rivers, with the goal of establishing the background needed for sedimentation engineering. Because of their relevance, the hydraulics of both reservoir sedimentation and turbidity currents also is considered. Emphasis is placed on noncohesive
sediment transport, where the material involved can be silt, sand, or gravel. When possible, the behavior of both uniform-sized material and sediment mixtures is analyzed.
Although such topics as cohesive sediment transport, debris and mud flows, alluvial fans,
river meandering, and sediment transport by wave action are not discussed here, it is
hoped that the material covered in this chapter will provide a firm foundation to tackle
problems in those.
For more information on sediment transport and sedimentation engineering, readers
are referred to Allen (1985), Ashworth et al. (1996), Bogardi (1974), Bouvard (1992),
Carling and Dawson (1996), Chang (1988), Coussot (1997), Freds0e and Deigaard
(1992), Garde and Ranga Raju (1985), Graf (1971), Jansen et al. (1979), Julien (1992),
Mehta (1986), Mehta et al. (1989a, 1989b), Morris and Fan (1998), Nakato and Ettema
(1996), National Research Council (1996), Nielsen (1992), National Research council
(1996), Parker and Ikeda (1989), Raudkivi (1990, 1993), Renard et al. (1997), Sieben
(1997), Simons and Senturk (1992), Sloff (1997), van Rijn (1997), Yalin (1972, 1992),
Yang (1996), and Wan and Wang (1994).

6.2 HYDRAULICSFORSEDIMENTTRANSPORT
6.2.1 Flow Velocity Distribution
Consider a steady, turbulent, uniform, open-channel flow having a mean depth H and a
mean flow velocity U (Fig. 6.1). The channel is extremely wide and its bottom has a mean
slope S and a surface roughness that can be characterized by an effective height ks
(Brownlie, 198Ib). When the bottom of the channel is covered with sediment having a
mean size or diameter D, the roughness height ks will be proportional to that diameter.
Because of the weight of the water, the flow exerts on the bottom a tangential force per
unit bed area known as the bed shear stress ib, which can be expressed as:
T* = PgHS

(6.1)

where p is the water density and g is the gravitational acceleration. With the help of the
boundary shear stress, it is possible to define the shear velocity u* as

FLOW

FIGURE 6.1 Definition diagram for open-channel flow over an credible bed.

M. = VVP

(6.2)

The shear velocity, and thus the boundary shear stress, provides a direct measure of the
intensity of flow and its ability to entrain and transport sediment particles. The size of the
sediment particles on the bottom determines the surface roughness, which in turn affects
the flow velocity distribution and its sediment transport capacity. Since flow resistance
and sediment transport rates are interrelated, the ability to determine the role played by
the bottom roughness is important.
Research has shown (Schlichting, 1979) that the flow velocity distribution is well represented by:
JL = I ]nz + const.

(6.3)

where u is the time-averaged flow velocity at distance z above the bed and K is known as
von Karman's constant and is equal to 0.4. For obvious reasons, the above law is known
as the logarithmic law of the wall. It strictly applies only in a thin layer near the bed. It is
empirically found to apply as a reasonable approximation throughout most of the flow in
many rivers.
If the bottom boundary is sufficiently smooth (a condition rarely satisfied in rivers),
turbulence will be drastically suppressed in an extremely thin layer near the bed. In this
region, a linear velocity profile will hold:
^- = ^
(6.4)
W*
V
where v is the kinematic viscosity of water. This law merges with the logarithmic law near
z 5V, where

5 V =11.6-^-

(6.5)

denotes the height of the viscous sublayer. In the logarithmic region, the constant of integration introduced above has been evaluated from data to yield
J. = Ih IWl+5.5
U* K ( V )

(6.6)

Most boundaries in river flow are rough. Let ks denote an effective roughness height.
If kjbv > 1, then no viscous sublayer will exist. The corresponding logarithmic velocity
profile is given by
t=Mt) + 8 - 5 = H 3 t)

(6 7)

As noted above, this relation often holds as a first approximation throughout the flow in
a river. It is by no means exact.
The conditions kjbv 1 for rough turbulent flow and kjbv 1 for smooth turbulent
flow can be rewritten to indicate that u*kjv should be much larger than 11.6 for turbulent
rough flow and much smaller than 11.6 for turbulent smooth flow. A composite form that
represents both ranges, as well as the transitional range between them, can be written as
i = Hf) +B *
with Bs as a function of Re* = u*kjv, which can be estimated with

(6 8)

'

B5 = 8.5 + [2.5 ]n(Re.) - 3]e--*v2

(6.9)

as proposed by Yalin (1992).

6.2.2 Relations for Channel Resistance
Most river flows are indeed hydraulically rough. Equation (6.7) can be used to obtain an
approximate expression for depth-averaged velocity U that is reasonably accurate for
many flows. Using the following integral:
H
U = -ludz
(6.10)
H o
but changing the lower limit slightly to avoid the fact that the logarithmic law is singular
at z = O, the following result is obtained:
Z-B(MH*

"

or, performing the integration

^ = KIJfU 6 =K IlJlIfI
.
l*J
I *J

(6.12)

This relation is known as Keulegan 's resistance relation for rough flow.
An approximation to Keulegan's relation is the Manning-Strickler power form
^= 8
.
IAJ

(6-13)

Between Eqs. (6.2) and (6.12), a resistance relation can be found for bed shear stress:
T6 = pC/t/*

(6.14)

where the friction coefficient Cf is given by

<HH"fF

(6

-i5)

IfEq. (6.13) is used instead of Eq. (6.12), the friction coefficient takes the form
c

,=Kf)T

It is useful to show the relationship between the friction coefficient Cf and the roughness parameters in open-channel flow relations commonly used in practice. Between Eqs.
(6.1) and (6.14), the following form of Chezy's law can be derived:
U=CJI112S112
where the Chezy coefficient Cc is given by the relation

(6.17)

[ fSvS \I2

(6.18)

A specific evaluation of Chezy's coefficient can be obtained by substituting Eq. (6.15) into
Eq. (6.18). It is seen that the coefficient is not constant but varies as the logarithm ofH/ks.
A logarithmic dependence is typically a weak one, partially justifying the common
assumption that Chezy's coefficient in Eq. (6.17) is a constant. Substituting Eq. (6.16) into
Eqs. (6.17) and (6.18), Manning's law is obtained:
U = !#2/351/2
n

(6.19)
^ '

1/6
=8^

(6-2)

where Manning's n is given by

The above relation is often called the Manning-Strickler form of Manning's n.

6.2.3 Fixed-Bed and Movable-Bed Roughness
It is clear that to use the above relations for channel flow resistance, a criterion for evaluating ks is necessary. Nikuradse (1933) proposed the following criterion: Suppose a
rough surface is subjected to a flow. The equivalent roughness ks of that surface is equal
to the diameter of sand grains that, when glued uniformly to a completely smooth wall and
then subjected to the same external conditions, yields the same velocity profile. Nikuradse
used sand glued to the inside of pipes to conduct this evaluation. Extending Nikuradse's
concept of equivalent grain roughness to the case of rivers and streams, ks can be assumed
to be proportional to a representative sediment size Dx,
ks = af>x

(6.21)

Suggested values of OC5, which have appeared in the literature, are listed in Table 6.1 (Yen,
1992). Different sizes of sediment have been suggested for Dx in Eq. (6.21). Statistically, D50
(the grain size for which 50% of the bed material is finer) is most readily available and
meaningful. Physically, a representative size larger than D50 is more meaningful to estimate
TABLE 6.1 Ratio of Nikuradse Equivalent Roughness Size and Sediment Size for Rivers.
Investigator
Ackers and White (1973)
Strickler (1923)
Keulegan (1938)
Meyer-Peter and Muller (1948)
Thompson and Campbell (1979)
Hammond et al. (1984)
Einstein and Barbarossa (1952)
Irmay (1949)
Engelund and Hansen (1967)
Lane and Carlson (1953)

Measure of Sediment Size, Dx

D35
D50
D50
D50
D50
D50
D65
D65
D65
D75

(X^ = ks/Dx
1.23
3.3
1
1
2.0
6.6
1
1.5
2.0
3.2

TABLE 6.1. (Continued)

Investigator

Measure of Sediment Size, Dx

Gladki (1979)
Leopold et al. (1964)
Limerinos (1970)
Mahmood(1971)
Hey (1979), Bray (1979)
Ikeda(1983)
Colosimo et al. (1986)
Whiting and Dietrich (1990)
Simons and Richardson (1966)
Kamphuis (1974)
van Rijn (1982)

D80
D84
D84
D84
D84
D84
D84
D84
D85
D90
D90

ens = ks/Dx
2.5
3.9
2.8
5.1
3.5
1.5
3~6
2.95
1
2.0
3.0

SOURCE: Adapted from Yen (1992)

flow resistance because of the dominant effect by large sediment particles.
In flow over a geometrically smooth, fixed boundary, the apparent roughness of the
bed ks can be computed using Nikuradse's approach. However, once the transport of bed
material has been instigated, the characteristic grain diameter and the thickness of the viscous sublayer no longer provide the relevant length scales. The characteristic length scale
in this situation is the thickness of the layer where the sediment particles are being transported by the flow, usually referred to as the bedload layer.
Once the bed shear stress ib exceeds the critical shear stress for particle motion Tc, the
apparent bed roughness ka can be estimated as follows (Smith and McLean, 1997):
(6 22)
* = * ^ + k
where OC0 = 26.3, ks is Nikuradse's fixed-bed roughness, and p5 is the bed sediment density. This approach is particularly suitable for sand bed rivers.
Under intense sediment transport conditions, bedforms, such as dunes, can develop. In
this situation, the apparent roughness also will be influenced by the form drag caused by
the presence of bedforms. Nikuradse's approach is valid only for grain-induced roughness.
Methods for flow resistance in the presence of both bedforms and grain roughness are presented later.

6.3

SEDIMENTPROPERTIES

6.3.1 Rock Types

The solid phase of the problem embodied in sediment transport can be any granular substance. In engineering applications, however, the granular substance in question typically
consists of fragments ultimately derived from rockshence the name sediment transport.
The properties of these rock-derived fragments, taken singly or in groups of many particles, all play a role in determining the transportability of the grains under fluid action. The

important properties of groups of particles include porosity and size distribution. The most
common rock type one is likely to encounter in the river or coastal environment is quartz.
Quartz is a highly resistant rock and can travel long distances or remain in place for long
periods without losing its integrity. Another highly resistant rock type that is often found
together with quartz is feldspar. Other common rock types include limestone, basalt, granite, and more esoteric types, such as magnetite. Limestone is not a resistant rock; it tends
to abrade to silt rather easily. Silt-sized limestone particles are susceptible to solution
unless the water is buffered sufficiently. As a result, limestone typically is not a major
component of sediments at locations distant from its source. On the other hand, it often
can be the dominant rock type in mountain environments.
Basaltic rocks tend to be heavier than most rocks composing the earth's crust and typically are brought to the surface by volcanic activity. Basaltic gravels are relatively common in rivers that derive their sediment supply from areas subjected to vulcanism in
recent geologic history. Basaltic sands are much less common. Regions of weathered
granite often provide copious supplies of sediment. Although the particles produced by
weathering are often in the granule size, they often break down quickly to sand size.
Sediments in the fluvial or coastal environment in the size range of silt, or coarser, are
generally produced by mechanical means, including fracture or abrasion. The clay minerals, on the other hand, are produced by chemical action. As a result, they are fundamentally different from other sediments in many ways. Their ability to absorb water means
that the porosity of clay deposits can vary greatly over time. Clays also display cohesivity, which renders them more resistant to erosion.
6.3.2 Specific Gravity
The specific gravity of sediment is defined as the ratio between the sediment density ps
and the density of water p. Some typical specific gravities for various natural and artificial sediments are listed in Table 6.2.
6.3.3

Size

Herein, the notation D is used to denote sediment size, the typical units of which are
millimeters (mm) for sand and coarser material or microns (JLI) for clay and silt.
Another standard way of classifying grain sizes is the sedimentological <3> scale,
according to which
TABLE 6.2 Specific Gravity of Rock Types
and Artificial Material
Rock type or
materia/
quartz
limestone
basalt
magnetite
plastic
coal
walnut shells

Specific gravity
p/p
2.60 ~ 2.70
2.60 ~ 2.80
2.70 ~ 2.90
3.20 ~ 3.50
1.00 ~ 1.50
1.30-1.50
1.30- 1.40

D = 2*

(6.23)

Taking the logarithm of both sides, it is seen that

O = - log2(D) = - Igg-

(6.24)

Note that the size O = O corresponds to D = 1 mm. The usefulness of the O scale will
become apparent upon a consideration of grain size distributions. The minus sign has been
inserted in Eq. (6.24) simply as a matter of convenience to sedimentologists, who are
more accustomed to working with material finer than 1 mm than they are with coarser
material. The reader should always recall that larger O implies finer material. The O scale
provides a simple way of classifying grain sizes into the following size ranges in descending order: boulders, cobbles, gravel, sand, silt, and clay. (Table 6.3).
Note that the definition of clay according to size (D < 2JJL) does not always correspond
to the definition of clay according to mineral. That is, some clay-mineral particles can be
coarser than this limit, and some silt-sized particles produced by grinding can be finer than
that. In general, however, the effect of viscosity makes it difficult to grind up particles in
water to sizes finer than 2 JJL.
In practical terms, there are several ways to determine grain size. The most popular
way for grains ranging from O= -4 to O = -4 (0.0625 to 16 mm) is with the use of
sieves. Each sieve has a square mesh, the gap size of which corresponds to the diameter
of the largest sphere that would fit through it. Thus, the grain size D so measured corresponds exactly to the diameter only in the case of a sphere. In general, the sieve size D
corresponds to the smallest sieve gap size through which a given grain can be fitted.
For coarser grain sizes, it is customary to approximate the grain as an ellipsoid. Three
lengths can be defined. The length along the major (longest) axis is denoted as a, the
length along the intermediate axis is denoted as b, and the length along the minor (smallest) axis is denoted as c. These lengths are typically measured with a caliper. The value b
is then equated to grain size D.
For grains in the silt and clay sizes, many methods (hydrometer, sedigraph, and so
forth) are based on the concept of equivalent fall diameter. That is, the terminal fall velocity vs of a grain in water at a standard temperature is measured. The equivalent fall diameter D is the diameter of the sphere having exactly the same fall velocity under the same
conditions. Sediment fall velocity is discussed in more detail below.
A variety of other more recent methods for sizing fine particles rely on blockage of
light beams. The blocked area can be used to determine the diameter of the equivalent circle: i.e., the projection of the equivalent sphere. It can be seen that all the above methods
can be expected to operate consistently as long as grains shape does not deviate too greatly from a sphere. In general, this turns out to be the case. There are some important exceptions, however. At the fine end of the spectrum, mica particles tend to be platelike; the
same is true of shale grains at the coarser end. Comparison with a sphere is not necessarily an especially useful way to characterize grain size for such materials.
6.3.4 Size Distribution
Any sample of sediment normally contains a range of sizes. An appropriate way to characterize these samples is by grain size distribution. Consider a large bulk sample of sediment of given weight. Let Pj(D)or p/O)denote the fraction by weight of material in
the sample of material finer than size >(O). The customary engineering representation of

TABLE 6.3 Sediment Grade Scale

Class Name

Size Range
Millimeters

Very large boulders

4,096 - 2,048
Large boulders
2,048 - 1,024
Medium boulders
1,024 - 512
Small boulders
512 - 256
Large cobbles
256 - 128
128-64
Small cobbles
Very coarse gravel
64-32
32- 16
Coarse gravel
Medium gravel
16-8
8-4
Fine gravel
Very fine gravel
4-2
Very coarse sand
2.000 - 1.000
Coarse sand
1.000 - 0.500
Medium sand
0.500 - 0.250
Fine sand
0.250 - 0.125
Very fine sand
0.125 - 0.062
Coarse silt
0.062 - 0.031
Medium silt
0.031 - 0.016
Fine silt
0.016 - 0.008
Very fine silt
0.008 - 0.004
Coarse clay
0.004 - 0.0020
Medium clay
0.0020 - 0.0010
Fine clay
0.0010 - 0.0005
Very fine clay
0.0005 - 0.00024
SOURCE: AdaptedfromVanoni, 1975.

-9- -8
-8
7
-7
6
-6 --5
-5- -4
-4- -3
-3- -2
-2- -1
-1-0
0-1
1 -2
2-3
3-4
4-5
5-6
6-7
7-8
8-9

Microns

Inches
160 - 80
80-40
40-20
20- 10
10-5
5-2.5
2.5 - 1.3
1.3 - 0.6
0.6 - 0.3
0.3 - 0.16
0.16 - 0.08

2,000 - 1,000
1,000 - 500
500 - 250
250 - 125
125 - 62
62-31
31 - 16
16-8
8-4
4-2
2- 1
1 -0.5
0.5 - 0.24

Approximate Sieve Mesh

Openings per Inch
Tyler
U.S. standard

2- 1/2
5
9
16
32
60
115
250

5
10
18
ac
35
60
120
230

C0
n>Qg

|.

I
2.

3
OI
2
S

the grain size distribution consists of a plot of pf *100 (percentage finer) versus log10(D):
that is, a semilogarithmic plot is used. The same size distribution plotted in sedimentological form would involve plotting /y 100 versus O on a linear plot.
The size distribution p/O) and size density /?(O) by weight can be used to extract useful statistics concerning the sediment in question. Let x denote some percentage, say 50%;
the grain size Ox denotes the size such that x percent of the weight of the sample is composed of finer grains. That is, O^ is defined such that
(6 25)
PW = jfe
It follows that the corresponding grain size of equivalent diameter is given by Dx, where

Dx = 2 -**

(6.26)

The most commonly used grain sizes of this type are the median size D50 and the size
Z)90: i.e., 90% of the sample by weight consists of finer grains. The latter size is especially useful for characterizing bed roughness.
The density p(O) can be used to extract statistical moments. Of these, the most useful
are the mean size Om and the standard deviation a. These are given by the relations.
O1n = /Op(O)JO;

a2 = (O - OJ2p(O)DO

(6.27a, b)

The corresponding geometric mean diameter Dg and geometric standard deviation og

are given as
Dg = 2-^;

(5g = 2

(6,28a,b)

Note that for a perfectly uniform material, a = O and ag = 1. As a practical matter, a sediment mixture with a value of Gg less than 1.3 is often termed well sorted and can be treated as a uniform material. When the geometric standard deviation exceeds 1.6, the material can be said to be poorly sorted (Diplas and Sutherland, 1988).
In fact, one never has the continuous function p(O) with which to compute the
moments of Eqs. (6.27a, and b). Instead, one must rely on a discretization. To this end, the
size range covered by a given sample of sediment is discretized using n intervals bounded by n + 1 grain sizes O1, O2,..., On + l in ascending order of O. The following definitions are made from / = 1 to n:
> = ^t + /+i)

(6-29a)

P1=Pj(Q)-Pf(^+1)

(6.29b)

Eqs. (6.27a and b) now discretize to

n
n
*m = E *tPt 2 = E (*i - ^m)2P,
(6.30)
I= 1
I= 1
In some cases, especially when the material in question is sand, the size distribution
can be approximated as gaussian on the O scale (i.e., log-normal in D). For a perfectly
Gaussian distribution, the mean and median sizes coincide:
* = *50 = ^84 + *,)

(6-31)

Furthermore, it can be demonstrated from a standard table of the Gauss distribution that
the size O displaced one standard deviation larger that Om is accurately given by O84; by
symmetry, the corresponding size that is one standard deviation smaller than Om is O16.
The following relations thus hold:
a = ^(O84 - O16)

(6.32a)

m = |(<*>84 + *16>

(6-32b)

Rearranging the above relations with the aid of Eqs. (6.28a and b) and Eqs. (6.31 and
6.32a),
.=f
Ds = (D84T)16)"*

(6.33b)

It must be emphasized that the above relations are exact only for a gaussian distribution
in O. This is not often the case in nature. As a result, it is strongly recommended that Dg
and Gg be computed from the full size distribution via Eqs. (6.3Oa and b) and (6.28a and
b) rather than the approximate form embodied in the above relations.
6.3.5 Porosity
The porosity Xp quantifies the fraction of a given volume of sediment that is composed of
void space. That is,
, _
volume of voids
P volume of total space
If a given mass of sediment of known density is deposited, the volume of the deposit
must be computed, assuming that at least part of it will consist of voids. In the case of
well-sorted sand, the porosity often can take values between 0.3 and 0.4. Gravels tend to
be more poorly sorted. In this case, finer particles can occupy the spaces between coarser
particles, thus reducing the void ratio to as low as 0.2. Because so-called open-work gravels are essentially devoid of sand and finer material in their interstices, they may have
porosities similar to sand. Freshly deposited clays are notorious for having high porosities. As time passes, the clay deposit tends to consolidate under its own weight so that
porosity slowly decreases.
The issue of porosity becomes of practical importance with regard to salmon spawning grounds in gravel-bed rivers, for example (Diplas and Parker, 1985). The percentage
of sand and silt contained in the sediment is often referred to as the percentage of fines in
the gravel deposit. When this fraction rises above 20 or 26 percent by weight, the deposit
is often rendered unsuitable for spawning. Salmon bury their eggs within the gravel, and
a high fines content implies a low porosity and thus reduced permeability. The flow of
groundwater necessary to carry oxygen to the eggs and remove metabolic waste products
is impeded. In addition, newly hatched fry may encounter difficulty in finding enough
pore space through which to emerge to the surface. All the above factors dictate lowered
survival rates. Chief causes of elevated fines in gravel rivers include road building and
clear-cutting of timber in the basin.

6.3.6

Shape

Grain shape can be classified in a number of ways. One of these, the Zingg classification
scheme, is illustrated here (Vanoni, 1975). According to the definitions introduced earlier,
a simple way to characterize the shape of an irregular clast (stone) is by lengths a, b, and
c of the major, intermediate, and minor axes, respectively. If the three lengths are equal,
the grain can be said to be close to a sphere in shape. If a and b are equal but c is much
larger, the grain should be rodlike. Finally, if c is much smaller than b, which in turn, is
much larger than a, the resulting shape should be bladelike.
6.3.7

Fall Velocity

A fundamental property of sediment particles is their fall velocity. The relation for terminal fall velocity in quiescent fluid vs can be presented as

*-[hy

where

Rff = ^VR^D

(6.35a)

RP=V-^-

(6.35b)

and the functional relation C0 = CD(Rp) denotes the drag curve for spheres. This relation
is not particularly useful because it is not explicit in V5; one must compute fall velocity by
trial and error. One can use the equation for C0 given below
C0 = ^ (1 + 0.152/?/'2 + 0.015 lRp)
P

(6.36)

and the definition

Rep = ^j^

(6.37)

to obtain an explicit relation for fall velocity in the form of Rf versus Rep. In Fig. 6.2, the
ranges for silt, sand, and gravel are plotted for v = 0.01 cm2/s (clear water at 2O0C) and
R = 1.65 (quartz). A good summary of relations for terminal fall velocity for the case of
nonspherical (natural) particles can be found in Dietrich (1982), who also proposed the
following useful fit:
Rf = expi-b, + b2\n(Rep) - b,[\n(RepW ~ b4[\n(RepW + b5\\n(Re^Y}

(6.38)

where b, = 2.891394, b2 = 0.95296, b3 = 0.056835, b4 = 0.002892, and b5 = 0.000245

6.3.8 Relation Between Size Distribution and Stream Morphology
The study of sediment properties and, in particular, size distribution is most relevant to the
context of stream morphology. The following discussion points out some of the more
interesting issues.

FIGURE 6.2 Sediment fall velocitydiagram

PERCENT FINER OY HEIGHT

In Fig. 6.3, several size distributions from the sand-bed Kankakee River in Illinois, are
shown (Bhowmik et al., 1980). The characteristic S shape suggests that these distributions
might be approximated by a gaussian curve. The median size D50 falls near 0.3 to 0.4 mm.
The distributions are tight, with a near absence of either gravel or silt. For practical purposes, the material can be approximated as uniform.
In Fig. 6.4, several size distributions pertaining to the gravel-bed Oak Creek in
Oregon, are shown (Milhous, 1973). In gravel-bed streams, the surface layer ("armor" or
"pavement") tends to be coarser than the substrate (identified as "subpavement" in the figure). Whether the surface or substrate is considered, it is apparent that the distribution
ranges over a much wider range of grain sizes than is the case in Fig. 6.3. More specifically, in the distributions of the sand-bed Kankakee River, O varies from about O to about

I linois - Indiana
State Line
3 31/79 - 4/21/79

PERCENT FIHER OY WEIGHT

GRAIN SIZE, mm

Illinois - Indiana
State Line
4/29/79 - 9/14/79

Pavement
Sub pavement
Composite of Pavement

and Subpavement

Sediment Size in Millimeters

FIGURE 6.4 Size distribution of bed material samples in Oak Creek. Oregon. Source: (Milhous, 1973)

PERCENTAGE FINER (Per cent)

3, whereas in Oak Creek, O varies from about -8 to about 3. In addition, the distribution
of Fig. 6.4 is upward-concave almost everywhere and thus deviates strongly from the
gaussian distribution.
These two examples provide a window toward generalization. A river can be loosely
classified as sand-bed or gravel-bed according to whether the median size D50 of the surface material or substrate is less than or greater than 2 mm. The size distributions of sandbed streams tend to be relatively narrow and also tend to be S shaped. The size distributions of gravel-bed streams tend to be much broader and to display an upward-concave
shape. Of course, there are many exceptions to this behavior, but it is sufficiently general
to warrant emphasis.
More evidence for this behavior is provided in Fig. 6.5. Here, the grain size distributions for a variety of stream reaches have been normalized using the median size D50.
Four sand-bed reaches are included with three gravel-bed reaches. All the sand-bed
distributions are S shaped, and all have a lower spread than the gravel-bed distributions.
The standard deviation is seen to increase systematically with increasing D50(White et al.,
1973).
The three gravel-bed size distributions differ systematically from the sand-bed distributions in a fashion that accurately reflects Oak Creek (Fig. 6.4). The standard deviation
in all cases is markedly larger than any of the sand-bed distributions, and the distributions
are upward-concave except perhaps near the coarsest sizes.

River
Miss, Tarbeii La.
Niobran
Miss, at Sl Louis
Mountain free*
Elbow
Aart

DfMENSIONLESS GRAIN SIZE COMPOSITION

(0,/O3^

FIGURE 6.5 Dimensionless grain-size distribution for different rivers (White et al., 1973)

6A

THRESHOLD CONDITION FOR SEDIMENT MOVEMENT

When a granular bed is subjected to a turbulent flow, virtually no motion of the grains is
observed at some flows, but the bed is mobilized noticeably at other flows. Factors that
affect the mobility of grains subjected to a flow are summarized below:

I grain placement
randomness ^
[turbulence
f fluid 1/lift mean & turbulent
forces on grain <
I drag
I gravity
In the presence of turbulent flow, random fluctuations typically prevent the clear definition of a critical, or threshold condition for motion: The probability for the movement of
a grain is never precisely zero (Lavelle and Mofjeld, 1987). Nevertheless, it is possible to
define a condition below which movement can be neglected for many practical purposes.
6.4.1 Granular Sediment on a Stream Bed
Figure 6.6 is a diagram showing the forces acting on a grain in a bed of other grains. When
critical conditions exist and the grain is on the verge of moving, the moment caused by
the critical shear stress Tc about the point of support is just equal to that of the weight of
the grain. Equating these moments gives (Vanoni, 1975):
Tc = (Js ~ J) Dcos (tan 9 - tanty)
C2a2

(6.39)

in which ys = specific weight of sediment grains, y = specific weight of water, D = diameter of grains, <(> is the slope angle of the stream, 6 = the angle of repose of the sediment, C1

FIGURE 6.6 Forces acting on a sediment particle on an inclined bed

and C2 are dimensionless constants, and av and a2 are lengths shown in Fig. 6.6. Any consistent set of units can be used in Eq. (6.39). For a horizontal bed, Eq. (6.39) reduces to
Tc = (Y, - J)D tan 6
C2#2

(6.40)

For an adverse slope (i.e., 4> < O),

C CL
Tc = -^- (Y, - J)D cos (tan 0 + tan <|>)
(6.41)
C2#2
Equations (6.39), (6.40), and (6.41) cannot be used to give ic because the factors C1, C2,
G1, and a2 are not known. Therefore, the relation between the pertinent quantities is
expressed by dimensional analysis, and the actual relation is determined from experimental data. Figure 6.7 is such a relation, first presented by Shields (1936) and carries his
name. The curve is expressed by dimensionless combinations of critical shear stress Tc,
sediment and water specific weights Y5 and Y, sediment size D, critical shear velocity w*c
= Vi/p and kinematic viscosity of water v.
These quantities can be expressed in any consistent set of units. Dimensional analysis
yields,

-^5-4^)
The Shields values of Tc* are commonly used to denote conditions under which bed
sediments are stable but on the verge of being entrained. Not all workers agree with the
results given by the Shields curve. For example, some workers give Tc* = 0.047 for the
dimensionless critical shear stress for values of R* = u*D/v in excess of 500 instead of
0.06, as shown in Fig. 6.7. Taylor and Vanoni (1972) reported that small but finite amounts
of sediment were transported in flows with values of i* given by the Shields curve.
The value of Tc to be used in design depends on the particular case at hand. If the situation is such that grains that are moved can be replaced by others moving from upstream,
some motion can be tolerated, and the Shields values can be used. On the other hand, if
grains removed cannot be replaced, as on a stream bank, the Shields value of ic are too
large and should be reduced.
The Shields diagram is not especially useful in the form of Fig. 6.7 because to find Tc,
one must know w* = VT/P. The relation can be cast in explicit form by plotting TC* versus Rep, noting the internal relation
_u*D
1_ =
v

u _ VfltfD
U
s D
L
v
VRgD

= (T*)i/2/fc

(6.43)

p p
where R =
is the submerged specific gravity of the sediment. A useful fit is given

byBrownlie(1981a):
T* = 0.22Re- + o.06 exp(- 17.77/te;006)

(6.44)

With this relation, the value of Tc* can be computed readily when the properties of the
water and the sediment are given.
The value of bed-shear stress ib for a wide rectangular channel is given by ib = jHS,
as shown earlier. The average bed-shear stress for any channel is given by ib = JR^S, in
which Rh = the hydraulic radius of the channel cross section.

Amber

Fully developed turbulent velocity profile

Turbulent boundary layer

Value of

Boundary Reynolds Number, R*

FIGURE 6.7 Shields diagram for initiation of motion. Source Vanoni (1975)

** (Shields)
Granite
Barite
Sand (Casey)
Sand (Kramer)
Sand (U.S. WES.)
Sand (Gilbert)
Sand (White)
Sand in air (White)
Steel shot (White)

6.4.2 Granular Sediment on a Bank

A sediment grain on a bank is less stable than one on the bed because the gravity force
tends to move it downward (Ikeda, 1982). The ratio of the critical shear stress iwc for a
particle on a bank to that for the same particle on the bed ic is (Lane, 1955)
V
L (tan Cb1 >
17 = 00 ^ 1 V 1 -MJ

<6-45>

where (J)1 is the slope of the bank and 0 is the angle of repose for the sediment. Values of 0
are given in Fig. 6.8 after Lane (1955) and also can be found in Simons and Senturk (1976).
6.4.3 Granular Sediment on a Sloping Bed

Angle of Repose, 9, in Degrees

Equation (6.39) shows that ic diminishes as the slope angle (|> increases. For extremely
small <|)'s, TC is given by Eq. (6.40). Taking the ratio between Eqs. (6.39) and (6.40) yields

Particle Size, d&9 In Inches

FIGURE 6.8 Angle of repose of granular material. (Lane,
1955)

<-=$)
T^ is the critical shear stress for sediment on a bed with a slope angle <|), and ico is the critical shear stress for a bed with an extremely small slope. The value of ico can be found
from the Shields diagram or with Eq. (6.44). Equation (6.46) is for positive <|), which is
positive for downward sloping beds. For beds with adverse slope, ty is negative and the
term tan <|)/tan 0 in Eq. (6.46) is positive.
6.4.4 Sediment Mixtures
Several authors have offered empirical or quasi-theoretical extensions of the above relations to the case of mixtures (e.g., Wilcock, 1988). Let D1 denote the characteristic grain
size of the ith size range in a mixture. Furthermore, let Dsg denote the geometric mean size
of the surface (exchange, active) layer. Most of the generalizations can be written in the
following form (Parker, 1990):
^ =^

(6.47)

Here
'--pfe

(6 48a)

*"pfe

(^b)

and

where ifed and ibcsg denote the values of the dimensioned critical shear stress required to
move sediment of sizes D1 and Dsg in the mixture, respectively, and P is an exponent taking a value given below;
P = 0.9

(6.49)

Figure 6.9 shows the similarity between four different published expressions having
the general form given by Eq. (6.47), which is of interest because it includes the effect of
hiding. For uniform material, the critical Shields stress is defined by Eq. (6.44). Consider
two flumes, one with uniform size D0 and the other with uniform size Db. For sufficiently coarse material (*Z)/v 1 or Rep 1), the critical Shields stress must be the same for
both sizes (Fig. 6.7). It follows from Eq. (6.42) that where ibca and ibcb denote the dimensioned boundary shear stresses for cases a and b respectively,

(D,\
*bcb = V^j

(6.50)

For the case of mixtures, on the other hand, it is seen from Eqs. (6.47) and (6.48) that
a
0 1
(6 51)
^1 = *Jff\
\ SgJ "" ^fzr]
\ SgJ '
Comparing Eqs. (6.50) and (6.51), it is seen that a finer particle (Db < Da, or alternatively, D1 < Dsg) is more mobile than a coarser particle. For example, suppose that one grain
size is four times coarser than another. If two uniform sediments are being compared, it

EgiazaroffCl965)
Hayashietal

Andrews

FIGURE 6.9 Critical shear stress for sediment mixture (Source: Misri et
al, 1983)
follows from Eq. (6.50) that the critical shear stress for the coarser material is four times
that of the finer material. In the case of a mixture, however, the critical shear stress for the
coarser material is only about 4-1, or 1.15 times that for the finer material.
A finer particle in a mixture is thus seen to be only a little more mobile than its coarser-sized brethren, where uniform beds of fine material are much more mobile than are uniform beds of coarser material. The reason is that finer particles in a mixture are relatively
less exposed to the flow; they tend to hide in the lee of coarser particles. By the same token,
a particle is relatively more exposed to the flow when most of its neighbors are finer.
A method to calculate the critical shear stress for motion of uniform and heterogeneous
sediments was proposed by Wiberg and Smith (1987) on the basis of the fluid mechanics
of initiation of motion, which takes into account both roughness and hiding effects.

6.5 SEDIMENTTRANSPORT
6.5.1 Sediment Transport Modes
The most common modes of sediment transport in rivers are bedload and suspended load.
In the case of bedload, the particles roll, slide, or saltate over each other, never deviating
too far above the bed. In the case of suspended load, the fluid turbulence comes into play
carrying the particles well up into the water column. In both cases, the driving force for
sediment transport is the action of gravity on the fluid phase; this force is transmitted to
the particles via drag.

The same phenomena of bedload and suspended load transport occur in a variety of
other geophysical contexts. Sediment transport is accomplished in the near-shore lake and
oceanic environment by wave action. Turbidity currents carry sediment into lakes, reservoirs, and the deep sea.
The phenomenon of sediment transport can sometimes be disguised in rather esoteric
phenomena. When water is supercooled, large quantities of particulate frazil ice can form.
As this water moves under a frozen ice cover, the phenomenon of sediment transport in
rivers is stood on its head. The frazil ice particles float rather than sink and thus tend to
accumulate on the bottom side of the ice cover rather than on the river bed. Turbulence
tends to suspend the particles downward rather than upward.
In the case of a powder snow avalanche, the fluid phase is air and the solid phase consists of snow particles. The dominant mode of transport is suspension. These flows are
close analogies of turbidity currents, insofar as the driving force for the flow is the action
of gravity on the solid phase rather than the fluid phase. That is, if all the particles drop
out of suspension, the flow ceases. In the case of sediment transport in rivers, it is accurate to say that the fluid phase drags the solid phase along. In the case of turbidity currents
and powder snow avalanches, the solid phase drags the fluid phase along.
Desert sand dunes provide an example for which the fluid phase is air, but the dominant mode of transport is saltation rather than suspension. Because air is so much lighter
than water, quartz sand particles saltate in long, high trajectories, relatively unaffected by
the direct action of turbulent fluctuations. The dunes themselves are created by the effect
of the fluid phase acting on the solid phase. They, in turn, affect the fluid phase by changing the resistance.
Among the most interesting sediment-transport phenomena are debris flows, slurries,
and hyperconcentrated flows. In all these cases, the solid and fluid phases are present in
similar quantities. A debris flow typically carries a heterogeneous mixture of grain sizes
ranging from boulders to clay. Slurries and hyperconcentrated flows are generally restricted to finer grain sizes. In most cases, it is useful to think of such flows as consisting of a
single phase, the mechanics of which are highly non-Newtonian.
The study of the movement of grains under the influence of fluid drag and gravity
becomes even more interesting when one considers the link between sediment transport
and morphology. In the laboratory, the phenomenon can be studied in the context of a variety of containers, such as channel and wave tanks, specified by the experimentalist. In the
field, however, the fluid-sediment mixture constructs its own container. This new degree
of freedom opens up a variety of intriguing possibilities.
Consider the river. Depending on the existence or lack of a viscous sublayer and the
relative importance of bedload versus suspended load, a variety of rhythmic structures can
form on the river bed. These include ripples, dunes, antidunes, and alternate bars. The first
three of these can have a profound effect on the resistance to flow offered by the river bed.
Thus, they act to control river depth. River banks themselves also can be considered to be
a self-formed morphological feature, thus specifying the entire container.
The container itself can deform in plan. Alternate bars cause rivers to erode their banks
in a rhythmic pattern, thus allowing for the onset of meandering. Fully developed river
meandering implies an intricate balance between sediment erosion and deposition. If a
stream is sufficiently wide, it will braid rather than meander, dividing into several intertwining channels.
Rivers create morphological structures at much larger scales as well. These include
canyons, alluvial fans, and deltas. Turbidity currents create similar structures in the oceanic environment. In the coastal environment, the beach profile itself is created by the interaction of water and sediment. On a larger scale, offshore bars, spits, and capes constitute
rhythmic features created by wave-current-sediment interaction. The boulder levees often
created by debris flows provide another example of a morphologic structure created by a
sediment-bearing flow.

The floodplains of most sand-bed rivers often contain copious amounts of silt and clay
finer than approximately 50 |JL This material is often called wash load because it moves
through the river system without being present in the bed in significant quantities.
Increased wash load does not cause deposition on the bed, and reduced wash load does
not cause erosion because it is transported well below capacity. This is not meant to imply
that the wash load does not interact with the river system. Wash load in the water column
exchanges with the banks and the floodplain rather than the bed. Greatly increased wash
load, for example, can lead to thickened floodplain deposits, with a consequent increase
in bankfull channel depth.
The emphasis here is the understanding of bedload and suspended load transport in
rivers, with the goal of providing the knowledge needed to do sound sedimentation engineering, particularly with problems involving stream restoration and naturalization.
6.5.2

Shields Regime Diagram

In the context of rivers, it is useful to have a way to determine what kind of sedimenttransport phenomena can be expected for different flow conditions and different characteristics of sediment particles. In Fig. 6.10, the ordinates correspond to bed shear stresses
written in the dimensionless form proposed by Shields
T*
T =

Tfc

pgRD

= HS
RD

(f) 52)
TO
(6
'

and the particle Rep, defined by Eq. (6.37) is used for the abscissa values. There are three
curves in the diagram which make it possible to know, for different values of (T*, Rep), if
the given bed sediment will go into motion, and if this is the case whether or not the prevailing mode of transport will be in suspension or as bedload. The diagram also can be used
to predict what kind of bedforms can be expected. For example, ripples will develop in the
presence of a viscous sublayer and fine-grained sediment. If the viscous sublayer is disrupted by coarse sediment particles, then dunes will be the most common type of bedform.
The Shields regime diagram also shows a clear distinction between the conditions
observed in sand-bed rivers and gravel-bed rivers at bankfull stage. If one wanted to model in the laboratory sediment transport in rivers, the experimental conditions would be different, depending on the river system in question. As could be expected, the diagram also
shows that in gravel-bed rivers, sediment is transported as bedload. In sand-bed rivers, on
the other hand, suspended load and bedload transport coexist most of the time.
The regime diagram is valid for steady, uniform, turbulent flow conditions, where the
bed shear stress lb can be estimated with Eq. (6.1). The ranges for silt, sand, and gravel
also are included. In the diagram, the critical Shields stress for motion was plotted with
the help of Eq. (6.44). The critical condition for suspension is given by the following ratio:
^- = 1

(6.53)

where u* is the shear velocity and vs is the sediment fall velocity. Equation (6.53) can be
transformed into
*; = ^

(6.54)

^ OD

^)

where

Suspension
No Suspension
Field
Sand-Bed
Rivers

Dunes ?
Flat?

Field
Gravel-Bed
Rivers

Motion
No Motion
Gravel

FIGURE 6.10 Shields regime diagram. (Source: Gary Parker)

and Rf is given by Eq. (6.35a) and can be computed for different values of Rep with the
help of Eq. (6.38).
Finally, the critical condition for viscous effects (ripples) was obtained with the help
of Eq. (6.5) as follows:
11.6-^-= 1
uD
which in dimensionless form can be written as

(6.56)

*-
Relations (6.44), (6.54), and (6.57) are the ones plotted in Fig. 6.10. The Shields regime
diagram should be useful for studies concerning stream restoration and naturalization
because it provides the range of dimensionless shear stresses corresponding to bankfull
flow conditions for both gravel- and sand-bed streams.

6.6

BEDLOAD TRANSPORT

6.6.1 The Bed Load Transport Function

Bedload particles roll, slide, or saltate along the bed. The transport thus occurs tangential
to the bed. In a case where all the transport is directed in the streamwise, or s direction,
the volume bedload-transport rate per unit width (n direction) is given by q; the units are
lengthYlength/per time, or Iength2/time. In general, q is a function of boundary shear stress
ib and other parameters; that is,
q = q(lb, other parameters)

(6.58)

In general, bedload transport is vectorial, with components qs and qn in the s and n directions, respectively.
6.6.2 Erosion Into and Deposition from Suspension
The volume rate of erosion of bed material into suspension per unit time per unit bed area
is denoted as E. The units of E are lengthVlengthVtime, or velocity. A dimensionless sediment entrainment rate E5 can thus be defined with the sediment fall velocity vs:
E = vfs

(6.59)

In general, E5 can be expected to be a function of boundary shear stress ib and other parameters. Erosion into suspension can be taken to be directed upward normal: i.e., in the
positive z direction.
Let c denote the volume concentration of suspended sediment (m3 of sediment/m3 of
sediment-water mixture), averaged over turbulence. The streamwise volume transport rate
of suspended sediment per unit width is given by
H
qs = J c udz
(6.60)
o

In a two-dimensional case, two components, qSs and qSn, result, where

H
qss = \~cudz
o

H
<lsn = I cvdz
O

(6.6Ia)

(6.6Ib)

Deposition onto the bed is by means of settling. The rate at which material is fluxed
vertically downward onto the bed (volume/area/time) is given by vscb, where ~cb is a nearbed value of c. The deposition rate D realized at the bed is obtained by computing the
component of this flux that is actually directed normal to the bed:
D = vscb

(6.62)

6.6.3 The Exner Equation of Sediment Mass Conservation for

Uniform Material
Consider a portion of river bottom, where the bed material is taken to have a (constant)
porosity A,p. Mass balance of sediment requires the following equation to be satisfied:
-^- [mass of bed material] = net mass bedload inflow rate
+ net mass rate of deposition from suspension.
A datum of constant elevation is located well below the bed level, and the elevation of the
bed with respect to such datum is given by TJ. Then, bed level changes as a result of bedload transport, sediment entrainment into suspension, and sediment deposition onto the
bed can be predicted with the help of
^V=-t-^-s^-,s)

(6.63)

To solve the Exner equation, it is necessary to have relations to compute bedload transport (i.e., qs and gn), near-bed suspended sediment concentration cb, and sediment entrainment into suspension E5. The basic form of Eq. (6.63) was first proposed by Exner (1925).
6.6.4 Bedload Transport Relations
A large number of bedload relations can be expressed in the general form
q* = <7*(T*, Rep> R)

(6.64)

Here, q* is a dimensionless bedload transport rate known as the Einstein number, first
introduced by H. A. Einstein in 1950 and given by
<f =

V# g D D
The following relations are of interest. In 1972, Ashida and Michiue introduced
<f = 17(T* - ty [(l*)m - (Tyi/2]

(6.65)

(6>66)

and recommend a value of Tc* of 0.05. It has been verified with uniform material ranging
in size from 0.3 mm to 7 mm. Meyer-Peter and Muller (1948) introduced the following:
q* = 8(T* - Tc*)3/2

(6.67)

where T* = 0.047. This formula is empirical in nature and has been verified with data for
uniform gravel.
Engelund and Freds0e (1976) proposed,
q* = 18.74(T* - Tc*) [(f) 1/2 - 0.7(Tp1/2]

(6.68)

where T* = 0.05. This formula resembles that of Ashida and Michiue because the derivation is almost identical.
Fernandez Luque and van Beek (1976) developed the following,
q* = 5.7(T* - TCT'2

(6.69)

where T* varies from 0.05 for 0.9 mm material to 0.058 for 3.3. mm material. The relation
is empirical in nature.
Wilson (1966):
<?* - 12(T* - Tc*)^

(6.70)

where T* was determined from the Shields diagram. This relation is empirical in nature;
most of the data used to fit it pertain to very high rates of bedload transport.
Einstein (1950):
q" = 4*(T*)

(6.71)

where the functionality is implicitly defined by the relation

. r(0.143/T*)~2

1-rV j
*-**=
,
l
f
f
i
s
.
(6-72>
l
V* -,413/,*)-2
+ 43'5<7
Note that this relation contains no critical stress. It has been used for uniform sand and
gravel.
Yalin (1963):
+
* = 0.635,(T*)4l-^
I
a2s -^lJ

(6.73)

where
a2 = 2A5(R + I)0-4 (Tc*)1/2;

s =T ~T*
^C

(6.74)

and T* is evaluated from a standard Shields curve. Two constants in this formula have been
evaluated with the aid of data quoted by Einstein (1950), pertaining to 0.8 mm and 28.6
mm material.
Parker (1978):
(T* - 0.03)45
<7*=11.2 IT ^3 )
(6.75)
developed with data sets pertaining to rough mobile-bed flow over gravel.
Several of these relations are plotted in Fig. 6.11. They tend to be rather similar in
nature. Scores of similar relations could be quoted.

To date, only few research groups have attempted complete derivations of the bedload
function in water. They are Wiberg and Smith (1989), Sekine and Kikkawa (1992), Garcia
and Nino (1992), Nino and Garcia, (1994, 1998), and Nino et al., (1994).
6.6.5

Bedload Transport Relation for Mixtures.

Relatively few bedload relations have been developed specifically in the context of mixtures (e.g., Bridge and Bennett, 1992). One of these is presented below as an example.
The relationship of Parker (1990) applies to gravel-bed streams. The data used to fit
the relation are solely from two natural gravel-bed streams: Oak Creek in Oregon and the
Elbow River in Alberta, Canada. The relation is surface-based; load is specified per unit
of fractional content in the surface layer. The surface layer is divided into N size ranges,

D = 0.5mm
^ D = 28.6mrn
EfNSTEfN (O=O 5mm)WIOERG & SMITH (D-O 5mm)
MEYER-PEIEH A MULLER

*
M
///S
'// /

FERNANDEZ LUOUE

FERNANDEZ LUOUE

MEYER-PETER & MULLER

EINSTEIN
WIBERG & SMITH

FIGURE 6.11 Bedload transport relations. (Parker, 1990)

each with a fractional content F1 by volume, and_a mean phi size (J)1; Di = 2~to- The arithmetic mean of the surface size on the phi scale and the corresponding arithmetic standard deviation C^ are given by
<j> = IFfo

Oj = Ufa - ^)2

(6.76a, b)

The corresponding geometric mean size Dsg and the geometric standard deviation Gsg of
the surface layer are given by
A* = 2-* ov = 2*

(6.77a,b)

In the Parker relation, the volume bedload transport per unit width of gravel in the ith
size range is given by the product qf. (no summation), where q. denotes the transport per
unit fraction in the surface layer. The total volume bedload transport rate of gravel per unit
width is qT, where
qT = ^q1F1

(6.78)

The relation does not apply to sand. Thus, before using the relation for a given surface
distribution, the sand content of the grain-size distribution must be removed and F1 must
be renormalized so that it sums to unity over all sizes in excess of 2 mm.
If p. denotes the fraction volume content of material in the ith size range in the bedload, it follows that
P, = ^

(6-79)

The parameter qi is made dimensionless as follows:

^ = (JPb;

<6-80>

A dimensionless Shields stress based on the surface geometric mean size is defined as
follows:
*-&-,

<6TO

Let tysgo denote a normalized value of this Shields stress, given by

where

^ = *
rsgo
T;SO = 0.0386

(6-82)
(6.83)

corresponds to a "near-critical" value of Shields stress. The Parker relation can then be
expressed in the form
Wl1 = 0.0218 G [CD(^(S1)]

(6.84a)

In the above relationship, g0 denotes a hiding function given by

S0(S) = 5ro.o95i;

5. = PL

(6.84b)

The parameter co is given by the relationship

C0 = i + _^.(co
(6.84c)
o - i)
a
$o
where (J^0 and CO0 are specified as functions of sgo in Fig. 6.12. The function G is specified as
5474(1 -0.853/(J))45

<|> > 1.65

2

G[<|)] = ' exp[l4.2($ -I)- 9.28(<|) - I) ]

- (|)MO

1 < <|> < 1.65

<|> < 1

(6.85)

and is shown in Fig. 6.13. Here, M0 = 14.2 and $ is a dummy variable for the argument
in Eq. (6.84) and is not to be confused with the grain-size scale.
An application of Eq. (6.84) to uniform material with size D results in the relation
* = -0218(TH<*y

(6 86)

'

where
?= VgRDD

^ =-^n
P8RD

(6 87)

and q denotes the volumetric sediment transport per unit width. In Fig. 6.11, Eq. (6.86) is
compared to several other relations and selected laboratory data for uniform material. The
figure is adapted from Figs. 6b and 7 in Wiberg and Smith (1989), where reference to the
data and equations can be found. The data pertain to 0.5 mm sand and 28.6 mm gravel.
Equation (6.86) shows a reasonable correspondence with the data and with several other
relations for uniform material.
The Parker relationship (Eq. 6.84) can be used to predict mobile or static armor in
gravel streams. Note that there is no formal critical stress in the formulation; instead for
$ < 1, the transport rates become extremely small. For the computation of bedload transport in poorly sorted gravel-bed rivers, the above formulation has been used to implement
a series of programs named "ACRONYM" (Parker, 1990). The program "ACRONYMl"
provides an implementation of the surface-based bedload transport equation presented in
Parker (1990). It computes the magnitude and size distribution of bedload transport over
a bed surface of given size distribution, on which a given boundary shear stress is
imposed. The program "ACRONYM2" inverts the same bedload transport equation,
allowing for calculation of the size distribution at a given boundary shear stress. The program was used to compute mobile and static armor size distributions in Parker (1990) and
Parker and Sutherland (1990).
The program "ACRONYM3" allows for the computation of aggradation or degradation to a specified active or static equilibrium final state. To this end, Parker's method
(1990) is combined with a resistance relation of the Keulegan type. In the program, both
constant width and water discharge are assumed.
The program "ACRONYM4" is directed toward the wavelike aggravation of self-similar form discussed in Parker (199Ia, 199Ib). It uses Parker's method and a resistance
relation of the Manning-Strickler type to compute downstream fining and slope concavity caused by selective sorting and abrasion.

FIGURE 6.12 Plots of CO0 and G(J)0 versus (J)^0, the asymptotes are noted on the plot. (Parker, 1990)

FIGURE 6.13 Plot of G and G7 versus ^50. (Parker, 1990)

Next Page

CHAPTER 7
RISK/RELIABILITY-BASED
HYDRAULIC ENGINEERING
DESIGN

Yeou-Koung Tung
Department of Civil Engineering
Hong Kong University of Science and Technology
Clear Water Bay Kowloon,
Hong Kong

7.1

INTRODUCTION

7.1.1 Uncertainties in Hydraulic Engineering Design

In designing hydraulic engineering systems, uncertainties arise in various aspects including, but not limited to, hydraulic, hydrologic, structural, environmental, and socioeconomical aspects. Uncertainty is attributed to the lack of perfect knowledge concerning the
phenomena and processes involved in problem definition and resolution. In general,
uncertainty arising because of the inherent randomness of physical processes cannot be
eliminated and one has to live with it. On the other hand, uncertainties, such as those associated with the lack of complete knowledge about processes, models, parameters, data,
and so on, can be reduced through research, data collection, and careful manufacturing.
Uncertainties in hydraulic engineering system design can be divided into four basic
categories: hydrologic, hydraulic, structural, and economic (Mays and Tung, 1992).
Hydrologic uncertainty for any hydraulic engineering problem can further be classified
into inherent, parameter, or model uncertainties. Hydraulic uncertainty refers to the uncertainty in the design of hydraulic structures and in the analysis of the performance of
hydraulic structures. Structural uncertainty refers to failure from structural weaknesses.
Economic uncertainty can arise from uncertainties in various cost items, inflation, project
life, and other intangible factors. More specifically, uncertainties in hydraulic design
could arise from various sources (Yen et al., 1986) including natural uncertainties, model
uncertainties, parameter uncertainties, data uncertainties, and operational uncertainties.
The most complete and ideal way to describe the degree of uncertainty of a parameter,
a function, a model, or a system in hydraulic engineering design is the probability density function (PDF) of the quantity subject to uncertainty. However, such a probability function cannot be derived or found in most practical problems. Alternative ways of expressing the uncertainty of a quantity include confidence intervals or statistical moments. In
particular, the second order moment, that is, the variance or standard deviation, is a measure of the dispersion of a random variable. Sometimes, the coefficient of variation,
defined as the ratio of standard deviation to the mean, is also used.

The existence of various uncertainties (including inherent randomness of natural

processes) is the main contributor to the potential failure of hydraulic engineering systems. Knowledge of uncertainty features of hydraulic engineering systems is essential for
assessing their reliability.
In hydraulic engineering design and analysis, the decisions on the layout, capacity, and
operation of the system largely depend on the system response under some anticipated
design conditions. When some of the components in a hydraulic engineering system are
subject to uncertainty, the system's responses under the design conditions cannot be
assessed with certainty. An engineer should consider various criteria including, but not
limited to, the cost of the system, failure probability, and consequences of failure, such
that a proper design can be made for the system.
In hydraulic engineering design and analysis, the design quantity and system output
are functions of several system parameters not all of which can be quantified with
absolute certainty. The task of uncertainty analysis is to determine the uncertainty features
of the system outputs as a function of uncertainties in the system model and in the stochastic parameters involved. Uncertainty analysis provides a formal and systematic
framework to quantify the uncertainty associated with the system output. Furthermore, it
offers the designer useful insights with regard to the contribution of each stochastic parameter to the overall uncertainty of the system outputs. Such knowledge is essential in
identifying the "important" parameters to which more attention should be given to better
assess their values and, accordingly, to reduce the overall uncertainty of the system outputs.
7.1.2 Reliability of Hydraulic Engineering Systems
All hydraulic engineering systems placed in a natural environment are subject to various
external stresses. The resistance or strength of a hydraulic engineering system is its ability to accomplish the intended mission satisfactorily without failure when subject to loading of demands or external stresses. Failure occurs when the resistance of the system is
exceeded by the load. From the previous discussions on the existence of uncertainties, the
capacity of a hydraulic engineering system and the imposed loads, more often than not,
are random and subject to some degree of uncertainty. Hence, the design and operation of
hydraulic engineering systems are always subject to uncertainties and potential failures.
The reliability, ps, of a hydraulic engineering system is defined as the probability of
nonfailure in which the resistance of the system exceeds the load; that is,
ps = P(L<R)

(7.1)

where P(*) denotes the probability. The failure probability, pf, is the compliment of the
reliability which can be expressed as
pf=P(L>R)= l-ps

(7.2)

In hydraulic engineering system design and analysis, loads generally arise from natural
events, such as floods and storms, which occur randomly in time and in space. A common
practice for determining the reliability of a hydraulic engineering system is to assess the
return period or recurrence interval of the design event. In fact, the return period is equal to
the reciprocal of the probability of the occurrence of the event in any one time interval. For
most engineering applications, the time interval chosen is 1 year so that the probability
associated with the return period is the average annual exceedance probability of the occurrence of the event. Flood frequency analysis, using the annual maximum flow series, is a
typical example of this kind of application. Hence, the determination of return period

depends on the time period chosen (Borgman, 1963). The main disadvantage of using the
return period method is that reliability is measured only in terms of time of occurrence of
loads without considering the interactions with the system resistance (Melchers, 1987).
Two other types of reliability measures that consider the relative magnitudes of resistance and anticipated load (called design load) are frequently used in engineering practice. One is the safety margin (SM), defined as the difference between the resistance (R)
and the anticipated load (L), that is,
SM = R-L

(7.3)

The other is called the safety factor (SF), a ratio of resistance to load, which is
defined as
SF = RIL

(7.4)

Yen (1979) summarized several types of safety factors and discussed their applications to
hydraulic engineering system design.
There are two basic probabilistic approaches to evaluate the reliability of a hydraulic
engineering system. The most direct approach is a statistical analysis of data of past failure
records for similar systems. The other approach is through reliability analysis, which considers and combines the contribution of each factor potentially influencing the failure. The
former is a lumped system approach requiring no knowledge about the behavior of the facility or structure nor its load and resistance. For example, dam failure data show that the overall average failure probability for dams of all types over 15m height is around 10~3 per dam
per year (Cheng, 1993). In many cases, this direct approach is impractical because (1) the
sample size is too small to be statistically reliable, especially for low probability/high consequence events; (2) the sample may not be representative of the structure or of the population; and (3) the physical conditions of the dam may be non-stationary, that is, varying with
respect to time. The average risk of dam failure mentioned above does not differentiate concrete dams from earthfill dams, arch dams from gravity dams, large dams from small dams,
or old dams from new dams. If one wants to know the likelihood of failure of a particular
10 -year-old double-curvature arch concrete high dam, one will most likely find failure
data for only a few similar dams, this is insufficient for any meaningful statistical analysis.
Since no dams are identical and dam conditions change with time, in many circumstances,
it may be more desirable to use the second approach by conducting a reliability analysis.
There are two major steps in reliability analysis: (1) to identify and analyze the uncertainties of each contributing factor; and (2) to combine the uncertainties of the stochastic
factors to determine the overall reliability of the structure. The second step, in turn, may
proceed in two different ways: (1) directly combining the uncertainties of all factors, or
(2) separately combining the uncertainties of the factors belonging to different components or subsystems to evaluate first the respective subsystem reliability and then combining the reliabilities of the different components or subsystems to yield the overall reliability of the structure. The first way applies to very simple structures, whereas the second way is more suitable for complicated systems. For example, to evaluate the reliability of a dam, the hydrologic, hydraulic, geotechnical, structural, and other disciplinary reliabilities could be evaluated separately first and then combined to yield the overall dam
reliability. Or, the component reliabilities could be evaluated first, according to the different failure modes, and then combined. Vrijling (1993) provides an actual example of
the determination and combination of component reliabilities in the design of the Eastern
Scheldt Storm Surge Barrier in The Netherlands.
The main purpose of this chapter is to demonstrate the usage of various practical
uncertainty and reliability analysis techniques through worked examples. Only the essential theories of the techniques are described. For more detailed descriptions of the methods and applications, see Tung (1996).

7.2

TECHNIQUES FOR UNCERTAINTY ANALYSIS

In this section, several analytical methods are discussed that would allow an analytical
derivation of the exact PDF and/or statistical moments of a random variable as a function
of several random variables. In theory, the concepts described in this section are straightforward. However, the success of implementing these procedures largely depends on the
functional relation, forms of the PDFs involved, and analyst's mathematical skill.
Analytical methods are powerful tools for problems that are not too complex. Although
their usefulness is restricted in dealing with real life complex problems, situations do exist
in which analytical techniques could be applied to obtain exact uncertainty features of
model outputs without approximation or extensive simulation. However, situations often
arise in which analytical derivations are virtually impossible. It is, then, practical to find
an approximate solution.
7.2.1 Analytical Technique: Fourier and Exponential Transforms
The Fourier and exponential transforms of a PDF, fx(x), of a random variable X are
defined, respectively, as
&x(s) = E[eisx] = I eisxfx(x) dx

(7.5a)

^x(s) = E[e] = I

(7.5b)

and

e**fx(x) dx

where i = V-I and() is the expectation operator; E[eisX] andE[e sX ] are called, respectively, the characteristic function and moment generating function, of the random variable
X. The characteristic function of a random variable always exists for all values of the arguments whereas for the moment generating function this is not necessarily true.
Furthermore, the characteristic function for a random variable under consideration is
unique. In other words, two distribution functions are identical if and only if the corresponding characteristic functions are identical (Patel et al., 1976). Therefore, given a characteristic function of a random variable, its probability density function can be uniquely
determined through the inverse transform as
/ = 2^/1 e-isx9x(s)

ds

(7-6>

The characteristic functions of some commonly used PDFs are shown in Table 7.1.
Furthermore, some useful operational properties of Fourier transforms on a PDF are given
in Table 7.2
Using the characteristic function, the r* order moment about the origin of the random
variable X can be obtained as

-^1,
Fourier and exponential transforms are particularly useful when random variables are
independent and linearly related. In such cases, the convolution property of the Fourier
transform can be applied to derive the characteristic function of the resulting random vari-

TABLE 7.1 Characteristic Functions of Some Commonly Used Distributions

Distribution

PDF,

fyx)

Characteristic Function

Binomial
Poisson
Uniform
Normal
Gamma
Exponential
ExtremeValue I

TABLE 7.2 Operation Properties of Fourier Transform on a PDF

Property

PDF

Standard
fax)
Scaling
f^ax)
Linear
af^x)
Translation 1
^fx(X)
Translation 2
/x(jc - a)
Source: From Springer (1979).

Random
Variable

Fourier
Transform

X
X
X
X
X

&x(s)
a~l&x(s/a)
ax(s)
&x(s + id)
e~ x(s)

able. More specifically, consider that W = X1 + X2 + . . . + XN and all Xs are independent

random variables with known PDF, ffie), j = 1,2, ... ,N. The characteristic function of
W then can be obtained as
9Jis) = &i(s) &2(s)' ' ' ^N(S)

(7.8a)

Vw(s) = V1(S) V2(S) - VJs)

(7.8b)

and

which is the product of the characteristic and moment generating functions of each
individual random variable. The resulting characteristic function or moment generating

function for W can be used in Eq. (7.7) to obtain the statistical moments of any order for
the random variable W. Furthermore, the inverse transform of $y, according to Eq.
(7.6), can be made to derive the PDF of W, if it is analytically possible.
7.2.2 Analytical Technique: Mellin Transform
When the random variables in a function W = g(X) are independent and nonnegative and
the function g(X) has a multiplicative form as
TV
W=g(X) = a0 II X?
(7.9)
i=i
the Mellin transform is particularly attractive for conducting uncertainty analysis (Tung,
1990). The Mellin transform of a PDFAM, where x is positive, is defined as
Mx(S) = M\fx(x)] = I X3^fx(X) dx = E (x*-l\ x>0

JO

(7.10)

where Mx(s) is the Mellin transform of the function fx(x) (Springer 1979). Therefore, the
Mellin transform provides an alternative way to find the moments of any order for nonnegative random variables.
Similar to the convolutional property of the exponential and Fourier transforms, the
Mellin transform of the convolution of the PDFs associated with multiple independent
random variables in a product form is simply equal to the product of the Mellin transforms of individual PDFs. In addition to the convolution property, the Mellin transform
has several useful operational properties as summarized in Tables 7.3 and 7.4.
Furthermore, the Mellin transform of some commonly used distributions are summarized in Table 7.5.
TABLE 7.3 Operation Properties of the Mellin Transform on a PDF
Property
Standard
Scaling
Linear
Translation
Exponentiation
Source: From Park (1987).

PDF
fx(x)
fx(x)
fx(ax)
afx(x)
xafx(x)
AC*0)

Random
Variable
X
X
X
X
X

Mellin
Transform
Mx(s)
a~sMx(s)
a Mx(s)
Mx(a + s)
a~l Mx(sla)

TABLE 7.4 Mellin Transform of Products and Quotients of Random Variables*

Random Variable

PDF Given

M^s)

W =X
A ( X ) M X
W = Xb
A(JC)
Mx(bs - b + 1)
W = UX
AW
Mx(2-s)
W = XY
A(JC), ,00
Mx(S)M^s)
W = XIY
A(*), S7OO
Mx(S)MfI-S)
W = aXbY<
A(JC), gfy)
a* ~ Mx(bs -b+ \)M^cs -c + 1)
Source: From Park (1987). a, b, c: constants ; X, Y, W: random variables.

TABLE 7.5 Mellin Transforms for Some Commonly Used Probability Density Functions
Probability

PDF, j^x)

Mellin Transform

Uniform

Standard Normal

Lognormal
Exponential
Gamma
Triangular

Weibull

Nonstandard beta

Standard beta
where Mx(k) for standard beta

Example 7.1. Manning's formula is frequently used for determining the flow capacity of storm sewer by
Q = 0.463 n~l D267 S05
where Q is flow rate (ftVs), n is the roughness coefficient (ft176), D is the sewer diameter
(ft), and S is pipe slope (ft/ft). Assume that all three model parameters are independent
random variables with the following statistical properties. Compute the mean and variance of the sewer flow capacity by the Mellin transform.

Parameter

Distribution

Uniform distribution with lower bound 0.0137 and upper

bound 0.0163
Triangular distribution with lower bound 2.853, mode 3.0,
and upper bound 3.147
Uniform distribution with bounds (0.00457, 0.00543)

D
S

Referring to Table 7.4, the Mellin transform of sewer flow capacity, MQ(s), can be
derived as
MQ(s) = 0.463s ~ l Mn(-s + 2) MD(2.61s - 1.67) Ms(0.5s + 0.5)
For roughness coefficient n having a uniform distribution, from Table 7.5, one obtains
n S+2
b~ ~ na~S +2
Mn(~s + 2) = (_s + 2)(Hb_na)

For sewer diameter D with a triangular distribution, one obtains

2
MD(2.67s - 1.67) - (^ - da)(2.61s - 1.67) (2.67s - 0.67)

d (ft 2.675-1.67 J 2.67s-1.67) J (^2-67s ~ L67 /2'67s ~ L67)

db-dm
~~d~^df
;

For sewer slope S with a uniform distribution, one obtains

^0.5.+ 0.5)=

fe,f5+o-5-(oa5+05

(0:55 + 0 . 5)( ;_ 0

Substituting individual terms into MQ(s) results in the expression of the Mellin transform of sewer flow capacity specifically for the distributions associated with the three stochastic model parameters.
Based on the information given, the Mellin transforms of each stochastic model parameter can be expressed as
,, f . 0.0163* - 0.0137s
^ =
00026^

MD(S)

2
[3.147(3.147-3.000 2.853(3.00* - 2.853*)]
~ (0.294),(5 + 1) [
0-147
~
0.147
\
M( .
M s)

0.00543* - 0.00457*
^
0.00086 5
The computations are shown in the following table:

s 1

0.463 "
M n (-j + 2)
MD(2.61s - 1.67)
Mn(0.55 + 0.5)

s=2

s=3

0.463
Mn(O) = 66.834
MD(3.67) = 18.806
M/1.50) = 0.0707

0.2144
MB(-1) = 4478.080
MD(6.34) = 354.681
Af/2.00) = 0.005

Therefore, the mean sewer flow capacity can be determined as

E(Q) = MQ(s = 2) = 0.463 Mn(O) MD(3.67) M5(LSO)
= 0.463(66.834)(18.806)(0.0707) - 41.14 fWs
The second moment about the origin of the sewer flow capacity is
E(Q2) = MQ(s = 3) = 0.4632 MB(-1) MD(6.34) Af/2.00)
= 0.4632 (4478.08) (354.681) (0.005) = 1702.40 ftVs.
The variance of the sewer flow capacity can then be determined as
Var(G) = E(Q2) - E\Q) = 1702.40 - 41.1372 = 10.15 (ftVs)2
with the standard deviation being
oe = VlO.147 = 3.19ft3/s
7.2.3 Approximate Technique: First-Order Variance Estimation
(FOVE) Method
The FOVE method, also called the variance propagation method (Bermouex, 1975), estimates uncertainty features of a model output based on the statistical properties of the
model's random variables. The basic idea of the method is to approximate a model involving random variables by the Taylor series expansion.
Consider that a hydraulic or hydrologic design quantity W is related to N random variables X19 X2, . . .,XN as
W = S(X) = S(X1, X 2 , . . . , X N )

(7.11)

where X = (X1, X2, . . ., XN)1, an TV-dimensional column vector of random variables, the
superscript t represents the transpose of a matrix or vector. The Taylor series expansion of
the function g(X) with respect to the means of random variables X = |U in the parameter
space can be expressed as
- ^ *i=i
iM
* dXjjv
l
[ dAf
>- w 4 i
* zi (
j =f
i [OX
i=

(X1 - ft) (Xj - ft) + e

where JLI,. is the mean of the /th random variable X1 and represents the higher order terms.
The first-order partial derivative terms are called the sensitivity coefficients, each representing the rate of change of model output W with respect to unit change of each vari-

able at (I.
Dropping the higher-order terms represented by e, Eq. (7.12) is a second-order
approximation of the model g(X). The expectation of model output W can be approximated as
N

r
i
E(W] g(n) + I ^ S [S|Cov [X" X;]

(7 13)

'

and the variance of W = g(X) can be expressed as

Var[W] -2 E (^f]1 (^} W1 - m) (*; - M

(7.14)
I Y Y V \dS(X)} \ 32S(X) } mx - , , W V - M W V _ -,I
l^^^^li^l
' ^(i *'(t ^
As can be seen from Eq. (7.14), when random variables are correlated, the estimation
of the variance of W using the second-order approximation would require knowledge
about the cross-product moments among the random variables. Information on the
cross-product moments are rarely available in practice. When the random variables are
independent, Eqs. (7.13) and (7.14) can be simplified, respectively, to

and

TV r
i
E[W] #00 + U
^SrI R&i ~ M*)2]
2
; = i L dX? I

VW -| [^l<* ii [^l [^l W - Wl

<7-15)

C-H9

Referring to Eq. (7.16), the variance of W from a second-order approximation, under

the condition that all random variables are statistically independent, would require knowledge of the third moment. For most practical applications where higher order moments
and cross-product moments are not easily available, the first order approximation is frequently adopted. In the area of structural engineering, the second order methods are commonly used (Breitung, 1984; Der Kiureghian et al.; 1987 Wen, 1987).
By truncating the second and higher-order terms of the Taylor series, the first-order
approximation of W at X = Ji is
E[W] ^g(Ji) = W

(7.17)

vartw] - 5; i Np] fer]Cov (x.' v = s'c^s

/ = i j = i V ' )v \ J Jv

< 7 - 18 >

and

in which s = V^W(JLI) is an TV-dimensional column vector of sensitivity coefficients evaluated

at |Li; C(X) is the variance-covariance matrix of the random vector X. When all random vari-

ables are independent, the variance of model output W can be approximated as

TV
Var(W) 2 sf a 2 - s<Ds,
(7.19)
/=i
in which a represents the standard deviation and D = diag (a2, a 2 ,,..., a^), a diagonal
matrix of variances of involved random variables. From Eq. (7.19), the ratio s2 a2/Var[W]
indicates the proportion of overall uncertainty in the model output contributed by the
uncertainty associated with the random variable X1..
Example 7.2. Referring to Example 7.1, the uncertainty associated with the sewer
slope due to installation error is 5 percent of its intended value 0.005. Determine the
uncertainty of the sewer flow capacity using the FOVE method for a section of 3 ft sewer
with a 2 percent error in diameter due to manufacturing tolerances. The roughness coefficient has the mean value 0.015 with a coefficient of variation 0.05. Assume that the correlation coefficient between the roughness coefficient n and sewer diameter D is -0.75.
The sewer slope S is uncorrelated with the other two random variables.
Solution: The first-order Taylor series expansion of Manning's formula about n =
Jin = 0.015, D = [L0 = 3.0, and S = \LS= 0.005, according to Eq. (7.12), is
Q 0.463 (0.015)-1 (3)2-67 (0.005)05 + (n - 0.015)
L a/i J
^i]'0-3-0'^!]'5-0-0005'
- 41.01 + [0.463 (-1) (0.015)-2 (3.O)267 (0.005)05] (n - 0.015)
+ [0.463 (2.67X0.015)-1 (3.O)167 (0.005)5] (D - 3.0)
+ [0.463 (0.5X0.015)-1 (3.O)267 (0.005)-05] (S - 0.005)
- 41.01 - 2733.99 (n - 0.015) + 36.50 (D - 3.0) + 4100.99 (S - 0.005)
Based on Eq. (7.17), the approximated mean of the sewer flow capacity is
ji fi 41.0ftVs
According to Eq. (7.18), the approximated variance of the sewer flow capacity is
02Q (2733.99)2 Var(n) + (36.5O)2 Var(D) + (4100.99)2 Var(S)
- 2(2733.99)(36.50) Cov(n, D) - 2(2733.99)(4100.99) Cov(n, S)
+ 2(36.50)(4100.99) Cov(D, S)
The above expression reduces to
a2 - (2733.99)2 Var(n) + (36.5O)2 Var(D) + (4100.99)2 Var(S)
- 2(2733.99)(36.50) Cov(n, D)
because Cov(n, S) = Cov(D, S) = O. Since the standard deviations of roughness, pipe

diameter, and slope are

an = (0.05)(0.015) = 0.00075
<TD - (0.02)(3.0) - 0.06
a5 = (0.05)(0.005) = 0.00025
the variance of the sewer flow capacity can be computed as
a2Q (2V33.99)2 (5.625 X 10~7) + (36.5O)2 (3.6 X 10~4)
+ (4100.99)2 (6.25 X 10~8) - 2 (2733.99)(36.50) (-3.375 X IQ-5)
= 2.052 + 2.192 + 1.032 + 6.74 = 16.79 ftVs2
Hence, the standard deviation of the sewer flow capacity is V16.79 =4.10 ftVs which is
10.0 percent of the estimated mean sewer flow capacity.
Without considering correlation between n and D, Gj = 16.79 6.74 = 10.05
which underestimates the variance of the sewer flow capacity. The percentages contribution of uncertainty of n, D, and S to the overall uncertainty of the sewer flow capacity under the uncorrelated condition are, respectively, 41.8 percent, 47.7 percent, and
10.5 percent. The uncertainty associated with the sewer slope contributes less significantly to the total sewer flow capacity uncertainty as compared with the other two random variables even though it has the highest sensitivity coefficient among the three.
This is because the variance of 5, Var(S), is smaller than the variances of the other two
random variables.
7.2.4 Approximate Technique: Rosenblueth's Probabilistic Point
Estimation (PE) Method
Rosenblueth 's probabilistic point estimation (PE) method is a computationally straightforward technique for uncertainty analysis. It can be used to estimate statistical moments
of any order of a model output involving several random variables which are either correlated or uncorrelated. Rosenblueth's PE method was originally developed for handling
random variables that are symmetric (Rosenblueth, 1975). It was later extended to treat
nonsymmetric random variables (Rosenblueth, 1981).
Consider a model, W = g(X), involving a single random variable X whose first three
moments or probability density function (PDF)/ probability mass function (PMF) are
known. Referring to Fig. 7.1, Rosenblueth's PE method approximates the original PDF or
PMF of the random variable X by assuming that the entire probability mass of X is concentrated at two points x_ and x+. Using the two point approximation, the locations of x_
and x+ and the corresponding probability masses p_ and/?+ are determined to preserve the
first three moments of the random variable X. Without changing the nature of the original
problem, it is easier to deal with the standardized variable, X = (X\i)/G, which has zero
mean and unit variance. Hence, in terms of X, the following four simultaneous equations
can be established to solve for jc'_, *'+, p_, andp + :
P+ + p_ = 1

(7.2Oa)

FIGURE 7.1 Schematic diagram of Rosenblueth's PE method in univariate case

(Tung, 1996)
P+X' + -p_*'_ = Mx, =0

(7.2Ob)

p+x'+ + /?_jc'_2 = (Ty = 1

(7.2Oc)

P+x'+-P_x>

_ = Y

(7.2Od)

in which x'_ = \x_ |i|/<7, jc'+ = |*+ - |i|/a, and y is the skew coefficient of the random
variable X. Solving Eqs. (7.20a-d) simultaneously, one obtains
Y
/
/Y\ 2
+
1+
^ =I V
(I)
+

(7 21a)

x'_=x'+-<f

(7.21b)

P+ = -T-V-TJt + + X _

(7.2Ic)

P^ = I-P+
(7.2Id)
When the distribution of the random variable X is symmetric, that is, y = O, then Eqs.
(7.21a-d) are reduced to x'_ = x\ = 1 andp_ = p+ = 0.5. This implies that, for a symmetric random variable, the two points are located at one standard deviation to either side
of the mean with equal probability mass assigned at the two points.
From x'_ and jc' + the two points in the original parameter space, x_ and X+, can respectively be determined as
x_ = \i-x'_a

(7.22a)

X+ = \L + X'+G

(7.22b)

Based on x_ and X+9 the values of the model W = g(X) at the two points can be computed, respectively, as w_ = g(x_) and W + = g(x+). Then, the moments about the origin
ofW = g(X) of any order can be estimated as
E[W-] = [i\m ~ p+ w+- + p_ w_

(7.23)

Unlike the FOVE method, Rosenblueth's PE estimation method provides an added

capability allowing analysts to account for the asymmetry associated with the PDF of a
random variable. Karmeshu and Lara Rosano (1987) show that the FOVE method is a first
order approximation to Rosenblueth's PE method.
In a general case where a model involves W correlated random variables, the
m* moment of the model output W = g(X^ X2, ..., XN) about the origin can be approximated as
E(W-) 2p(81t82i...iWO [w(61?52)...)6AO]i

(7.24)

in which the subscript 8- a sign indicator and can only be + or representing the random variable X1 having the value of Jt1+= p.,-+JtV+ a. or xt_ = ji.-jt'^a,, respectively; the
probability mass at each of the 2N points, p(51t 52,..., 5Ao can be approximated by

N
with

N-I ,

p^...^ =n
1=1 M + E ( E
i = 1 \j = i + 1M /
_

(7 25)

P/ 2N

"JFi]
where p.. is the correlation coefficient between random variables Xi and Xj. The number of
terms in the summation of Eq. (7.24) is 2N which corresponds to the total number of possible combinations of + and for all N random variables.
Example 7.3 Referring to Example 7.2, assume that all three model parameters in
Manning's formula are symmetric random variables. Determine the uncertainty of sewer
flow capacity by Rosenblueth's PE method.
Solution: Based on Manning's formula for the sewer, Q = 0.463 n~l D261 S 05 , the standard deviation of the roughness coefficient, sewer diameter, and pipe slope are
Cn = 0.00075; G0 = 0.06; Gs = 0.00025
With N = 3 random variables, there are a total of 23 = 8 possible points to be considered
by Rosenblueth's method. Since all three random variables are symmetric, their skew
coefficients are equal to zero. Therefore, according to Eqs. (7.21a-b), ri_ n'+ = U _ =
D'+ = S'_ = S1+ = 1 and the corresponding values of roughness coefficient, sewer diameter, and pipe slope are
n+ = JiIn + ri+ Gn = 0.015 + (1)(0.00075) = 0.01575
/i_ = M* ~ ri- Vn = 0.015 - (1X0.00075) = 0.01425
D+ = [I0 + D'+ G0 =3.0 + (1)(0.06) = 3.06 ft
D_ = \iD- U_ G0 = 3.0 - (1)(0.06) = 2.94 ft
S+ = \is + 5"+ G5 = 0.005 + (1)(0.00025) = 0.00525
S_ = \LS- S'_ G8 = 0.005 - (1)(0.00025) = 0.00475
Substituting the values of n_,n + ,>_,>+, 5_, and S+ into Manning's formula to compute the corresponding sewer capacities, one has, for example,

Q+++ = 0.463 (/I+)-1 (D+)M (S+)05

- 0.463 (0.01575)-1 (3.06)267 (0.00525)05 = 42.19 fWs
Similarly, the values of sewer flow capacity for the other seven points are given in the
following table:
Point
1
2
3
4
5
6
7
8

n
+
+
+
+
-

D
+
+
-

S
+
+

+
+
-

Q(ft3s)
42.19
40.14
37.92
36.07
46.64
44.36
41.91
39.87

+
-

p
0.03125
0.03125
0.21875
0.21875
0.21875
0.21875
0.03125
0.03125

Because the roughness coefficient and sewer diameter are symmetric, correlated random variables, the probability masses at 23 = 8 points can be determined, according to
Eqs. (7.25-7.26) as
P+++ = P = d + PnD + Pns +
p++_ = p__+ = (1 + PnD - pns P+^+ = p_ + _ - (1 - pnD + pns /?_ ++ = P+^ = (l~ pnD - Pns +

Pi*)/8 = d - 0.75 + O + 0)/8 = 0.03125

pDS)/8 = (1 - 0.75 - O - 0)/8 = 0.03125
pD5)/8 = (1 + 0.75 + O - 0)/8 = 0.21875
Pi)/8 = d + 0.75 - O + 0)/8 = 0.21875

The values of probability masses also are tabulated in the last column of the above
table. Therefore, the mth order moment about the origin for the sewer flow capacity can be
calculated by Eq. (7.24). The computations of the first two moments about the origin are
shown in the following table in which columns (l)-(3) are extracted from the above table.
Point
(1)
1
2
3
4
5
6
7
8
Sum

<2(ft3)
(2)
42.19
40.14
37.92
36.07
46.64
44.36
41.91
39.87

p
(3)
0.03125
0.03125
0.21875
0.21875
0.21875
0.21875
0.03125
0.03125
1.00000

QXp
(4)
1.318
1.254
8.295
7.890
10.203
9.704
1.310
1.246
41.219

Q2
(5)
1780.00
1611.22
1437.93
1301.04
2175.29
1967.81
1756.45
1589.62

Q2Xp
(6)
55.625
50.351
314.547
284.603
475.845
430.458
54.889
49.676
1715.99

3
From the above table, \IQ = E(Q) = 41.22 ft3 and E(Q2) = 1715.994 /ft
I V
J . Then, the
variance of the sewer flow capacity can be estimated as
\s/

/ft 3 V
Var(0 = E(Q2) - faQ)2 = 1715.994 - (41.219)2 = 16.988
Vs/
Hence, the standard deviation of sewer flow capacity is V16.988 = 4.12 fWs.
Comparing with the results in Example 7.2, one observes that Rosenblueth's PE method
yields higher values of the mean and variance for the sewer flow capacity than those
obtained by the FOVE method.
7.2.5 Approximate Technique: Hair's Probabilistic
Point Estimation (PE) Method
To avoid the computationally intensive nature of Rosenblueth's PE method when the
number of random variables is moderate or large, Harr (1989) proposed an alternative
probabilistic PE method which reduces the required model evaluations from 2N to 2N and
greatly enhances the applicability of the PE method for uncertainty analysis of practical
problems. The method is a second-moment method which is capable of taking into account
of the first two moments (that is, the mean and variance) of the involved random variables
and their correlations. Skew coefficients of the random variables are ignored by the
method. Hence, the method is appropriate for treating normal and other symmetrically
distributed random variables. The theoretical basis of Harr's PE method is built on orthogonal transformations of the correlation matrix.
The orthogonal transformation is an important tool for treating problems with correlated random variables. The main objective of the transformation is to map correlated random variables from their original space to a new domain in which they become uncorrelated. Hence, the analysis is greatly simplified.
Consider Af multivariate random variables X = (X1, X2, ..., XN)1 having a mean vector
JLlx = (JUt1, |I2, ..., jn^)' and the correlation matrix R(X)
1

Pi2 Pi3
Pii * P23

- PIAT
PIN

C (X') = R ( X ) =

PM PATC P*3

Note that the correlation matrix is a symmetric matrix, that is, pj;/. = p{j for i = j.
The orthogonal transformation can be made using the eigenvalue-eigenvector decomposition or spectral decomposition by which R(X) is decomposed as
R(X) = Cf X') = V A V'

(7.27)

where V is an N X N eigenvector matrix consisting of Af eigenvectors as V = (V1, V2, ...,

\N) with V1. being the Ith column eigenvector and A = diag(A<1, X2, ..., KN) is a diagonal
eigenvalues matrix.
In terms of the eigenvectors and eigenvalues, the random vector in the original parameter space can be expressed as
X = n + D^ VA"2 Y

(7.28)

in which Y is a vector of N standardized random variables having O as the mean vector

and the identity matrix, I, as the covariance matrix, and D is a diagonal matrix of variances of N random variables. The transformed variables, Y, are linear functions of the
original random variables, therefore, if all the original random variables X are normally
distributed, then the standardized transformed random variables, Y, are independent standard normal random variables.
For a multivariate model W = g (X1, X2, ..., XN) involving Nrandom variables, Harr's
method selects the points of evaluation located at the intersections of N eigenvector axes
with the surface of a hypersphere having a radius of VTV in the eigenspace as
xi = Ji VTV7D1/2 V1, / = 1, 2, ..., TV

(7.29)

in which xi represents the vector of coordinates of the N random variables in the parameter space corresponding to the ith eigenvector V 1 ; |il = (U1, U 2 ,..., uN)r, a vector of means
of TV random variables X.
Based on the 2TV points determined by Eq. (7.29), the function values at each of the 2TV
points can be computed. Then, the mm moment of the model output W about the origin
can be calculated according to the following equations:

wm + wm
pm(x } + Qm (r }
wrn=^i_^_^_ = S (xi+) -r g (x._) ^

fof

.= ^ ^

^ N-9m=i9 2, ...

(7.30)

TV

]>><
E[W] = Un(W) =
, for m = 1, 2,...

(7.31)

Alternatively, the orthogonal transformation can be made to the covariance matrix.

Example 7A. Referring to Example 7.2, determine the uncertainty of the sewer flow
capacity using Harr's PE method.
Solution: From the previous example, statistical moments of random parameters in
Manning's formula,
Q = 0.46Sn- 1 D 267 S 05
are

[In = 0.01500; \JLD = 3.00; Ji5 = 0.00500

On = 0.00075; OD = 0.06; G5 = 0.00025
From the given correlation relation among the three random variables, the correlation
matrix can be established as

1.00 -0.75 000

R(n, D, S) = -0.75 1.00 000
000 000 1.00

The corresponding eigenvector matrix and eigenvalue matrix are, respectively,

v

V = [V1 V2 V3] =

n V i2 Vi3
0.7071
V
V
2i 22 23 = -0.7071
V
3i V32 V 33j L o.OOOO

0.7071 0.0000
0.7071 0.0000
0.0000 1.0000.

and

A = diag (A1, A2, A3) = diag (1.75, 0.25, 1.00)

According to Eq. (7.29), the coordinates of the 2 X 3 = 6 intersection points corresponding to the three eigenvectors and the hypersphere with a radius Vs" can be determined as
^

<* O O

xi = M-D
\LS

V3

0.015

GD O V1.
O O G5

0.00075 O

= 3.0 V3
0.005

O
O

0.06 O
O
0.00025

v/

, f o r / = 1,2,3

The resulting coordinates at the six intersection points from the above equation are listed in column (2) of the table given below. Substituting the values of x in column (2) into
Manning's formula, the corresponding sewer flow capacities are listed in column (3).
Column (4) lists the value of Q2 for computing the second moment about the origin later.
After columns (3) and (4) are obtained, the averaged value of Q and Q2 along each eigenvector are computed and listed in columns (5) and (6), respectively.
Point
(1)

x - (TI, D, S)
(2)

Q
(3)

Q2
(4)

1+
12+
23+
3-

(0.01592,2.9265,0.00500)
(0.01408,3.0735,0.00500)
(0.01592,3.0735,0.00500)
(0.01408,2.9265,0.00500)
(0.01500,3.00,0.00543)
(0.01500,3.00,0.00457)

36.16
46.61
41.22
40.89
42.74
39.20

1307.82
2172.14
1699.09
1671.99
1826.45
1537.17

~Q
(5)

~Q2
(6)

41.39

1739.98

41.05

1685.54

40.97

1681.81

The mean of the sewer flow capacity can be calculated, according to Eq. (7.31), with
m = 1, as
^

_ A1 Q1 + A 2 Q 2 + A3 Q3
A1 + A2 + A3
=

1.75 (41.39) + 0.25 (41.05) + 1.00 (40.97)

^
= 41.22 ftVs

The second moment about the origin is calculated as:

^*
=

_ X 1 Qf + ^ 2 Ql + ^ 3 Q ?
K1 + K2 + ^3

1.75 (1727.11) + 0.25(1673.12) + 1.00(1669.54)

^^ ^

The variance of the sewer flow capacity then can be calculated as

Var(0 = E(Qi) - (JLi2)2 - 1716.05 - (41.22)2 =

16.82

(ft3/s)2.

Hence, the standard deviation of sewer flow capacity is V16.82 = 4.10 ftVs.
Comparing with the results in Examples 7.2 and 7.3, one observes that the mean and variance of the sewer flow capacity computed with Hair's PE method lie between those computed with the FOVE method and Rosenblueth's method.

7.3 RELIABILITY ANALYSIS METHODS

In a multitude of hydraulic engineering problems uncertainties in data and in theory,
including design and analysis procedures warrant a probabilistic treatment of the problems. The risk associated with the potential failure of a hydraulic engineering system is
the result of the combined effects of inherent randomness of external loads and various
uncertainties involved in the analysis, design, construction, and operational procedures.
Hence, to evaluate the probability that a hydraulic engineering system would function as
designed requires performing uncertainty and reliability analyses.
As discussed in Sec. 7.1.2, the reliability, ps, is defined as the probability of safety (or
non-failure) in which the resistance of the structure exceeds the load, that is,
ps = P(L^ R).

Conversely, the failure probability, pf, can be computed as

pf=P(L>R)=l-p,
The above definitions of reliability and failure probability are equally applicable to
component reliability as well as total system reliability. In hydraulic design, the resistance
and load are frequently functions of a number of random variables, that is, L = g(XL) =
g(Xl9 X2, ..., XJ and R = h(XR) = h(Xm + 19 Xn + 29...,Xn) whereX1, X2, ..., Xn are random
variables defining the load function, g(XL), and the resistance function, h(XR).
Accordingly, the reliability is a function of random variables
ps = P Ig(X1) < h(XR)]

(7.32)

As discussed in the preceding sections, the natural hydrologic randomness of flow and
precipitation are important parts of the uncertainty in the design of hydraulic structures.
However, other uncertainties also may be significant and should not be ignored.
7.3.1 Performance Functions and Reliability Index
In the reliability analysis, Eq. (7.32) can alternatively be written, in terms of a performance function, W(X) = W(XL, XR)9 as
Next Page

CHAPTER 8
HYDRAULIC DESIGN FOR
ENERGY GENERATION

H. Wayne Coleman
C. Y. Wei
James E. Lindell
Harza Company
Chicago, Illinois

8.1

INTRODUCTION

This chapter describes the design aspects of hydraulic structures related to the production of hydroelectric power. These structures include headrace channels; intakes;
conveyance tunnels; surge tanks; penstocks; penstock manifolds; draft-tube exits; tailtunnels, including tail-tunnel surge tanks and outlets; and tailrace channels. The procedures provided in this chapter are most suitable for developing the preliminary
designs of hydraulic structures related to the development of the hydroelectric projects. To finalize designs, detailed studies must be conducted: for example, economic
analysis for the determination of penstock diameters, computer modeling of hydraulic
transients for surge tank design, and studies of physical models of intake and its
approach.

8.2

HEADRACECHANNEL

An open-channel called the headrace channel or power channel (canal) is sometimes

required to connect a reservoir with a power intake when the geology or topography is not
suitable for a tunnel or when an open-channel is more economical. The channel can be
lined or unlined, depending on the suitability of the foundation material and the projects
economics. Friction factors for various linings used for design are as follows:
Manning's n
Lining

Minimum.

Maximum

Unlined rock

0.030

0.035

Shotcrete

0.025

0.030

Formed concrete

0.012

0.016

Grassed earth

0.030

0.100

Headrace channels are generally designed and sized for a velocity of about 2 m/s
(6.6 ft/s) at design flow conditions. Economic considerations may result in some variation
from this velocity, depending on actual project conditions.
Channel sections are normally trapezoidal because this shape is easier to build for
many different geologic conditions. The bottom width should be at least 2 m (6.6 ft) wide.
Side slopes are determined according to geologic stability as follows: earth, 2H: 1V or flatter; and rock, IH: IV or steeper. The channel's proportionsbottom width versus depth
are largely a matter of construction efficiency. In general, the minimum bottom width
reduces excavation, but geologic conditions may require a wider, shallower channel. The
channel slope will result from the conveyance required to produce design velocity for
design flow.
Channel bends should have a center-line radius of 3W to 5W or more, where W is the
water surface width of the design flow. For this radius, head loss and the rise in the water
surface at the outer bank (superelevation) will be minimal. If the radius must be reduced,
the following formula can be used to estimate head loss hL:
hL = K6^

(8.1)

where Kb = 2 (W/RC), W = channel width, Rc = center-line radius, and V = mean velocity.

Superelevation will be as follows (Chow, 1959):
2W V2
AZ =T

(8.2)

where AZ = rise in water surface above mean flow depth.

Freeboard must include allowances for the following conditions: (1) static conditions with maximum reservoir level (unless closure gates are provided to isolate the
channel from the reservoir), (2) water surface rise (superelevation) caused by flow
around a curve, and (3) surge resulting from shut-off of flow downstream or sudden
increase of flow upstream. A forebay is provided at the downstream end of the headrace channel to facilitate one or more of the following: (1) low approach velocity to
intake, (2) surge reduction, (3) sediment removal (desanding), or (4) storage. The forebay should be designed to maintain the approach flow conditions to the intake as
smoothly as possible. As the minimum requirement, a small forebay should be provided to facilitate good entrance conditions to the intake. It should include a smooth transition to a section with a velocity not exceeding 0.5 m/s (1.64 ft/s) at the face of the
intake structure
A larger forebay could be required for upsurge protection during rapid closure of turbine gates for load rejection. The size would be determined on the basis of the freeboard
allowance for the entire headrace channel and on a hydraulic transient analysis of the
channel, if necessary.
Surge calculations should consider maximum and minimum friction factors, depending on which is more critical for the case under study. Hydraulic transient (surge) studies
are generally performed using a one-dimensional, unsteady open-channel-flow simulation
program. The computer model developed should be capable of simulating the operation
of various hydraulic structures, the effect of the forebay, and operation of the power plant.
Several advanced open-channel flow-simulation programs have been described by Brater
et al. (1996).

Exhibit 8.1 Sun Koshi hydroelectric project, Nepal.

(a) A view of the desanding basin (looking upstream) showing concrete
guide vanes.(
(b) Layout Of the desanding basin.
A large forebay is required if it will be used for diurnal storage-say, for a power peaking operation. In such a case, maximum and minimum operating levels would include the
required water volume, with the intake located below the minimum level. Such a forebay
also could accommodate the other three functions described above.
When the flow carries too much sediment and its removal is required to protect the turbines, a still larger forebay would be provided to function as a desanding basin (also
known as a desilting basin or desander). However, the desanding basin is more likely to
be located at the upstream end of the headrace channel. Exhibit 8.1 Illustrates a desending basin. The basin can be sized using the following equation (Vanoni, 1977):

Fall velocity i-f (m s ')

Particle diameter d (mm)

FIGURE 8.1 Settling velocity as a function of particle diameter.
(Dingman, 1984)
_ LVs
P = (I- eVD ) X 100%

(8.3)

where P = percentage of sediment of a particular size to be retained by the basin, L =

basin length, Vs = settling (fall) velocity of suspended particles, V = mean flow velocity,
and D = depth of the desanding basin. The settling velocity Vs for each particular sand
particle size can be estimated from Fig. 8.1. A separate sluicing outlet (or outlets) would
be provided to flush the desanding basin intermittently.

8.3

INTAKES

Most power intakes are horizontal, a few are vertical, and very few are inclined. Figures
8.2, 8.3, and 8.4 are examples of the three types of intakes. The horizontal intake is usually connected to a tunnel or penstock on a relatively small slope (up to 2-3 percent). The
vertical intake is frequently used in pumped-storage projects when the upper reservoir is
on high ground, such as a mountain top, and a vertical shaft-tunnel is the obvious choice.
An inclined intake is used when the topography, geology, or type of dam dictate a steeper slope for the downstream tunnel or penstock. Exhibits 8.2 and 8.3 are examples of
intakes on an arch dam and on a pump-storage upper reservoir.

El 54 O
Max. operating
T.W El 50 O-

Intake

Min. operating
TW El. 43 5 -

Units (150 o.c )

Flow

GENERA TOR
HALL

El 3325
Trashrock guide
Emergency gate slot

Penstock 5.O ID.

Service gate slot

ISO T bridge crane

-Draft tube gantry crone
Service Bay
EI.S75
Max. T.W. El 6.5
Normal T.W. 1.61
Min. T.W
EI.48
distributor
El. .5

Note: Dimensions in meters.

FIGURE 8.2 A typical horizontal intake. (Harza Engineering Co.)

Draft tubs gate

7 1217.25

e/ 73727

/. 64728

FIGURE 8.3 A typical vertical intake. (Harza Engineering Co.)

Dasigrt
flood
surcharge
El. 781
Max. power pool
El. 770.0
Normal min. El. 700.0
Extreme min. pool
(Drawdown limit)
El. 6OO O
Trashrack blocKout

Axis of dam
El.785.0

Service gate blackout

Appro*. 22O ft. (Unit I)
Appro*. 190 ft. (Unit 2)

penstock
E1.512

Stair house

Units (65 -O" o. C )

450 TPH. crone
Porcelain Al. insulated wall panel
Glass block panel
Porctlain Al. insulated wall panel
Max Operating
T.W. El. 475.0
El 4750
Normal Operating (2 Units)
T.W. El 430 O
MIn Operating
T.W. El. 427.0
distributor El. 417O
Draft tube gats

FIGURE 8.4 A typical inclined intake. (Harza Engineering Co.)

Spillway

Powerhouse

Main dam

Exhibit 8.2 (a)

Exhibit 8.2 Karun hydroelectric project, Iran

(a) A view of the dam (looking downstream) showing spillway
crest, radial gates, power intakes, and diversion tunnel entrace
structure.
(b) Layout of dam showing spillway, intake and powerhouse.
A variation on the three basic intake types is a tower structure, sometimes required for
selective withdrawal of water. The tower includes openings with tmshmcks and bulkheads at various levels, which permit water to be withdrawn from different depths to control temperature or water quality. Computer modeling of a reservoir's temperature and
water-quality structure is generally required to finalize the required opening sites.
Descriptions of several reservoir-simulation models can be found in Brater et al. (1996).
Figure 8.5 is an example of a multilevel intake tower structure for selective withdrawal.Exhibit 8.3 illustrates the intake structure for a pumped storage project.
Trashracks for power intakes are designed for a velocity of about 1 m/s (3.3 ft/s) when
the intake is accessible for cleaning. If a trashrack is not accessible for cleaning, the allowable velocity is approximately 0.5 m/s (1.6 ft/s). Trashrack bar spacing is dictated by turbine protection requirements, but clear spacing of 5cm (2 in) is typical. Although head
loss through trashracks depends heavily on the amount of clogging, the following can be
used for a clean trashrack, (U.S. Bureau of Reclamation, 1987);

33 TON CRANE

FOLLOWER
TRAiHWACK

NORMAL MAX
OPERATING WL.EL.20OO

EL.2057 MIGMMT LIFT

INTAKE HOUSE

POST'TCNSIONCO
ANCHOM(TYR)

HEATED ICE BOOM

3'-00IA. HOLE (TYR)
SHUT ten

AMCHOMS (TYP.)
TRASHRACK(TYP)
FLOW

HYDRAULIC HOIST

MINIMUM OPERATING
W.L.EL.I8SO
INTAKE OATC
12-0 W 24-0H

ENDOFPIER
BULKHEAD GATE
I4'-OW 32'-OH
FIGURE 8.5 A typical multi-level intake tower structure for selective withdrawal. (Harza Engineering Co.)
hL = K,^

(8.4)

A
(A Y
where Vn = velocity based on the net area, K = 1.45 0.45 -r hp L An = net area of
A
s {AgJ
trashrack and support structure, and Ag = gross area of trashrack and support structure.
An intake gate is generally provided when the power tunnel or penstock is long or
when a short penstock does not have a turbine inlet valve. This gate is provided for emergency closure against flow in case of runaway conditions at the turbine or penstock rupture. The effective area of the gate is usually about the same as that of the power tunnel
or penstock, but it is rectangular in shape, with a height that is the same as the conduit's
diameter and a width that is 0.8 X the conduit diameter. A bulkhead (or stop-logs) is provided upstream of the intake gate for servicing the gate. The trashrack slot might be used
for this function by first pulling the trashrack.

Exhibit 8.3 Rocky mountain pumped storage project, Georgia.

(a) Intake structure of the upper reservoir.
(b) Close-up view of the upper reservoir intake structure.
(c) Profile of the project including upper reservoir intake, power tunnel, and power
house.

Ar , ,at, Md access 6uiMn

0,rt,ub.g,

H 737.27

Exhibit 8.3 (c)

A hydraulic study is generally conducted for emergency closure of the intake gate.
The maximum turbine flow or runaway flow should be considered. The runaway flow
may be 50 percent higher than the normal turbine flow for a propeller turbine. In the
hydraulic study, the water levels and pressures, as well as flow into and from the gate
well, as a function of gate position are investigated (Fig. 8.6). With this information, critical gate loads can be determined for the gate and hoist. The gate also may be used for
penstock filling. A minimum gate opening of 10 to 15 cm (4-6 in) is usually specified
for this, but a special hydraulic study must be made to determine potential gate load and
vibration if the gate opens continuously by accident. In such cases, a generous gate well
or air vent must be provided downstream of the gate to provide relief once the tunnel or
penstock fills.
The head loss for a bulkhead or gate slot, including top opening, is generally about 0.1
of the local velocity head at the slot. The transition length (m or ft) Lt from gate section
to tunnel or penstock should be approximately:

WIRE ROPES
FACE OF
ROLLER TRACK
TRASHRACK

SEAL
EL.545

TOP SEAL

PENSTOCK
EL.5I2

FEET

FIGURE 8.6 A typical intake gate arrangement. (Harza Engineering Co.)

A =^

(8-5)

where V = tunnel/penstock velocity (m/s or ft/s), D = tunnel/penstock diameter

(m or ft), and C = 3.00 for units in metric systems or = 9.84 for units in English systems. The variation of velocity in the transition section should be as close to linear as
practicable.
Overall head loss for an intake includes trashrack, bellmouth (0.1 X V1IIg), gate slots,
and transition. The potential vortex formation for an intake should be checked using Fig.
8.7. Note that when the intake Froude number (V/VgD) exceeds 0.5, submergence
requirements increase dramatically, and the vortex formation is difficult to predict. In this
case, a physical model study should be carried out.

8.4

TUNNELS

When the powerhouse is situated a considerable distance from the intake and when geologic conditions permit, a tunnel is often used to convey the flow for power generation.
The size of the tunnel is dictated by economics: that is, construction cost is added to the
cost of head loss (loss of generating revenue) to obtain the minimum combined cost. This
determination is usually obtained by trial and error because the process does not lend itself

Horizontal intake
Vertical intake
Intakes with
vortex problems
Intakes with
no problems

FIGURE 8.7 Intake submergence and vortex formation.

(Gulliver and Arndt, 1991)

to a simple formula. The resulting tunnel velocity with the economic diameter is usually
in the range of 3 to 5 m/s (10 to 17 ft/s).
The shape of the excavated tunnel normally will approximate a square bottom and a circular top. The diameter of the circular top (or the width of the square bottom) should be larger than the required diameter. If the tunnel is lined with concrete, its cross section is likely to
be circular or have a square or trapezoidal bottom. If it is unlined or lined with shotcrete, the
excavated shape will remain, with some smoothing by filling the larger overbreak sections.
Lining is an economic consideration, balancing the cost of the lining with the power
loss caused by friction. Even an unlined tunnel will have lined sections, such as portals,
and sections where rock needs extra support for geologic stability. Friction factors for
design are as follows:
Manning's n
Lining
Minimum Maximum
Unlined
0.030
0.035
Shotcrete
0.025
0.030
Formed concrete 0.012
0.016
Minimum friction corresponds to new conditions and is used for turbine-rating and
pressure-rise calculations. Maximum friction corresponds to aging and is used for economic-diameter and pressure-drop calculations. Tunnel slope is dictated by construction
suitability and geology, with a minimum of 1:1000 for drainage during dewatered condi-

Exhibit 8.4 (a) Bath County pumped storage project, Virginia.

(a) Surge tank openings during construction (44-ft inside diamenter and 300-ft deep)

imping

Bl*2J^o"PO'
Sta'sssSfcr'oiBaisM
200 7. mm.

Exhibit 8.4 Bath County pumped storage project, Virginia. (Continued)

(b) Profile of the project including upper reservor intake, gate structure, surge tanks, power tunnel and powerhouse

tions. Tunnel horizontal bends generally have large radii for convenience of construction.
Vertical bends at shafts usually have a minimum radius of 3D to minimize head loss and
to provide constructibility.

8.5

SURGETANKS

Surge tanks generally are used near the downstream end of tunnels or penstocks to reduce
changes in pressure caused by hydraulic transients (waterhammer) resulting from load
changes on the turbines (ASCE, 1989; Chaudhry, 1987; Gulliver and Arndt, 1991; Moffat
et al., 1990; Parmakian, 1955; Rich, 1951; Wylie and Streeter, 1993; Zipparro and Hasen,
1993). A surge tank should be provided if the maximum rise in speed caused by maximum
load rejection cannot be reduced to less than 60 percent of the rated speed by other practical methods, such as increasing the generator's inertia or the penstock's diameter or by
decreasing the effective closing time of the wicket gates. In general, the provision of a
surge tank should be investigated if
I>,jj > 3 to 5 for units in m/s and m or
n

> 10 to 20 for units in ft/s and ft,

(8.6)

where L1. is the length of a penstock segment and V1 is the velocity for the segment
(Dingman, 1984). The term ^L1V1 is computed from the intake to the turbine and Hn is the
minimum net head.
Surge tanks normally are located as close as possible to the powerhouse for maximum
effectiveness and may be free-standing or excavated in rock. The tanks are usually vented to atmosphere or can be pressurized as air chambers. The latter is not used frequently
because of requirements of size, air compressors, and air tightness. Exhibit 8.4 illustrates
a pumped storage project with a surge tank. Figure 8.8 shows typical installations of surge
tanks for controlling hydraulic transients.
Surge tanks usually are simple cylindrical vertical shafts or towers, but other geometric designs are used when the surge amplitude is to be limited. For instance, an enlarged
chamber can be used at the top if upsurge might cause the water level to rise above the
ground surface. Similarly, an enlargement or lateral tunnel or chamber is sometimes used
near the bottom of the shaft if downsurge would caused the water level to drop below the
tunnel crown. When the geometry is a cylinder, analysis is relatively simple and can be
performed using design charts. If the geometry is more complicated, a hydraulic transient
simulation model is required to carry out the study (Chaudhry, 1987; Wylie and Streeter,
1993; Brater et al., 1996).
Hydraulic stability for a surge tank assures that surging is limited and brief after load
changes (Rich, 1951; Parmakian, 1955; U.S. Bureau of Reclamation 1980; Zipparro and
Hasen, 1993). The minimum cross-sectional area of a simple cylindrical surge tank
required for stability can be determined using the Thoma formula:
AL
^=2^H

(8 7)

'

where AST = minimum tank area, A = tunnel area between reservoir and surge tank, L =
tunnel length between reservoir and surge tank, g = gravitational acceleration, c =

COUNTY FOWCRPUUtT
2100 WBATH
funtf^Ster99
fitapinf t1985)
on Buck Cw*
MNOMU
FIGURE 8.8(a) Typical vented surge tank installation. Bath County powerplant (1985): 2100 MW pumped storage development on Back Creek, Virginia.

HW 1. 10250
r>n/wc*
FlOKf

Cl. 1035.0

lnttkt control building

InIfHt oil* Hot

EI tee o

30'-0' cti tit chtmtr

ElBtSOQ

0. 940
T.W. El. 900
1 674.1
O. 159.53

4.600' OMOOSE RIVER POWERPLANT (1987)

12 MW Development on Moose River
NEW YORK
FIGURE 8.8(b) Typical pressurized surge tank installation. Moose River powerplant (1987): 12 MW
development on Moose River, New York. (Harza Engineering Co.)

A#
1 ( Mi \
head loss coefficient = -F^- = y- T/2/^~~ , where A// = minimum head loss from

reservoir to surge tank, including tunnel velocity head W2g, and H - minimum net operating head on turbine.
For a simple surge tank (without an orifice), increase the diameter obtained from the
Thoma formula by 50 percent. For a typical surge tank with a restricted orifice, increase
the diameter by 25 percent. These increases are necessary to provide damping of the oscillation in a reasonable period of time.
Maximum upsurge in a cylindrical surge tank can be determined from Fig. 8.9. For a
given tank size, the optimum size of the orifice is based on the balanced head design so
that the maximum tunnel pressure below the surge tank equals the maximum upsurge
level.
Maximum downsurge in a cylindrical surge tank can be determined from Fig. 8.10.
Here again, the size of the orifice should be based on balanced head design as a first
attempt. However, since downsurge may differ from upsurge, and the required orifice size
may be different for the two purposes, shaping the orifice (i.e., changing the discharge
coefficient) by rounding the top or bottom may satisfy the two area requirements approximately.
For maximum upsurge, use the maximum normal headwater, minimum head loss
between reservoir and surge tank, and maximum plant flow. Assume full plant load-rejection (tripout) in the shortest reasonable time.
For maximum downsurge, use the minimum normal headwater, maximum head loss,
and accept load from 50 percent to 100 percent in the shortest reasonable time. At some
projects, such as pumped-storage plants, the load acceptance is criterion is more extreme;
full load acceptance, is O percent to 100 percent in the shortest reasonable time. The controlling criterion will be used to design the orifice on downsurge. When the surge tank
geometry is complex (noncylindrical), a computer model should be used to determine the
limiting surge levels. (See Brater et al., 1996, for available computer models). Freeboard

SURGE RATIO

F Cross-sectionol oreo of surge tonk...(sq.ft.)

L -Length of pipeline between reservoir ond surge tonk...(ft.)
ft Cross-sectionoi oreo of pipe.._(sq ft)
Accelerotion
grovity...(ft
SAg - Moi
imum surgeo<from
operotmgper
level$ecdueperIosec.)
instontoneous stopping of flow Oo(ft)
For turbine operotion__.5 Upsurge (ft.) For pump operotion.._S Oownsurge (ft.)
(Noreturn flow permitted through the pump)
Hf Pipe f net i on loss +velocity heod ony other pipe losses between surge tonk ond
reservoir ossocioted with OQ--(M.)
H|? Throttling loss for flow into 01 out of surge tonk ossocioted with Q0. .(ft.)

BALANCED DESIGN

MAXIMUM SURGE IN SURGE TANK DUE TO

INSTANTANEOUS STOPPING OFFLOW Qo
FIGURE 8.9 Maximum surge in surge tank due to instantaneous stopping of flow.
(Parmakian, 1955)

SURGE RATIO

F - Cross- iectionol oreo ot surge tonk...(sq.ft.)

A Cross-sectionol oreo of pipe...(sq.ft.)
Accelerotion
grovity...(ft.
perdue
sec.topermstonfoneous
sec.)
Sgg " Moiimum
surgeoffrom
stotic level
storting of Oe (ft.)
For turbine operotion...Sg Oownsurge (ft) For pump operotion...S- Upsurge (ft)
OMf 8 - Pipe
Flow demonded
by
the
turbine
or
dischorged
by
the
pump...l
eu. ft. persurge
sec.)tonk ond
friction loss + velocity heod + ony other pipe losses between
1
reservoir lojs
ossocioted
Hf. Throttling
for flowwith
intoQore.(ft.)
out of surge tonk ossocioted with 0( (ft.)

BALANCEODESIGN

MAXIMUM SURGE IN SURGE TANK DUE TO

INSTANTANEOUS STARTING OFFLOW Ot
FIGURE 8.10 Maximum surge in a surge tank resulting from instantaneous starting
of flow. (Parmakian, 1955)

for the surge tank is 10 percent of the computed rise in the water level in the surge tank
for upsurge and 15 percent of the drop in the water level for downsurge to maintain, submergence of the tank invert or the orifice to avoid admitting air into the penstock.
Pressurized air chambers are often used in pumping plants for surge protection. They
are used occasionally for power plants when the generating flow is not excessive. The
hydraulic characteristics of the chambers are complicated by the compressibility effects
of air and temperature, and the analysis does not lend itself to simple formulas and charts.
A computer model is required to verify performance. Fig. 8.8(B) shows a typical air chamber design for a hydropower plant.

8.6

PENSTOCKS

A penstock generally refers to a steel conduit or steel-lined tunnel connecting a reservoir

or surge tank to a powerhouse (ASCE, 1989, 1993; U.S. Bureau of Reclamation, 1967;
Chaudhry, 1987; Gulliver, and Arndt, 1991; Warnick et al., 1984; Wylie and Streeter,
1993; Zipparro and Hasen, 1993). It is used when the internal pressure is high enough to
make a concrete-lined tunnel or unlined rock tunnel uneconomical, particularly where
rock cover is low.
Penstock size is usually governed by project economics. The economical diameter is
determined by the minimum combined cost of construction and energy reduction caused
by head loss in the penstock. The energy loss decreases as the diameter of the penstock
increases while construction cost increase. As with tunnels, the most economical diameter can be determined more accurately by a trial-and-error procedure. The following variables are generally considered (U.S. Bureau Reclamation, 1967; Gulliver and Arndt,
1991):
1.
2.
3.
4.
5.
6.

Cost of pipe
Value of energy loss
Plant efficiency
Minor loss factor
Average head
Waterhammer effect

7. Surface roughness (friction factor)

8. Weight of steel penstock
9. Design discharge
10.Allowable hoop stress

For the assessment of a preliminary design or a feasibility level, the most economical
diameter can be estimated using the following formula (Moffat et al., 1990).
CP0-43
D. = -JpW

(8.8)

where De = the most economical penstock diameter (m or ft), H = the rated head (m or
ft), P = the rated capacity of the plant (kW or hp), and C = 0.52 (for metric units) or =
3.07 (for English units). If the project is a small hydropower installation, the following
simple equation can be used (Warnick et al., 1984).
D6 = CQ0*

(8.9)

where De = the most economical penstock diameter (m or ft), Q = the design discharge
(m3/s or ftVsec), and C = 0.72 (for metric units) or = 0.40 (for English units).

For large hydroelectric projects with heads varying from approximately 60 m (190 ft)
to 315 m (1,025 ft) and power capacities ranging from 154 MW to 730 MW, the following equation can be used (Warnick et al., 1984).
O?0-43
. = -%r

(8-10)

where D6 = the most economical penstock diameter (m or ft), p = the rated turbine capacity (kW or hp), h = the rated net head (m or ft), and C = 0.72 (for metric units) or = 4.44
(for English units).
The maximum velocity in the penstock is normally kept lower than 10 m/s (33 ft/s).
To determine the minimum thickness of the penstock, based on the need for stiffness, corrosion protection, and handling requirements, the following formula can be used (U.S.
Bureau of Reclamation, 1967; Warnick et al., 1984).

D+K
*min = 400

(8 U)

'

where tmin = the minimum thickness of the penstock (mm or in), D = penstock diameter
(mm or in), and K = 500 (for metric units) or = 20 (for English units). After determining
the economic diameter, check for the operating stability of the generating unit-penstock
combination using the following steps (Chaudhry, 1987; U.S. Bureau of Reclamation,
1980; Warnick et al., 1984).
1. Determine the mechanical starting time in seconds for the unit T1n as
lm

JGD^V-_
36XlO4P

^'^

or

(WR2W
~ 1.6 X 106P7

(8 13)

'

where GD = flywheel effect of the turbine and generator rotating parts used in
metric system (kg-m2), WR2 = flywheel effect of the turbine and generator rotating
parts in English system (lb-ft2) = 5.932 GD2, G = weight of rotating parts (kg), D
= 2 X radius of gyration of the rotating parts (m), W = weight of rotating parts
(Ib), R = radius of gyration of the rotating parts (ft), N = turbine speed (rpm), P =
maximum turbine output (kW), and P1 = maximum turbine output (hp).
Tm is the time for torque to accelerate the rotating mass from zero to rotational
speed. Together, the turbine runner in water, connecting shafts, and the generator
develop the flywheel effect WR2 or GD2. The WR2 can be determined using on the
following formulas:
WRL^ = 23,800 (^)

(8.14)

and

KVA\5/4
-JJST)
'

(8.15)

where Pd = turbine rated output (hp) and kVA = generator rated output (kilovoltamperes).
2. Determine the water column starting time for the penstock Tw as follows:
Z(LV)
= ~W~

<8-16)

where ^LV) = summation of product of length (measured from nearest open water
surface) and velocity for each segment of penstock from intake or surge tank to tailrace (m2/s or ftVsec), g = gravitational acceleration (m/s2 or ft/sec2), and H = minimum net operating head (m or ft).
3. In general, TJTj should be maintained greater than 2 for good operating stability
and to have reasonably good responses to load changes. If Tm/Tw2 is less than 2,
there are three possible solutions:
Increase WR2 or GD2 for the generator; this is relatively inexpensive for
increases of up to 50%.
Increase the penstock diameter; this is probably not economical, except for a
narrow range.
Add a surge tank or move the surge tank closer to the powerhouse.
A combination of these three possible solutions may be the most cost-effective solution. The following friction factors are recommended for designing steel penstocks:

Exhibit 8.5 A typical steel penstock branch structure

being fabricated

Penstock Age
New
Old

Manning's n
0.012
0.016

Use the value for new penstock to calculating turbine-rating and pressure-rise. To calculate pressure drop use the higher values.
Design pressure is determined on the basis of the turbine's characteristics and the closure rates of the wicket gates or needle valves. For Pelton turbines, closure rates are slow,
and design pressure rise is usually of the order of 20 percent of the static pressure head.
For Francis turbines, design pressure rise is usually 30 to 40 percent of the static pressure head, depending on the cost of steel lining required. A fast closure is desirable to
minimize speed rise and the potential for runaway conditions in the turbine. Detailed
pressure conditions are determined by a computer model that includes the water conductors and surge tank as well as the turbine discharge-speed characteristics and generator inertia. Many computer programs capable of simulating hydraulic transients are
described in Wylie and Streeter, 1993. Such computer simulation studies are often
required of turbine or governor manufacturers now as a part of the specifications.
Ultimately, the predicted pressure conditions are verified in the load rejection tests during unit start-up.
The profile for a free-standing penstock is based on the topographic and geologic conditions of the ground. In other cases, the penstock may consist of shaft and tunnel sections
that are largely lined with concrete, with a relatively short section of steel-lined penstock
near the powerhouse.
If the penstock is free-standing, the risk of penstock rupture is greater than it is for the
shaft and tunnel system. If there is a long tunnel section upstream of the free-standing penstock, an emergency closure valve is often added near the tunnel outlet. A hydraulic transient study is necessary to determine closure conditions (by accident or because of penstock rupture). A vent must be provided to admit air just downstream of the valve for penstock rupture and must be large enough to prevent collapse of the penstock from internal
subatmospheric pressure caused by water-column separation. A free-standing penstock
also requires small air inlet-outlet valves at local high points to remove air during filling
and admit air during dewatering.
8.6.1 Penstock Branches
A penstock often delivers water to more than one turbine. In such cases, the penstock is
branched in various ways to subdivide the flow.Exhibit 8.5 illustrates a typical steel penstock branch structure. When the powerhouse is normal to the penstock, several configurations are possible (Fig. 8.11). If the powerhouse is at an angle with the penstock, a manifold is used (Fig. 8.12).
Head losses in branches and manifolds depend on precise geometry and often are
developed by model studies. However, for a typical well-designed layout, the following
head loss coefficients can be used to estimate the head loss hb from the main into a
branch:
hb = Kb^

(8.17)

where V = branch velocity (m/s or ft/s); g = gravitational acceleration (m/s2 or ft/sec2); and
Kb = head loss coefficient 0.2 for symmetrical bifurcation, 0.3 for symmetrical trifurcation, and 0.2 for manifold branch.

MAIN PENSTOCK
SINGLE
BIFURCATION

MAIN PENSTOCK
TRIFURCATION

BRANCH
(TYPICAL)

2 UNITS

BRANCH
(TYPICAL)

3 UNITS

MAIN PENSTOCK

MAIN PENSTOCK

BOUBLE
Y-BRANCHING

BRANCH
(TYPICAL)

3 UNITS

DOUBLE
BIFURCATION
BRANCH
(TYPICAL)

4 UNITS

FIGURE 8.11(a) Schematic penstock branch configurations for powerhouse normal to the penstock.
The diameters of branched penstocks are usually determined so that the velocity is
increased significantly relative to the main penstock. Here again, the branch size is determined by economics so that construction and material costs added to cost of energy loss
are at a minimum. The lower limit for the size of the branch is the size of the turbine inlet
that is normally provided by the turbine manufacturer. If a turbine inlet valve is provided,
its diameter will either be equal to the inlet diameter or be between the inlet diameter and
the penstock branch diameter. This valve is usually a spherical type, and, as such, no head
loss occurs in the fully open position. Friction losses in the branch penstocks are calculated using the same friction factors used for the main penstock and the conduit lengths
up to the net head taps in the turbine inlet.

SINGLE BIFURCATION
TRIFURCATION

DOUBLE Y-BRANCHING
FIGURE 8.11(b) Configurations for single bifurcated, double y-branching, and trifurcated penstocks.
(Harza Engineering Co.).
8.7

DRAFT-TUBEEXITS

Draft-tubes are designed by considering the turbine's characteristics. The net head for the
turbine is based on pressure taps at the spiral-case inlet and near the draft-tube exit.
Therefore, any head losses which occur after the draft-tube pressure taps are subtracted
from the turbine net head. Because the exit head loss is generally considered to be the
average velocity head at the end of the draft-tube, a longer draft-tube with expansion to a
larger area would, in theory, reduce this loss. In actuality, however the flow is not uniform

MAIN PENSTOCK
MAIN PENSTOCK
BRANCH
(TYPICAL)

BRANCH
(TYPICAL)

3 UNITS

2 UNITS

MAIN PENSTOCK

BRANCH
(TYPICAL)

4 UNITS
FIGURE 8.12(a) Schematic penstock manifold configurations for a powerhouse
oriented at an angle with the penstock.

FIGURE 8.12(b) Penstock manifold for an installation with six units. (Harza Engineering Co.)

at this point; it is highly turbulent and swirling, and the true exit loss is difficult to define.
Current thinking is that further extension of the draft-tube is not economical. The rule of
thumb is to end the draft-tube when the mean velocity is about 2m/s and to base the exit
head loss on this velocity.
A trashrack is usually provided at the end of the draft-tube at a pumped-storage project to prevent entry of coarse debris during the pumping mode. However, during the generating mode with the trashrack in place, the trashrack is subject to vibration caused by
the concentration of flow and by swirling. This complicates the design of the trashrack
and increases its cost. The analysis of the rack is a combined hydraulic and structural one.
The hydraulic loadings consist of drag forces on rack bars that are dependent on velocity
patterns along with pulsation of pressure caused by swirling flow. The data on hydraulic
conditions can be obtained from a physical model (usually the model from the pump-turbine manufacturer) because fully developed mathematical models are not readily available to predict these forces. A structural mathematical model is then applied using the
hydraulic loadings obtained from the hydraulic model tests. By trial and error, the
trashrack is designed to withstand the flow-induced vibrations.

8.8

TAIL-TUNNELS

An underground power plant will have a tail-tunnel to deliver the flow to the downstream
river or lake. For a pumped-storage project, this tunnel provides flow both ways, because
it acts as the inlet tunnel during pumping.
For a conventional hydroelectric plant with generating only, the tunnel is usually pressurized.. However, if the turbines are the Pelton type, the tunnel is likely to be free flow
to maintain freeboard on the turbine. For a pumped-storage plant, the tunnel is most likely to be pressurized, because it must deliver water both ways. If the tunnel is pressurized
and is long enough, a surge chamber will be required to prevent large fluctuations of pressure on the turbines during load changes.
The number of tail-tunnels, usually one or two, is based on economics and constructability. From an operational standpoint, two tunnels are desirable to allow partial
operation of the plant even during maintenance or inspection of one of the tunnels.
However, two tunnels are usually more expensive than one, and usually only one will be
used unless its size becomes unmanageable. The limiting size is dictated by available
equipment and tunneling methods. These factors must be evaluated carefully when estimating the costs of one tunnel versus two tunnels.
A manifold is used to collect the flow from the individual draft-tubes and guide the
flow through a transition section to the tail-tunnel proper. This manifold is similar in concept to the penstock manifold, but generally the velocities are much lower. The velocity
at the end of the draft-tube is typically 2 m/s (7 ft/s) and 3 m/s (10 ft/s) at the tail-tunnel.
Therefore, head losses are not significant and the flow conditions are generally acceptable. A typical tail-tunnel manifold design is shown on Fig. 8.13.
8.8.1 Tail-Tunnel Surge Tanks
When an underground power plant has a significant length of pressurized tail-tunnel, a
surge tank is likely to be required. The procedures for sizing and determining extreme
surges are similar to the procedures used for surges in the head-tunnel, using the hydraulic
characteristics of the tail-tunnel instead of the head-tunnel. (Refer to Sec. 8.5). Figure 8.14
shows a typical tail-tunnel surge chamber.

GENERAL PLAN

FIGURE 8.13 A typical tail-tunnel manifold arrangement. (Harza Engineering Co.)

8.8.2 Tail-Tunnel Outlet Structures

The tail-tunnel outlet structure is typically a bulkhead structure, which might incorporate
some flow spreading for energy recovery. The spreading of the flow is an economic decision based on construction costs and the value of energy loss. Figure 8.15 shows a typical structure of a tail-tunnel outlet. If the project is the pumped-storage type, the outlet
structure will incorporate trashracks at the face of the structure, and the velocity at the
trashracks will be approximately 1.0 m/s (3.3 ft/s), because the racks tend to be self-cleaning during the generating mode.

8.9

TAILRACECHANNELS

If the outlet structure is a significant distance from the receiving waterway, a tailrace
channel will be required (Fig. 8.16). The sizing of the channel will be similar to that of the
headrace channel. (Refer to Sec. 8.2).

TAIL-TUNNEL SURGE CHAMBER

FIGURE 8.14 A typical tail-tunnel surge chamber. (Harza Engineering Co.)

STEEL CMATMC COVEK

PLAN EL. 1670.0

ROAO
STEEL CHATMC COVCH
2.0 FREEBOARD

FON TRASHRACK
(SLOT
SEE NOTE
31

BULKHEAD SLOT

INTAKE /OUTLET SECTION

BULKHEAD SHAFT SECTION

BULKHEAD SHAFT PLAN

LOWER RESERVOIR

INTAKE /OUTLET PLAN

NOTCS:
L ALL DIMENSIDATUM
ONS ANO2000
ELEVATONS
AAE M KTERS.SCA LEVEL.
UCOUCRBANEAN
2. PKOJCCT
ALL CONCKETEWLLFORCL.HAVE
HEADRACE
TAlRACE VNAfER
CONDUCTORS
A MMUUUANO COUPRESSI
E STRENClH
3. THE TRASHRACK BAR SP*CNC SHALL BC EQUAL TO 20 CU.
SCALE O IO 20 30 METERS
1 i SOO
THE ISRAEL ELECTRIC CORPORATION LIMITED
PARSA PUMPED-STORAGE PROJECT
WATER CONDUCTORS
INTAKE /OUTLET STRUCTURFQ

FIGURE 8.15 A typical tail-tunnel outlet structure. (Harza Engineering Co.)

General Plan of the Guri Project

Left embankment dam
(Final Stag*)
Left gravity dorr
(Final Stag*)
Exi
loft
gravistitnygdam
Flow
Raising of right
Existingdam
right
gravity

Extensiondamto main
gravity

Powerhouse No. I
Substoae U
Substag*

Powerhouse Na 2
Final Stage

Ridam
ght (Final
embankmentStage)
Trestle

FIGURE 8.16 Tailrace channels of the Guri Project. (EDELCA, Venezuela)

REFERENCES
American Society of Civil Engineer (ASCE), Civil Engineering Guidelines for Planning and
Designing Hydroelectric Developments: Vol. 2 Waterways, American Society of Civil Engineers,
New York, 1989.
American Society of Civil Engineer (ASCE), Steel Penstock, ASCE Manuals and Reports on
Engineering Practice No. 79, American Society of Civil Engineers, New York, 1993.
Brater, E. R, King,
H. W., J. E. Lindell, and C. Y. Wei, Handbook of Hydraulics, 7th ed., McGraw-Hill, New York,
1996.
Chaudhry, M. H., Applied Hydraulic Transients, 2nd ed., Van Nostrand Reinhold, New York, 1987.
Chow, V. T., Open-Channel Hydraulics, McGraw-Hill, New York, 1959.
Dingman, S. L., Fluvial Hydrology, W. H. Freeman, New York, 1984.
Gulliver, J. S., and R. E. A. Arndt, Hydropower Engineering Handbook, McGraw-Hill, New York,
1991.
Henderson, F. M., Open Channel Flow, Macmillan, New York, 1966.
Moffat, A. I. B., C. Nalluri, and R. Narayanan, Hydraulic Structures, Unwin Hyman, London, UK,
1990.
Parmakian, J., Waterhammer Analysis, Dover Publications, New York, 1955.
Rich, G. R., Hydraulic Transients, Dover Publications, New York, 1951.
U. S. Army Corps of Engineer (USAGE), Hydraulic Design Criteria, U.S. Army Corps of
Engineers, Waterways Experiment Station, Vicksburg, MS, 1988.
U.S. Bureau of Reclamation, Selecting Hydraulic Reaction Turbines, Engineering Monograph
No.20, Department of the Interior, 1980.

U. S. Bureau of Reclamation, Design of Small Dams, U.S. Department of the Interior, Denver, Co,
1987.
U. S. Bureau of Reclamation Welded Steel Penstocks, Engineering Monograph No.3, U.S.
Department of the Interior, Denver, Co, 1967.
Vanoni, V. A., ed., Sedimentation Engineering, American Society of Civil Engineers, New York
1977.
Warnick, C. C., H. A. Mayo Jr., J. L. Carson, and L. H. Sheldon, Hydropower Engineering,
Prentice-Hall, NJ, 1984.
Wylie, E. B., and V. L. Streeter, Fluid Transients in Systems, Prentice-Hall, Englewood Cliffs, NJ,
1993.
Zipparro, V. J., and H. Hasen, Davis' Handbook of Applied Hydraulics, 4th ed., McGraw-Hill, New
York, 1993.

CHAPTER 9
HYDRAULICS OF WATER
DISTRIBUTION SYSTEMS

Kevin Lansey
Department of Civil Engineering and Engineering Mechanics
University of Arizona
Tucson, Arizona
Larry W. Mays
Department of Civil and Enviromental Engineering
Arizona State University
Tempe, Arizona

9.1

INTRODUCTION

In developed countries, water service is generally assumed to be reliable and utility customers expect high-quality service. Design and operation of water systems require an
understanding of the flow in complex systems and the associated energy losses. This
chapter builds on the fundamental flow relationships described in Chap. 2 by applying
them to water distribution systems. Flow in series and parallel pipes is presented first and
is followed by the analysis of pipe networks containing multiple loops. Water-quality
modeling is also presented. Because solving the flow equations by hand for systems
beyond a simple network is not practical, computer models are used. Application of these
models is also discussed.
9.1.1 Configuration and Components of Water Distribution Systems
A water distribution system consists of three major components: pumps, distribution storage, and distribution piping network. Most systems require pumps to supply lift to overcome elevation differences and energy losses due to friction. Pump selection and analysis
is presented in Chap. 10. Storage tanks are included in systems for emergency supply or
for balancing storage to reduce energy costs. Pipes may contain flow-control devices,
such as regulating or pressure-reducing valves. A schematic of a distribution system is
shown in Fig. 9.1.
The purpose of a distribution system is to supply the system's users with the amount
of water demanded under adequate pressure for various loading conditions. A loading
condition is a spatial pattern of demands that defines the users' flow requirements. The
flow rate in individual pipes results from the loading condition and is one variable that
describes the networks hydraulic condition. The piezometric and pressure heads are other
descriptive variables. The piezometric or hydraulic head is the surface of the hydraulic
grade line or the pressure head (p/y) plus the elevation head (z):

Tank 2
Reservoir 1

Pump 9

FIGURE 9.1 Network schematic (from EPANET User's Menual, Rossman, 1994)

h = ?-+z

(9.1)

Because the velocity is relatively small compared to the pressure in these systems, the
velocity head typically is neglected. Heads are usually computed at junction nodes.
A junction node is a connection of two or more pipes or a withdrawal point from the
network. A. fixed-grade node (FGN) is a node for which the total energy is known, such
as a tank.
The loading condition may remain constant or vary over time. A distribution system is in
steady state when a constant loading condition is applied and the system state (the flow in all
pipes and pressure head at all nodes) does not vary in time. Unsteady conditions, on the other
hand, are more common and hold when the system's state varies with time. Extended-period
simulation (EPS) considers time variation in tank elevation conditions or demands in discrete
time periods. However, within each time period, the flow within the network is assumed to
be in steady state. The only variables in the network that are carried between time steps of an
EPS are the tank conditions that are updated by a conservation of mass relationship.
Dynamic modeling refers to unsteady flow conditions that may vary at a point and
between points from instant to instant. Transient analysis is used to evaluate rapidly varying changes in flow, such as a fast valve closure or switching on a pump. Gradually varied
conditions assume that a pipe is rigid and that changes in flow occur instantaneously along
a pipe so that the velocity along a pipe is uniform but may change in time. Steady, extended period simulation, and gradually temporally varied conditions are discussed in this
chapter. Transient analysis is described in Chap. 12.

9.1.2 Conservation Equations for Pipe Systems

The governing laws for flow in pipe systems under steady conditions are conservation of
mass and energy. The law of conservation of mass states that the rate of storage in a system is equal to the difference between the inflow and outflow to the system. In pressurized water distribution networks, no storage can occur within the pipe network, although
tank storage may change over time. Therefore, in a pipe, another component, or a junction node, the inflow and outflow must balance. For a junction node,
2Gin - SG0Ut = <7ext

(9-2)

where Qm and <2out are the pipe flow rates into and out of the node and gext is the external
demand or supply.
Conservation of energy states that the difference in energy between two points is equal
to the frictional and minor losses and the energy added to the flow in components between
these points. An energy balance can be written for paths between the two end points of a
single pipe, between two FGNs through a series of pipes, valves and pumps, or around a
loop that begins and ends at the same point. In a general form for any path,
I hu + S hpj =A
(9-3)
%
where hLi is the head loss across component i along the path, hp . is the head added by
pump J9 and A is the difference in energy between the end points of the path.
Signs are applied to each term in Eq. (9.3) to account for the direction of flow. A common convention is to determine flow directions relative to moving clockwise around the
loop. A pipe or another element of energy loss with flow in the clockwise direction would
be positive in Eq. (9.3), and flows in the counterclockwise direction are given a negative
sign. A pump with flow in the clockwise direction would have a negative sign in Eq. (9.3),
whereas counterclockwise flow in a pump would be given a positive sign. See the Hardy
Cross method in Sec. 9.2.3.1 for an example.

9.1.3 Network Components

The primary network component is a pipe. Pipe flow (Q) and energy loss caused by
friction (hL) in individual pipes can be represented by a number of equations, including
the Darcy-Weisbach and Hazen-Williams equations that are discussed and compared in
Sec. 2.4.2. The general relationship is of the form
hL = KQn

(9.4)

where K is a pipe coefficient that depends on the pipes diameter, length, and material and
n is an exponent in the range of 2. K is a constant in turbulent flow that is commonly
assumed to occur in distribution systems.
In addition to pipes, general distribution systems can contain pumps, control valves,
and regulating valves. Pumps add head hp to flow. As shown in Fig. 9.2, the amount of
pump head decreases with increasing discharge. Common equations for approximating a
pump curve are
hp=AQ> + BQ + hc

(9.5)

Pump head

Pump curve
Horsepower
curve

Flow rate
FIGURE 9.2 Typical pump curve

or
hp = hc- CQ-

(9.6)

where A, B, C, and m are coefficients and hc is the maximum or cutoff head. A pump
curve can also be approximated by the pump horsepower relationship (Fig. 9.2) of
the form
H-&

where Hp is the pump's water horsepower. Further details about pumps and pump selection are discussed in Chap. 10.
Valves and other fittings also appear within pipe networks. Most often, the head loss
in these components is related to the square of the velocity by
*= *%=K^

(9 8)

where hm is the head loss, and Kv is an empirical coefficient. Table 2.2 lists Kv values for
a number of appurtenances.
Pressure-regulating valves (PRVs) are included in many pipe systems to avoid
excessive pressure in networks covering varying topography or to isolate pressure
zones for reliability and maintaining pressures. Pressure regulators maintain a constant pressure at the downstream side of the valve by throttling flow. Mathematical
representation of PRVs may be discontinuous, given that no flow can pass under certain conditions.

Pipel

Pipe 2

Pipe 3

Pipel

Pipe 2
Pipe 3

FIGURE 9.3 Pipe systems. A: Series pipe system (not to scale)

B: Branched pipe
9.2

STEADY-STATE HYDRAULIC ANALYSIS

9.2.1 Series and Parallel Pipe Systems

The simplest layouts of multiple pipes are series and parallel configurations (Fig. 9.3). To
simplify analysis, these pipes can be converted to an equivalent single pipe, that have the
same relationship between head loss and flow rate as the original complex configuration.
Series systems, as shown in the Fig. 9.3A, may consist of varying pipe sizes or types.
However, because no withdrawals occur along the pipe, the discharge through each pipe
is the same. Since the pipes are different, head losses vary between each segment. The
total head loss from a to Ms the sum of the head losses in individual pipes,
^L = IX, = IXen'
(9-9)
Mp
Mp
where Ip are the set of pipes in the series of pipes. Assuming turbulent flow conditions and
a common equation, with the same nt for all pipes, a single equivalent pipe relationship
can be substituted:
hL = KeQ"
(9.10)
where Ke is the pipe coefficient for the equivalent pipe. Kt can be determined by combining Eqs. (9.9) and (9.10):
K1 = K1 + K2 + K3 + ... = 2) K1
(9.11)
U
P
Note that no assumption was made regarding Q, so Ke is independent of the flow rate.

Problem. For the three pipes in series in Fig. 9.3, (1) find the equivalent pipe coefficient, (2) calculate the discharge in the pipes if the total head loss is 10 ft, and
(3) determine the piezometric head at points b, c, and d if the total energy at the inlet
(pt. a) is 95 ft?
Solution. For English units, the K coefficient for the Hazen-Williams equation is
K

=c^^

<9-12)

where 0 is a unit constant equal to 4.73, L and D are in feet, and C is the Hazen-Williams
coefficient. Substituting the appropriate values gives K1 = 0.229, K2 = 0.970, and K3 =
2.085. The equivalent Ke is the sum of the individual pipes (Eq. 9.11), or K6 = 3.284.
Using the equivalent loss coefficient, the flow rate can be found by Eq. 9.10, or hL =
K6Q1*5. For hL equals 10 feet and K6 equals 3.284, the discharge is 1.83 cfs.
This relationship and K6 can be used for any flow rate and head loss. Thus, if the flow
rate was 2.2 cfs, the head loss by Eq. 9.10 would be hL = 3.284 (2.2)1-85 = 14.1 ft.
The energy at a point in the series pipes can be determined by using a path head-loss
equation of the form of Eq. (9.3). The total energy at Point, b is the total energy at the
source minus the head loss in the first pipe segment, or
H

b = Ha~ hL,i = 95- K1Q1-*5 = 95 - 0.229(1.83)!85 - 94.3 ft

Similarly, the head losses in the second and third pipes are 2.97 and 6.38 ft., respectively. Thus, the energy at c and d are 91.33 and 84.95 ft, respectively.
Two or more parallel pipes (Fig. 9.3B) can also be reduced to an equivalent pipe with
a similar K6. If the pipes are not identical in size, material, and length, the flow through
each will be different. The energy loss in each pipe, however, must be the same because
they have common end points, or
hA~hB = hL>l = ^2 = hL.

(9.13)

Since flow must be conserved, the flow rate in the upstream and downstream pipes
must be equal to the sum of the flow in the parallel pipes, or
G = d + G2 + - = S Qm
(9-14>
meMp
where pipe m is in the set of parallel pipes, Mp. Manipulating the flow equation (Eq. 9.4),
the flow in an individual pipe can be written in terms of the discharge by Q = (hL/K)1/n
Substituting this in Eq. (9.14) gives
(Hr 1 V1 (hj O VS (hr -I VS
+
+
+
2 HIT
hr
hr
(Kl)
(K2j
(Kl) -

<9-15)

As is noted in Eq. (9.13), the head loss in each parallel pipe is the same. If the same n is
assumed for all pipes, Eq. (9.15) can be simplified to
\( \ \/n ( 1 Vn ( 1 Vn
1
^r-, ( 1 Vn
( 1 Vn
+
+
o-Hfe)
(il nil
-H" sji) -"-(T] "">

Dividing by hLl/n isolates the follwing equivalent coefficient:

^, ( 1 \ln

( 1 \ln
=

n)
SJsJ ra
Because the K values are known for each pipe based on their physical properties,
K6 can be computed, then substituted in Eq. (9.10) to determine the head loss across the
parallel pipes, given the flow in the main pipe.

Problem. Determine the head loss between points A and B for the three parallel pipes.
The total system flow is 0.2 m3/s. Also find the flow in each pipe.
Solution. The head-loss coefficient K for each pipe is computed by Eq. (9.12), with </>
equal to 10.66 for SI units and L and D in meters, or K1 = 218.3, K2 = 27.9, and K3 =
80.1. The equivalent Ke is found from Eq. 9.17:
( 1 ^T85
( 1 ^T85
( 1 ^L85
( 1 "p5
+
+
=
=a313
K
K
hr
hr
hr
hr
1*1 J
( 2j
( 3j
(Ke)
or Ke = 8.58. By Eq. (9.10), the head loss is hL = K6Q^5 = 8.58*(0.2y** = 0.437 m.
The flow in each pipe can be computed using the individual pipe's flow equation and
K. For example, Q1 = (V^i)17185 = (0.437/218.3)1/185 = 0.035 mVs. Similarly, Q2 and Q3
are 0.105 and 0.060 m3/s, respectively. Note that the sum of the flows is 0.2 m3/s, which
satisfies conservation of mass.
9.2.2

Branching Pipe Systems

The third basic pipe configuration consists of branched pipes connected at a single junction node. As shown in Fig. 9.4, a common layout is three branching pipes. Under steady
conditions, the governing relationship for this system is conservation of mass applied at
the junction. Since no water is stored in the pipes, the flow at the junction must balance
Reservoir 1
H = 100 m
Reservoir 2
H - 60m
Pipel
Pipe 2
Junction
with
pressure, P
Pipe 3
Reservoir 3
H = 40m

FIGURE 9.4 Branched pipe system.

G1 + Q2 ~ G3 = O

(9.18)

where the sign on the terms will come from the direction of flow to or from the node. In
addition to satisfying continuity at the junction, the total head at the junction is unique.
Given all the pipe characteristics for each system in Fig. 9.4, the seven possible
unknowns are the total energy at each source (3), the pipe flows (3), and the junction
node's total head P(I). Four equations relating these variables are available: conservation
of mass (Eq. 9.18) and the three energy loss equations. Thus, three of the seven variables
must be known. Two general problems can be posed.
First, if a source energy, the flow from that source and one other flow or source energy is known, all other unknowns can be solved directly. For example, if the flow and
source head for reservoir 1 and pipe 1 are known, the pipe flow equation can be used to
find P by the following equation (when flow is toward the junction):
HA-? = hu = K1Qi"

(9-19)

If a flow is the final known (e.g., Q2), Q3 can be computed using Eq. (9.18). The source
energies can then be computed using the pipe flow equations for Pipes 2 and 3, in the form
of Eq. (9.19), with the computed P.
If the final known is a source head, the discharge in the connecting pipe can be computed using the pipe equation in the form of Eq. (9.19). The steps in the previous paragraph are then repeated for the last pipe.
In all other cases when P is unknown, all unknowns can be determined after P is
computed. P is found most easily by writing Eq. 9.18 in terms of the source heads.
From Eq. (9.19),
/n
= sign(Hsl-p{}-^^}
(9.20)
I Ki J
where a positive sign indicates flow to the node. Substituting Eq. (9.20) for each pipe in
Eq. (9.18) gives
Qj

F(P) = sign(Hsl - P/^V^f

+ sign(Hs2 Pf^f^+
I Kl
J
I K2 J
(IH,-Pl^f
<9-21)
sign(Hs3-P)\ *K3
T =O
V
i
)
If a pipe's flow rate is known, rather than the source head, the flow equation is not substituted; instead the actual flow value is substituted in Eq. (9.21). The only unknown in
this equation is P and it can be solved by trial and error or by a nonlinear equation solution scheme, such as the Newton-Raphson method.
The Newton-Raphson method searches for roots of an equation F(X). At an initial estimate of x, a Taylor series expansion is used to approximate F:
O = F(JC) + ^ L Ax + ^
L Ac2 +....
dx
dx2

(9.22)

where Ax is the change in x required to satisfy F(x). Truncating the expansion at the firstorder term gives a linear equation for Ax:
A*=--^
(9.23)
^/3x1,
The estimated x is updated by x = x + Ax. Since the higher order terms were dropped
from Eq. (9.22), the correction may not provide an exact solution to F(JC). So several iter-

ations may be necessary to move to the solution. Thus, if Ax is less than a defined criteria,
the solution has been found and the process ends. If not, the updated x is used in Eq. (9.23)
and another Ax is computed. In the three reservoir case, x=P and the required gradient
3F/3P is:
& = - L r p ^ i - ^ n ' - - 1 ' , P ^ 2 - P I i ' - " 1 ' , ri/^-pn'-- 1 '"!
+
+
3P
n[( K1 }
( K2 J
[ K3 }
\
= -(
\
+
1
+
I
InJT1IGiI11-1 ^2IG2I"-1 W^3IG3I--1 J

(924)}
^

F(P) is computed from Eq. (9.21) using the present estimate of P. AP is then computed
using AP = -F(P)/(dF/dP), and P is updated by adding AP to the previous P. The iterations continue until AP is less than a defined value. The Newton-Raphson method also can
be used for multiple equations, such as the nodal equations (Sec. 9.2.3.3). A matrix is
formed of the derivatives of each equation and the update vector is calculated.
Problem. Determine the flow rates in each pipe for the three-pipe system shown in Fig.
9.4. The friction factors in the table below assume turbulent flow conditions through a
concrete pipe (e = 0.08 cm).
Solution. Using the Darcy-Weisbach equation (n = 2), the K coefficients are computed using
8/L
K_
^2D5g

Pipe
D (cm)
L(m)
/[]
K
Reservoir elevation, H (m)

1
80
1000
0.0195
4.9
100

2
40
600
0.0235
113.8
60

3
40
700
0.0235
132.7
40

Iteration 1
In addition to the three discharges, the energy at the junction P also is unknown. To
begin using the Newton-Raphson method, an initial estimate of P is assumed to be 80 m,
and Eq. (9.21) is evaluated as follows:
F(P = 80.) = L(IOO - 8O)(Ii^f) + L(60 - 8O)MfI
I
^ 4.y
) ype\ i
^ 113.8 ) ypei
(
^
+ (40 - 80)f l 4 ^~^ Q i f
= 2.020 - 0.419 - 0.549= 1.052 mVs
^
I 132.7 ) J,ipe3
which states that flow enters the node at more than 1.052 m3/s, than leaves through pipes
2 and 3 with P = 80 m. Therefore, P must be increased. The correction is computed by
Eq. (9.23) after computing 3F/3P using Eq. (9.24):

aF
(
i
i
3P = " |V2*4.9*I2.020K2-1) + 2*113.8*I-0.419K2-1)

i
^
2*132.7*I-0.549|(2-1)J-

-(0.0505 + 0.0105 + 0.0069) = -0.0679

The correction is then

F(P)
1.052
=~W7= -TT00679 = 15'5m3P/P
The P for the next iteration is then P = 80 + 15.5 = 95.5 m.
The following iterations give
AP

Iteration 2: F(P = 95.5 m) = -0.247; ^F/^pP = 955 = -0.120; AP = -2.06 m,

P = 93.44 m
Iteration 3:F(P = 93.44 m) = -0.020; ^/3^ = 9344= -0.102; AP = -0.20 m,
P = 93.24 m.
Iteration 4: F(P = 93.24 m) = 7.xlO~4; ^F/^PP = 9324 = -0.101; AP = -0.006 m,
P = 93.25 m.
Stop based on F(P) or AP, with P = 93.25 m.
Problem. In the same system, the desired flow in pipe 3 is 0.4 m3/s into the tank. What are
the flows in the other pipes and the total energy required in Tank 3?
Solution. First, P is determined with Q1 and Q2 using Eq. 9.21. Then H53 can be calculated by the pipe flow equation. Since Q3 is known, Eq. (9.21) is
j_

j_

F(P) = Q1 + Q2 + G3 = sign(Hsl - P)P 1 "*! + si&*H* - P)PV-^F

~0-4 =
\ ^l J
\ K2 J
Iteration 1
Using an initial trial of P equal to 90 m, F(P) = 0.514 mVs. When evaluating Eq.
(9.24), only the first two terms appear since the flow in pipe 3 is defined, or
3F _ _( *
,
1 I
_ _Q QOQ
3P
U^iIGiI nK2\Q2\) P = wm

The correction for the first iteration is then - (0.514/-0.080) = 6.42 m, and the new
P is 96.42 m. The next two iterations are
Iteration 2: F(P = 96.42 m) = -0.112; ^F/3p|P = 9642m = -0.127; AP = -0.88 m,
P = 95.54 m
Iteration 3: F(P = 95.54 m) = -0.006; ^/3P P = 9554. = -0.115; AP = -0.05 m,
P = 95.49 m
To determine Hs3, the pipe flow equation (Eq. 9.20) is used with the known discharge, or
(\Ha - 95.49Iy2
Q3= -0.4 = sign(Ha - 95.49)1
^ J = HA = 74.26 m

9.2.3

Pipe Networks

A hydraulic model is useful for examining the impact of design and operation decisions.
Simple systems, such as those discussed in Sees. 9.2.1 and 9.2.2, can be solved using a
hand calculator. However, more complex systems, require more effort even for steady
state conditions, but, as in simple systems, the flow and pressure-head distribution
through a water distribution system must satisfy the laws of conservation of mass and
energy (Eqs. 9.2 and 9.3). These relationships have been written in different ways to solve
for different sets of unknowns.
Using the energy loss-gain relationships for the different components, the conservation
equations can be written in three forms: the node, loop, and pipe equations. All are nonlinear and require iterative solution schemes. The form of the equations and their common
solution methods are described in the next four sections. Programs that implement these
solutions are known as network solvers or simulators and are discussed in Sec. 9.5.
9.2.3.1 Hardy Cross method. The Hardy Cross method was developed in 1936 by Cross
before the advent of computers. Therefore, the method is amenable to solution by hand
but, as a result, is not computationally efficient for large systems. Essentially, the method
is an application of Newton's method to the loop equations.
Loop equations. The loop equations express conservation of mass and energy in
terms of the pipe flows. Mass must be conserved at a node, as discussed in Sec. 9.2.2 for
branched pipes. For all Nj junction nodes in a network, it can be written as
2 Qi =
(9-25>
Uj
Conservation of energy (Eq. 9.3) can be written for closed loops that begin and end at
the same point (AE = O) and include pipes and pumps as
E KG" ~ S CAfrC*. + BtPQiP + Clp) = O
(9.26)
ie/L
ip*Jp
This relationship is written for N1 independent closed loops. Because loops can be nested in the system, the smallest loops, known as primary loops, are identified, and each pipe
may appear twice in the set of loops at most. The network in Fig. 9.1 contains 3 primary
loops.
Energy also must be conserved between points of known energy (fixed-grade nodes). If
NfFGNs appear in a network, A^-I independent equations can be written in the form of
E K& - S (**<% + BiPQiP + <V = AEFGN
(9-27)
ieIL
ip*Ip
where A"FGN is the difference in energy between the two FGNs. This set of equations is
solved by the Hardy Cross method (Cross, 1936) by successive corrections to the pipe
flows in loops and by the linear theory method by solving for the pipe flows directly (Sec.
9.2.3.2).
Solution method. To begin the Hardy Cross method, a set of pipe flows is assumed
that satisfies conservation of mass at each node. At each step of the process, a correction
AQL is determined for each loop. The corrections are developed so that they maintain conservation of mass (Eq. 9.25), given the initial set of flows. Since continuity will be preserved, those relationships are not included in the next steps.
The method then focuses on determining pipe flows that satisfy conservation of energy. When the initial flows are substituted in Eqs. (9.26) and (9.27), the equations are not

likely to be satisfied. To move toward satisfaction, a correction factor AQL is determined

for each loop by adding this term to the loop equation or for a general loop
2 K1(Q1 + AQ1)" - 2 (Aip(Qip + AQLy + Bip(Qip + AQ1) + Cip) = AE (9.28)
ieIL
ipelp
Note that AE equals zero for a closed loop and signs on terms are added as described in
Sec. 9.1.3. Expanding Eq. 9.28 and assuming that AQ1 is small so that higher order terms
can be dropped gives
2 K1Q1 + n^ K1Qr1
/e/L
*e/L

&QL - &,<% + B!pQlp + Cip) +

ip*Ip

X dAipQipAQL + BlpAQL) = AE
ipeip

(9.29)

Given Qik the flow estimates at iteration k, Eq. (9.29) can be solved for the correction
for loop L as
AO = - &K>Ql- S C-V&* + BQ,* + cip} - AE)
L
1
^
'^

!*"" I + Sk2^e**+ **)|

*>/;,

(9.30)

In this form, the numerator of Eq. (9.30) is the excess head loss in the loop and should
equal zero by conservation of energy. The terms are summed to account for the flow direction and component. The denominator is summed arithmetically without concern for
direction. Most texts present networks with only closed loops and no pumps. Equation
(9.30) simplifies to this case by dropping the pump terms and setting AE to zero, or
-S K0U

-2 hu

WL = -TT
=
!,*,<&-'
n^hu/Qa
C
' L
/e/^

-F(ft)
=~aW
'*Q u

(9-31>

Comparing Eq. (9.31) with Eq. (9.23) shows that the Hardy Cross correction is essentially a Newton's method.
The AQ1 corrections can be computed for each loop in sequence and can be applied
before moving to the next loop (Jeppson, 1974) or corrections for all loops can be determined and applied simultaneously. Once the correction has been computed, the estimates
for the next iteration are computed by
Q*+i = Qi* + ^QL

(9-32)

Qk+l is then used in the next iteration. The process of determining corrections and
updating flows continues until the AQL for each loop is less than some defined value. After
the flows are computed, to determine the nodal heads, head losses or gains are computed
along a path from fixed-grade nodes to junction nodes.

The Hardy Cross method provides an understanding of principles and a tool for solving small networks by hand. However, it is not efficient for large networks compared with
algorithms presented in the following sections.
Problem. List the loop equations for the network shown in Fig. 9.5 using the direction of
flow shown. Then determine the flow in each pipe and the total energy at Nodes 4 and 5.
Solution The loop equations consist of conservation of mass at the five junction nodes
and the loop equations for the two primary loops and one pseudo-loop. In the mass balance equations, inflow to a node is positive and outflow is negative.
Node 1.
Node 2.
NodeS.
Node 4.
Node 5.
Loop I.
Loop II.
Pseudo-loop.

Q1 - Q2 - Q5 = O
Q2 + Q3 - Q6 = 2
-Q3 + Q4 -Q7 = O
Q5 - G8 = 1
Q6 + Q1 + Q, = 2
\2 + hL 6 - hLi8 - hL>5 = O = K2Qi + K6Ql - K8Qi - K5Qi
hLJ - hL>6 - hL 3 = O = K1Qt - K6Ql - K3QI
h L4 - hp + h L3 - h L2 - hLl - EFGN2 + E FGN , = O
= K4Ql - (ApGJ + BpQ4 + Cp + K3QI -K2Ql - K1Gf
~ EpGN,2 + EFGNJ

Because the Darcy-Weisbach equation is used, n equals 2. The loop equations assume
that flow in the clockwise direction is positive. Flow in Pipe 5 is moving counterclockwise and is given a negative sign for loop I. Flow in pipe 6 is moving clockwise relative
to loop I (positive sign) and counter clockwise relative to Loop II (negative sign).
Although flow is moving clockwise through the pump in the pseudo-loop, hp is given a
negative sign because it adds energy to flow. To satisfy conservation of mass, the initial
set of flows given below is assumed, where the values of K for the Darcy-Weisbach equation are given by

K
KDW

- ^ - 8^
~ TtDg ~ VD^;

The concrete pipes are 1 ft in diameter and have a friction

flow.
Pipe
1
2
3
4
5
K
1.611 2.417
2.417
1.611
3.222
Q
2.5
1.0
1.5
2.5
1.5

33)
(933)
factor of 0.032 for turbulent
6
3.222
0.5

7
4.028
1.0

8
2.417
0.5

Also, A^ = -6, Bp = O, and Cp = 135'.

Iterahtion 1. To compute the correction for the pseudo-loop, the numerator of Eq. (9.30) is
K, - (ApQl + Cp + K3Gi - K2Gi - K1G? - EFGN,2 + EFGNJ =
1.611(2.5)2 - (-6(2.5)2 + 135) + 2.417(1.5)2 - 2.417(1.O)2 - 1.611(2.5)2 - (10 + 100)
= -4.48

FIGURE 9.5 Example network (Note all pipes have diameters of 1 ft and friction factors equal to 0.032).

The denominator is
TiK4Q4 + 2ApQ4 + nK3Q3 + nK2Q2 + TiK1Q1
= 2 (1.611(2.5))+I2(-6)2.5I + 2(2.417(1.5)) + 2 |2.417(1.0) +2 1.611(2.5) = 58.20.
Thus, the correction for the pseudo-loop AQPL is
*.--%->
The correction for Loop I is computed next. The numerator of Eq. (9.30) is
K2Ql + K6Gi - K8Gi - K5Q52 =
2.417(1.O) + 3.222(0.5)2-2.417(0.5)2-3.222(1.5)2 = -4.63
2

and the denominator is

nK2Q2 + UK6Q6 + nK8<28 =
2(2.417(1.0))+ 2(3.222(0.5))+2(2.417(0.5))+2(3.222(1.5)) = 20.14
Thus, the correction for Loop 1, AQ1 is
Ag1 = -1^f1 = 0.230
Finally to adjust loop II from the numerator of Eq. (9.30) is
K7G72 - K6Ql - K3Ql =
4.028(1.O) - 3.222(0.5)2 -2.417(1.5)2 = -2.22
2

and the denominator is

nK7Q7+ nK6Q6 + nK3Q3 = 2(4.028(1.O)) + 2(3.222(0.5)) + 2(2.417(1.5)) = 18.53
Thus, the correction for the loop II, AQ11, is
.--*-<
The pipe flows are updated for Iteration 2 as follows:
Pipe
A<2

1
-0.077

2
3 4 a n d pump
5
6
7
8
0.230 0.0770.077
-0.230 0.2300.120
-0.230
-0.077 0.120
0.120
2.42 1.15 1.46 2.58 1.27 0.61 1.12
0.27

Because the flow direction for Pipe 1 is counterclockwise relative to the pseudo-loop,
the correction is given a negative sign. Similarly, Pipe 2 receives a negative correction for
the pseudo-loop. Pipe 2 is also in Loop I and is adjusted with a positive correction for that
loop since flow in the pipe is in the clockwise direction for Loop I. Pipes 3 and 6 also
appear in two loops and receive two corrections.

Iteration 2. The adjustment for the pseudo-loop is

A(

__ _ K4Qj - (AnQl + C0) + K3Qj - K2Q* - K,g + (EfCM2 -E^1) _

nK4<24 + 2Apg4 + nK3g3 + nK2Q2 + nK.fi!

1.611(2.58)2 - (-6(2.58)2 + 135) + 2.417(1.46^ - 2.417(1.15)^ - 1.611(2.42)^ -10+100

2(1.611(2.58)) + 2| - 6(2.58)1 + 2(2.417(1.46)) + 2(2.417(1.15)) + 2(1.611(2.42))
--&)"
In the correction for loop I, the numerator of eq. (9.30) is
K2Gi + K6Gi - K8Gi - K5Gi =
2

2.417(1.15) + 3.222(0.61)2 -2.417(0.27)2 - 3.222(1.27)2 = -0.978

and the denominator is
nK2G2 + nK6G6 + nK8G8 + nK5G5 =
2 (2.417(1.15)) + 2 (3.222(0.61)) + 2 (2.417(0.27)) + 2(3.222(1.27)) = 18.98
T^us the correction for loop, AG1 is
AG1 = -

~899878) = 0.052

Finally, to correct loop II, the numerator of Eq. (9.30) is

K7Q? - K6Gi ~ K3Gi = 4.028(1.12)2 - 3.222(0.61)2 -2.417(1.46)2 = -1.30
and the denominator is
nK7Q2 + HK6Gi + nK3Gi =
2*[4.028(1.12)] + 2[3.222(0.6I)] + 2[2.417(1.46)] = 20.01
Thus, the correction for the pseudo-loop, AGn is
AQ

~ Solf1 = -065

The pipe flows are updated for iteration 3 as follows:

Pipe
A0

1
-0.030

2.39

2
3 4 a n d pump
5
0.052+ 0.0300.030
-0.052
(-0.030) 0.065
1.17
1.43
2.61
1.22

6
0.0520.065
0.60

7
0.065

8
-0.052

1.18 0.22

Iteration 3. The corrections for iteration 3 are 0.012, 0.024, and 0.021 for the pseudo-loop, loop I and loop II, respectively. The resulting flows are as follows:

Pipe
AQ
Q

1
-0.012

2.38

2
0.024+
(-0.012)
1.18

3 4 a n d pump
5
0.012- 0.012 -0.024
0.024
1.42
2.62
1.20

6
7
8
0.024- 0.024 -0.024
0.024
0.60
1.20 0.20

After two more iterations, the changes become small, and the resulting pipe flows are
as follows. Note that the nodal mas balance equations are satisfied at each iteration.
Pipe
Q

1
2.37

2
1.19

3
4 and pump
1.41
2.63

5
1.18

6
0.60

7
8
1.21 0.18

The total energy at Nodes 4 and 5 can be computed by path equations from either FGN
to the nodes. For example, paths to Node 4 consist of Pipes 1 and 5 or of pipes 4 (with the
pump), 7, and 8. For the path with pipes 1 and 5, the equation is
100 - K1Q12- K5Q52= 100 - 1.611(2.37)2 - 3.222(1.19)2 = 100 - 9.05 - 4.56 - 86.39m
For the path containing pipes 4, 7 and 8 the result is
10 + (135 - 6(2.63)2) - 1.611(2.63)2 - 4.028(1.22)2 + 2.417(0.19)2 =
10 + 93.50 - 11.14 - 6.00 + 0.09 = 86.45m
This difference can be attributed to rounding errors. Note that pipe 8 received a positive sign in the second path equation. Because the flow in Pipe 8 is the opposite of the path
direction, the energy along the path is increasing from Nodes 5 to 4. The total energy at
Node 5 can be found along pipes 4 and 7 or 86.36 m or along the path of Pipes 1-2-6, giving (100 - 9.05 - 3.42 - 1.16 = 86.37m).
9.2.3.2 Linear theory method. Linear theory solves the loop equations or Q equations
(Eqs. 9.25 to 9.27). Np equations (Nj + N1 + Nf -1) can be written in terms of the Np
unknown pipe flows. Since these equations are nonlinear in terms of Q, an iterative procedure is applied to solve for the flows. Linear theory, as described in Wood and Charles
(1972), linearizes the energy equations (Eqs. 9.26 and 9.27) about Qi>k+1, where the subscript k+\ denotes the current iteration number using the previous iterations Q1k as known
values. Considering only pipes in this derivation, these equations are

and

Ea*+i = * far all Nj nodes

^J

(9.34)

^
K1QIk1 Qi,k+i = fr al1 N\ dosed loPs
e/
<L

(9-35)

2^ KiQi kl Qik+i = ^EFGN for all Nf I independent pseudoloops (9.36)

L
These equations form a set of linear equations that can be solved for the values
of Q1-^+1. The absolute differences between successive flow estimates are computed and
compared to a convergence criterion. If the differences are significant, the counter k is
updated and the process is repeated for another iteration. Because of oscillations in the
flows around the final solution, Wood and Charles (1972) recommended that the average of the flows from the previous two iterations should be used as the estimate for the

next iterations. Once the pipe flows have been determined, the nodal piezometric heads
can be determined by following a path from a FGN and accounting for losses or gains
to all nodes.
Modified linear theory Newton method. Wood (1980) and his collaborators at the
University of Kentucky developed the KYPIPE program but essentially modified the
original linear theory to a Newton's method. However, rather than solve for the change in
discharge (AQ), Qk+l *s determined.
To form the equations, the energy equations (Eq. 9.3) are written in terms of the
current estimate of Qk, including pipes, minor losses and pumps, as
K
/(ft) - E
Qi+ Ee/ Kim Ql + E < A *G? + B*& + CP - AE
(9-37>
ie/
L
<m m
ip^p
where for simplicity the subscripts /, /m, and ip denoting the pipe, minor loss component,
and pump, respectively, are dropped from the flow terms and k again denotes the iteration
counter. This equation applies to both closed loops (AE = O) and pseudo-loops (AE =
A"FGN), but, in either case, f(Qk) should equal zero at the correct solution.
To move toward the solution, the equations are linearized using a truncated Taylor
series expansion:

f(Qk+l) =f(Qk)

^Qk (G^+I - G4) =k) + G4(G4+I -

ft)

(9-38)

Note that/and Q are now vectors of the energy equations and pipe flow rates, respectively, and Gk is the matrix of gradients that are evaluated at Qk. Setting Eq. (9.38) to zero
and solving for Qk+l gives
or

0=f(Qk) + Gk(Qk+l-Qk)
GkQk+v = GkQk+f(Qk)

(9.39)

This set of (N1 + Nf-l) equations can be combined with the Adjunction equations in
Eq. (9.34) that also are written in terms of Qk+l to form a set of Np equations. This set of
linear equations is solved for the vector of Np flow rates using a matrix procedure. The
values of Qk+1 are compared with those from the previous iteration. If the largest absolute
difference is below a defined tolerance, the process stops. If not, Eq. (9.39) is formed
using Qk+l and another iteration is completed.
9.2.3.3 Newton-Raphson method and the node equations. The node equations are the
conservation of mass relationships written in terms of the unknown nodal piezometric
heads. This formulation was described in Sec. 9.2.2 for branching pipe system. In Fig. 9.4,
if P and the pipe flows are unknown, the system is essentially a network with one junction node with three FGNs. In a general network, Nj junction equations can be written in
terms of the Nj nodal piezometric heads. Once the heads are known, the pipe flows can be
computed from the pipe's head-loss equation.
Other network components, such as valves and pumps, are included by adding junction nodes at each end of the component. Node equations are then written using the flow
relationship for the component.
Solution method. Martin and Peters (1963) were the first to publish an algorithm
using the Newton-Raphson method for solving the node equations for a pipe network.
Shamir and Howard (1968) showed that pumps and valves could be incorporated and
unknowns other than nodal heads could be deternined by the method. Other articles have
been published that attempt to exploit the sparse matrix structure of this problem.

At iteration k, the Newton-Raphson method is applied to the set of junction equations

F(hk) for the nodal heads hk. After expanding the equations and truncating higher order
terms, the result is
3F
F(hk) + _ AJi4 = O
(9.40)
where F is the set of node equations evaluated at hk, the vector of nodal head estimates at
iteration k. dF/dh is the Jacobian matrix of the gradients of the node equations with
respect to the nodal heads. This matrix is square and sparse because each nodal head
appears in only two nodal balance equations. The unknown corrections A/^ can be determined by solving the set of linear equations:
r)F
F(hk)= -^- AA4
(9.41)
The nodal heads are then updated by:
A4+1 = hk + Uik

(9.42)

As in previous methods, the magnitude of the change in nodal heads is examined to

determine whether the procedure should end. If the heads have not converged, Eq. (9.41)
is reformulated with hk+l and another correction vector is computed. If the final solution
has been found, the flow rates are then computed using the component relationships with
the known heads.
As in all formulations, at least one FGN must be hydraulically connected to all nodes
in the system. Some convergence problems have been reported if poor initial guesses are
made for the nodal heads. However, the node equations result in the smallest number of
unknowns and equations of all formulations.
Problem. Write the node equations for the system in Fig. 9.5.
Node 1:
1
/1AA - /J,/'100-/I
. n - ;/I /l/*
-^lO"
_L sign(h
n - /II )NP^-^l'l"
1 Il^
22 ^
,WgAi(IOO
M
+ SIgH(H
+
* = On
1)
2
1)
4
1
V K\
J
\ K2 J
(. K5 )
Node 2 (note that the right-hand side is equal to the external demand):

sign(hl - /J^f
( K2 )

1
***> - 4
( ^
K3 f)

2
***> - 4^f
( A6 ) =

^5 - jyf^f
( K7 J

-^v - 414
( V^f=
K
) o

Node 3:
Sign(h2

- />( 3fv^f
K
3 )

Node 4:
sign(hl - J^f
(. K5 J

Si8n(H5 - J^V^l"
= 1
( As ;

Node 5:
sign(/,2 - hf^f
( K6

sig*/, - /yf^f + sign^ - J^f = 2

( A8 )
( K1 )

New node for the pump:

<*-.-v(^H^):=o
The first term in the pump node equation is the outflow from the pump toward Node
3 in Pipe 4. The second term is the discharge relationship for the pump, written in terms
of the total energy at the outlet of the pump hpd.
Because the pump relationship is different from that for Pipe 4, this new node with
zero demand was added at the outlet of the pump (assuming that the pump inlet is the
tank). This type of node must be added for every component (valve, pipe, or pump); therefore, one must know the precise location of the component. For example, if a valve
appears within a pipe, to be exact in system representation, new nodes would be added on
each side of the valve, and the pipe would be divided into sections upstream and downstream of the valve.
In summary, six equations can be written for the system to determine six unknowns
(the total energy for Nodes 1 to 5 and for the pump node). Using the solution from the
Hardy Cross method gives the following nodal heads, the values of which can be confirmed to satisfy the node equations:
Node
1
2
3
4
5
Pump
Total head (m)
90.95
87.54
92.35
86.45
86.38
103.50
Pipe
1
Pipe flow 2.37
(mVs)

2
1.19

3
4
1.41 2.63

5
1.18

6
7
8
Pump
0.60 1.22 0.18 2.63

9.2.3.4 Gradient algorithm Pipe equations. Unlike the node and loop equations, the
pipe equations are solved for Q and h simultaneously. Although this requires a larger set
of equations to be solved, the gradient algorithm by Todini and Pilati (1987) has been
shown to be robust to the extent that this method is used in EPANET (see Sec. 9.5.3).
To form the pipe equations, conservation of energy is written for each network component in the system in terms of the nodal heads. For example, a pipe equation is
ha-hb = KQn

(9.43)

and, using a quadratic approximation, a pump equation is

hb-ha = AQ* + BQ+ C

(9.44)

where ha and hb are the nodal heads at the upstream and downstream ends of the component. These equations are combined with the nodal balance relationships (Eq. 9.2) to form
TV7 + Np equations with an equal number of unknowns (nodal heads and pipe flows).
Solution method. Although conservation of mass at a node is linear, the component
flow equations are nonlinear. Therefore, an iterative solution scheme, known as the gradient algorithm, is used. Here the component flow equations are linearized using the previous flow estimates Qk. For pipes,
KQi~l QM + (ha - A4) = O

(9.45)

In matrix form, the linearized equations are

A12/* + A11G + A10A0 = O

(9.46)

AnQ - qm = O

(9.47)

and

where Eq. 9.46 is the linearized flow equations for each network component and Eq. 9.47
is the nodal flow balance equations. A12 (= A217) is the incidence matrix of zeros and ones
that identify the nodes connected to a particular component and A10 identifies the
fixed grade nodes. A11 is a diagonal matrix containing the linearization coefficients
(e.g.,\KQs-i DDifferentiating eqs. 9.46 and 9.47 gives:
[JVA11 A 12 IFdBl JdEl
[A21 O J U J [dq\

*'->

where dE and dq are the residuals of equations 9.2 and 9.43-44 evaluated at the present solution, Qk and hk. N is a diagonal matrix of the exponents of the pipe equation
(n). Eq. 9.48 is a set of linear equations in terms of dQ and dh. Once solved Q and h
are updated by
and

G ft+ i = Qk + dQ

(9.49)

hk+l = hk + dh

(9.50)

Convergence is checked by evaluating dE and dq and additional iterations are completed

as necessary.
Todini and Pilati (1987) applied an alternative efficient recursive scheme for solving
for Qk+l and hk+l. The result is
Vi = -(A2A-1Al1 A1^1{A17N^(Qk+^Al^ + (qat-A2lQl)}

(9.51)

then using hk+l, Qk+l by is determined:

G4+1 = (l-tf-OG* -tf-1 AT1 (A1A+1 +A10H0)

(9.52)

where A11 is computed at Qk. Note that N and A11 are diagonal matrices so the effort for
inversion is negligible. Yet, one full matrix must be inverted in this scheme.
Problem. Write the pipe equations for the network in Fig. 9.5.
Solution. The pipe equations include mass balance equations for each node in the system. The network contains five junction nodes plus an additional node downstream of the
pump. The pump is considered to be a link and is assumed to be located directly after the
FGN.
Conservation of energy equations are written for each pipe and pump link. Eight pipe
equations and one pump equation are written. The total number of equations is then 15,
which equals the 15 unknowns, including 8 pipe flows, 1 pump flow, and 6 junction node
heads, including the additional nodal head at the pump outlet hp.
Node 1:

Q1 - Q2 - Q5 = O

Pipe 1: 100 -H1 = K,Q>{

Node 2:
Node 3:
Node 4:
Node 5:
Pump Node:
Pump:
fcp

Q2 + Q3 - Q6 = 2
- Q3 + Q4 - Q1 = O
5 - G8 = 1
S6 + C7 + G8 = 2
Gp - G4 = O
- 10 = 135 - 6Q2p

Pipe 2:
Pipe 3:
Pipe 4:
Pipe 5:
Pipe 6:
Pipe 7:
Pipe 8:

H 1 - H 2 = K2Qi
H 3 - H 2 = K3Q]
hp-H3 = K4Q4
H 1 - H 4 = K5Qn5
H 2 - H 5 = K6Ql
H 3 - H 5 = K1Qf1
/I4 - /I5 = K,Ql

9.2.3.5 Comparison of solution methods. All four methods are capable of solving the
flow relationships in a system. The loop equations solved by the Hardy Cross method are
inefficient compared with the other methods and are dropped from further discussion. The
Newton-Raphson method is capable of solving all four formulations, but because the node
equations result in the fewest equations, they are likely to take the least amount of per iteration. In applications to the node equations, however, possible convergence problems may
result if poor initial conditions are selected (Jeppson, 1974).
Linear theory is reportedly best for the loop equations and should not be used for the
node or loop equations with the AG corrections, as used in Hardy Cross (Jeppson, 1974).
Linear theory does not require initialization of flows and, according to Wood and Charles
(1972), always converges quickly.
A comparative study of the Newton-Raphson method and the linear theory methods
was reported by Holloway (1985). The Newton-Raphson scheme was programmed in two
codes and compared with KYPIPE that implemented the linear theory. For a 200-pipe network, the three methods converged in eight or nine iterations, with the Newton-Raphson
method requiring the least amount of computation time.
Salgado, Todini, and O'Connell (1987) compared the three methods for simulating
a network under different levels of demand and different system configurations. Four
conditions were analyzed and are summarized in Table 9.1. Example A contains 66
pipes and 41 nodes but no pumps. Example B is similar to Example A, but 6 pumps are
introduced and a branched connection has been added. Example C is the same network
as in Example B with higher consumptions, whereas Example D has the same network
layout but the valves are closed in two pipes. Closing these pipes breaks the network
into two systems. The results demonstrate that all methods can simulate the conditions,
but the gradient method for solving the pipe equations worked best for the conditions
analyzed.
All comparisons and applications in this chapter are made on the basis of assuming
reasonably sized networks. Given the speed and memory available in desktop computers, it is likely that any method is acceptable for these networks. To solve extremely
large systems with several thousand pipes, alternative or tailored methods are necessary. Discussion of these approaches is beyond the scope of this chapter. However,
numerical simulation of these systems will become possible, as discussed in Sec. 9.5
on network calibration, but good representation of the system with accurate parameters may be difficult.
9.2.3.6 Extended-period simulation. As noted earlier, time variation can be considered
in network modeling. The simplest approach is extended-period simulation, in which a
sequence of steady-state simulations are solved using one of the methods described earlier in this section. After each simulation period, the tank levels are updated and demand
and operational changes are introduced.

TABLE 9.1 Comparison of solution methods

Example

Special conditions

Low velocities

Pumps and branched

network

Example B with high

demand

Closed pipes

Node equations

Solution method:
Loop equations

Converged
Iterations = 16,
T = 70 s
Converged
Iterations =12,
T = 92 s
Converged
Iterations =13,
T=IOOs
Converged
Iterations = 21,
T = 155 s
Some heads not
available

Converged
Iterations =17,
T = 789 s
Slow convergence
Iterations =13,
T = 962 s
Slow convergence
Iterations =15,
T=IIlOs
Converged
Iterations = 21,
T = 1552 s
Some heads not
available

Pipe equations
Converged
Iterations =16,
T = 30 s
Converged
Iterations =10,
T = 34 s
Converged
Iterations =12,
T = 39 s
Converged
Iterations =19,
T = 57 s

Source: Modified from Todini and Pilati (1987).

Tank levels or water-surface elevations are used as known energy nodes. The levels
change as flow enters or leaves the tank. The change in water height for tanks with constant geometry is the change in volume divided by the area of the tank, or
J. = QI&
T
AT
AT
where QT and VT are the flow rate and volume of flow that entered the tank during the period, respectively; A? is the time increment of the simulation; AT is the tank area; and AHr
is the change in elevation of the water surface during period T. More complex relationships are needed for noncylindrical tanks. With the updated tank levels, the extended-period simulation continues with these levels as known energy nodes for the next time step.
The process continues until all time steps are evaluated. More complex unsteady analysis
are described in the next section.

9.3

UNSTEADY FLOW IN PIPE NETWORK ANALYSIS

In steady state analysis or within an extended-period simulation, changes in the distributions of pressure and flow are assumed to occur instantaneously after a change in external
stimulus is applied. Steady conditions are then reached immediately. In some cases, the
time to reach steady state and the changes during this transition may be important.
Recently, work has proceeded to model rapid and gradual changes in flow conditions.
Rapid changes resulting in transients under elastic column theory are discussed in Chap.
12. Two modeling approaches for gradually varied unsteady flow under a rigid-column
assumption are described in this section.

9.3.1 Governing Equations

In addition to conservation of mass, the governing equations for unsteady flow under rigid
pipe assumptions are developed from conservation of momentum for an element
(Fig. 9.6). Conservation of momentum states that the sum of the forces acting on the volume of fluid equals the time rate of change of momentum, or
2F = F 1 -F 2 -F 7 =^l

(9.54)

where F1 and F2 are the forces on the ends of the pipe element, Ff is the force caused by
friction between the water and the pipe, and m and v are the mass and velocity of the fluid
in the pipe element.
The end forces are equal to the force of the pressure plus the equivalent force caused
by gravity or for the left-hand side of the element:
F1 = jA^+z] = JAh1

(9.55)

The friction force is the energy loss times the volume of fluid, or
Ff=jAhL.

(9.56)

The change of momentum can be expanded to

/yALvl
J(UiV) _ d(pVv) _( 8 ) = yLd(Av=yLc
dt
dt
dt
g dt
g dt
y
where the mass is equal pV = -AL,
in which all terms are constants with respect to time
O
and can be taken out of the differential. Note that under the rigid-water-column assumption, the density is a constant as opposed to elastic-water-column theory. Substituting
these terms in the momentum balance gives

FIGURE 9.6 Force balance on a pipe element

JA(H1 - H 2 - H 1 ) = ^-^

(9.58)

Assuming that a steady state friction loss relationship can be substituted for hL and dividing each side by ^A,
*.-*,-*? =
(9-59)
With conservation of mass (Eq. 9.2), this ordinary differential equation and its extensions for loops have been used to solve for time-varying flow conditions.
9.3.2 Solution Methods
9.3.2.1 Loop formulation. Holloway (1985) and Islam and Chaudhry (1998) extended
the momentum equation (Eq. 9.59) to loops as follows:
2 (*-*>-S ^=S ^f1
i/p
ie/p
ielp $ '
Separating variables and integrating over time gives
P+^IV.
1 P+AT^
1
f^+Ar^ L.
1 [s (*-**>}*-i g/4H
|^de'

(9-60)

(9 6i)

At any instant in time, the head loss around a closed loop must equal zero, so the first
term can be dropped. Dropping this term also eliminates the nodal piezometric heads as
unknowns and leaves only the pipe flows.
One of several approximations for the friction loss term can be used:
KQ+"\Q\*-lbt

(9.62)

K[(Q^ + Q)\Q+* + 01"-!/2"JAf

(9.63)

K[(Qt+*> IQ*+*!"-1 + Q\Q\-*) /2]Af

(9.64)

Holloway (1985) obtained results using Eq. (9.62), known as the integration approximation that compared favorably with the other two nonlinear forms. Using this form in
Eq. (9.61),
(9 65)
^-^T
&-^, K&+"\Q\*-i Ar = X ^T
W
'
8 l
A
MP
Mp
Mp 8 i
This equation is written for each loop and is used with the nodal conservation of mass
equations to given Np equations for the Np unknown pipe flows. Note that these equations
are linear in terms of Qt+At and can be solved at each time step in sequence using the previous time step for the values in the constant terms.

9.3.2.2 Pipe formulation with gradient algorithm. An alternative solution method

developed by Ahmed and Lansey (1999) used the momentum equation for a single pipe
(Eq. 9.59) and the nodal flow balance equations to form a set of equations similar to those
developed in the gradient algorithm. An explicit backward difference is used to solve the
equations. The right-hand side of Eq. 9.59 is written in finite difference form as

f -3;*^
The left-hand side of Eq. 9.9 is written in terms on the unknowns h and Q at time step
t + At. After substituting and rearranging a general algebraic equation for pipe between
two nodes results in
\Kt\Q^ I"-1 Ar - -^U <2rA'+ iTA'~ ^TA'] & = \ - -^r] Qi
(9-67)
L
&A'J
L A'J
Np equations of this form can be written for each pipe or other component. With the Nj
nodal flow balance equations, a total of Nj + Np equations can be written in terms of an
equal number of unknown pipe flows and nodal heads. Given an initial condition at time
t, the pipe flows and nodal heads at time t + At by solving Eq. 9.67 and Eq. 9.2 The new
values are then used for the next time step until all times have been evaluated. Unlike the
loop formulation, in the form above, Eq. 9.67 is nonlinear with respect to the unknowns.
In addition, like the loop equation, the time step will influence the accuracy of the results.

9.4

WATER-QUALITY MODELING

Interest in water quality in distribution systems heightened with the passing of the 1986
amendment to the Safe Drinking Water Act. This amendment required that standards must
be developed for chlorine levels not only at the point of disinfection but also at the most
distant point of withdrawal. Thus, modeling the fate and transport of dissolved substances
in networks with emphasis on chlorine became necessary. As a result, methods of analysis and computer programs implementing these methods, such as EPANET (Rossman,
1994), have been developed.
Since the velocity in pipes is relatively high, constituents in the water are assumed to
move completely with the flow, that is, by advective transport. This assumption allows the
use of explicit numerical modeling schemes to solve for constituent movement within the
system. As in hydraulic analysis, steady and unsteady transport models have been developed. Both models use conservation of mass as the basic governing equation describing
mixing and movement. Because advective transport dominates, the pipe flow rates are
critical in estimating transport in the system. In most unsteady water-quality models,
extended-period simulation has been used to account for demand and operational changes
(Sec. 9.2.3.6) Although water quality analysis considering slow transients using rigidwater-column theory for the flow analysis has been performed by Chaudhry and Islam
(1998), it will not be discussed here.
As water moves through the network, constituent (with emphasis on chlorine) decay is
generally assumed to follow first-order kinetics, or
ct = c0e-V

(9.68)

where CQ and ct are the constituent concentrations at times O and t, respectively, t is time
and kt is the first-order decay coefficient, which is defined by
"--[^ + ScWl
<*">
where RH is the hydraulic radius of the pipe, and kb, kw, and kf are the bulk flow-decay constant, the wall reaction rate constant, and a mass transfer coefficient that is dependent on
the Reynold's number, respectively.

9.4.1 Steady State Modeling

Given a steady flow distribution, the contribution from different sources or the concentration of a constituent at withdrawal nodes can be determined by solving a set of linear
algebraic equations. Under the assumption that complete mixing occurs at a junction
node, the general conservation of mass equation under these conditions states that the
mass of constituent entering the junction equals the mass leaving the junction, or
2 QfI. + Q8C= 2 Qf0
(9.70)
iel
Jelk
e
where CI- is the constituent concentration in incoming pipey, C0 is the concentration in all
outgoing pipes, and Cs is the constituent concentration in the incoming source water. Qj is
the volumetric flow rate in incoming pipe j and Qs is the external-source flow rate. Q. is
the outgoing flow from the node in pipe i. If the junction is a demand node, the external
demand is included in set I6. Given steady flow, the total inflow must equal the total outflow. Substituting the flow balance and solving for the concentration in all flows leaving
the node, C0 gives
2 QPt + QSCS
C0 = J^
Ea
e

(9.71)

One constituent mass balance equation can be written for each node. Since the flow
rates are defined by the hydraulic relationships, C8 is known, and the CI for one node is
the outflow from another node, the system of equations can be solved for the Af unknown
c;s.
A steady-state model provides the concentrations at all points in the network under
steady flow and concentrations. By modeling each source concentration independently
in a series of simulations, the model also can be used to determine the relative source
contribution at any point under the same conditions.
9.4.2 Dynamic Analysis
Steady flow conditions for water quality provide information regarding movement of dissolved substances but are likely to be less useful for predicting point concentrations under
normal operations. Unsteady analysis, also known as dynamic modeling, provides a more
realistic picture and better estimates of constituent movement under time-varying flow
conditions.
Dynamic modeling can solve several types of problems. In addition to determining the
variation in concentration at a point over time, it can be used to determine the age of or
average travel time for water at some location and time. Finally, as with steady models,
the relative source contributions providing flow to a point can be computed.
9.4.2.1 Governing equations. To determine the fate and transport of dissolved substances under unsteady conditions, the primary governing equation is the one-dimensional advection equation that is solved in conjunction with the assumption of complete mixing at a node. The advection equation is

=-i,+ *(C'.)

(9.72)

where C1 is the constituent concentration in pipe / at location x and time t\ U1 is the velocity in pipe z, and R(C1) is the reaction/decay function. The decay relationship for firstorder kinetics R = ktc is used when modeling chlorine and possibly other nonconservative substances. For conservative substances, such as fluoride, the reaction relationship
is simply zero. Finally, when modeling water age, R is equal to one and the concentration C is interpreted as the water age with new water entering the system having concentration equal to zero.
Tanks act as sources or sinks in the system with variable water quality, depending on
the history of inflow and outflow as well as on the reactions in the tank. The simplest
water-quality relationship for a tank assumes that the water is mixed completely. In this
case, the variation in constituent concentration is
^^
= 2 G1-CE1. - 2 QjC7 + R(C1)
01
ieIT
j*0T

(9.73)

where VT and CT are the storage volume and constituent concentration within the tank at
time r, respectively. Pipes in the set of 1T provide inflows Q1 to the tank, and pipes in the
set OT receive flows Q. from the tank. CE is the concentration at the exit of the pipe as it
enters the tank. R is the reaction relationship for water in the tank.
9.4.2.2 Solution methods Eulerian methods. Rossman and Boulos (1996) compared
the different solution methods for solving the unsteady water-quality problem. This section generally follows their notation and terminology. Dynamic models can be classified
spatially as Eulerian or Lagrangian models and temporally as time driven or event driven.
Eulerian methods define a grid of either points or volume segments within a pipe. Flow
and the associated constituents are tracked through this fixed grid. Chaudhry and Islam
(1998) used a finite-difference method with a fixed-point grid, and Grayman et al. (1988),
and its extension by Rossman et al. (1993), have developed the discrete-volume method
(DVM). The following discussion focuses on the DVM as it has been implemented in the
EPANET model (Rossman, 1994).
For a given hydraulic condition, the DVM divides each pipe into equally sized,
completely mixed, volume segments. The number of segments for a particular pipe is
computed by
n

'

= J^ = ^L
ut Ar

(9.74)
'

where L1 and u( are the length of and flow velocity in pipe z, respectively; tti is the travel time
for water to pass through pipe /; and Ar is the duration of the water quality time step. A small
Ar provides the highest numerical accuracy at the expense of higher computation times.
When the flow conditions change in the network (i.e., u changes), the grid must be redefined.
At each water-quality step, four operations are completed, as shown in Fig. 9.7. First,
the present constituent masses are reduced to account for the decay reactions. Next, the
elements from each segment are advanced to the next downstream segment. Third, if the
segment is the most downstream in a pipe, the flow is mixed with the flow from other
pipes that enter the node using Eq. (9.71). Finally, the flow from the node is passed to the
first segments of pipes leaving the node.
These operations are repeated for each water-quality time step until the flow distribution
changes. Pipes are then resegmented, and the process is repeated for that hydraulic condi-

Original mass

After reaction

Transport to downstream node

Transport along link

Transport out of node

FIGURE 9.7 Computational steps of discrete volume method (From Rossman and Boulos (1996))

tion. When the pipes are divided for different flow conditions, the number of segments may
be different and some numerical blending occurs. As a result, the accuracy of DVM and
finite difference methods depends on the selection of the water-quality time step Af.
Lagrangian methods. Unlike Eulerian methods which use a fixed grid, Lagrangian
methods track segments of water as they move through a network. As the front or leading
edge of the segment reaches a node, it is combined with other incoming segments. The
segments leaving the node are developed with constituent levels determined by Eq. (9.71)
(Fig. 9.8). Two approaches have been used to define when segments are combined and
transported through a pipe.
Liou and Kroon (1987) applied this type of model using a defined time step to
determine when to combine segments. During each time step, the total mass of constituent
and volume of water that reaches a node is computed. The average nodal concentration is
computed, and new segments emanating from the node are introduced. To avoid adding
too many new segments, they are created only when the concentration difference between
the new and the previous segment in a link is above a threshold. When more than one segment in a link reaches a downstream node in one time step, artificial mixing will occur.
Rather than combine segments at defined time intervals, the second Lagrangian
approach is an event-driven method (Boulos et al., 1994, 1995; Hart et al., 1987;
El-Shorbagy and Lansey, 1994; and Shah and Sinai, 1985). Event methods combine segments each time a front reaches a node, thus avoiding artificial mixing. Since defined
times are not used, the projected times when a front reaches a downstream node are corn-

FIGURE 9.8 Water quality transport for the Lagrangian methods for a conservative substance at
three different times. The flowrates in the two inflow pipes are equal and the flowrate in the outgoing pipe is then twice the flow in either inflow pipe. A: water quality at time t: flow is to the left,
and the constituent level equals the average of the inflow concentrations, or (4+l)/2 = 2.5 B: The
water quality at time t + At some time after the front concentration 2 in the vertical pipe reached
the node. For some time, the inflow concentrations were 2 and 4, or an average outflow concentration of 3, C: Water quality at some later time: Two elements have developed downtream. The element with a concentration of 3.5 developed when the inflows of 3 and 4 mixed at node. The final
element closest to the node with concentration 4 developed when the inflow with concentrations of
3 and 5 mixed at the node.

puted for the present flow condition. The water-quality conditions at nodes remain constant until the next segment front reaches a node. At that time, new segments are generated in pipes that carry flow from the node that the first front reaches. The concentration in
these segments is computed by Eq. (9.71) and is recorded with the transition time.
Projection times are then updated, and the process continues when the next closest front
reaches a node or the hydraulic condition changes. If the flow condition changes, new projection times are computed. Event-driven models avoid numerical dispersion; however,
the method can result in a large number of segments. To save computer memory, segments
can be combined according to the difference in concentration between adjacent segments.
Further error may result during flow reversals for reactive constituents.
Comparison of methods. Rossman and Boulos (1996) conducted numerical experiments comparing the alternative methods described in the previous sections, and reached
the following conclusions:
1. The numerical accuracy of all methods is similar, except that the Eulerian methods had
occasional problems. All methods can represent observed behavior adequately in real
systems.
2. Network size is not always an indicator of solution time and computer memory
requirements.
3. Lagrangian methods are more efficient in both time and memory requirements than
Eulerian methods when modeling chemical constituents.
4. The time-dependent Lagrangian method are most efficient in computation time for
modeling water age, whereas the Eulerian methods are the most memory efficient.
Overall, Rossman and Boulos concluded that the time-based Lagrangian method was
the most versatile unless computer memory was limiting for modeling water age for large
networks. In which case, Eulerian methods were preferable.
9.5 COMPUTER MODELING OF WATER
DISTRIBUTION SYSTEMS
Because the numerical approaches for analyzing distribution systems cannot be completed by hand except for the smallest systems, computer-simulation models have been developed. These models solve the system of nonlinear equations for the pipe flows and nodal
heads. In addition to the equation solver, many modeling packages have sophisticated
input preprocessors, which range from spreadsheets to tailored full-page editors, and output postprocessors, including links with computer-aided drafting software and geographic information systems. Although these user interfaces ease the use of the simulation models, a dependable solver and proper modeling are crucial for accurate mathematical models of field systems.
An array of packages is available, and the packages vary in their level of sophistication. The choice of a modeling package depends on the modeling effort. Modeling needs
range from designing subdivisions with fewer than 25 pipes to modeling large water utilities that possibly involve several thousand links and nodes. Users should select the package that best suit their objectives.

9.5.1 Applications of Models

Clark, et al. (1988) identified a series of seven steps that are necessary to develop and
apply a water distribution simulation model:
1. Model selection: Definition of modeling requirements including the model's
purpose. The desired use of a model imprtant must be understood when selecting one
(hydraulic or water quality) because the necessary accuracy of the model and the level
of detail required will vary, depending on its expected use.
2. Network representation: Determination of how the components of a system will be
represented in the numerical model. Step 2 includes skeletonizing the piping system
by not including some pipes in the model or making assumptions regarding the parameter values for pipes, such as assuming that all pipes of a certain type have the same
roughness value. The degree of model simplification depends on what problems the
model will be used to help address.
3. Calibration: Adjustment of nonmeasurable model parameters, with emphasis on the
pipe roughness coefficients, so that predicted model results compare favorably with
observed field data (see Sec. 9.5.2). This step also may require reexamination the network representation.
4. Verification: Comparison of model results with a second set of field data (beyond that
used for calibration) to confirm the adequacy of the network representation and parameter estimates.
5. Problem definition: Identification of the design or operation problem and incorporation of the situation in the model (e.g., demands, pipe status or operation decisions).
6. Model application: Simulation of the problem condition.
7. Display/analysis of results: Presentation of simulation results for modeler and other
decision-makers in graphic or tabular form. Results are analyzed to determine
whether they are reasonable and the problem has been resolved. If the problem is not
resolved, new decisions are made at step 5 and the process continues.
9.5.2 Model Calibration
Calibration, step 3 above, is the process of developing a model that represents field conditions for the range of desired conditions. The time, effort, and money expended for data
collection and model calibration depend on the model's purpose. For example, a model
for preliminary planning may not be extremely accurate because decisions are at the planning level and an understanding of only the major components is necessary. At the other
extreme, a model used for engineering decisions regarding a system that involves pressure
and water-quality concerns may reguire significant calibration efforts to provide precise
predictions. All models should be calibrated before they are used in the decision-making
process.
The calibration process consists of data collection, model calibration, and model
assessment. Data collection entails gathering field data, such as tank levels, nodal pressures, nodal elevations, pump head and discharge data, pump status and flows, pipe flows,
and, when possible, localized demands. These data are collected during one or more loading conditions or over time through automated data logging. Rossman et al., (1994) discussed using water-quality data for calibration. To ensure that a calibration will be successful, the number of measurements must exceed the number of parameters to be estimated in the model. If this condition is not satisfied, multiple sets of parameters that

match the field observations can be found: that is, a unique solution may not be determined. Each set may give dramatically different results when predicting under other conditions.
During model calibration, field data are compared with model estimates and model
parameters are adjusted so that the model predictions match the field observations. Two
stages of model calibration are desirable. The first stage is a gross study of the data and
the model predictions. The intent is to insure that the data are reasonable and that major
modeling assumptions are valid. For example, this level would determine if valves
assumed to be open are actually closed or if an unexpectedly high withdrawal, possibly
caused by leakage, is occurring. Walski (1990) discussed this level of calibration.
[TITLE]
EPANET Example network 1
[JUNCTIONS]
Elevation
Demand
ID
ft
gpm
10
710
O
11
710
150
12
700
150
13
695
100
21
700
150
22
695
200
23
690
150
31
700
100
32
710
100

[CONTROLS]
Li^~9~6pEiiWN6DE2~Bl^wn6
LINK 9 CLOSED IF NODE 2 ABOVE 140
[PATTERNS]
ID Multipliers
1 1.0 1.2
1 1.0 0.8

1.4
0.6

1.6 1.4 1.2

0.4 0.6 0.8

[QUALITY]
[TANKS]
Elev. Init. Min. Max. Diam
ID
ft Level Level Level
ft
2
850 120 100 150 50.5
9
800
[PIPES]
ID
""To
11
12
21
22
31
110
111
112
113
121
122

Head
Node
To
11
12
21
22
31
2
11
12
13
21
22

Tail
Node
ii
12
13
22
23
32
12
21
22
23
31
32

Length Diam.
ft
in.
io53o""i~8
5280 14
5280 10
5280 10
5280 12
5280 6
200
18
5280 10
5280 12
5280 8
5280 8
5280 6

Rough
Coeff.
Too""
100
100
100
100
100
100
100
100
100
100
100

[PUMPS]
ID
9

Head
Node
9

Tail
Node
10

Design
ft
250

H-Q
gpm
1500

Nodes
2
9
2

Initial
Concen. mg/1
05
1.0
1.0

32

[REACTIONS]
GLOBAL BULK - .5
GLOBAL WALL -1
[TIMES[
DURATION 24
PATTERN TIME STEP

[OPTIONS]
QUALITY
MAP

[END]

FIGURE 9.9 EPANET input file for example network (Figure 9.1)

Chlorine
Net !.map

; Bulk decay coeff.

; Wall decay coeff.

;24 hour simulation

period
;2 hour pattern time
period

; Chlorine analysis
; Map coordinates file

EPANET
Hydraulic and Water Quality
Analysis for Pipe Networks
Version 1.0
EPANET Example Network 1
Input data File
Verification File
Hydraulics File
Map File
Number of Pipes
Number of Nodes
Number of Tanks
Number of Pumps
Number of Valves
Headloss Formula
Hydraulic Timestep
Hydraulic accuracy
Maximum Trials
Quality Analysis
Minimum Travel Time
Maximum Segments per Pipe
Specific Gravity
Kinematic Viscosity
Chemical Diffusivity
Total Duration
Reporting Duration
All Nodes
All Links

Node Results at 0.00 hrs:

Elev.
Demand
Node
ft.
gpm
10
710.00
0.00
11
710.00
150.00
12
700.00
150.00
13
695.00
100.00
21
700.00
150.00
22
695.00
200.00
23
690.00
150.00
31
700.00
100.00
32
710.00
100.00
2
850.00
765.06
9
800.00
-1865.06

net 1. inp
Net 1. map
12
11
2
1
O
Hazen-Williams
1.00 hrs
0.001000
40
Chlorine
6.00 min
100
1.00
l.lOe-005 sq ft/sec
1.3e-008 sq ft/sec
24.00 hrs

Grade
ft
1004.50
985.31
970.07
968.86
971.55
969.07
968.63
967.35
965.63
970.00
800.00

Pressure
psi
127.61
119.29
117.02
118.66
117.66
118.75
120.73
115.84
110.77
52.00
0.00

FIGURE 9.10 EPANET output file for example network (figure 9.1)

Chlorine
rng/L
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
1.00
Tank
1.00
Reservoir

Link results at 0.00 hrs.

Start
Link
Node
10
10
11
11
12
12
21
21
22
22
31
31
110
2
11
11
112
12
113
13
121
12
122
22
9
9

End
Node
11
12
13
22
23
32
12
21
22
23
31
32
10

Diameter
in
18.00
14.00
10.00
10.00
12.00
6.00
18.00
10.00
12.00
8.00
8.00
6.00

Flow
gpm
1865.06
1233.57
129.41
190.71
120.59
40.77
-765.06
481.48
189.11
29.41
140.77
59.23
1865.06

Node Results at 1.00 hrs

Elev. demand Grade
node
ft
gpm
10
710.00
0.00
11
710.00
150.00
12
700.00
150.00
13
695.00
100.00
21
700.00
150.00
22
695.00
200.00
23
690.00
150.00
31
700.00
100.00
32
710.00
100.00
2
850.00
747.57
9
800.00
-1847.57

Pressure
ft
1006.92
988.05
973.13
971.91
974.49
972.10
971.66
970.32
968.63
973.06
800.00

Link Results at 1.00 hrs.

Start
End
Link
Node
Node
10
10
11
11
11
12
12
12
13
21
21
22
22
22
23
31
31
32
110
2
12
111
11
21

Diameter
in
18.00
14.00
10.00
10.00
12.00
6.00
18.00
10.00

FIGURE 9.10 (Continued)

Velocity Headloss
fps
/1000ft
2.35
1.82
2.57
2.89
0.53
0.23
0.78
0.47
0.34
0.08
0.46
0.33
0.96
0.35
1.97
2.61
0.54
0.19
0.19
0.04
0.90
0.79
0.67
0.65
96 hp -204.50

Chlorine
psi
128.65
120.48
118.35
119.98
118.94
120.07
122.04
117.13
112.06
53.32
0.00

Flow
gpm
1847.49
1219.82
130.19
187.26
119.81
40.42
-747.49
477.68

mg/1
1.00
0.45
0.44
0.44
0.43
0.44
0.45
0.41
0.40
0.97
1.00

Velocity
fps
2.33
2.54
0.53
0.76
0.34
0.46
0.94
1.95

Pump

Tank
Reservoir

Headloss
/1000ft
1.79
2.83
0.23
0.45
0.08
0.32
0.34
2.57

Link Results at 1.00 hrs (continued)

Start
End
Link
Node
Node
112
12
22
113
13
23
121
21
31
122
22
32
9
9
10
Node Results at 2.00 hrs
Elev.
Node
ft
10
710.00
11
710.00
12
700.00
13
695.00
21
700.00
22
695.00
23
690.00
31
700.00
32
710.00
2
850.00
9
880.00
Link results at 2.00 hrs.
Start
Link
Node
10
10
11
11
12
12
21
21
22
22
31
31
110
2
111
11
112
12
113
13
121
21
122
22
9
9

Diameter
in
12.00
8.00
8.00
6.00

Demand
gpm
0.00
180.00
180.00
120.00
180.00
240.00
180.00
120.00
120.00
516.44
-1836.44

End
Node
11
12
13
22
23
32
12
21
22
23
31
32
10

Flow
gpm
192.14
30.19
140.42
59.58
1847.49

Grade
ft
1008.43
989.77
976.09
974.02
975.41
973.81
973.33
969.96
968.13
976.06
800.00

Diameter
in
18.00
14.00
10.00
10.00
12.00
6.00
18.00
10.00
12.00
8.00
8.00
6.00

Flow
gpm
1836.44
1163.77
173.00
150.47
127.00
42.20
-516.44
492.67
294.33
53.00
162.20
77.80
1836.44

Velocity
fps
0.55
0.19
0.90
0.68
97 hp

Pressure
psi
129.31
121.22
119.63
120.90
119.34
120.81
122.77
116.98
111.85
54.62
0.00

Headless
/1000ft
0.20
0.05
0.79
0.66
-206.92

Chlorine
mg/L
1.00
0.87
0.81
0.37
0.76
0.38
0.40
0.34
0.31
0.94
1.00

Velocity
Headless
fps
/1000ft
2.32
1.77
2.43
2.59
0.71
0.39
0.61
0.30
0.36
0.09
0.48
0.35
0.65
0.17
2.01
2.72
0.83
0.43
0.34
0.13
1.04
1.03
0.88
1.08
97 hp -208.43

Pump

Tank
Reservoir

Pump

FIGURE 9.10 (Continued)

After the model representation is determined to be reasonable, the second stage of
model calibration begins with the adjustment of individual model parameters. At this
level, the two major sources of error in a model are the demands and the pipe roughness
coefficients. The demands are uncertain because water consumption is largely unmonitored in the short term, is highly variable, and because the water is consumed along a pipe,
whereas it is modeled as a point of withdrawal. Because pipe roughnesses vary over time
and are not directly measurable, they must be inferred from field measurements.

Adjustment of these terms and others, such as valve settings and pump lifts, can be
made by trial and error or through systematic approaches. Several mathematical modeling methods have been suggested for solving the model calibration problem (Lansey and
Basnet, 1991).
Once a model is believed to be calibrated, an assessment should be completed. The
assessment entails a sensitivity analysis of model parameters to identify which parameters
have a strong impact on model predictions and future collection should emphasize
improving. The assessment also will identify the predictions (nodal pressure heads or tank
levels) that are sensitive to calibrated parameters and forecasted demands. Model assessments can simply be plots of model predictions versus parameter values or demand levels,
or they can be more sophisticated analyses of uncertainty, as discussed in Araujo (1992)
and Xu and Coulter (1998).
9.5.3 Model Results
Water-distribution simulation models require the model parameters, such as pipe and
pump characteristics, nodal demands, and valve settings, to solve the appropriate set of
equations and display the nodal peizometric heads, pipe flow rates water quality predictions, and other results, such as pipe head loss and pipe velocities. No standard format is
used between models. Abbreviated input and output files are shown in Figs. 9.9 and 9.10
for a sample system shown in Fig. 9.1. These files are for the EPANET code and are used
because the EPANET program is in the public domain and it models both flow and water
quality in an extended-period simulation format.
The constituent, chlorine, is reactive and results are shown for a selected subgroup of
nodes. As in most models, the constituent levels along a pipe are not provided. Finally,
tank concentrations, although not shown directly, can be found by examining the concentration closest to the tank node when flow is exiting the tank.

REFERENCES
Ahmed, L, Application of the gradient method for analysis of unsteady flow in water networks,
Master's thesis (Civil Engineering), University of Arizona, Tucson, 1997.
Araujo, J.V., A statistically based procedure for calibration of water systems, Doctoral dissertation,
Oklahoma State University, Stillwater, 1992.
Ahmed, L, and K. Lansey, "Analysis of unsteady flow in networks using a gradient algorithm
based method," ASCE Specialty Conference on Water Resoures, Tempe, AZ, June, 1999.
Boulos, P. E, T. Altman, P. A. Jarrige, and F. Collevati, "An event-driven method for modeling contaminant propagation in water networks," Journal of Applied Mathematical Modeling, 18(2): 8492, 1994.
Boulos, P E, T. Altman, P. A. Jarrige, and E Collevati, "Discrete simulation approach for network
water quality models," Journal of Water Resources Planning and Management, 121(1): 49-60,
1995.
Clark, R., Grayman, W. and R. M. Males, "Contaminant propagation in distribution systems" J. of
Environmental Eng., 114(4): 1988.
Cross, H. "Analysis of flow in networks of conduits or conductors," Bulletin No. 286, University
of Illinois Engineering Experimental Station, Urbana, IL 1936.
El-Shorbagy, W, and K. Lansey, "Non-conservative water quality modeling in water systems,"
Proceedings of the AWWA Specialty Conference on Computers in the Water Industry, Los
Angeles, April 1994.
Grayman, W. M., R.M. Clark, and R.M. Males, "Modeling Distribution System Water Quality:
Dynamic Approach," Journal of Water Resources Planning and Management, 114(3),: 295312, 1988.

Hart, F. L., J. L. Meader, and S.-M. Chiang, "CLNETA simulation model for tracing chlorine
residuals in a potable water distribution network," AWTK4 Distribution System Symposium
Proceedings, American Water Works Association, Denver, CO, 1987.
Holloway, M. B., Dynamic Pipe Network Computer Model, Doctoral dissertation, Washington State
University, Pullman, WA, 1985.
Islam R., and M. H. Chaudhry, "Modeling of constituent transport in unstead flows in pipe networks,"/. of Hydraulics Division, 124 (11): 1115-1124, 1998.
Jeppson, R. W., Analysis of Flow in Pipe Networks, Ann Arbor Science, Ann Arbor, MI, 1974.
Lansey, K., and C. Basnet, "Parameter Estimation for Water Distribution Systems," Journal of
Water Resources Planning and Management, 117(1): 126-144, 1991.
Liou, C. P., and J. R. Kroon, "Modeling the Propagation of Waterborne Substances in Distribution
networks," Journal of the American Water Works Association, 79(11): 54-58, 1987.
Martin, D. W., and G. Peters, "The Application of Newton's Method to Network Analysis by
Digital computer," Journal of the Institute of Water Engineers, 17: 115-129, 1963.
Rossman, L. A., "EPANETUsers manual," EPA-600/R-94/057, U.S. Environmental Protection
Agency, Risk Reduction Engineering Laboratory, Cincinnati, OH, 1994.
Rossman, L., and P. Boulos, "Numerical methods for modeling water quality in distribution systems: A comparison," Journal of Water Resources Planning and Management, 122(2): 137146, 1996.
Rossman, L. A., P. F. Boulos, and T. Altman, "Discrete Volume Element Method for Network
Water Quality Models," Journal of Water Resources Planning and Management, 119(5): 505517,1993.
Rossman, L. A., R. M. Clark, and W. M. Grayman, "Modeling Chlorine Residuals in DrinkingWater Distribution Systems," Journal of Environmental Engineering, American Society of Civil
Engineers, 120(4): 303-320, 1994.
Salgado, R., E. Todini, and P. E. O'Connell, "Comparison of the gradient method with some traditional methods for the analysis of water supply distribution networks," Proceeediigs International
Conference on Computer Applications for Water Supply and Distribution 1987, Leicester
Polytechnic, UK, September 1987.
Shah, M. and G. Sinai, "Steady State Model for Dilution in Water Networks." Journal of
Hydraulics Division, 114(2), 192-206, 1988.
Shamir, U., and C. D. Howard, "Water Distribution System Analysis," Journal of Hydraulics.
Division, 94(1). 219-234, 1965.
Todini, E., and S. Pilati, "A gradient method for the analysis of pipe networks," International
Conference on Computer Applications for Water Supply and Distribution 1987, Leicester
Polytechnic, UK, September 1987.
Walski, T., "Hardy-Cross meets Sherlock Holmes or model calibration in Austin, Texas," Journal
of the American Water Works Association, 82:34-38, March, 1990.
Wood, D. J., User's ManualComputer Analysis of Flow in Pipe Networks Including Extended
Period Simulations, Department of Civil Engineering, University of Kentucky, Lexington,
KY, 1980.
Wood, D., and C. Charles, "Hydraulic Network Analysis Using Llinear Theory," Journal of
Hydraulic Division, 98, (HY7): 1157-1170, 1972.
Xu, C., and I. Goulter, "Probabilistic Model for Water Distribution Reliability," Journal of Water
Resources Planning and Management, 124(4): 218-228, 1998.

CHAPTER 10
PUMP SYSTEM
HYDRAULIC DESIGN

B. E. Bosserman
Boyle Engineering Corporation
Newport Beach, CA

W.1
10.1.1

PUMP TYPES AND DEFINITIONS

Pump Standards

Pump types are described or defined by various organizations and their respective publications:
Hydraulics Institute (HI), American National Standard for Centrifugal Pumps for
Nomenclature, Definitions, Application and Operation [American National Standards
Institute (ANSI)THI 1.1-1.5-1994]
American Petroleum Institute (API), Centrifugal Pumps for Petroleum, Heavy Duty
Chemical, and Gas Industry Services, Standard 610, 8th ed., August 1995
American Society of Mechanical Engineers (ASME), Centrifugal Pumps,
Performance Test Code PTC 8.2-1990
In addition, there are several American National Standards Institute (ANSI) and
American Water Works Associations (AWWA) standards and specifications pertaining to
centrifugal pumps:
ANSI/ASME B73.1M-1991, Specification for Horizontal End Suction Centrifugal
Pumps for Chemical Process.
ANSI/ASME B73.2M-1991, Specification for Vertical In-Line Centrifugal Pumps for
Chemical Process.
ANSI/ASME B73.5M-1995, Specification for Thermoplastic and Thermoset
Polymer Material Horizontal End Suction Centrifugal Pumps for Chemical
Process.
ANSI/AWWA E 101-88, Standard for Vertical Turbine PumpsLineshaft and
Submersible Types.

10.1.2 Pump Definitions and Terminology

Pump definitions and terminology, as given in Hydraulics Institute (HI) 1.1-1.5-1994
(Hydraulics Institute, 1994), are as follows:
Definition of a centrifugal pump. A centrifugal pump is a kinetic machine converting
mechanical energy into hydraulic energy through centrifugal activity.
Allowable operating range. This is the flow range at the specified speeds with the
impeller supplied as limited by cavitation, heating, vibration, noise, shaft deflection,
fatigue, and other similar criteria. This range to be defined by the manufacturer.
Atmospheric head (hatm). Local atmospheric pressure expressed in ft (m) of liquid.
Capacity. The capacity of a pump is the total volume throughout per unit of time at
suction conditions. It assumes no entrained gases at the stated operating conditions.
Condition points
Best efficiency point (BEP). The best efficiency point (BEP) is capacity and head at
which the pump efficiency is a maximum.
Normal condition point. The normal condition point applies to the point on the rating
curve at which the pump will normally operate. It may be the same as the rated condition point.
Rated condition point. The rated condition applies to the capacity, head, net positive
suction head, and speed of the pump, as specified by the order.
Specified condition point. The specified condition point is synonymous with rated
condition point.
Datum. The pump's datum is a horizontal plane that serves as the reference for head
measurements taken during test. Vertical pumps are usually tested in an open pit with the
suction flooded. The datum is then the eye of the first-stage impeller (Fig. 10.1).
Optional tests can be performed with the pump mounted in a suction can. Regardless
of the pump's mounting, its datum is maintained at the eye of the first-stage impeller.
Elevation head (Z). The potential energy of the liquid caused by its elevation relative
to a datum level measuring to the center of the pressure gauge or liquid surface.
Friction head. Friction head is the hydraulic energy required to overcome frictional
resistance of a piping system to liquid flow expressed in ft (m) of liquid.
Gauge head (hg). The energy of the liquid due to its pressure as determined by a pressure gauge or other pressure measuring device.
Head. Head is the expression of the energy content of the liquid referred to any arbitrary datum. It is expressed in units of energy per unit weight of liquid. The measuring
unit for head is ft (m) of liquid.
High-energy pump. High-energy pump refers to pumps with heads greater than 650 ft
(200 m) per stage and requiring more than 300 hp (225 KW) per stage.
Impeller balancing
Single-plane balancing (also called static balancing). Single-plane balancing refers
to correction of residual unbalance to a specified maximum limit by removing or

HGL

Datum

Gauge

Pump

Gauge

Absolute zero pressure

FIGURE 10.1 Terminology for a pump with a positive suction head.

adding weight in one correction plane only. This can be accomplished statically using
balance rails or by spinning.
Two-plane balancing (also called dynamic balancing). Two plane-balancing referes
to correction of residual unbalance to a specified limit by removing or adding weight
in two correction planes. This is accomplished by spinning on appropriate balancing
machines.
Overall efficiency (T|OA). This is the ratio of the energy imparted to the liquid (Pw) by
the pump to the energy supplied to the (Pmo/); that is, the ratio of the water horsepower to
the power input to the primary driver expressed in percent.
Power
Electric motor input power (P1110,). This is the electrical input power to the motor.
Pump input power (Pp). This is the power delivered to the pump shaft at the driver to
pump coupling. It is also called brake horsepower.
Pump output power (PJ. This is the power imparted to the liquid by the pump. It is
also called water horsepower.
P. = ^^

(U-S. units)

(10.1)

"=

Qx

x
^
366 *

<S-L units>

(10-2>

where
Q
H
S
Pw

=
=
=
=

flow in gal/min (U.S.) or mVhr (SI)

head in feet (U.S.) or meters (SI)
specific gravity
power in a horsepower (U.S.) or kilowatt (SI)

Pump efficiency (r|p). This is the ratio of the energy imported to the liquid (Pw) to the
energy delivered to the pump shaft (Pp) expressed in percent.
Pump pressures
Field test pressure. The maximum static test pressure to be used for leak testing a
closed pumping system in the field if the pumps are not isolated. Generally this is
taken as 125 percent if the maximum allowable casing working pressure. In cases
where mechanical seals are used, this pressure may be limited by the pressure-containing capabilities of the seal.
Note: Seesure of the pump to 125 percent of the maximum allowable casing working
pressure on the suction split=case pumps and certain other pump types.
Maximum allowable casing working pressure. This is the hcase pumps and certain
other pump types.
Maximum allowable casing working pressure. This is the highest pressure at the specified pumping temperature for which the pump casing is designed. This pressure shall
be equal to or greater than the maximum discharge pressure. In the case of some
pumps (double suction, vertical turbine, axial split case can pumps, or multistage, for
example), the maximum allowable casing working pressure on the suction side may be
different from that on the discharge side.
Maximum suction pressure. This is the highest section pressure to which the pump
will be subjected during operation.
Working pressure (pd). This is the maximum discharge pressure that could occur in the
pump, when it is operated at rated speed and suction pressure for the given application.
Shut off. This is the condition of zero flow where no liquid is flowing through the
pump, but the pump is primed and running.
Speed. This is the number of revolutions of the shaft is a given unit of time. Speed is
expressed as revolutions per minute.
Suction conditions
Maximum suction pressure. This is the highest suction pressure to which the pump
will be subjected during operation.
Net positive suction head available (NPSHA). Net positive suction head available is
the total suction head of liquid absolute, determined at the first-stage impeller datum,
less the absolute vapor pressure of the liquid at a specific capacity:

NPSHA = hsa - hvp

(10.3)

where
hsa = total suction head absolute = hatm + hs

(10.4)

orNPSHA = hatm + hs-hvp

(10.5)

Net positive suction head required (NPSHR). This is the amount of suction head, over
vapor pressure, required to prevent more than 3 percent loss in total head from the first
stage of the pump at a specific capacity.
Static suction lift (Is). Static suction lift is a hydraulic pressure below atmospheric at
the intake port of the pump.
Submerged suction. A submerged suction exists when the centerline of the pump inlet
is below the level of the liquid in the supply tank.
Total discharge head (hd). The total discharge head (hd) is the sum of the discharge
gauge head (hgd) plus the velocity head (hvd) at point of gauge attachment plus the elevation head (Z4) from the discharge gauge centerline to the pump datum:
Total head (H]. This is the measure of energy increase per unit weight of the liquid,
imparted to the liquid by the pump, and is the difference between the total discharge
head and the total suction head.
This is the head normally specified for pumping applications since the complete characteristics of a system determine the total head required.
hd = hgd + hvd + Zd
(10.6)
Total suction head (hs), closed suction test. For closed suction installations, the pump
suction nozzle may be located either above or below grade level.
Total suction head (hs), open suction. For open suction (wet pit) installations, the first
stage impeller of the bowl assembly is submerged in a pit. The total suction head (hs)
at datum is the submergence (Zw). If the average velocity head of the flow in the pit is
small enough to be neglected, then:
hs = Zw

(10.7)

where
Zw = vertical distance in feet from free water surface to datum.
The total suction head (hs), referred to the eye of the first-stage impeller is the algebraic sum of the suction gauge head (hvs) plus the velocity head (hvs) at point of gauge
attachment plus the elevation head (Z5) from the suction gauge centerline (or manometer
zero) to the pump datum:
hs = hgs + hvs + Z3

(10.8)

The suction head (hs) is positive when the suction gauge reading is above atmospheric pressure and negative when the reading is below atmospheric pressure by an amount
exceeding the sum of the elevation head and the velocity head.
Velocity head (hv). This is the kinetic energy of the liquid at a given cross section.
Velocity head is expressed by the following equation:

h, = ~

(10.9)

where v is obtained by dividing the flow by the cross=sectional area at the point of gauge
connection.
10.1.3 Types of Centrifugal Pumps
The HI and API standards do not agree on these definitions of types of centrifugal pumps
(Figs. 10.2 and 10.3). Essentially, the HI standard divides centrifugal pumps into two
types (overhung impeller and impeller between bearings), whereas the API standard
divides them into three types (overhung impeller, impeller between bearings, and vertically suspended). In the HI standard, the "vertically suspended" type is a subclass of the
"overhung impeller" type.
I

Overhung
Impeller

Close coupled
single or two
stage

End suction
(Including submersibies)
In-line
In-line ANSI B73.2

Centrifugal

frame mounted
Separately
coupled
single or
two stage

Centerline support
API-610
Frame mounted
ANSI B73.1
Wet pit volute
Axial flow impeller (propeller)
volute type (horiz. or vertical)

Sealless

Kinetic

Canned motor
Magnetic drive

Impeller
between
bearing

Separately
coupled
single stage

Axial (horiz.) split case

Separately
coupled
multistage

Axial (horiz.) split case

Peripheral
Regenerative turbine

Radial (vertical) split case

Radial (vertical) split case
Single stage
Multistage

Side channel
Special effect
FIGURE 10.2 Kinetic type pumps per ANSI/HI-1.1-1.5-1994.

Reversible centrifugal
Rotating casing (pitot)

CENTRIFUGAL PUMP

Overhung
Flexibly
Coupled
Horizontal
Foot
Mounted

Rigidly
Coupled

Vertical
In-Line
Bearing
Frame

Centerline
Mounted

OH1 OH2

OH3

TYPES

Vertically Suspended

Between Bearings
Close
Coupled

Vertical
Vertical

1 & 2 Stage Multistage

Axially
Split

High
In-Line opeea
Integral
Gear
In-Line

OH4 OH5

OH6

Radially
Radially
Split
Split
Axially
Split

Discharge
Thru
Column

Single Double
Casing Casing
Axial
Diffuser
Flow
Volute
BB1

BB2 BB3 BB4

FIGURE 10.3 Pump class type identification per API 610.

Double
Casing

Single
Casing
Separate
Discharge
(Sump)
Line CantiShaft lever

BB5 VS1 VS2 VS3 VS4

Diffuser

Volute

VS5 VS6

VS7

NPSH0

Efficiency

Discharge, gal/min
FIGURE 10.4 Typical discharge curves.

10.2

PUMPHYDRAULICS

10.2.1 Pump Performance Curves

The head that a centrifugal pump produces over its range of flows follows the shape of a
downward facing or concave curve (Fig. 10.4). Some types of impellers produce curves
that are not smooth or continuously decreasing as the flow increases: that is, there may be
dips and valleys in the pump curve.
10.2.2

Pipeline Hydraulics and System Curves

A system curve describes the relationship between the flow in a pipeline and the head
loss produced; see Fig. 10.5 for an example. The essential elements of a system curve
include:
The static head of the system, as established by the difference in water surface elevations between the reservoir the pump is pumping from and the reservoir the pump is
pumping to,
The friction or head loss in the piping system. Different friction factors representing
the range in age of the pipe from new to old should always be considered.
The system curve is developed by adding the static head to the headlosses that occur
as flow increases. Thus, the system curve is a hyperbola with its origin at the value of the
static head.
The three most commonly used procedures for determining friction in pipelines are the
following:
10.2.2.1 Hazen-Williams equation. The Hazen-Williams procedure is represented by
the equation:
V = 1.318C/?0635054(U.S. units)

(10.1Oa)

where: V= velocity, (ft/s), C= roughness coefficient, R hydraulic radius, (ft), and S = friction head loss per unit length or the slope of the energy grade line (ft/ft).
In SI units, Eq. (10.10a)is
V = 0.849CR0 63S0 54

(10.1Ob)

where V = velocity (m/s), C = roughness coefficient, R = hydraulic radius, (m) and, S

= friction head loss per unit length or the slope of the energy gradeline in meters per
meter.
A more convenient form of the Hazen-Williams equation for computing headloss or
friction in a piping system is
4.72 /Q\ L85
^=D^()

(10"lla)

Head

System H-Qcurves for different C

values

Friction, fitting
and valve losses

Total static
head

Discharge
FIGURE 10.5 Typical system head-capacity curves.

where H1= headless, (ft), L = length of pipe, (ft), D = pipe internal diameter, (ft), Q =
flow, (ftVs), and C= roughness coefficient or friction factor.
In SI units, The Hazen-Williams equation is.
H

<-

/ L W151Q Y 8 5 10.74L (Q\*

=
(WM)(CD^)
-D^(C)

nftllM
(iailb)

where HL = head loss, (m), Q = flow, (m3/s), D = pipe diameter, (m), and L = pipe
length, (m).
The C coefficient typically has a value of 80 to 150; the higher the value, the smoother
the pipe. C values depend on the type of pipe material, the fluid being conveyed (water or
sewage), the lining material, the age of the pipe or lining material, and the pipe diameter.
Some ranges of values for C are presented below for differing pipe materials in Table
10.1.
TABLE 10.1 Hazen-Williams Coefficents
Pipe Material
PVC
Steel (with mortar lining)
Steel (unlined) 120 to 140
Ductile iron (with mortar lining)

C Value for Water

C Value for Sewage

135-150
120-145
110-130
100-140

130-145
120-140
110-130
100-130

TABLE 10.1 (Continued)

Pipe Material
Asbestos cement
Concrete pressure pipe
Ductile iron (unlined)

C Value for Water

C Value for Sewage

120-140
130-140
80-120

110-135
120-130
80-110

AWWA Manual Mil, Steel PipeA Guide for Design and Installation (AWWA,
1989), offers the following relationships between C factors and pipe diameters for water
service:
C

= 140 + O.lld for new mortar-lined steel pipe (U.S. units)

(10.12)

= 140 + 0.0066929d (SI units, d in (mm)

C

= 130 + 0.16J (U.S. units) for long-term considerations of lining

(10.13)

deterioration, slime buildup, and so on.

= 130 + 0.0062992d (SI units, d in mm),
where
C = roughness coefficient or friction factor (See Table 10.1)
d = pipe diameter, inches or millimeters, as indicated above.
10.2.2.2 Manning's equation. Manning's procedure is represented by the equation
V = ^- R2W2 (U.S. units)
n

(10.14)

V=-R2W2 (SI units),

n
where V=velocity, (f/s or m/s), n=roughness coefficient, R =Hydraulic radius, (ft or m),
and S = friction head loss per unit length or the slope of the energy grade line in feet per
foot or meters per meter.
A more convenient form of the Manning equation for computing head loss or friction
in a pressurized piping system is
H1 = 4'66^6/(3ng)2
10.29L(Hg)2
=
pi

(U.S. units)

(10.15)

. x
/OT
SIumts
(

>

where n=roughness coefficient, H1 = head loss (ft or m), L = length of pipe (ft or m),
D=pipe internal diameter (ft or m), and Q=flow (cuVs or m3/s).
Values of n are typically in the range of 0.010 - 0.016, with n decreasing with
smoother pipes.

10.2.2.3 Darcy-Weisbach equation. The Darcy-Weisbach procedure is represented by

the equation
HL=f~

(10.16)

where / = friction factor from Moody diagram, g = acceleration due to gravity = 32.2
(ft/s) (U.S. units) = 9.81 m/s2 (SI units), HL = head loss (ft or m), L = length of pipe (ft
or m), D = pipe internal diameter (ft or m), and V = velocity (ft/s or m/sec).
Sanks et al., (1998) discuss empirical equations for determining/values. A disadvantage of using the Darcy-Weisbach equation is that the values for/depend on both roughness (EID) and also on the Reynolds number (Re):
R=

VD

(10.17)

where R = reynolds number (dimensionless), V = fluid velocity in the pipe (ft/s or m/s),
D = pipe inside diameter (ft or m), and v = kinematic viscosity (ft2/s or m2/s)
Values for/as a function of Reynold's number can be determined by the following
equations:
64
R less than 2000:/=
(10.18)
K
R = 2000^000: -= = 2 loglo fe + j^\

(10.19)

0.25
r
(ElD 5.74U
(10.20)
log
+
[ \Tr R55-jJ
where EID = roughness, with E = absolute roughness, feet or meters, and D = pipe diameter, (ft or m).
R greater than 4000: / =

Equation 10.19 is the Colebrook-White equation, and Eq. 10.20 is an empirical equation developed by Swamee and Jain, in Sanks et al., (1998). For practical purposes,/values for water works pipelines typically fall in the range of 0.016 to about 0.020.
10.2.2.4 Comparisons off, C9 and n. The Darcy-Weisbach friction factor can be compared to the Hazen-Williams C factor by solving both equations for the slope of the
hydraulic grade line and equating the two slopes. Rearranging the terms gives, in SI units,
/=(c^)(^i)

(10-21a)

where v is in m/s and D is in m. In U.S. customary units, the relationship is

/=(i)(^w)

<10-21b)

where v is in fps and D is in feet (Sanks et al., 1998).

For pipes flowing full and under pressure, the relationship between C and n is
)0.037
n = Ll2
(1(X22a)
cs^
in SI units, where D is the inside diameter ID in m. In U.S. customary units, the equation
is
7)0.037
" = L07cs^
dO-22b)
where D is the (ID) in ft.
10.2.3 Hydraulics of Valves
The effect of headlosses caused by valves can be determined by the equation for minor
losses:
hL = K

(10.23)

where hL = minor loss (ft or m), K = minor loss coefficient (dimensionless), V = fluid
velocity (ft/s or m/s), and g = acceleration due to gravity (= 32.2 ft-s/s or 9.81 m-s/s).
Headless or pressure loss through a valve also is determined by the equation
Q = Cv VAP (U.S. units) = 0.3807CV VAP

(S.I. units)

(10.24)

where Q = flow through valve (gal/m or m /s), Cv = valve capacity coefficient, and AP
= pressure loss through the valve (psi or kPa)
The coefficient Cv varies with the position of the valve plug, disc, gate, and so forth.
Cy indicates the flow that will pass through the valve at a pressure drop of 1 psi. Curves
of Cv versus plug or disc position (0-90,with O being in the closed position) must be
obtained from the valve manufacturer's catalogs or literature.
Cv and K are related by the equation
Cv =29.85 -^=

(U.S. units)

(10.25)

where d = valve size, (in).

Thus, by determining the value for Cv from the valve manufacturer's data, a value for
K can then be calculated from Eq. (10.25). This K value can then be used in Eq. (10.23)
to calculate the valve headloss.
10.2.4

Determination of Pump Operating PointsSingle Pump

The system curve is superimposed over the pump curve; (Fig. 10.6). The pump operating
points occur at the intersections of the system curves with the pump curves. It should be

observed that the operating point will change with time. As the piping ages and becomes
rougher, the system curve will become steeper, and the intersecting point with the pump
curve will move to the left. Also, as the impeller wears, the pump curve moves downward.
Thus, over a period of time, the output capacity of a pump can decrease significantly. See
Fig. 10.7. for a visual depiction of these combined effects.
10.2.5 Pumps Operating in Parallel
To develop a composite pump curve for pumps operating in parallel, add the flows together
that the pumps provide at common heads (Fig. 10.8). This can be done with identical pumps
(those having the same curve individually) as well as with pumps having different curves.
10.2.6 Variable-Speed Pumps
The pump curve at maximum speed is the same as the one described above. The point on
a system-head curve at which a variable=speed pump will operate is similarly determined
by the intersection of the pump curve with the system curve. What are known as the pump
affinity laws or homologous laws must be used to determine the pump curve at reduced
speeds. These affinity laws are described in detail in Chap 12. For the discussion here, the
relevant mathematical relationships are Sanks et al., (1998).
1 = !L.
Q2 n2

(10.26)

IL = (Zk)2
H2
n2

(10.27)

- = (^
(10.28)
H
*2
2
where Q=flow rate, H=head, Ppower, n=rotational speed, and subscripts 1 and 2 are
only for corresponding points. Equations (10.26) and (10.27) must be applied simultaneously to ensure that Point 1 "corresponds" to Point 2. Corresponding points fall on parabolas through the original. They do not fall on system H-Q curves. These relationships,
known collectively as the affinity laws, are used to determine the effect of changes in
speed on the capacity, head, and power of a pump.
The affinity laws for discharge and head are accurate because they are based on actual
tests for all types of centrifugal pumps, including axial-flow pumps. The affinity law for
power is not as accurate because efficiency increases with an increase in the size of the pump.
When applying these relationships, remember that they are based on the assumption that
the efficiency remains the same when transferring from a given point on one pump curve to
a homologous point on another curve. Because the hydraulic and pressure characteristics at
the inlet, at the outlet, and through the pump vary with the flow rate, the errors produced by
Eq. (10.28) may be excessive, although errors produced by Eqs. (10.26) and (10.27) are
extremely small. See Fig. 10.9 for an illustration of the pump curves at different speeds.
Example. Consider a pump operating at a normal maximum speed of 1800 rpm, having a head-capacity curve as described in Table 10.2. Derive the pump curve for operating speeds of 10001600 rpm at 200-rpm increments.

System head
capacity curve
Pump head
capacity curve

Pump operating point

Efficiency

Discharge, gpm
FIGURE 10.6 Determining the operating point for a single-speed pump with a fixed value of hstat

Head

Original performance

Performance
after wear

Capacity
FIGURE 10.7 Effect of impeller wear
The resulting new values for capacity (Q) and head (H) are shown in Table 10.2. The
values are derived by taking the Q values for the 1800 rpm speed and multiplying them
by the ratio O1M2) and by taking the H values for the 1800 rpm speed and multiplying
them by the ratio (H1In1)2.

10.3

CONCEPTOFSPECIFICSPEED

10.3.1 Introduction: Discharge-Specific Speed

The specific speed of a pump is defined by the equation:
, = H
<10-29)
where Ns = specific speed (unitless), n = pump rotating speed (rpm), Q = pump discharge
flow (gal/mm, mVs, L/s, m3h) (for double suction pumps, Q is one-half the total pump
flow, and H = total dynamic head (ft or m) (for multistage pumps, H is the head per stage),
The relation between specific speeds for various units of discharge and head is given
in Table 10.3, wherein the numbers in bold type are those customarily used (Sanks et al.,
1998).
Pumps having the same specific speed are said to be geometrically similar. The specific speed is indicative of the shape and dimensional or design characteristics of the pump
impeller (HI, 1994). Sanks et al. (1998) also gives a detailed description and discussion of
impeller types as a function of specific speed. Generally speaking, the various types of

Ratio
Vn2

1,800

1,00

1,00

200

1,000

180

2,000

160

3,000

130

1,600

0,889

0,790

158

889

142

1,778

126

2,667

103

1,400

0,777

0,605

121

778

109

1,556

97

2,333

79

1,200

0,667

0,444

89

667

80

1,333

71

2,000

58

1,000

0,555

0,309

62

556

1,111

49

1,667

40

Speed
(rpm)

Ratio
U1Xn2

1,800

1,00

Ratio
(Vn2)2

HeadCapacity at Various Points

Point 2
I
Point 3
Q (gpm) H(feet) I Q (gpm) H(feet)

Speed
(rpm)

Ratio
(Vn2)"

1,00

Point 1
Q (gpm) H(feet)

Point 1
Q (gpm) H(feet)

HeadCapacity at Various Points

Point 3
Point 2
I
Q (gpm) H(feet) I Q (gpm) H(feet)

Point 4
Q (gpm) H(feet)

Point 4
Q (gpm) H(feet)

60

63

55

126

49

189

40

56

44

112

39

168

32

1,600

0,889

0,790

1,400

0,777

0,605

36

49

33

98

30

147

24

1,200

0,667

0,444

27

42

24

84

22

126

18

1,000

0,555

0,309

19

33

17

70

15

105

12

Two pumps
in parallel

Operating
point

One pump
Flow from one pump
with both pumps operating

i
System head <

Flow from one pumpi

with only one pump I
operating
I

Capacity (unitless)*
FIGURE 10.8 Pumps operating in parallel

System-head
curve
Head. H

Head Opacity
at

Peed
Head capacity at
speed

Friction
losses

Static pressure or head

Capacity,
FIGURE 10.9 Typical Discharge Curves for a Variable Speed Pump
impeller designs are as follows:
Type of Impeller
Radial-vane
Mixed-flow
Axial-flow

Specific Speed Range (U.S. Units)

500 -4200
4200-9000
9000-15,000

10.3.2 Suction-Specific Speed

Suction-specific speed is a number similar to the discharge specific and is determined by
the equation
*

nQQ.50
NPSH/*

Q0m
Ua3UJ

where S = suction-specific speed (unitless) n = pump rotating speed (rpm) Q = pump

discharge flow as defined for Eq. (10.29).
NPSHR = net positive suction head required, as described in Sec. 10.4
The significance of suction-specific speed is that increased pump speed without proper suction head conditions can result in excessive wear on the pump's components
(impeller, shaft, bearings) as a result of excessive cavitation and vibration (Hydraulics
Institute, 1994). That is, for a given type of pump design (with a given specific speed),
there is an equivalent maximum speed (n) at which the pump should operate.
Rearranging Eq. (10.30) results in
S X NPSH,075
=
05T*-

(1

'31)

Equation (10.31) can be used to determine the approximate maximum allowable pump
speed as a function of net positive suction head available and flow for a given type of

TABLE 10.3 Equivalent Factors for Converting Values of Specific Speed Expressed in One Set
of Units to the Corresponding Values in Another Set of Units
Quantity
N
Q
H

(rev/min,
IJs,
m)
1.0
31.62
0.527
0.612
12.98

Expressed in Units of
(rev/min,
(rev/min,
(rev/min,
gal/mn,
m3/s,
m3/h,
m)
m)
ft)
1.898
1.633
0.0316
51.64
60.0
1.0
0.0167
1.0
0.861
1.162
0.0194
1.0
24.63
0.410
21.19

(rev/min,
ft3/s,
ft)
0.0771
2.437
0.0406
0.0472
1.0

Source: Sanks, et al 1998

For example, if the specific speed is expressed in metric units (e.g., N = rev/min, Q = m3/s, and H =
m), the corresponding value expressed in U.S. customary units (e.g., N = rev/min, Q = gal/min, and H
= feet) is obtained by multiplying the metric value by 51.64.
pump (i.e., a given suction-specific speed). Inspection of Eq. (10.31) reveals that, for a
given specific speed, the following pump characteristics will occur:
The higher the desired capacity (Q), the lower the allowable maximum speed. Thus, a
properly selected high-capacity pump will be physically larger beyond what would be
expected due solely to a desired increased capacity.
The higher the NPSH4, the higher the allowable pump speed.

70.4 NETPOSITIVESUCTIONHEAD
Net positive suction head, or NPSH, actually consists of two concepts:
the net positive suction available (NPSHA), and
the net positive suction head required (NPSH^).
The definition of NPSHA and NPSH^, as given by the Hydraulics Institute (1994),
were presented in Sec. 10.1.
10.4.1 Net Positive Suction Head Available
Figure 10.1 visually depicts the concept of NPSHA. Since the NPSH4 is the head available
at the impeller, friction losses in any suction piping must be subtracted when making the
calculation. Thus, the equation for determining NPSH4 becomes
NPSH, = halm + hs- hvp - hL
where:

(10.32)

hatm = atmospheric pressure (ft or m).

hs = static head of water on the suction side of the pump (ft or m) (hs is negative if
the water surface elevation is below the eye of the impeller).
hvp = vapor pressure of water, which varies with both altitude and temperature (ft or
m), and
hL = friction losses in suction piping (ft or m), typically expressed as summation of
velocity heads (KV1IIg) for the various fittings and pipe lengths in the suction
piping.
Key points in determining NPSHA are as follows (Sanks et al., 1998):
the barometric pressure must be corrected for altitude,
storms can reduce barometric pressure by about 2 percent, and
the water temperature profoundly affects the vapor pressure.
Because of uncertainties involved in computing NPSH4, it is recommended that the
NPSHA be at least 5 ft (1.5 m) greater than the NPSH7J or 1.35 times the NPSH7J as a factor
of safety (Sanks et al., 1998). An example of calculating NPSH4 is presented in Section
10.5.
10.4.2 Net Positive Suction Head Required by a Pump
Hydraulics Institute (1994) and Sanks et al., (1998) have discussed the concept and
implications of NPSH7J in detail. Their discussions are presented or summarized as follows. The NPSH7J is determined by tests of geometrically similar pumps operated at
constant speed and discharge but with varying suction heads. The development of cavitation is assumed to be indicated by a 3 percent drop in the head developed as the suction inlet is throttled, as shown in Fig. 10.10. It is known that the onset of cavitation
occurs well before the 3 percent drop in head (Cavi, 1985). Cavitation can develop substantially before any drop in the head can be detected, and erosion indeed, occurs more
rapidly at a 1 percent change in head (with few bubbles) than it does at a 3 percent
change in head (with many bubbles). In fact, erosion can be inhibited in a cavitating
pump by introducing air into the suction pipe to make many bubbles. So, because the 3
percent change is the current standard used by most pump manufacturers to define the
NPSH7J, serious erosion can occur as a result of blindly accepting data from catalogs. In
critical installations where continuous duty is important, the manufacturer should be
required to furnish the NPSH7J test results. Typically, NPSH7J is plotted as a continuous
curve for a pump (Fig. 10.11). When impeller trim has a significant effect on the
NPSH7J, several curves are plotted.
The NPSH required to suppress all cavitation is always higher than the NPSH7J
shown in a pump manufacturer's curve. The NPSH required to suppress all cavitation
at 40 to 60 percent of a pump's flow rate at BEP can be two to five times as is necessary to meet guaranteed head and flow capacities at rated flow (Fig. 10.10; Taylor,
1987). The HI standard (Hydraulics Institute, 1994) states that even higher ratios of
NPSHA to NPSH7J may be required to suppress cavitation: It can take from 2 to 20
times the NPSH7J to suppress incipient cavitation completely, depending on the
impeller's design and operating capacity.
If the pump operates at low head at a flow rate considerably greater than the capacity
at the BEP, Eq. (10.33) is approximately correct:

NPSH7J at operating point

NPSH7, at BEP

/ Q at operating point \
\
Q at BEP
)

(10.33)

where the exponent n varies from 1.25 to 3.0, depending on the design of the impeller. In
most water and wastewater pumps, n lies between 1.8 and 2.8. The NPSH7J at the BEP
increases with the specific speed of the pumps. For high-head pumps, it may be necessary
either to limit the speed to obtain the adequate NPSH at the operating point or to lower
the elevation of the pump with respect to the free water surface on the suction side i to
increase the NPSHA.
10.4.3 NPSH Margin or Safety Factor Considerations
Any pump and piping system must be designed such that the net positive suction head
available (NPSHA) is equal to, or exceeds, the net positive suction head required
(NPSH^) by the pump throughout the range of operation. The margin is the amount by
which NPSHA exceeds NPSH^ (Hydraulics Institute, 1994). The amount of margin
required varies, depending on the pump design, the application, and the materials of
construction.
Practical experience over many years has shown that, for the majority of pump
applications and designs, NPSH^ can be used as the lower limit for the NPSH available. However, for high=energy pumps, the NPSH7J may not be sufficient. Therefore,
the designer should consider an appropriate NPSH margin over NPSH7J for high-energy pumps that is sufficient at all flows to protect the pump from damage caused by
cavitation.

Head

Point of cavitation inception

NPSHR (at 3% drop)

NPSH1 (at point of inception)

NPSH
FIGURE 10.10 Net positive suction head criteria as determined from pump test results.

NPSHR to
eliminate
cavitation

NPSHR to deliver
capacities at 3%
reduction of head
on H- Q

NPSHR, ft

Head and efficiency (scales omitted)

Efficiency

Capacity, % of Q at B.E.P.
FIGURE 10.11 NPSH Required to suppress visible cavitation.
10.4.4 Cavitation
Cavitation begins to develop in a pump as small harmless vapor bubbles, substantially
before any degradation in the developed head can be detected (Hydraulics Institute, 1994).
This is called the point of incipient cavitation (Cavi, 1985; Hydraulics Institute, 1994).
Studies on high-energy applications show that cavitation damage with the NPSHA
greater than the NPSH^ can be substantial. In fact, there are studies on pumps that show
the maximum damage to occur at NPSH4 values somewhere between O and 1 percent head
drop (or two to three times the NPSH^), especially for high suction pressures as required
by pumps with high impeller-eye peripheral speeds. There is no universally accepted relationship between the percentage of head drop and the damage caused by cavitation. There
are too many variables in the specific pump design and materials, properties of the liquid
and system. The pump manufacturer should be consulted about NPSH margins for the
specific pump type and its intended service on high-energy, low-NPSHA applications.
According to a study of data contributed by pump manufacturers, no correlation exists,
between the specific speed, the suction specific speed, or any other simple variable and
the shape of the NPSH curve break-off. The design variables and manufacturing variables
are too great. This means that no standard relationship exists between a 3, 2, 1 or O percent head drop. The ratio between the NPSH required for a O percent head drop and the
NPSH^ is not a constant, but it generally varies over a range from 1.05 to 2.5. NPSH for
a O, 1, or 2 percent head drop cannot be predicted by calculation, given NPSH^.
A pump cannot be constructed to resist cavitation. Although a wealth of literature is
available on the resistance of materials to cavitation erosion, no unique material property
or combination of properties has been found that yields a consistent correlation with
cavitation damage rate (Sanks et al., 1998).

70.5 CORRECTED PUMP CURVES

Figures 10.6 and 10.9 depict "uncorrected" pump curves. That is, these curves depict a
pump H-Q curve, as offered by a pump manufacturer. In an actual pumping station design,
a manufacturer's pump must be "corrected" by subtracting the head losses that occur in
the suction and discharge piping that connect the pump to the supply tank and the pipeline
system. See Table 10.4 associated with Fig. 10.12 in the following sample problem in performing these calculations. The example in Table 10.4 uses a horizontal pump. If a vertical turbine pump is used, minor losses in the pump column and discharge elbow also must
be included in the analysis. This same example is worked in U.S. units in Appendix 10.A
to this chapter.
Problem
L Calculation of minor losses. The principal headless equation for straight sections of
pipe is:
(10 lla)

H^)W*

where L = length (m), D pipe diameter (m), Q = flow (mVs), C = Hazen-Williams

friction factor.

TABLE 10.4 Calculate Minor Losses

Item in
Fig. 10.12

Description

Pipe Size
mm

1
Entrance
300
0.30
2
90 elbow
300
0.30
3
4.5 m of straight pipe
300
0.30
4
30 elbow
300
0.30
5
2 m of straight pipe
300
0.30
6
Butterfly valve
300
0.30
7
1.2m of straight pipe
300
0.30
8
300 mm X 200 mm reducer
200
0.20
9
150 mm X 250 mm increaser
250
0.25
10
1 m of straight pipe
250
0.25
11
Pump control valve
250
0.25
12
1 m of straight pipe
250
0.25
13
Butterfly valve
250
0.25
14
0.60 m of straight pipe
250
0.25
15
90 elbow
250
0.25
0.30
16
1.5m of straight pipe
250
0.25
17
Tee connection
250
0.25
Typical K values. Different publications present other values.
tReasonable value for mortar-lined steel pipe. Value can range from 130 to 145.

Friction Factor
K*
C+
1.0
0.30
140
0.20
140
0.46
140
0.25
0.25
140
0.80
140
0.46
140
140
0.50

HORIZONTAL SPLIT
CASE PUMP

FOREBAY

PLAN VIEW

HIGH WATER
LEVEL
LOW WATER
LEVEL

PUMP

FLOOR

SECTION VIEW
FIGURE 10.12

Piping system used in example in Table 10.4.

The principal headless equation for fittings is

*. = 2X

where K = fitting friction coefficient, V = velocity (m/s), g = acceleration due to gravity (m-s/s)
Sum of K values for various pipe sizes:
K300 = 1.96
K200 = 0.25
^250 = 2.31
Sum of C values for various pipe sizes:

Pipe lengths for 300-mm pipe: L = IHm

Pipe lengths for 250-mm pipe: L = 4.1 m
Determine the total headloss:

Convert V2/2g terms to Q2 terms:

Therefore,

TABLE 10.5 Convert Pump Curve Head Values to Include Minor Piping Losses

Q
Us
O
63
126
189
252

m3/s
O
0.063
0.126
0.189
0.252

H(m)
Uncorrected
60
55
49
40
27

Corrected
60
54.74
47.99
37.74
23.01

2. Modification of pump curve. Using the above equation for H1, a "modified"
pump curve can then be developed (see Table 10.5)
The H values as corrected must then be plotted. The operating point of the pump is
the intersection of the corrected H-Q curve with the system curve.
3. Calculation of NPSHA Using the data developed above for calculating the minor
losses in the piping, it is now possible to calculate the NPSHA for the pump. Only the
minor losses pertaining to the suction piping are considered: items 1-8 in Fig. 10.12. For
this suction piping, we have:
K300,-. = 1-96
Km m = 0.25
Sum of the C values: pipe length for 300-mm pipe is L = 7.7 m.
Determine the headloss in the suction piping:

Tabla 10.6 Computation of NPSHA:

Condition

Flow
(m3/s)

hs
(m)

hatm
(m)

hvp
(m)

High=static suction head O

0.06
0.12
0.18
0.24
0.30
Low-static suction head O
0.06
0.12
0.18
0.24
0.30

9.0
9.0
9.0
9.0
9.0
9.0
1.0
1.0
1.0
1.0
1.0
1.0

10.35
10.35
10.35
10.35
10.35
10.35
10.35
10.35
10.35
10.35
10.35
10.35

0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24
0.24

HL at Flow NPSH at Flow

(m)
(m)
0.00
0.12
0.53
1.20
2.12
3.29
0.00
0.12
0.53
1.20
2.12
3.29

19.11
18.99
18.58
17.91
16.99
15.81
11.11
10.99
10.58
9.91
8.99
7.82

- 3.10 <2185 + 19.99 Q2 + 12.90 Q2

= 3.10Q185 + 32.89 Q2
For Fig. 10.12, assume that the following data apply:
High-water level = elevation 683 m
Low-water level = elevation 675 m
Pump centerline elevation = 674 m
Therefore:
Maximum static head = 683 674 = 9 m
Minimum static head = 675 674 = 1 m
Per Eq. (10.32), with computation of NPSHA shown in table 10.5
NPSHA = hatm + hs- hvp - hL
For this example, use
hatm
hvp
hs
hs

10.6

=
=
=
=

10.35m
0.24 mat 150C
9 m maximum
1 m minimum

HYDRAULICCONSIDERATIONSINPUMPSELECTION

10.6.1 Flow Range of Centrifugal Pumps

The flow range over which a centrifugal pump can perform is limited, among other things,
by the vibration levels to which it will be subjected. As discussed in API Standard 610
(American Petroleum Institute, n.d.), centrifugal pump vibration varies with flow, usually being a minimum in the vicinity of the flow at the BEP and increasing as flow is
increased or decreased. The change in vibration as flow is varied from the BEP depends
on the pump's specific speed and other factors. A centrifugal pump's operation flow range
can be divided into two regions. One region is termed the best efficient or preferred operating region, over which the pump exhibits low vibration. The other region is termed the
allowable operating range, with its limits defined as those capacities at which the pump's
vibration reaches a higher but still "acceptable" level.
ANSI/HI Standard 1.1-1.5 (Hydraulics Institute, 1994) points out that vibration can be
caused by the following typical sources:
1. Hydraulic forces produced between the impeller vanes and volute cutwater or diffuser
at vane-passing frequency.
2. Recirculation and radial forces at low flows. This is one reason why there is a definite
minimum capacity of a centrifugal pump. The pump components typically are not
designed for continuous operation at flows below 60 or 70 percent of the flow that
occurs at the BEP.

3. Fluid separation at high flows. This is one reason why there is also a definite maximum
capacity of a centrifugal pump. The pump components typically are not designed for
continuous operation of flows above about 120 to 130 percent of the flow that occurs
at the BEP.
4. Cavitation due to net positive suction head (NPSH) problems. There is a common misconception that if the net positive suction head available (NPSH4) is equal to or greater
than the net positive suction head required (NPSH^) shown on a pump manufacturer's
pump curve, then there will be no cavitation. This is wrong! As discussed in ANSI/HI
1.1-1.5-1994 (Hydraulics Institute, 1994 and also discussed by Taylor (1987), it takes
a suction head of 2 to 20 times the NPSH^ value to eliminate cavitation completely.
5. Flow disturbances in the pump intake due to improper intake design.
6. Air entrainment or aeration of the liquid.
7. Hydraulic resonance in the piping.
8. Solids contained in the liquids, such as sewage impacting in the pump and causing
momentary unbalance, or wedged in the impeller and causing continuous unbalance.
The HI standard then states:
The pump manufacturer should provide for the first item in the pump design and establish limits for low flow. The system designer is responsible for giving due consideration
to the remaining items.
The practical applications of the above discussion by observing what can happen in a
plot of a pump curve-system head curve as discussed above in Fig. 10.6. If the intersection of the system curve with the pump H-Q curve occurs too far to the left of the BEP
(i.e., at less than about 60 percent of flow at the BEP) or too far to the right of the BEP
(i.e., at more than about 130 percent of the flow at the BEP), then the pump will eventually fail as a result of hydraulically induced mechanical damage.
10.6.2 Causes and Effects of Centrifugal Pumps Operating Outside
Allowable Flow Ranges
As can be seen in Fig. 10.6, a pump always operates at the point of intersection of the system curve with the pump H-Q curve. Consequently, if too conservative a friction factor is
used in determining the system curve, the pump may actually operate much further to the
right of the assumed intersection point so that the pump will operate beyond its allowable
operating range. Similarly, overly conservative assumptions concerning the static head in
the system curve can lead to the pump operating beyond its allowable range. See Fig. 10.13
for an illustration of these effects. The following commentary discusses the significance of
the indicated operating points 1 through 6 and the associated flows Q1 through Q6.
Q1 is the theoretical flow that would occur, ignoring the effects of the minor head losses in the pump suction and discharge piping. See Fig. 10.12 for an example. Q1 is
slightly to the right of the most efficient flow, indicated as 100 units.
Q2 is the actual flow that would occur in this system, with the effects of the pump suction and discharge piping minor losses included in the analysis. Q2 is less than Q1, and
Q2 is also to the left of the point of most efficient flow. As shown in Fig. 10.7, as the
impeller wears, this operating point will move even further to the left and the pump
will become steadily less efficient.

Q1 and Q2 are the flows that would occur assuming that the system head curve that is
depicted is "reasonable," that is, not unrealistically conservative. If, in fact, the system
head curve is flatter (less friction in the system than was assumed), then the operating
point will be Q3 (ignoring the effects of minor losses in the pump suction and discharge
piping). If these minor losses are included in the analysis, then the true operating point
is Q4. At Q3, the pump discharge flow in this example is 130 percent of the flow that
occurs at the BEP. A flow of 130 percent of flow at the BEP is just at edge of, and may
even exceed, the maximum acceptable flow range for pumps (see discussion in Sec.
10.6.1). With most mortar-lined steel or ductile-iron piping systems, concrete pipe, or
with plastic piping, reasonable C values should almost always be in the range of 120
145 for water and wastewater pumping systems. Lower C usually would be used only
when the pumping facility is connected to existing, old unlined piping that may be
rougher.
If the static head assumed was too conservative, then the actual operating points would
be Q5 or Q6. Q5 is 150 percent of the flow at the BEP. Q6 is 135 percent of the flow at
the BEP. In both cases, it is most likely that these flows are outside the allowable range
of the pump. Cavitation, inadequate NPSH4, and excessive hydraulic loads on the
impeller and shaft bearings may likely occur, with resulting poor pump performance
and high maintenance costs.
10.6.3

Summary of Pump Selection

In selecting a pump, the following steps should be taken:

1. Plot the system head curves, using reasonable criteria for both the static head range and
the friction factors in the piping. Consider all feasible hydraulic conditions that will
occur:
a. Variations in static head
b. Variations in pipeline friction factor (C value)
Variations in static head result from variations in the water surface elevations (WSE)
in the supply reservoir to the pump and in the reservoir to which the pump is pumping. Both minimum and maximum static head conditions should be investigated:
Maximum static head. Minimum WSE in supply reservoir and maximum WSE in
discharge reservoir.
Minimum static head. Maximum WSE in supply reservoir and minimum WSE in
discharge reservoir.
2. Be sure to develop a corrected pump curve or modified pump curve by subtracting the
minor losses in the pump suction and discharge piping from the manufacturer's pump
curve (Table 10.5 and Fig. 10.13). The true operating points will be at the intersections
of the corrected pump curve with the system curves.
3. Select a pump such that the initial operating point (intersection of the system head
curve with pump curve) occurs to the right of the BEP. As the impeller wears, the
pump output flow will decrease (Fig. 10.7), but the pump efficiency will actually
increase until the impeller has worn to the level that the operating point is to the left
of the BEP.
For a system having a significant variation in static head, it may be necessary to
select a pump curve such that at high static head conditions the operating point is to
the left of the BEP. However, the operating point for the flows that occur a majority
of the time should be at or to the right of the BEP. Bear in mind that high static head

Pump head capacity

curve per manufacturer
System head
capacity curve
(assumed
conservative
friction value)

Pump head^^^
capacity curve,
corrected for minor
losses in pump piping
Theoretical Pump
operating point (T)
Actual Pump
operating point (2)

hstat
hstat
(actual)

System head
capacity curve
(actual friction
value)

(assunned)
Allowable operating
range of pump

Discharge as % of flow at the Best Efficiency Point (BEP)

FIGURE 10.13 Determining the operating point for a single-speed pump.

conditions normally only occur a minority of the time: the supply reservoir must be
at its low water level and the discharge reservoir must simultaneously be at its maximum water levelconditions that usually do not occur very often. Consequently,
select a pump that can operate properly at this conditionbut also select the pump
that has a BEP which occurs at the flow that will occur most often. See Fig. 10.14
for an example.
4. In multiple-pump operations, check the operating point with each combination of
pumps that may operate. For example, in a two-pump system, one pump operating
alone will produce a flow that is greater than 50 percent of the flow that is produced
with both pumps operating. This situation occurs because of the rising shape of the
system head curve; see Fig. 10.8. Verify that the pump output flows are within the
pump manufacturer's recommended operating range; see Fig. 10.13.
5. Check that NPSHA exceeds the NPSH7J for all the hydraulic considerations and operating points determined in Steps 1 and 3.

10.7 APPLICATION OF PUMP HYDRAULIC ANALYSIS

TO DESIGN OF PUMPING STATION COMPONENTS
10.7.1 Pump Hydraulic Selections and Specifications
10.7.1.1 Pump operating ranges Identify the minimum, maximum, and design flows for
the pump based on the hydraulic analyses described above. See Fig. 10.14 as an example.
The flow at 100 units would be defined as the design point.
There is a minimum flow of 90 units.
There is a maximum flow of 115 units.
In multiple-pump operation, the combination of varying static head conditions and the
different number of pumps operating in parallel could very likely result in operating
points as follows (100 units = flow at BEP; see accompaning Table 10.7).
Table 10.7 Pump Operating Ranges
Operating Flow Condition

Flow
(per Pump)

Minimum

70

Normal 1

100

Normal 2

110

Maximum 1

115

Maximum 2

125

Comments
Maximum static head condition,
all pumps operating
Average or most frequent operating
condition: fewer than all pumps
operating, average static head
condition. Might also be the case
of all pumps operating, minimum
static head condition.
Fewer than all pumps operating,
minimum static head condition
Maximum static head condition,
one pump operating
Minimum static head condition,
one pump operating

System head
capacity curves
Pump head
capacity curve

Operating point at
maximum static
head

Most efficient operating point

(equals 100 units of flow)

hstat
hstat
(minimum)

Operating point
at minimum
static head
Efficiency

Discharge, as % of flow at Best Efficiency Point (BEP)

FIGURE 10.14 Determining the operating points for a single-speed pump with variation in values of hstat.

Some observations of the above example are:

The flow range of an individual pump is about 1.8:1 (125 -T- 70).
The pump was deliberately selected to have its most efficient operating point
(Q= 100) at the most frequent operating condition, not the most extreme condition. This
will result in the minimum power consumption and minimum power cost for the system.
The pump was selected or specified to operate over all possible conditions, not just one
or two conditions.
In variable-speed pumping applications, the minimum flow can be much lower than
what is shown in these examples. It is extremely important that the minimum flow be identified in the pump specification so that the pump manufacturer can design the proper combination of impeller type and shaft diameter to avoid cavitation and vibration problems.
10.7.1.2 Specific pump hydraulic operating problems. Specific problems that can occur
when operating a centrifugal pump beyond its minimum and maximum capacities include
(Hydraulics Institute, 1994):
Minimum flow problems. Temperature buildup, excessive radial thrust, suction recirculation, discharge recirculation, and insufficient NPSH4.
Maximum flow problems. Combined torsional and bending stresses or shaft deflection
may exceed permissible limits; erosion drainage, noise, and cavitation may occur
because of high fluid velocities.
10.7.2

Piping

Having selected a pump and determined its operating flows and discharge heads or pressures, it is then desirable to apply this data in the design of the piping. See Fig. 10.12 for
typical piping associated with a horizontal centrifugal pump.
10.7.2.1 Pump suction and discharge piping installation guidelines. Section 1.4 in the
Hydraulic Institute (HI) publication ANSI/HI 1.1-1.5 (1994) and Chap. 6 in API
Recommended Practice 686 (1996) provide considerable discussion and many recommendations on the layout of piping for centrifugal pumps to help avoid the hydraulic problems discussed above.
10.7.2.2 Fluid velocity. The allowable velocities of the fluid in the pump suction and discharge piping are usually in the following ranges:
Suction:

3-9 ft/s (4-6 ft/s most common)

Discharge:

5-15 ft/s (7-10 ft/s most common)

1.0-2.7 m/s (1.2-1.8 m/s most common)

1.5-4.5 m/s (2-3 m/s most common)

Bear in mind that the velocities will vary for a given pump system as the operating
point on a pump curve (i.e., intersection of the pump curve with the system curve) varies
for the following reasons:

1. Variation in static heads, as the water surface elevations in both the suction and discharge reservoirs vary
2. Long-term variations in pipeline friction factors (Fig. 10.5)
3. Long-term deterioration in impeller (Fig. 10.7)
4. Variation in the number of pumps operating in a multipump system (Fig. 10.8).
A suggested procedure for sizing the suction and discharge piping is as follows:
1. Select an allowable suction pipe fluid velocity of 3-5 ft/s (1.0-1.5 m/s) with all pumps
operating at the minimum static head condition. As fewer pumps are used, the flow
output of each individual pump will increase (typically by about 20 to 40 percent with
one pump operating compared to all pumps operating) with the resulting fluid velocities in the suction piping also increasing to values above the 3-5 ft/s (1.0-1.5 m/s)
nominal criteria;
2. Select an allowable discharge pipe fluid velocity of 5-8 ft/s (1.5-2.4 m/s) also with all
pumps operating at the minimum static head condition. As discussed above, as fewer
pumps are used, the flow output of each individual pump will increase with the resulting fluid velocities in the discharge piping also increasing in values above the 5-8 ft/s
(1.5-2.4 m/s) nominal criteria.
10.7.2.3 Design of pipe wall thickness (pressure design) Metal pipes are designed for
pressure conditions by the equation for hoop tensile strength:
t=

^E

(10 34)

'

where
t = wall thickness, in or mm
D = inside diameter, in or mm (although in practice, the outside diameter is often
conservatively used, partly because the ID is not known initially and because it
is the outside diameter (OD) that is the fixed dimension: ID then varies with the
wall thickness)
P = design pressure (psi or kPa)
S = allowable design circumferential stress (psi or kPa)
E = longitudinal joint efficiency
The design value for S is typically 50 percent of the material yield strength, for "normal" pressures. For surge or transient pressures in steel piping systems, S is typically
allowed to rise to 70 percent of the material yield strength (American Water Works
Association 1989).
The factor E for the longitudinal joint efficiency is associated with the effective
strength of the welded joint. The ANSI B31.1 (American Society for Mechanical
Engineers, 1995) and B31.3 (American Society for Mechanical Engineers, 1996) codes
for pressure piping recommend the values for E given in Table 10.8

TABLE 10.8 Weld Joint Efficiencies

Type of Longitudinal Joint
Arc or gas weld (steel pipe)
Single-butt weld
Double-butt weld
Single-or double-butt weld with 100% radiography
Electric resistance weld (steel pipe)
Furnace butt weld (steel pipe)
Most steel water pipelines
Ductile iron pipe

Weld Joint
Efficiency Factor (E)
0.80
0.90
1.00
0.85
0.60
0.85
1.0

The wall thickness for plastic pipes [polyvinyl chloride (PVC), high-density polyethylene (HDPE), and FRP] is usually designed in the United States on what is known as the
hydrostatic design basis or HDB:
,,.^.x
where Pt = total system pressure (operating + surge), t = minimum wall thickness (in),
D = average outside diameter (in), HDB = hydrostatic design basis (psi) anh F = factor
of safety (2.50-4.00)
10.7.2.4 Design of pipe wall thickness (vacuum conditions). If the hydraulic transient
or surge analysis (see Chap. 12) indicates that full or partial vacuum conditions may
occur, then the piping must also be designed accordingly. The negative pressure required
to collapse a circular metal pipe is described by the equation:
(10 36)
^ = (T^SP (^
'
where AP = difference between internal and external pipeline pressures (psi or kPa), E =
modulus of elasticity of the pipe material (psi or kPa), fi = Poisson's ratio, SF = safety factor (typically 4.0), e = wall thickness (in or m) anh D = outside diameter (in or m)

Because of factors such as end effects, wall thickness variations, lack of roundness,
and other manufacturing tolerances, Eq. (10.3b) for steel pipe is frequently adjusted in
practice to
AP = 50,000,000 MB
SF
\DJ

(1037)

10.7.2.5 Summary of pipe design criteria. The wall thickness of the pump piping system is determined by consideration of three criteria:
1. Normal operating pressure [Eq. (10.34)], with S = 50 percent of yield strength
2. Maximum pressure due to surge (static + dynamic + transient rise), using Eq. (10.34)
with S = 70 percent of yield strength (in the case of steel pipe)
3. Collapsing pressure, if negative pressures occur due to surge conditions (Eq. 10.36).

10.8 IMPLICATIONS OF HYDRAULIC TRANSIENTS

IN PUMPING STATION DESIGN
Hydraulic transient, or surge, analysis is covered in detail in Chap. 12. Surge or hydraulic
transient effects must be considered in pump and piping systems because they can cause
or result in (Sanks et al., 1998):

rupture or deformation of pipe and pump casings,

pipe collapse,
vibration,
excessive pipe or joint displacements, or
pipe fitting and support deformation or even failure.

The pressures generated due to hydraulics, thus, must be considered in the pipe design,
as was discussed in Sec. 10.7, above.
10.8.1 Effect of Surge on Valve Selection
At its worst, surges in a piping can cause swing check valves to slam closed violently
when the water column in the pipeline reverses direction and flows backward through the
check valve at a significant velocity before the valve closes completely. Consequently, in
pump and piping systems in which significant surge problems are predicted to occur,
check valves or pump control valves are typical means to control the rate of closure of the
valve. Means of controlling this rate of closure include
Using a valve that closes quickly, before the flow in the piping can reverse and attain
a high reverse velocity.
Providing a dashpot or buffer on the valve to allow the valve clapper or disc to close
gently.
Closing the valve with an external hydraulic actuator so that the reverse flowing water
column is gradually brought to a halt. This is frequently done with ball or cone valves
used as pump control valves.
The pressure rating of the valve (both the check valve or pump control valve and the
adjacent isolation) should be selected with a pressure rating to accommodate the predicted surge pressures in the piping system.
10.8.2 Effect of Surge on Pipe Material Selection
Metal piping systems, such as steel and ductile iron, have much better resistance to
surge than do most plastic pipes (PVC, HDPE, ABS, and FRP). The weakness of plastic
pipes with respect to surge pressures is sometimes not adequately appreciated because the
wave velocity (a) and, hence, the resulting surge pressures are significantly lower than is
the case with metal piping systems. Since the surge pressures in plastic piping are lower
than those in metal piping systems, there is sometimes a mistaken belief that the entire
surge problem can then be neglected. However, plastic piping systems inherently offer
less resistance to hydraulic transients than do metal piping systems, even with the lower
pressures. This is particularly the case with solvent or adhesive welded plastic fittings.

HDPE has better resistance to surge pressures than other plastic piping systems. In
addition, the joints are fusion butt welded, not solvent welded, which results in a stronger
joint. However, HDPE is still not as resistant to surge effects as a properly designed steel
or ductile iron piping system.

REFERENCES
American Petroleum Institute, Centrifugal Pumps for Petroleum, Heavy Duty Chemical, and Gas
Industry Services, API Standard 610, 8th ed American Petroleum Institute, Washington, DC.
American Society of Mechanical Engineers (ASME), B31.1, Power Piping, ASME, New York, 1995.
American Society of Mechanical Engineers (ASME), B31.3, Process Piping, ASME, New York, 1996
American
Water Works Association, Steel PipeA Guide for Design and Installation, AWWA Mil,
3rd ed., American Water Works Association, Denver, CO 1989.
Cavi, D., "NPSHR Data and Tests Need Clarification," Power Engineering, 89:47-50, 1985.
Hydraulics Institute, American National Standard for Centrifugal Pumps for Nomenclature,
Definitions, Applications, and Operation, ANSI/HI 1.1-1.5-1994, Hydraulics Institute,
Parsippany, NJ, 1994.
American Petroleum Institute, Recommended Practices for Machinery Installation and Installation
Design, Practice 686, 1st ed. Washington, DC, 1996.
Sanks, R. L., et al., Pumping Station Design, 2nd ed., Butterworths, 1998.
Taylor., "Pump Bypasses Now More Important," Chemical Engineering, May 11, 1987.

APPENDIX 10. A
PUMP SYSTEM
HYDRAULIC DESIGN

Calculation of minor losses and NPSH4 in piping and modification of a pump curve
(U.S. units)
Part 1. Calculation of Minor Losses
Principal headless equations
For straight sections of pipe: H1 = ^jjj ( |

[Sec Eq. (10.11)]

where L = length in feet, D = pipe diameter (ft), Q - flow (ft3/s) anh C = Hazen-Williams friction
factor
For fittings: HL = ^K ^L

[Sec Eq. (10.23)]

where K = fitting friction coefficient, V = velocity in (ft/s), anh g = acceleration due to gravity
[(ft-s)/s]
Sum of K values for various pipe sizes:
Kn = 1.96
KS = 0.25
K10 = 2.31

Sum of C values for various pipe sizes:

Pipe lengths for 12-in pipe: L = 26 ft
Pipe lengths for 10-in pipe: L- 13 ft.
Determine the total headloss:
V2
V2
Vl
L = HL 12in + HL 10in+ K\2 ~^T + ^1O "^T + K* ^g

HL^
H

LM

_ 4.72(26) / 6 V*_ O o m 3
0-0""
(12/12)4.86 ^ 140j

_ 4.72(13) /6V-"- 0 0 1 S 9015y3

3<iC?1.M
(io/12)6 ^T40J

V2
Convert terms to Q2 terms
^g

Friction Factor
Item in
Fig. 10.12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

Description
Entrance
90 elbow
15 ft of straight pipe
30 elbow
7 ft of straight pipe
Butterfly valve
4 ft of straight pipe
12 in X 8 in reducer
6 in X 10 in increaser
3 ft of straight pipe
Pump control valve
3 ft of straight pipe
Butterfly valve
2 ft of straight pipe
90 elbow
5 ft of straight pipe
Tee connection

Pipe Size
(in)
12
12
12
12
12
12
12
8
10
10
10
10
10
10
10
10
10

K*

1.0
0.30
140
0.20
140
0.46

Typical K values. Different publications present other values.

Reasonable value for mortar-lined steel pipe. Value can range from 130 to 145.

Therefore,

C+

140
0.25
0.25
140
0.80
140
0.46
140
0.30
140
0.50

= 0.013139 g185 + 0.015936 <2185 + 0.04933 Q2 + 0.12058 Q2 + 0.031859 Q2

= 0.0291 g1-85 + 0.202 Q2
Part 2: Modification of Pump Curve
Using the above equation for HL, a "modified" pump curve can then be developed by converting pump curve head values to include minor piping losses:

H (ft)

GPM

CFS

Uncorrected

Corrected

O
1000
2000
3000
4000

O
2.228
4.456
6.684
8.912

200
180
160
130
90

200
178.87
151.52
120.0
72.29

The H values as corrected must then be plotted. The operating point of the pump is the
intersection of the corrected H-Q curve with the system curve.
Part 3: Calculation of NPSHA
Using the data developed above for calculating the minor losses in the piping, it is now
possible to calculate the NPSHA for the pump. Only the minor losses pertaining to the suction piping are considered: Items 1-8 in Fig. 10.12. For this suction piping, we have: ^T12
= 1.96, ^T8 = 0.25, sum of C values, Pipe length for 12-in pipe: L = 26 ft.
Determine the headloss in the suction piping

Compute NPSHA:
Condition

Flow
(fP/s)

h,
(ft)

Ji01n
(ft)

hvp
(ft)

High-static suction head

O
2
4
6
8
10

29
29
29
29
29
29

33.96
33.96
33.96
33.96
33.96
33.96

0.78
0.78
0.78
0.78
0.78
0.78

0.00
0.37
1.47
3.28
5.81
9.05

62.18
61.81
60.71
58.90
56.37
53.13

Low-static suction head

O
2
4
6
8
10

5
5
5
5
5
5

33.96
33.96
33.96
33.96
33.96
33.96

0.78
0.78
0.78
0.78
0.78
0.78

0.00
0.37
1.47
3.28
5.81
9.05

38.18
37.81
36.71
34.90
32.37
29.13

= 0.013139 2185 + 0.081189<22

For Fig. 10.12, assume that the following data apply
High-water level = Elevation 2241 ft
Low-water level = Elevation 2217 ft
Pump centerline elevation = 2212 ft
Therefore:
Maximum static head = 2241 - 2212 = 29 ft.
Minimum static head = 2217 - 2212 = 5 ft.
Per Eq. (10.31),
NPSHA = hatm + hs- hvp - hL
For this example, use
hatm=
hvp =
hs =
hs =

33.96 ft
0.78 ft at 6O0F
29 ft maximum
5 ft minimum

HL at Flow NPSHAatlow
(ft)
(ft)

CHAPTER 11
WATER DISTRIBUTION
SYSTEM DESIGN

Mark A. Ysusi
Montgomery Watson
Fresno, California

11.1

INTRODUCTION

The primary purpose of a water distribution system is to deliver water to the individual consumer in the required quantity and at sufficient pressure. Water distribution systems typically carry potable water to residences, institutions, and commercial and industrial establishments. Though a few municipalities have separate distribution systems, such as a highpressure system for fire fighting or a recycled wastewater system for nonpotable uses, most
municipal water distribution systems must be capable of providing water for potable uses
and for nonpotable uses such as fire suppression and irrigation of landscaping.
The proper function of a water distribution system is critical to providing sufficient
drinking water to consumers as well as providing sufficient water for fire protection.
Because these systems must function properly, the principals of their planning, design,
and construction need to be understood. This chapter focuses on the critical elements of
planning and design of a water distribution system. The information presented primarily
discusses typical municipal water distribution systems; however, the hydraulic and design
principles presented can be easily modified for the planning and design of other types of
pressure distribution systems such as fire protection and recycled wastewater.
11.1.1 Overview
Municipal water systems typically consist of one or more sources of supply, appropriate
treatment facilities, and a distribution system. Sources of supply include surface water,
such as rivers or lakes, groundwater, and in some instances, brackish or sea water. The
information contained in this chapter is limited to the planning and design of distribution
systems and does not address issues related to identifying and securing sources of supply
or designing and constructing appropriate water treatment facilities. Water distribution
systems usually consist of a network of interconnected pipes to transport water to the consumer, storage reservoirs to provide for fluctuations in demand, and pumping facilities.
11.1.2 Definitions
Many of the frequently used terms in water distribution system planning and design are
defined here.

Average day demand. The total annual quantity of water production for an agency
or municipality divided by 365.
Maximum day demand. The highest water demand of the year during any 24-h period.
Peak hour demand. The highest water demand of the year during any 1-h period.
Peaking factors. The increase above average annual demand, experienced during a
specified time period. Peaking factors are customarily used as multipliers of average day
demand to express maximum day and peak hour demands.
Distribution pipeline or main. A smaller diameter water distribution pipeline that
serves a relatively small area. Water services to individual consumers are normally placed
on distribution pipelines. Distribution system pipelines are normally between 150 and 400
mm (6-16 in.).
Transmission pipeline or main. A larger diameter pipeline, designed to transport larger quantities of water during peak demand periods. Water services for small individual
consumers are normally not placed on transmission pipelines. Transmission mains are
normally pipelines larger than 400 mm (16 in.).

77.2

DISTRIBUTIONSYSTEMPLANNING

The basic question to be answered by the water distribution system planner/designer is,
"How much water will my system be required to deliver and to where?" The answer to
this question will require the acquisition of basic information about the community
including historical water usage, population trends, planned growth, topography, and
existing system capabilities, to name just a few. This information can then be used to plan
for logical extension of the existing system and to determine improvements necessary to
provide sufficient water at appropriate pressure.
11.2.1 Water Demands
The first step in the design of a water distribution system is the determination of the quantity of water that will be required, with provision for the estimated requirements for the
future.
In terms of the total quantity, the water demand in a community is usually estimated
on the basis of per capita demand. According to a study published by the U.S. Geological
Survey, the average quantity of water withdrawn for public water supplies in 1990 was
estimated to about 397 L per day per capita (Lpdc) or 105 gal per day per capita (gpdc).
The withdrawals by state are summarized in Table 11.1. The reported water usage shown
in Table 11.1 illustrates a wide variation. Per capita water use varies from a low use in
Pennsylvania of just over 60 gpcd to over 200 gpcd in Nevada. These variations depend
on geographic location, climate, size of the community, extent of industrialization, and
other influencing factors unique to most communities. Because of these variations, the
only reliable way to estimate future water demand is to study each community separately, determining exiting water use characteristics and extrapolating future water demand
using population trends.
In terms of how the total water use is distributed within a community throughout the
day, perhaps the best indicator is land use. In a metered community, the best way to determine water demand by land use is to examine actual water usage for the various types of
land uses. The goal of examining actual water usage is to develop water "duties" for the

TABLE 11.1 Estimates Use of Water in the United States in 1990

State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Distrit of Columbia
Florida
Geogia
Hawaii
Idho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississipi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota

L/Capita/per day
379
299
568
401
556
549
265
295
678
420
435
450
704
341
288
250
326
265
469
220
397
250
291
560
466
326
488
435
806
269
284
511
450
254
326
189
322
420
235
254
288
307

gal/Capita/day
100
79
150
106
147
145
70
78
179
111
115
119
186
90
76
66
86
70
124
58
105
66
77
148
123
86
129
115
213
71
75
135
119
67
86
50
85
111
62
67
76
81

TABLE 11.1 (Continued)

State
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Puerto Rico
Virgin Islands
United Stated total
Source: Solley et al (1993)

L/Capita/per day
322
541
825
303
284
522
280
197
617
182
87
397

gal/Capita/day
85
143
218
80
75
138
74
52
163
48
23
105

various types of land uses that can be used for future planning. Water duties are normally
developed for the following land uses:
Single, family residential (some communities have low-medium-and, high-density
zones)
Multifamily residential
Commercial (normally divided into office and retail categories)
Industrial (normally divided into light and heavy categories and separate categories for
very high users
Public (normally divided into park, or open space, and schools)
Water duties are normally expressed in gallons per acre per day. Table 11.2 shows typical
water duties in the western United States. It should be noted that the definitions of land
use terms like "low-density residential," "medium-density residential," and so on, will
vary by community and should be examined carefully.
Another method of distributing water demand is to examine the water usage for individual users. This is particularly the case when an individual customer constitutes a significant portion of the total system demand. Table 11.3 presents water use for many different establishments. Although the rates vary widely, they are useful in estimating total
water use for individual users when no other data are available.
11.2.2 Planning and Design Criteria
To effectively plan and design a water distribution system, criteria must be developed and
adopted against which the adequacy of the existing and planned system can be compared.
Typical criteria elements include the following:
Supply
Storage

TABLE 11.2 Typical Water Duties*

Land Use

Low

Water Duty, (gal/day/acre)

High
Average

Low-density residential
400
3,300
1,670
Medium-density residential
900
3,800
2,610
High-density residential
2,300
12,000
4,160
Single family residential
1,300
2,900
2,300
Multifamily residential
2,600
6,600
4,160
Office commercial
1,100
5,100
2,030
Retail commercial
1,100
5,100
2,040
Light industrial
200
4,700
1,620
Heavy industrial
200
4,800
2,270
Parks
400
3,100
2,020
Schools
400
2,500
1,700
Source: Adapted from Montgomery Watson study of data of 28 western U.S. cities.
Note: gal X 3.7854 = L.
TABLE 11.3 Typical Rates of Water Use for Various Establishments

User
Airport, per passenger
Assembly hall, per seat
Bowling alley, per alley
Camp
Pioneer type
Children's, central toilet and bath
Day, no meals
Luxury, private bath
Labor
Trailer with private toilet and bath,
per unit (2 1/2 persons)
Country clubs
Resident type
Transient type serving meals
Dwelling unit, residential
Apartment house on individual well
Apartment house on public water
supply, unmetered
Boardinghouse
Hotel

Range of Flow
L/person
gal/person
or unit/day
or unit/day
10-20
6-10
60-100

3-5
2-3
16-26

80-120
160-200
40-70
300^00
140-200
500-600

21-32
42-53
11-18
79-106
37-53
132-159

300-600
60-100

79-159
16-26

300-400
300-500

79-106
79-132

150-220
200-400

40-58
53-106

TABLE 11.3. (Continued)

User
Lodging house and tourist home
Motel
Private dwelling on individual well
or metered supply
Private dwelling on public water supply,
unmetered
Factory, sanitary wastes, per shift
Fairground (based on daily attendance)
Institution
Average type
Hospital
Office
Picnic park, with flush toilets
Restaurant (including toilet)
Average
Kitchen wastes only
Short order
Short order, paper service
Bar and cocktail lounge
Average type, per sear
Average type 24 h, per seat
Tavern, per seat
Service area, per counter seat (toll road)
Service area, per table seat (toll road)
School
Day, with cafeteria or lunchroom
Day, with cafeteria and showers
Boarding
Self-service laundry, per machine
Store
First 7.5 m ( 25 ft) of frontage
Each additional 7.5 m of frontage
Swimming pool and beach, toilet and shower
Theater
Indoor, per seat, two showings per day
Outdoor, including food stand, per car
(3 1/3 persons)
Source: Adapted from Metcalf and Eddy (1979).

Range of Flow
L/person
gal/person
or unit/day
or unit/day
120-200
400-600
200-600

32-53
106-159
53-159

400-800

106-211

40-100
2-6

11-26
1-2

400-600
700-1200
40-60
20^0

106-159
185-317
11-16
5-11

25^0
10-20
10-20
4-8
8-12
120-180
160-220
60-100
1000-1600
600-800

7-11
3-5
3-5
1-2
2-3
32^8
42-58
16-26
264-423
159-211

40-60
60-80
200-400
1000-3000

11-16
16-21
53-106
264-793

1600-2000
1400-1600
40-60

423-528
370-423
11-16

10-20

3-5

10-20

3-5

 Fire demands
Distribution system analysis
Service pressures
11.2.2.1 Supply. In determining the adequacy of water supply facilities, the source of
supply must be large enough to meet various water demand conditions, and be able to
meet at least a portion of normal demand during emergencies such as power outages and
disasters. At a minimum, the source of supply should be capable of meeting the maximum
day system demand. It is not advisable to rely on storage to make up any shortfall in supply at maximum day demand. The fact that maximum day demand may occur several days
consecutively must be considered by the system planner/designer. It is common for communities to provide a source of supply that meets the maximum day demand, with the
additional supply to meet peak hour demand coming from storage. Some communities
find it more economical to develop a source of supply that not only meets maximum day
but also peak hour demand.
It is also good practice to consider standby capability in the source of supply. If the
system has been designed where the entire capacity of the supply is required to meet the
maximum demand, any portion of the supply that is placed out of service due to malfunction or maintenance will result in a deficient supply. For example, a community that
relies primarily on groundwater for its supply should, at a minimum, be able to meet its
maximum day demand with at least one of its largest wells out of service.
11.2.2.2 Storage. The principal function of storage is to provide reserve supply for (1)
operational equalization, (2) fire suppression reserves, and (3) emergency needs.
Operational storage is directly related to the amount of water necessary to meet peak
demands. The intent of operational storage is to make up the difference between the consumers' peak demands and the system's available supply. It is the amount of desirable
stored water to regulate fluctuations in demand so that extreme variations will not be
imposed on the source of supply. With operational storage, system pressures are typically
improved and stabilized. The volume of operational storage required is a function of the
diurnal demand fluctuation in a community and is commonly estimated at 25 percent of
the total maximum day demand.
Fire storage is typically the amount of stored water required to provide a specified fire
flow for a specified duration. Both the specific fire flow and the specific time duration varies
significantly by community. These values are normally established through the local fire
marshall and are typically based on guidelines established by the Insurance Service Office,
a nonprofit association of insurers that evaluate relative insurance risks in communities.
Emergency storage is the volume of water recommended to meet demand during emergency situations such as source of supply failures, major transmission main failures, pump
failures, electrical power outages, or natural disasters. The amount of emergency storage
included with a particular water system is an owner option, typically based on an assessment
of risk and the desired degree of system dependability. In considering emergency storage, it
is not uncommon to evaluate providing significantly reduced supplies during emergencies.
For example, it is not illogical to assume minimal demand during a natural disaster.
11.2.2.3 Fire demands. The rate of flow to be provided for fire flow is typically dependent on the land use and varies by community. The establishment of fire flow criteria
should always be coordinated with the local fire marshall. Typical fire flow requirements
are shown in Table 11.4.
11.2.2.4 Distribution system analysis. In evaluating an existing system or planning a
proposed system, it is important to establish the criteria of operational scenarios against

TABLE 11.4 Typical Fire Flow Requirements

Land Use
Single-family residential
Multi family residential
Commercial
Industrial
Central business district

Fire Flow Requirements, gal/m*

500-2,000
1,500-3,000
2,500-5,000
3,500-10,000
2,500-15,000

Note: gal X 3.7854 = L.

which the system will be compared. Any system can be shown to be inadequate if the
established criteria is stringent enough. Most systems are quite capable of meeting
the average day conditions. It is only when the system is stressed that deficiencies begin
to surface. The degree to which the system will be realistically stressed is the crux of
establishing distribution system analysis criteria. In evaluating a system it is common to
see how the system performs under the following scenarios:
Peak hour demand
Maximum day demand plus fire flow
Evaluating the system at peak hour demand gives the designer a look at system-wide
performance. Placing fire flows at different locations in the system during a "background" demand equivalent to maximum day demand will highlight isolated system
deficiencies. Obviously, it is possible for fires to occur during peak hour demand, but
since this simultaneous occurrence is more unlikely than for a fire to occur sometime
during the maximum day demand, this is not usually considered to be an appropriate
criteria for design of the system.
11.2.2.5 Service pressures. There are differences in the pressures customarily maintained in the distribution systems in various communities. It is necessary that the water
pressure in a consumer's residence or place of business be neither too low nor to high.
Low pressures, below 30 psi, cause annoying flow reductions when more than one waterusing device is in service. High pressures may cause faucets to leak, valve seats to wear
out quickly, or hot water heater pressure relief valves to discharge. Additionally, abnormally high pressures can result in water being wasted in system leaks. The Uniform
Plumbing Code requires water pressures not exceed 80 psi at service connections, unless
the service is provided with a pressure reducing device. Another pressure criteria, related
to fire flows, commonly requires a minimum of 20 psi at the connecting fire hydrant used
for fighting the fire. Table 11.5 presents typical service pressure criteria.
11.2.3 Peaking Coefficients
Water consumption changes with the seasons, the days of the week, and the hours of the day.
Fluctuations are greater in (1) small than in large communities, and (2) during short rather
than during long periods of time. Variations in water consumption are usually expressed as
ratios to the average day demand. These ratios are commonly called peaking coefficients.
Peaking coefficients should be developed from actual consumption data for an individual
community, but to assist the reader, Table 11.6 presents typical peaking coefficients.

TABLE 11.5 Typical Service Pressure Criteria

Condition

Service Pressure Criteria (psi)

Maximum pressure
Minimum pressure during maximum day
Minimum pressure during peak hour
Minimum pressure during fires

65-75
30-40
25-35
20

Note: psi X 6.895 = kPa.

TABLE 11.6 Typical Peaking Coefficients
Ratio of Rates

U.S. Range

Common Range

Maximum day: average day

Peak hour: average day

1.5-3.5:1
2.0-7.0:1

1.8-2.8:1
2.5^.0:1

11.2.4 Computer Models and System Modeling

Modeling water distribution systems with computers is a proven, effective, and reliable
technology for simulating and analyzing system behavior under a wide range of hydraulic
conditions. The network model is represented by a collection of pipe lengths interconnected in a specified topological configuration by node points, where water can enter and
exit the system. Computer models utilize laws of conservation of mass and energy to
determine pressure and flow distribution throughout the network. Conservation of mass
dictates that for each node the algebraic sum of flows must equal zero. Conservation of
energy requires that along each closed loop the accumulated energy loss must be zero.
These laws can be expressed as nonlinear algebraic equations in terms of either pressures
(node formulation) or volumetric flow rates (loop and pipe formulation). The nonlinearity reflects the relationship between pipe flow rate and the pressure drop across its length.
Due to the presence of nonlinearity in these equations, numerical solution methods are
iterative. Initial estimated values of pressure or flow are repeatedly adjusted until the difference between two successive iterates is within an acceptable tolerance. Several numerical iterative solution techniques have been suggested, from which the newtonian method
is the most widely used. See chapter 9 for more details on modeling.
11.2.4.1 History of computer models. Prior to computerization, tedious, and time-consuming manual calculations were required to solve networks for pressure and flow distribution. These calculations were carried out using the Hardy-Cross numerical method of
analysis for determinate networks. Only simple pipeline systems consisting of a few loops
were modeled and under limited conditions because of the laborious effort required to obtain
a solution. The first advent of computers in network modeling was with electric analogues,
followed by large mainframe digital computers to smaller microcomputers. The computational power of a laptop computer today is vastly superior to the original computing
machines that would fill several floors in an office building and at a fraction of the cost.
Many of the early computer models did not have interactive on-screen graphics, thus
limiting the ability of engineers to develop and interpret model runs. The user interface
was very rudimentary and often an afterthought. Input was either by punch cards or formatted American Standard Code for Information Interchange (ASCII) files created with a

text editor. Errors were commonplace and just getting a data file that would run could
involve days if not weeks of effort, depending on the size and complexity of the network
being modeled. Model output was usually a voluminous tabular listing of key network
results. Interpretation of the results was time consuming and typically involved hand plotting of pressure contours on system maps.
Because of the widespread use of microcomputers during the last two decades, network modeling has taken on new dimensions. Engineers today rely on computer models
to solve a variety of hydraulic problems. The use of interactive on-screen graphics to enter
and edit network data and to color code and display network maps, attributes, and analysis results has become commonplace in the water industry. This makes it much easier for
the engineer to construct, calibrate, and manipulate the model and visualize what is happening in the network under various situations such as non-compliance with system performance criteria. The engineer is now able to spend more time thinking and evaluating
system improvements and less time on flipping through voluminous pages of computer
printouts, thus leading to improved operation and design recommendations. The new generation of computer models have greatly simplified the formidable task of collecting and
organizing network data and comprehending massive results.
11.2.4.2 Software packages. Many of the software packages available offer additional
capabilities beyond standard hydraulic modeling such as water quality assessment (both
conservative and reactive species), multiquality source blending, travel time determination,
energy and power cost calculation, leakage and pressure management, fire flow modeling,
surge (transient) analysis, system head curve generation, automated network calibration,
real-time simulation, and network optimization. Some sophisticated models can even
accommodate the full library of hydraulic network components including pressure regulating valves, pressure sustaining valves, pressure breaker valves, flow control valves, float
valves, throttle control valves, fixed- and variable speed pumps, turbines, cylindrical and
variable cross-sectional area tanks, variable head reservoirs, and multiple inlet/outlet tanks
and reservoirs. Through their predictive capabilities, computer network models provide a
powerful tool for making informed decisions to support many organization
programs and policies. Modeling is important for gaining proper understanding of system
dynamic behavior, operator training, optimizing the use of existing facilities, reducing operating costs, determining future facility requirements, and addressing water quality distribution issues.
There is an abundance of network modeling software in the marketplace today. Some
are free and others can be purchased at a nominal cost. Costs can vary significantly
between models depending on the range of the features and capabilities provided. The
four major sources of computer models include consulting firms, commercial software
companies, universities, and government agencies. Many of the programs available from
these sources have been on the market for several years with established track records.
Most of the recent computer models however, provide very sophisticated and intuitive
graphical user interfaces and results presentation environment, as well as direct linkages
with information management systems such as relational databases and geographical
information systems. Table 11.7 lists the names, addresses, and phone numbers of network
modeling software vendors, along with their primary modeling product.
11.2.4.3 Development of a system model. As indicated above, the computerized tools
available to the engineer today are impressive and powerful. Once an appropriate software
is selected, data must be then input to the software in order to develop a computer model
of the water system under study. Input data includes the physical attributes of the system
including pipe sizes and lengths, topography, reservoir and pump characteristics, as well
as the anticipated nodal demands.

TABLE 11.7 Distribution System Modeling Software

Software

Vendor

Address

City

State

Country

Zip Code

Telephone

Fax

AQUA
AVWater
BOSS EMS
CYBERNET
EPANET
FAAST
FLOW NETWORK
ANALYSIS
H2ONET
HYDRONET
InWater
KYPIPE

Computer Modeling, Inc.

CEDRA
BOSS International
Haestad Methods, Inc.
USEPA
Faast Software
Kelix Software Systems

2121 Front Street

65 West Broad St.
6612 Mineral Point Road
37 Brookside Road
26 W. Martin Luther King Drive
3062 East Avenue
11814 Coursey Blvd. Suite 220

Cuyahoga Falls
Rochester
Madison
Waterbury
Cincinnati
Livermore
Baton Rouge

OH
NY
WI
CT
OH
CA
LA

USA
USA
USA
USA
USA
USA
USA

44221
14614
53705-4200
06708
45268
94550-4738
70816

(330) 929-7886
(716) 232-6998
1-800-488-4775
1-800-727-6555
(513)569-7603
(510)455-8086
(504) 769-6785

(330) 929-2756
(716) 262-2042
(608) 258-9943
(203) 597-1488
(513) 569-7185
(510)455-8087

300 N. Lake Ave. Suite 1200

P.O. Box 8128
One Madison Industrial Park
Civil Engineering Software Center,
University of Kentucky
SAFEGE Consulting Engrs. P.O. Box 727 Pare de !Tie,
15-27 rue du Port
Engineered Software, Inc. 4531 IntelcoLoop
Watercom Pty Ltd.
105 Queen Victoria St.
CEDEGER
1417 rue Michelin
RJN Computer Services, Inc. 200 W. Front Street
Charles Howard & Assoc. Ltd. 852 Fort St. 2nd Floor
Stoner Associates, Inc.
P.O. Box 86
TDH Engineering
607 Ninth St.
Utah State University
Utah State University, Dept of Civil
& Environmental Engineering
Lewis Publishers - 1990
2000 Corporate Blvd. NW

Pasadena
Truckee
Huntsville
Lexington

CA
CA
AL
KY

USA
USA
USA
USA

91101
92162
35898
40506

(626) 568-6868
(530) 582-1525
1-800-345-4856
(606) 257-3436

(626) 568-6619
(530) 582-8579
(205) 730-6109
(606) 257-8005

France

92007

01133146147181

01133147247202

98503
H7L4S2
60187
V8W 2H7
17013
20707
84322-4100

(360) 412-0702
612-9587-5384
(514)629-8888
(630) 682-4700
(250) 385-0206
(717) 243-1900
(301)490-4515
(801) 797-2943

(360) 412-0672
612-9587-5384
(514) 382-3077
(630) 682-4754
(250) 385-7737
(717) 243-5564
(301)490-4515
(801)750-1185

Syntex Systems Corporation

Municipal Hydraulics Ltd.
The Pitometer Associates
WRc
Expertware Dev. Corp.

PICCOLO
PIPE-FLO
PIPES
RINCAD
RJN CASS WORKS
SDP
Stoner Workstation
TDHNET
USU-NETWK
WADISO
Water Distribution
Systems: Simulation
and Sizing
Water Works
WATERAVGRAPH
WaterMax
WATNET
WATSYS

MW Soft, Inc.
Tahoe Designs Software
Intergraph
University of Kentucky

800-1188 West Georgia Street

2474 Pylades Drive RR#3
20 N. Wacker Drive, Suite 1530
8 Neshaminy Interplex, Suite 219
27 Linden Avenue

NANTERRE Cedex
Lacey
Bexley, NSW 2217
Laval
Wheaton
Victoria
Carlisle
Laurel
Logan

WA

Quebec
IL
BC
PA
MD
UT

USA
Australia
Canada
USA
Canada
USA
USA
USA

Boca Raton

FL

USA

33431-9868

1-800-272-7737

1-800-374-3401

Vancouver
Ladysmith
Chicago
Trevose
Victoria

B.C.
B.C.
IL
PA
B.C.

Canada
Canada
USA
USA
Canada

V6E 4A2
VOR2EO
60606
19053
V8V 4C9

(604) 688-8271
(250) 722-3810
(312) 236-5655
(215) 244-9972
(250) 384-5955

(604) 688-1286
(250) 722-3088
(312) 580-2691
(215) 244-9977
(250) 383-1692

Development of the nodal demands normally involves distributing the average day
flow throughout the system in proportion to land use. This is commonly accomplished
by determining a demand area for each node, measuring the area of each different land
use within the demand area, multiplying the area of each land use within the demand area
by its respective average day water duty (converted to gal/min or L/sec), and summing
the water duties for each land use within the demand area and applying the sum at the
node. In the past this effort required extensive mapping and determining the land use
areas by planimeter or hand measurement. Today, with the advent of graphical information system software (GIS), development of nodal demands is normally an activity
involving computer based mapping. The elements of the system, the demand areas, and
the land uses are all mapped in separate layers in the GIS software. The GIS software
capability of "polygon processing" intersects the different layers and automatically computes the land use sums with the various demand areas. When the water duties are multiplied by their respective land use, the result is the average day system demand, proportioned to each node by land use. The water system computer model is then used to apply
global peaking factors as described above.

11.3 PIPELINE PRELIMINARY DESIGN

The purpose of performing the water system planning tasks as outlined above is to develop a master plan for correcting system deficiencies and providing for future growth.
Normally the system improvements are prioritized and a schedule or capital improvement
program is developed based upon available (now or future) funding. As projects leave the
advanced planning stage, they begin the process of preliminary design. During preliminary design, the considerations of pipeline routing (alignment), subsurface conflicts, and
rights-of-way are considered.
11.3.1 Alignment
In deciding upon an appropriate alignment for a pipeline, important considerations
include right-of-way (discussed further below), constructability, access for future
maintenance, and separation from other utilities. Many communities adopt standardized locations for utility pipelines (such as water lines will generally be located 15 ft
north and east of the street centerline). Such standards compliment alignment considerations.
11.3.2 Subsurface Conflicts
A critical element of developing a proposed pipeline alignment is an evaluation of subsurface conflicts. To properly evaluate subsurface conflicts it will be necessary for the
designer to identify the type, size, and accurate location of all other underground utilities
along the proposed pipeline alignment. This information must be considered in the design
and accurately placed on the project plans so that the contractor (or whoever is constructing the line) is completely aware of potential conflicts.
It is good practice to thoroughly investigate potential utility conflicts. For example, it
is not enough to simply determine that the proposed pipeline route will cross an electrical
conduit. The exact location and dimensions of electrical conduits also need to be deter-

mined and the proposed water pipeline designed accordingly. What is shown on a utility
company plat as a single line representing an electrical conduit may turn out to be a major
electrical line with several conduits encased in concrete having a cross section 2 ft wide
and 4 ft deep! Or, what is shown as a buried 3/4-inch telephone line may turn out to be a
fiber-optic telecommunications cable that, if severed during construction, will result in
exorbitant fines being levied by the communications utility.
Another water pipeline alignment consideration is the lateral separation of the line
from adjacent sanitary sewer lines. Many state and local health officials require a minimum of 10 ft of separation (out-to-out) between potable water and sanitary sewer lines.
11.3.3 Rights-of-Way
The final location of a pipeline can only be selected and construction begun once appropriate rights-of-way are acquired. Adequate right-of-way both for construction and for
future access are necessary for a successful installation. Water lines are commonly located in streets and roadways dedicated to the public use. On occasion, it is necessary to
obtain rights-of-way for transmission-type pipelines across private lands. If this is the
case, it is very important to properly evaluate the width of temporary easement that will
be required during construction and the width of permanent easement that will be required
for future access. If a pipeline is to be installed across private property, it is also very
important for the entity that will own and maintain the pipeline to gain agreements that no
permanent structures will be constructed within the permanent easements and to implement a program of monitoring construction on the private property to ensure that access
to the pipeline is maintained. Otherwise, as the property changes hands in the future, the
pipeline stands a good chance of becoming inaccessible.

77.4 PlPlNGMATERlALS
The types of pipe and fittings commonly used for pressurized water distribution systems are
discussed in this section. The types of pipeline materials are first presented and then factors
effecting the types of materials selected by the designer are presented in Section 11.4.7. The
emphasis throughout this section is on pipe 100 mm (4 in.) in diameter and larger.
References to a standard or to a specification are given here in abbreviated form - code
letters and numbers only such as American National Standards Institute (ANSI) B36.10.
Double designations such as ANSI/AWWA C115/A21.15 indicate that American
Waterworks Association (AWWA) C115 is the same as ANSI A21.15. Most standards are
revised periodically, so it is advisable for the designer to obtain the latest edition.
11.4.1 Ductile Iron Pipe (DIP)
Available in sizes 100-1350 mm (4-54 in.), DIP is widely used throughout the United
States in water distribution systems. On the East Coast and in the Midwest, DIP is
commonly used for both smaller distribution mains and larger transmission mains. On
the West Coast, DIP is generally used for distribution pipelines 40 mm (16 in) and
smaller, with alternative pipeline materials often selected for larger pipelines due to
cost. Detailed descriptions of DIP, fittings, joints, installation, thrust restraint, and
other factors relating to design as well as several important ANSI/AWWA specifications are contained in the Ductile Iron Pipe Research Association (DIPRA) handbook
(DIPRA, 1984).

11.4.1.1 Materials. DIP is a cast-iron product. Cast-iron pipe is manufactured of an

iron alloy centrifugally cast in sand or metal molds. Prior to the early 1970s, most castiron pipe and fittings were gray iron, a brittle material that is weak in tension. But now
all cast-iron pipe, except soil pipe (which is used for nonpressure plumbing applications) is made of ductile iron. Ductile iron is produced by the addition of magnesium to
molten low-sulfur base iron, causing the free graphite to form into spheroids and making it about as strong as steel. Regular DIP (AWWA C151) has a Brinell hardness
(BNH) of about 165.
Tolerances, strength, coatings and linings, and resistance to burial loads are given in
ANSI/AWWA C151/A21.51.
11.4.1.2 Available sizes and thicknesses. DIP is available in sizes from 100 to 1350 mm
(4-54 in). The standard length is 5.5 m (18 ft) in pressure ratings from 1380 to 2400 kPa
(200-350 lb/in2).
Thickness is normally specified by class, which varies from Class 50 to Class 56 (see
DIPRA, 1984 or ANSI/AWWA C150/A21.50). Thicker pipe can be obtained by special
order.
11.4.1.3 Joints. For DIP, rubber gasket push-on and mechanical are the most commonly
used for buried service. These joints allow for some pipe deflection (about 2-5 depending on pipe size) without sacrificing water tightness. Neither of these joints are capable of
resisting thrust across the joint and require thrust blocks or some other sort of thrust
restraint at bends and other changes in the flow direction.
Flanged joints (AWWA C115 or ANSI B 16.1) are sometimes used at fitting and valve
connections. Grooved end joints (AWWA C606) are normally used for exposed service and
are seldom used for buried service. Flanged joints are rigid and grooved end joints are flexible. Both are restrained joints and do not typically require thrust restraint. Other types of
restrained joints such as restrained mechanical joints are also available for buried service.
Various types of ductile iron pipe joints are shown in Fig. 11.1.
11.4.1.4 Gaskets. Gaskets for ductile iron push-on and mechanical joints described in
AWWA Cl 11 are vulcanized natural or vulcanized synthetic rubber. Natural rubber is suitable for water pipelines but deteriorates when exposed to raw or recycled wastewater.
Gaskets for DIP flanges should be rubber, 3.2 mm (1/8 in) thick.
Gaskets for grooved end joints are available in ethylene propylene diene monomer
(EPDM), nitrile (Buna-N), halogenated butyl rubber, Neoprene, silicone, and

FIGURE 11.1 Couplings and joints for ductile iron pipe: (a) flexible coupling;
(b) mechanical joint; (c) push-on joint; (d) ball joint. Adapted from Sanks et al,. (1989).

fluorelastomers. EPDM is commonly used in water service and Buna-N in recycled

wastewater.
11.4.1.5 Fittings. Some standard ductile or gray iron fittings are shown in Fig. 11.2. A
list of standard and special fittings is also given in Table 11.8. Ductile iron fittings are normally only available in standard configurations as described in AWWA CIlO. Greater cost
and longer delivery times can be expected for special fittings. Fittings are designated by
the size of the openings, followed (where necessary) by the deflection angle. A 90 elbow
for 250 mm (10 in) pipe would be called a 250 mm (10 in) 90 bend (or elbow). Reducers,
reducing tees, or reducing crosses are identified by giving the pipe diameter of the largest
opening first, followed by the sizes of other openings in sequence. Thus, a reducing tee
on a 300 mm (12 in) line for a 150 mm (6 in) fire hydrant run might be designated as a
300 mm X 150 mm X 300 mm (12 in X 6 in X 12 in ) tee.
Standard ductile iron fittings are commonly available in flanged, mechanical joint, and
push-on ends. It is considered good practice to include sufficient detail in construction
plans and specifications to illustrate the type of joints that are expected at connections.
Failure to detail a restrained joint when one is required by the design, could result in an
unstable installation.
11.4.1.6 Linings. Considering its low cost, long life, and sustained smoothness, cementmortar lining for DIP in water distribution systems is the most useful and common.
Standard thicknesses for shop linings specified in AWWA C104 are given in Table 11.9.
Pipe can also be lined in place with the thicknesses given in Table 11.10. Because the

True wye

Wye branch

Flange and flare

Flange and flare 90 bend

90 bend
reducing

Cross

45 bend

22J bend

90 bend

Tee

Base tee

I1ibend

Base bend

Mechanical joint and plain end

wall pipe

Reducer

Flange and flange

wall pipe

FIGURE 11.2 Ductile iron flanged fittings. Adopted from Sanks et al,. (1989).

TABLE 11.8 Ductile Iron and Gray Cast-Iron Fittings,

Flanged, Mechanical Joint, or Bell and Spigot*
Standard Fittings

Special Fittings

Bends (90, 45, 22.5, 11.25)

Base bends
Caps
Crosses
Blind flanges
Offsets
Plugs
Reducers
Eccentric reducers
Tees
Base tees
Side outlet tees
Wyes

Reducing bends (90)

Flared bends (90, 45)
Flange and flares
Reducing tees
Side outlet tees
Wall pipes
True wyes
Wye branches

*Size from 100 to 350 mm (4-54 in).

standard, shop-applied mortar linings are relatively thin, some designers prefer to specify
shop linings in double thickness. The designer should also be careful in specifying mortar lining thickness to match the pipe inside diameter (ID) with system valve IDs, particularly with short-body butterfly valves where the valve vane protrudes into the pipe. If the
pipe ID is too small, the valve cannot be fully opened.

TABLE 11.9 Thickness of Shop-Applied Cement-Mortal Linings

Lining Thickness
Nominal Pipe Diameter
mm
in

Ductile Iron Pipe*

mm
in

Steel Pipe*
mm
in

100-250
4-10
1.6
1/16
6.4
300
1-2
1.6
1/16
7.9
350-550
14-22
2.4
3/32
7.9
600
24
2.4
3/32
9.5
750-900
30-36
3.2
1/8
9.5
1050-1350
42-54
3.2
1/8
12.7
*Single thickness per AWWA C104. Linings of double thickness are also readliy available.
tPerAWWAClOS.

1/4
5/16
5/16
3/8
3/8
1/2

TABLE 11.10 Thickness of Cement-Mortal Linings of Pipe in Place per AWWA C602
Nominal Pipe Diameter
mm
in
100-250
300
350-550
600-900
1050-1350
1500
1650-2250
>2250

4-10
12
14-22
24-36
42-54
60
66-90
>90

DIP or Gray Cast Iron

(new or Old Pipe)
mm
in
3.2
4.8
4.8
4.8
6.4

1/8
3/16
3/16
3/16
1/4

Steel Pipe
Old Pipe
New Pipe
mm
in
mm
in
6.4
6.4
7.9
9.5
9.5
9.5
12.7
12.7

1/4
1/4
5/16
3/8
3/8
3/8
1/2
1/2

4.8
4.8
6.4
6.4
9.5
9.5
11.1
12.7

3/16
3/16
1/4
1/4
3/8
3/8
7/16
1/2

Although cement-mortar lining is normally very durable, it can be slowly attacked by

very soft waters with low total dissolved solids content (less than 40 mg/L), by high-sulfate waters, or by waters undersaturated in calcium carbonate. For such uses, the designer should carefully investigate the probable durability of cement mortar and consider the
use of other linings. Other linings and uses are shown in Table 11.11. In general, the cost
of cement mortar is about 20 percent of that of other linings, so other linings are not justified except where cement mortar would not provide satisfactory service.
11.4.1.7 Coatings. Although DIP is relatively resistant to corrosion, some soils (and
peat, slag, cinders, muck, mine waste, or stray electric current) may attack the pipe. In
these applications, ductile iron manufacturers recommend that the pipe be encased in
loose-fitting, flexible polyethylene tubes 0.2 mm (0.008 in) thick (see ANSI/AWWA
C105/A21.5). These are commonly known as "baggies." An asphaltic coating approximately 0.25 mm (0.001 in) thick is a common coating for ductile iron pipe in noncorrosive soils. In some especially corrosive applications, a coating such as adhesive, hotapplied extruded polyethylene wrap may be required.

TABLE 11.11 Linings for Ductile Iron and Steel Pipe

Lining Material

Reference Standard

Recommended Service

Cement mortar

AWWA C104, C205

Glass
Epoxy
Fusion-bonded epoxy
Coal-tar epoxy
Coal-tar enamel
Polyurethane
Polyethylene

None
AWWA C210
AWWA C213
AWWA C210
AWWA C203
None
ASTM D 1248

Potable water, raw water and sewage, activated

and secondary sludge
Primary sludge, very aggressive fluids
Raw and potable water
Potable water, raw water and sewage
Not recommended for potable water
Potable water
Raw sewage, water
Raw sewage

In corrosive soils, the following coatings may be appropriate for protecting the pipe:

Adhesive, extruded polyethylene wrap

Plastic wrapping (AWWA C105)
Hot-applied coal-tar enamel (AWWA C203)
Hot-applied coal-tar tape (AWWA C203)
Hot-applied extruded polyethylene [ASTM D 1248 (material only)]

Coal-tar epoxy (MIL-P-23236)

Cold-applied tape (AWWA C209)
Fusion-bonded epoxy (AWWA C213)
Each of the above coatings is discussed in detail in the referenced specifications. Each
coating system has certain limited applications and should be used in accordance with the
NACE standards or as recommended by a competent corrosion engineer.
11.4.2 Polyvinyl Chloride (PVC) Pipe
In the United States, where it is used in both water and wastewater service, polyvinyl chloride (PVC) is the most commonly used plastic pipe for municipal water distribution systems.
Because of its resistance to corrosion, its light weight and high strength to weight ratio, ease
of installation, and its smoother interior wall surface, PVC has enjoyed rapid acceptance for
use in municipal water distribution systems since the 1960s. There are several other types of
plastic pipe, but PVC is the most common plastic pipe selected for use in municipal systems
and will be the only type of plastic pipe addressed in this section. There are also several different PVC pipe specifications. Only those having AWWA approval will be addressed in this
section, since only those should be used for municipal water distribution systems. Highdensity polyethylene pipe (HDPE) is discussed in Sec. 11.4.5.
11.4.2.1 Materials. PVC is a polymer extruded under heat and pressure into a thermoplastic that is nearly inert when exposed to most acids, alkalis, fuels, and corrosives, but
it is attacked by ketones (and other solvents) sometimes found in industrial waste waters.
Basic properties of PVC compounds are detailed in ASTM D 1784. ASTM D 3915 covers performance characteristics of concern, or cell classification, for PVC compounds
to be used in pressure pipe applications. Generally, PVC should not be exposed to direct
sunlight for long periods. The impact strength of PVC will decrease if exposed to sunlight
and should not be used in above-ground service.
In North America, PVC pipe is rated for pressure capacity at 230C (73.40F). The
pressure capacity of PVC pipe is significantly related to its operating temperature. As
the temperature falls below 230C (73.40F), such as in normal buried service, the pressure capacity of PVC pipe increases to a level higher than its pressure rating or class.
In practice, this increase is treated as an unstated addition to the working safety factor
but is not otherwise considered in the design process. On the other hand, as the operating temperature rises above 230C (73.40F), the pressure capacity of PVC pipe decreases to a level below its pressure rating or class. Thermal derating factors, or multipliers,
are typically used if the PVC pipe will be used for higher temperature services.
Recommended thermal derating factors are shown in Table 11.12. The pressure rating
or class for PVC pipe at service temperature of 270C (8O0F) would need to multiplied

TABLE 11.12 Thermal Derating Factors for PVC Pressure Pipes

and Fittings*
Maximum Service
Temperature
0
0
C
F
27
32
38
43
49
54
60

Multiply the Pressure

Rating
or Pressure Class at
0
73.4 F (230C) by These Factors

0.88
0.75
0.62
0.50
0.40
0.30
0.22

80
90
100
110
120
130
140

Source: Handbook of PVC Pipe, 1991.

by a thermal derating factor of 0.88. The pressure rating or class for PVC pipe at service temperature of 6O0C (14O0F) would need to multiplied by a thermal de-rating factor of 0.22.
11.4.2.2 Available sizes and thicknesses. AWWA C900 covers PVC pipe in sizes 100
to 300 mm (4-12 in.). AWWA C905 covers PVC pipe in sizes 350-900 mm (14-36
in). There are important differences in these two specifications that should be understood by the designer. AWWA C900 PVC pipe is manufactured in three "pressure
classes" (100, 150, and 200). The pressure class selected is typically the highest normal operating system pressure in psi. AWWA C900 PVC pipe design is based on a
safety factor of 2.5 plus an allowance for hydraulic transients (surge). AWWA C905
does not provide for "pressure classes" but refers to PVC pressure pipe in terms of
"pressure rating." As with pressure class, pressure rating also refers to system pressure
in psi. While AWWA C905 covers six pressure rating categories (100, 125, 160, 165,
200, and 235), the most commonly available pressure ratings are 165 and 235. The
design of AWWA C905 PVC pipe is based on a safety factor of 2.0 and does not
include an allowance for surge. In view of this important difference between the two
specifications, designers often specify higher pressure ratings of C905 PVC pipe than
system pressure would tend to indicate in order to allow for the reduced factor of
design safety.
Both C900 and C905 contain required pipe dimension ratios. Dimension ratios define
a constant ratio between the outside diameter and the wall thickness. For a given dimen-

TABLE 11.13 Pressure Class versus DR-AWWA C900

DR

Pressure Class at
Safety Factor = 2.5, psi (kPa)

14
18
25

200 (1380)
150 (1030)
100 (690)

TABLE 11.14 Pressure Rating versus DR-AWWA C905

DR

Pressure Rating at
Safety Factor = 2.0 psi (kPa)

18
21
25
26
32.5
41

235*
200
165*
160
125
100

*Most commonly used ratings for municipal systems.

sion ratio, pressure capacity and pipe stiffness remain constant, independent of pipe size.
Table 11.13 presents dimension ratios (DR) with corresponding pressure classes as
defined in AWWA C900. Table 11.14 presents dimension ratios with corresponding pressure ratings as defined in AWWA C905.
11.4.2.3 Joints. For PVC pipe, a rubber gasket bell and spigot type joint is the most commonly joint used for typical, municipal buried service. The bell and spigot joint allows for
some pipe deflection (Handbook of PVC Pipe, 1991) without sacrificing water tightness.
This joint is not capable of resisting thrust across the joint and requires thrust blocks or
some other sort of thrust restraint at bends and other changes in the flow direction.
Mechanical restraining devices are commonly used to provide restraint at PVC pipe joints
where necessary.
PVC pipe joints are specified in ASTM D 3139. At connections to fittings and other
types of piping, it is also common to detail a plain end (field-cut pipe) PVC pipe. Plain
end pipes are used to connect to mechanical joint ductile iron fittings and to flange
adapters.
11.4.2.4 Gaskets. Gaskets for PVC joints are specified in ASTM F 477. As with gaskets
for DIP, gaskets for PVC pipe are vulcanized natural or vulcanized synthetic rubber.
Natural rubber is suitable for water pipelines but deteriorates when exposed to raw or
recycled wastewater. EPDM is commonly used in water service and nitrile (Buna N), in
recycled wastewater.
11.4.2.5 Fittings. AWWA C900 and C905 PVC pipe for municipal use are manufactured
in ductile iron pipe OD sizes, so ductile iron fittings, conforming to AWWA CIlO, are
used in all available sizes. See Sec. 11.4.1.5 for discussion on ductile iron fittings.
Although not widely used, PVC fittings, in configurations similar to ductile iron fittings,
are also available for smaller line sizes. AWWA C907 covers PVC pressure pipe fittings
for pipe sizes 100-200 mm (4-8 in.) in pressure classes 100 and 150.
11.4.2.6 Linings and Coatings. PVC pipe does not require lining or coating.
11.4.3 Steel Pipe
Steel pipe is available in virtually any size from 100 m through 3600 mm (4-144 in) for
use in water distribution systems. Though rarely used for pipelines smaller than 400 mm
(16 in.), it is widely used in the western United States for transmission pipelines in sizes

larger than 600 mm (24 in.). The principal advantages of steel pipe include high strength,
the ability to deflect without breaking, the ease of installation, shock resistance, lighter
weight than ductile iron pipe, the ease of fabrication of large pipe, the availability of special configurations by welding, the variety of strengths available, and the ease of field
modification.
11.4.3.1 Materials. Conventional nomenclature refers to two types of steel pipe: (1)
mill pipe and (2) fabricated pipe.
Mill pipe includes steel pipe of any size produced at a steel pipe mill to meet finished
pipe specifications. Mill pipe can be seamless, furnace butt welded, electric resistance
welded, or fusion welded using either a straight or spiral seam. Mill pipe of a given size
is manufactured with a constant outside diameter and a variable internal diameter depending on the required wall thickness.
Fabricated pipe is steel pipe made from plates or sheets. It can be either straight or spiral-seam fusion welded pipe, and it can be specified in either internal or external diameters. Spiral-seam, fusion-welded pipe may either mill pipe or fabricated pipe.
Steel pipe may be manufactured from a number of steel alloys with varying yield and
ultimate tensile strengths. Internal working pressure ratings vary from 690 to 17,000 kPa
(100-2500 lb/in2) depending on alloy, diameter, and wall thickness. Steel piping in water
distribution systems should conform to AWWA C200, in which there are many ASTM
standards for materials (see ANSI B31.1 for the manufacturing processes).
11.4.3.2 Available sizes and thicknesses. Sizes, thicknesses, and working pressures for
pipe used in water distribution systems range from 100 m to 3600 mm (4-144 in) as
shown in Table 4-2 of AWWA Mil (American Water Works Association). The standard
length of steel water distribution pipe is 12.2 m (40 ft).
Manufacturers should be consulted for the availability of sizes and thicknesses of steel
pipe. Table 4-2 of AWWA Mil allows a great variety of sizes and thicknesses.
According to ANSI B36.10,
Standard weight (STD) and Schedule 40 are identical for pipes up to 250 mm (10 in).
Standard weight pipe 300 mm (12 in) and larger have walls 0.5 mm (3/8 in) thick. For
standard weight pipe 300 mm (12 in) and smaller, the ID approximately equals the
nominal pipe diameter. For pipe larger than 300 mm (12 in), the outside diameter (OD)
equals the nominal diameter.
Extra strong (XS) and Schedule 80 are identical for pipes up to 200 mm (8 in). All larger sizes of extra strong-weight pipe have walls 12.7 mm (1/2 in) thick.
Double extra strong (XXS) applies only to steel pipe 300 mm (12 in) and smaller,
There is no correlation between XXS and schedule numbers. For wall thicknesses of
XXS, which (in most cases) is twice that of XS, see ANSI B36.10.
For sizes of 350 mm (14 in) and larger, most pipe manufacturers use spiral welding
machines and, in theory, can fabricate pipe to virtually any desired size. In practice, however, most steel pipe manufacturers have selected and built equipment to produce given
ID sizes. Any deviation from manufacturers' standard practices is expensive so it is
always good practice for the designer to consult pipe manufacturers during the design
process. To avoid confusion, the designer should also show a detail of the specified pipe
size on the plans or tabulate the diameters in the specifications. For cement-mortar lined
steel pipelines, AWWA C200, C205, C207, and C208 apply.
Steel pipe must sometimes either be reinforced at nozzles and openings (tees, wye
branches) or a greater wall thickness must be specified. A detailed procedure for deter-

FIGURE 11.3 Reinforcement

for steel pipe openings.
(a) collar plate; (b) wrapper
plate; (c) crotch plates. Adopted
from Sanks et al,. (1989)
mining whether additional reinforcing is required is described in Chap. II and Appendix
H of ANSI B31.3. If additional reinforcement is necessary, it can be accomplished by a
collar or pad around the nozzle or branch, a wrapper plate, or crotch plates. These reinforcements are shown in Figure 11.3 and the calculations for design are given in AWWA
Mil (American Water Works Association, 1989).

May be welded inside or outside, or both

inside and outside when required.
A. Lap-Welded Slip Joint
C. Double-Butt Weld Joint
RUBBER GASKET
E. Fabricated Rubber Gasket Joint

B. Single-Butt Weld Joint

BUTT STRAP
O. Butt Strap Joint
RUBBER GASKET
F. Rolled-Groove Rubber Gasket Joint

Field-welded restraint bar (alternative

typical for joint types G, H, and I)
RUBBER GASKET
H. Carnegie-Shape Rubber Gasket Joint
For restraint, this weld-on RUBBER GASKET
bar can also be used on joint types E, F, H, and I
G. Tied Rubber Gasket Joint

RUBBERGASKET
CARNEGIESHAPE
I. Carnegie-Shape Rubber Gasket Joint With Weld-On Bell Ring

FIGURE 11.4 Welded and rubber-gasketed joints for steel pipe. AWWA Mil (1989)

11.4.3.3 Joints. For buried service, bell and spigot joints with rubber gaskets or mechanical couplings (with or without thrust harnesses) are common. Welded joints are also common for pipe 600 mm (24 in) and larger. Linings are locally destroyed by the heat of welding, so the ends of the pipe must be bare and the linings field applied at the joints. The
reliability of field welds is questionable without careful inspection, but when properly
made they are stronger than other joints. A steel pipeline project specification involving
field welding should always include a carefully prepared section on quality assurance and
testing of the welds. Different types of steel pipe joints are shown in Fig. 11.4.
11.4.3.4 Gaskets. Gaskets for steel flanges are usually made of cloth-inserted rubber
either 1.6 mm (1/16 in) or 3.2 mm (1/8 in) thick and are of two types:
ring (extending from the ID of the flange to the inside edge of the bolt holes);
full face (extending from the ID of the flange to OD).
Gaskets for mechanical and push-on joints for steel pipe are the same as described in
Section 11.4.1.4 for ductile iron pipe.
11.4.3.5 Fittings. For steel pipe 100 mm (4 in) and larger, specifications for steel fittings
can generally be divided into two classes, depending on the joints used and the pipe size:
Flanged, welded (ANSI B16.9)

TABLE 11.15 Steel Fittings

Mitered Fittings
Crosses
Two-piece elbows, 0-30 bend
Three-piece elbows, 31-60 bend
Four-piece elbows, 61-90 bend
Four-piece, long radius elbows
Laterals, equal diameters
Laterals, unequal diameters
Reducers
Eccentric reducers
Tees
Reducing tees
True wyes

Wrought Fittings
Caps
45 elbows
90 elbows, long radius
90 elbows, short radius
90 reducing elbows, long
radius
Multiple-outlet fittings
Blind flanges
Lap joint flanges
Slip-on flanges
Socket-type welding flanges
Reducing flanges
Threaded flanges
Welding neck flanges
Reducers
Eccentric reducers
180 returns, long radius
Saddles
Reducing outlet tees
Split tees
Straight tees
True wyes

Tee

Long elbow

Short elbow

45 bend

Cross

30 bend

FIGURE 11.5 Typical mitered steel fittings. Adopted

from Sanks et al,. (1989)

Cap

Welding neck flange

Eccentric
reducer

180 return

90 long
radius elbow

Reducing outlet tee

45 elbow

90 short
radius elbow

Concentric
reducer

Straight tee

Reducing elbow,
long radius

Lap joint flange

Blind flange
FIGURE 11.6 Wrought (forged) steel fittings for use with
welded flanges. Adopted from Sanks et al,. (1989)
Fabricated (AWWA C208).
Fittings larger than 100 mm (4 in) should conform to ANSI B 16.9 ("smooth" or
wrought) or AWWA C208 (mitered). Threaded fittings larger than 100 mm (4 in.) should
be avoided. The ANSI B16.9 fittings are readily available up to 300-400 mm (12-16 in)
in diameter. Mitered fittings are more readily available and cheaper for larger fittings.
The radius of a mitered elbow can range from 1 to 4 pipe diameters. The hoop tension
concentration on the inside of elbows with a radius less than 2.5 pipe diameters may

exceed the safe working stress. This tension concentration can be reduced to safe levels
by increasing the wall thickness, as described in ANSI B31, AWWA C208, and Piping
Engineering (Tube Turns Division, 1974). Design procedures for mitered bends are
described in ANSI B31.1 and B31.3. Types of steel fittings are shown in Table 11.15 and
in Figures 11.5 and 11.6.
11.4.3.6 Linings and coatings. Cement mortar is an excellent lining for steel pipe.
Tables 11.9 and 11.10 show required thicknesses for steel pipe.
Steel pipe can also be coated with cement mortar. Recommended mortar coating thicknesses are shown in AWWA C205. These thicknesses, however, are often thinner than
those required to provide adequate protection. Many designers specify a minimum cement
mortar coating thickness of at least 19 mm (3/4 in).
In corrosive soils, the following coatings may be appropriate for protecting steel pipe:

Hot-applied coal-tar enamel (AWWA C203)

Cold-applied tape system (AWWA C214)
Fusion-bonded epoxy (AWWA C213)
Coal-tar epoxy (AWWA C210)
Hot-applied extruded polyethylene [ASTM D 1248 (material only)]

As another alternative, epoxy-lined and -coated steel pipe can be used. Because this
lining is only 0.3-0.6 mm (0.12-0.20 in) thick, the ID of the bare pipe is only slightly
reduced by such linings. Epoxy-lined steel pipe is covered by AWWA C203, C210, and
C213 standards. Before specifying epoxy lining and coating, pipe suppliers must be consulted to determine the limitations of sizes and lengths of pipe that can be lined with
epoxy. Flange faces should not be coated with epoxy if flanges with serrated finish per
AWWA C207 are specified.
11.4.4 Reinforced Concrete Pressure Pipe (RCPP)
Several types of RCPP are manufactured and used in North America. These include steel
cylinder (AWWA C300), prestressed, steel cylinder (AWWA C301), noncylinder (AWWA

TABLE 11.16 General Description of Reinforced Concrete Pressure Pipe

Type of Pipe

AHWA
Standard

Steel
Cylinder

Reinforcement

Design
Basis*

Steel cylinder
Prestressed, steel cylinder
Noncylinder
Pretensioned, steel cylinder

C300
C301
C302
C303

X
X
None
X

Mild reinforcing steel

Prestressed wire
Mild reinforcing steel
Mild reinforcing steel

Rigid
Rigid
Rigid
Semirigid

"Rigid" and "cemirigid" are terms used in AWWA M9 and are intended to differentiate between two design theories. Rigid pipe does not depend on the passive resistance of the soil adjacent to the pipe for support of vertical
loads. Semirigid pipe requires passive soil resistance for vertical load support. The terms "rigid" and "semirigid"
as used here should not be confused with the definitions stated by Marston in Iowa State Experiment Station
Bulletin No. 96.

Grout Placed After

Installation

Steel Spigot Ring

Steel Bell Ring

Rod Reinforcement or
Welded Wire Fabric

Rubber Gasket
Steel Cylinder and Supplemental
Cement Mortar Placed in Rod or Wire Fabric Interior
Field or Other Protection Reinforcement if Necessary

FIGURE 11.7 Cross section of AWWA C300 pipe. AWWA M9 (1995)

C302), and pretensioned, steel cylinder (AWWA 303). Some of these types are made for
a specific type of service condition and others are suitable for a broader range of service
conditions. A general description of RCPP types is shown in Table 11.16. The designer
should be aware that not all RCPP manufactures make all of the pipe types listed.
11.4.4.1 Steel cylinder pipe, AWWA C300. Prior to the introduction of prestressed steel
cylinder pipe (AWWA C301), in the early 1940s, most of the RCPP in the United States
was steel cylinder type pipe. New installations of steel cylinder pipe have been declining
over the years as AWWA C301 and C303 pipes have gained acceptance. Steel cylinder
pipe is manufactured in diameters of 750-3600 mm (30-44 in). Standard lengths are
3.6-7.2 meters (12-24 ft).
AWWA C300 limits the reinforcing steel furnished in the cage(s) to no less 40 percent
of the total reinforcing steel in the pipe. The maximum loads and pressures for this type
of pipe depend on the pipe diameter, wall thickness, and strength limitations of the concrete and steel. The designer should be aware that this type of pipe can be designed for
high internal pressure, but is limited in external load capacity.
A cross section of AWWA C300 pipe and a typical joint configuration is shown in
Fig. 11.7.
11.4.4.2 Prestressed steel cylinder pipe, AWWA C301. Prestressed steel cylinder pipe
has been manufactured in the United States since 1942 and is the most widely used type
of concrete pressure pipe, except in the western United States. Due to cost considerations,
AWWA C301 pipe is often used for high-pressure transmission mains, but it also has been
for distribution mains and for many other low and high pressure uses.
A distressing number of failures of this pipe occurred in the United States primarily
during the 1980s. The outer shell of the concrete cracked, allowing the reinforcement to
corrode and subsequently fail. These failures have resulted in significant revisions to the
standards covering this pipe's design. Even so, the designer should not necessarily depend
solely on AWWA specifications or on manufacturers' assurances, but should make a careful analysis of internal pressure (including waterhammer) and external loads. Make certain that the tensile strain in the outer concrete is low enough so that cracking will either
not occur at all or will not penetrate to the steel under the worst combination of external
and internal loading.

Prestressing Wire and Wire Fabric Around

Bell or Thicker Bell Ring and Wire Fabric
Grout Joint After Installation

Concrete Core
Steel Spigot Ring

Cement - Mortar Coating

Prestress Wire
Steel Cylinder

Steel Bell Ring

Rubber Gasket
Cement Mortar Placed in Field
or Other Protection

A. Lined cylinder pipe

Grout Joint After
Installation

Concrete Core

Steel Spigot Ring

Cement - Mortar Coating

Rubber Gasket

Prestress Wire

Steel Bell Ring

Steel Cylinder

Cement Mortar Placed in

ReId or Other Protection
B. Embedded cylinder pipe
FIGURE 11.8 Cross section of AWWA C301 pipe. AWWA M9 (1995)

Prestressed cylinder pipe has the following two general types of fabrication: (1) a steel
cylinder lined with a concrete core, or (2) a steel cylinder embedded in concrete core.
Lined cylinder pipe is commonly available in IDs from 400 to 1200 mm (16-48 in). Sizes
through 1500 mm (60 in) are available through some manufacturers. Embedded cylinder
pipe is commonly available in inside diameters 1200 mm (48 in) and larger. Lengths are
generally 4.9-7.3 m (16-24 ft), although longer units can be furnished.
AWWA C304, Standard for Design of Prestressed Concrete Cylinder Pipe covers the
design of this pipe. The maximum working pressure for this pipe is normally 2758 kPa
(400 psi). The design method is based on combined loading conditions (the most critical
type of loading for rigid pipe) and includes surge pressure and live loads.
Cross sections of AWWA C301 pipe (lined and embedded) and typical joint configurations are shown in Fig. 11.8.

Steel Cages
Reinforcing Steel Skirt

Grout
InstalJoilantitoAfter
n

Steel Skirt

Steel Bell Ring

Cement
SteelRiSpi
acedProtecti
inMortar
Fieldoorn
ng got Rubber Gasket PlOther

A. AWWA C302-type pipe with steel joint rings

Steel Reinforcing
Cages

Rubber Gasket
B. AWWA C302-type pipe with concrete joint rings
FIGURE 11.9 Cross section of AWWA C302 pipe. AWWA M9 (1995)
11.4.4.3 Noncylinder pipe, AWWA C302. The maximum working pressure of noncylinder pipe is 379 kPa (55 psi) and is generally not suitable for typical municipal systems.
Noncylinder pipe is commonly furnished in diameters of 300 to 3600 mm (12-144 in),
but larger diameters can be furnished if shipping limitations permit. Standard lengths are
2.4-7.3 m (8-24 ft) with AWWA C302 limiting the maximum length that can be furnished
for each pipe size.
Cross sections of AWWA C302 pipe with steel and concrete joint ring configurations
are shown in Fig. 11.9.
11.4.4.4 Pretensioned steel cylinder, AWWA C303. Pretensioned steel cylinder, or
commonly called concrete cylinder pipe (CCP), is manufactured in Canada and in the
western and southwestern areas of the United States. It is commonly available in diameters of 300-1350 mm (12-54 in). Standard lengths are generally 7.3 to 12.2 m(24-40
ft). With maximum pressure capability up to 2758 kPa (400 psi), the longer laying
length, and the overall lighter handling weight, AWWA C303 is a popular choice among
many designers for various applications including municipal transmission and distributions mains.
Manufacture of CCP begins with a fabricated steel cylinder with joint rings which is
hydrostatically tested. A cement-mortar lining is then placed by the centrifugal process
inside the cylinder. The nominal lining thickness is 13 mm (1/2 in) for sizes up to and
including 400 mm (16 in), and 19 mm (3/4 in) for larger sizes. After the lining is cured,
the cylinder is wrapped, typically in a helical pattern, with a smooth, hot-rolled steel bar,
using a moderate tension in the bar. The size and spacing of the bar, as well as the thickness of the steel cylinder, are proportioned to provide the required pipe strength. The
cylinder and bar wrapping are then covered with a cement slurry and a dense mortar coating that is rich in cement.

Rod Reinforcement
Grout Joint After
Steel Bell Ring
Installation

Cement - Mortar Coating

Steel Spigot Ring

Mortar or Concrete Lining

Steel Cylinder
Cement Mortar Placed in
Field or Other Protection
Rubber Gasket

FIGURE 11.10 Cross section of AWWA C303 pipe. AWWA M9 (1995)

The design of CCP is based on a semirigid pipe theory in which internal pressure and
external load are designed for separately but not in combination. Since the theory of semirigid pipe design for earth loads above the pipe is based on the passive soil pressure adjacent to the sides of the pipe, the design must be closely coordinated with the installation
conditions.
A cross section of AWWA C303 pipe and a typical joint configuration is shown in
Fig. 11.10.
11.4.5 High-Density Polyethylene (HOPE) Pipe
Polyethylene pressure pipe has been used in the United States by various utilities in urban
environments for several years. Nearly all natural gas distribution pipe installed in the
United States since 1970 is polyethylene. It has only recently, however, become available
as an AWWA-approved transmission and distribution system piping material. AWWA
Standard C906, Polyethylene Pressure Pipe and Fittings, 4 in. through 63 in., for Water
Distribution, became effective March 1, 1992. AWWA Standard C901, Polyethylene
Pressure Pipe, Tubing, and Fittings, 1/2 in. through 3 in., for Water Service, has been in
effect since 1978. Prior to 1992, polyethylene pipe use in municipal water distribution
systems was normally limited to water services. Since AWWA approval in 1992, however, polyethylene pipe is now being used in transmission and distribution system applications. Because of its resistance to corrosion, its light weight and high strength to weight
ratio, resistance to cracking, smoother interior wall surface, and its demonstrated resistance to damage during seismic events, HDPE pipe is gaining acceptance for use in
municipal water systems
11.4.5.1 Materials. Low density polyethylene was first introduced in the 1930's and
1940s in England and then in the United States. This first material was commonly used
for cable coatings. Pipe grade resins were developed in the 1950s and have evolved to
today's high density, extra high molecular weight materials. AWWA C906 specifies
several different resins but, today all HDPE water pipe, manufactured in the United
States, is made with a material specified in ASTM D 3350 by a cell classification
345434C.

TABLE 11.17 Pressure Class versus DR-AWWA C906

DR*
11
13.5
17
21

Pressure Class,
Safety Factor = 2.0,psi (kPa)
160
130
100
80

(1100)
(900)
(690)
(550)

*These DRs are from the standard dimension ratio series established
by ASTM F 412.

HDPE pipe is rated for pressure capacity at 230C (73.40F). Being a thermoplastic, the
pressure capacity of HDPE pipe is related to its operating temperature. Through the normal range of municipal water system temperatures, 0C-24 (32 F-75 F), the pressure
rating of HDPE remains relatively constant. As the operating temperature rises above
230C (73.40F), however, the pressure capacity of HDPE pipe decreases to a level below
its pressure class. The pressure rating for HDPE pipe at service temperature of 6O0C
(14O0F) would be about half its rating at 230C (73.40F).
11.4.5.2 Available sizes and thicknesses. AWWA C906 covers HDPE pipe in sizes 100
-1600 mm (4-63 in). The design, according to AWWA C906, of HDPE pipe is similar to
AWWA C900 PVC pipe in that HDPE pipe is rated according to "pressure classes." The
pressure classes detailed in AWWA C906 include allowance for pressure rises above
working pressure due to occasional positive pressure transients not exceeding two times
the nominal pressure class and recurring pressure surges not exceeding one and one-half
times the nominal pressure class. AWWA C906 lists HDPE pipe sizes according to the
IPS (steel pipe) and the ISO (metric) sizing systems. Ductile iron pipe sizes are also
available.
As with AWWA C900 and C905 (PVC pipe), AWWA C906 contains dimension ratio
(outside diameter to wall thickness) specifications. Table 11.17 presents dimension ratios
with corresponding pressure classes as defined in AWWA C906 for commonly available
HDPE pipe.
11.4.5.3 Joints. HDPE pipe can be joined by thermal butt-fusion, flange assemblies, or
mechanical methods as may be recommended by the pipe manufacturer. HDPE is not to
be joined by solvent cements, adhesives (such as epoxies), or threaded-type connections.
Thermal butt-fusion is the most widely used method for joining HDPE piping. This procedure uses portable field equipment to hold pipe and/or fittings in close alignment while
the opposing butt-ends are faced, cleaned, heated and melted, fused together, and then
cooled under fusion parameters recommended by the pipe manufacturer and fusion equipment supplier.
For each polyethylene material there exists an optimum range of fusion conditions,
such as fusion temperature, interface pressure, and cooling time. Thermal fusion should
only be conducted by persons who have received training in the use of the fusion equipment according to the recommendations of the pipe manufacturer and fusion equipment
supplier. In situations where different polyethylene piping materials must be joined by
thermal butt-fusion process, both pipe manufacturers should be consulted to determine the
appropriate fusion procedures. ASTM D 2657 covers thermal butt-fusion of HDPE pipe.

HDPE pipe is normally joined above ground and then placed in the pipeline trench. The
thermal butt-fusion joint is not subject to movement due to thrust and does not require
thrust restraint such as thrust blocks.
Flanged and mechanical joint adapters are available for joining HDPE pipe to valves
and ductile iron fittings. The designer should always consult the pipe manufacturer to
insure a proper fit between pipe and fittings. The designer should also make sure that,
when connecting to a butterfly valve, the valve disc will freely swing to the open position
without hitting the face of the stub end or flange adapter.
11.4.5.4 Gaskets. Gaskets are not necessary for HDPE pipe using thermal butt-fusion
joints.
11.4.5.5 Fittings. AWWA C906 HDPE pipe for municipal use is manufactured in ductile
iron pipe OD sizes, so ductile iron fittings, conforming to AWWA CIlO, can be used in
all available sizes. See Sec. 11.4.1.5 for discussion on ductile iron fittings. Although not
widely used, HDPE fittings, in configurations similar to ductile iron fittings, are also
available. AWWA C906 covers HDPE pressure pipe fittings.
11.4.5.6 Linings and coatings. HDPE pipe does not require lining or coating.
11.4.6 Asbestos-Cement Pipe (ACP)
Asbestos-cement pipe (ACP), available in the United States since 1930, is made by mixing Portland cement and asbestos fiber under pressure and heating it to produce a hard,
strong, yet machinable product. It is estimated that over 480,000 km (300,000 mi) of ACP
is now in service in the United States.
In the late 1970s, attention was focused on the hazards of asbestos in the environment
and, particularly, in drinking water. There was significant debate of the issue with one set
of experts advising of the potential dangers and a second set of experts claiming that pipes
made with asbestos do not result in increases in asbestos concentrations in the water.
Studies have shown no association between water delivered by ACP and any general disease, however, the general fear that resulted from the controversy had a tremendous negative impact on ACP use in the United States. The debate on the health concerns of using
ACP along with the introduction of PVC pipe into the municipal water system market, has
reduced the use of ACP significantly in the past several years.
11.4.6.1 Available sizes and thicknesses. ACP is available in diameters of 100-1050 mm
(4-42 in). Refer to ASTM C 296 and AWWA C401, C402, and C403 for thickness and
pressure ratings and AWWA C401 and C403 for detailed design procedures. AWWA C401
for 100-400 mm (4-16 in) pipe is similar to AWWA C403 for 450-1050 mm (18 -42 in)
pipe. The properties of asbestos-cement for distribution pipe (AWWA C400) and transmission pipe (AWWA C402) are identical. However, under AWWA C403 (transmission
pipe) the suggested minimum safety factor is 2.0 for operating pressure and 1.5 for external loads, whereas the safety factors under AWWA C402 (distribution pipe) are 4.0 and
2.5, respectively. So the larger pipe has the smaller safety factors.
Section 4 in AWWA C403 justifies this on the basis of surge pressure in large pipe
tending to be less than those in small pipes. However, surge pressures are not necessarily a function of pipe diameter (Chap. 12). The operating conditions, including surge
pressures, for any proposed pipeline installation should be closely evaluated before the
pipe class is selected. It is the engineer's design prerogative to select which of the safety factors should apply. AWWA C400 specifies that safety factors should be no less than

4.0 and 2.5 if no surge analysis is made. The low safety factors given in AWWA C403
should only be used if all loads (external, internal, and transient) are carefully and accurately evaluated.
11.4.6.2 Joints and fittings. The joints are usually push-on, twin-gasketed couplings,
although mechanical and rubber gasket push-on joints can be used to connect ACP to ductile iron fittings.
Ductile iron fittings conforming to ANSI/AWWA C110/A21.10 are used with ACP,
and adapters are available to connect ACP to flanged or mechanical ductile iron fittings.
Fabricated steel fittings with rubber gasket joints can also be used.
11.4.7 Pipe Material Selection
Buried piping for municipal water transmission and distribution must resist internal pressure, external loads, differential settlement, and corrosive action of both soils and, potentially, the water it carries. General factors to be considered in the selection of pipe include
the following:
Service conditions
-

Pressure (including surges and transients)

Soil loads, bearing capacity of soil, potential settlement
Corrosion potential of soil
Potential corrosive nature of some waters

Availability
- Local availability and experienced installation personnel
- Sizes and thicknesses (pressure ratings and classes)
- Compatibility with available fittings
Properties of the pipe
-

Strength (static and fatigue, especially for waterhammer)

Ductility
Corrosion resistance
Fluid friction resistance (more important in transmission pipelines)

Economics
- Cost (installed cost, including freight to job site and installation)
- Required life
- Cost of maintenance and repairs
The items listed above are general factors relating to pipe selection to be considered during the design of any pipeline. Since most municipal water system projects are
either let out to competitive bid or are installed as a part of private land development,

TABLE 11.18. Comparison of Pipe for Municipal System Service

Pipe

Advantages

DisavantagesAimitations
2

Yield strength: 290,000 kPa (42,000 lb/in );

E= 1663 106 kPa (24 X 106 lb/in2); ductile, elongation * 10%; good corrosion
resistance, wide variety available fittings
and joints; available sizes: 100-1350 mm
(454 in); ID, wide range of available thicknesses, good resistance to waterhammer,
high strength for supporting earth loads
Yield strengths: 207,000-414,000
kPa
Steel
(30,000-60,000 lb/in2); ultimate strengths:
338,000-518,000 kPa (49,000-75,000
lb/in22); E = 207 X 106 kPa (30 X 106
lb/in ); ductile, elongation varies from
17 to 35%, pressure rating to 17,000 kPa
(2500 lb/in2); diameters to 3.66 m (12 ft);
widest variety of available fittings and
joints, custom fittings can be mitered and
welded, excellent resistance to waterhammer, low cost, high strength for supporting
earth loads
Polyvinyl
Tensile strength (hydrostatic design basis)
chloride
- 26,400 kPa (4000 lb/in2);
(PVC)
E = 2,600,000 kPa (400,000 lb/in2); light
weight, very durable, very smooth, liners
and wrapping not required, can use ductile
iron fittings with adapters, diameters from
High-density 100 to 375 mm (4-36 in)
design basis)
polyethylene Tensile strength (hydrostatic
(HDPE)
- 11,000 kPa (1600
lb/in2); E = 896,000
2
kPa (130,000 lb/in ); lightweight, very
durable, very smooth, liners and wrapping
not required, can use ductile iron fittings,
diameters from 100 to 1600 mm (4 to 63 in)
Reinforced
concrete
Several types available to suit different
pressure
conditions, high strength for supporting
earth loads, wide variety of sizes from
(RCPP)
300 to 3600 mm (24-144 in)
Ductile iron
(DIP)

Maximum2 pressure = 2400 kPa

(350 lb/in ); high cost especially for
long freight hauls, no diameters
above 1350 mm (54 in); difficult to
weld, may require wrapping or
cathodic protection in corrosive soils
Poor corrosion resistance unless both
lined and coated or wrapped, may
require cathodic protection in
corrosive soils, higher unit cost in
smaller diameters

Maximum2 pressure = 2400 kPa

(350 lb/in ); waterhammer not
included in AWWA C905; limited
resistance to cyclic loading, unsuited
for outdoor use above ground

Maximum2 pressure = 1750 kPa

(250 lb/in ); relatively new product,
750 mm (30 in) is largest size
available for municipal system
pressures, thermal butt-fusion joints,
requires higher laborer skill
Attacked by soft water, acids, sulfides,
sulfates, and chlorides, often requires
protective coatings; waterhammer can
crack outer shell, exposing reinforcement to corrosion and destroying its
strength with time; maximum pressure
= 138OkPa (200 lb/in2)
Attacked by soft water, acids,
Asbestoscement (ACP) Yield strength: not applicable; design based sulfates; requires thrust blocks at
on crushing strength, see ASTM C 296 and elbows tees, and dead ends;
maximum2 pressure =1380 kPa
C 500;
E = 23,500,000 kPa (3,400,000
(200 lb/in ) for pipe up to 400 mm
lb/in2); rigid, lightweight in long lengths,
low cost; diameters from 100 to 1050 mm (16 in); health hazards of asbestos in
potable water service are
(4-42 in), compatible with cast-iron fittings, pressure ratings from 1600 to 3100 controversial
kPa (225- 450 lb/in2) for large pipe 450

the designer will find that installed cost, lacking specific service conditions that
require otherwise, will tend to dictate pipe selection. For example, steel pipe and reinforced concrete pressure pipe are both available in 300 mm (12 in) diameter. However,
the installed cost of ductile iron pipe or PVC pipe is typically lower (typical municipal use) in the 300 mm (12 in) size. Therefore, if the service conditions do not require
the high pressure capabilities of steel or reinforced concrete pressure pipe, the logical
choice for 300 mm pipe (12 in) will optionally be ductile iron or PVC. Conversely, if
the proposed pipeline is 900 mm (36 in) in diameter, the installed cost of both steel and
reinforced concrete pressure pipe, depending on location, tend to be much more competitive.
A general comparison of the various types of pipe used in municipal water systems is
shown in Table 11.18.
77.5 PIPELINEDESIGN
This section will address typical issues that are addressed during the design of water distribution and transmission pipelines. Pressure pipelines must primarily be able to resist
internal pressures, external loads (earth and impact loads), forces transferred along the
pipe when pipe-to-soil friction is used for thrust restraint, and handling during construction. Each of these design issues will be discussed and appropriate formulas presented.
11.5.1 Internal Pressures
The internal pressure of a pipeline creates a circumferential tension stress, frequently
termed hoop stress, that governs the pipeline thickness. In other words, the pipe must be
thick enough to withstand the pressure of the fluid within. The internal pressure used in
design should be that to which the pipe may subjected during its lifetime. In a distribution
system this pressure may be the maximum working pressure plus and allowance for surge.
It may be also the pipeline testing pressure or the shutoff head of an adjacent pump. In a
transmission pipeline, the pressure is measured by the vertical distance between the pipe
centerline and the hydraulic grade line. Potential hydraulic grade lines on transmission
pipelines should be carefully considered. The static hydraulic grade line is potentially
much higher than the dynamic grade line if a downstream valve is closed.
Hoop tensile stress is given by the equation
*= ?
<!>
where s = allowable circumferential stress in kPa (lb/in2) p = pressure in kPa (lb/in2) D
= outside diameter of pipe in mm (in) t = thickness of the pipe in mm (in)
It should be noted that this equation is the basis for determining the circumferential stress
in steel and reinforced pressure pipe and for determining the pressure classes and pressure
ratings for virtually all other different types of pressure pipe.
11.5.2 Loads on Buried Pipe
Buried pipes must support external superimposed loads, including the weight of the soil
above plus any live loads such as wheel loads due to vehicles or equipment. The two
broad categories for external structural design are rigid and flexible pipe. Rigid pipe sup-

Next Page

CHAPTER 12
HYDRAULIC TRANSIENT
DESIGN FOR
PIPELINE SYSTEMS

C. Samuel Martin
School of Civil and Environmental Engineering
Georgia Institute of Technology
Atlanta, Georgia

12.1

INTRODUCTION TO WATERHAMMER AND SURGING

By definition, waterhammer is a pressure (acoustic) wave phenomenon created by relatively sudden changes in the liquid velocity. In pipelines, sudden changes in the flow
(velocity) can occur as a result of (1) pump and valve operation in pipelines, (2) vapor
pocket collapse, or (3) even the impact of water following the rapid expulsion of air out
of a vent or a partially open valve. Although the name waterhammer may appear to be a
misnomer in that it implies only water and the connotation of a "hammering" noise, it has
become a generic term for pressure wave effects in liquids. Strictly speaking, waterhammer can be directly related to the compressibility of the liquidprimarily water in this
handbook. For slow changes in pipeline flow for which pressure waves have little to no
effect, the unsteady flow phenomenon is called surging.
Potentially, waterhammer can create serious consequences for pipeline designers if not
properly recognized and addressed by analysis and design modifications. There have been
numerous pipeline failures of varying degrees and resulting repercussions of loss of property and life. Three principal design tactics for mitigation of waterhammer are (1) alteration
of pipeline properties such as profile and diameter, (2) implementation of improved valve
and pump control procedures, and (3) design and installation of surge control devices.
In this chapter, waterhammer and surging are defined and discussed in detail with reference to the two dominant sources of waterhammerpump and/or valve operation.
Detailed discussion of the hydraulic aspects of both valves and pumps and their effect on
hydraulic transients will be presented. The undesirable and unwanted, but often potentially possible, event of liquid column separation and rejoining are a common justification for
surge protection devices. Both the beneficial and detrimental effects of free (entrained or
entrapped) air in water pipelines will be discussed with reference to waterhammer and
surging. Finally, the efficacy of various surge protection devices for mitigation of waterhammer is included.

12.2

FUNDAMENTALS OF WATEFtHAMMER AND SURGE

The fundamentals of waterhammer, an elastic process, and surging, an incompressible

phenomenon, are both developed on the basis of the basic conservational relationships of
physics or fluid mechanics. The acoustic velocity stems from mass balance (continuity),
while the fundamental waterhammer equation of Joukowsky originates from the application of linear momentum [see Eq. (12.2)].
12.2.1 Definitions
Some of the terms frequently used in waterhammer are defined as follows.
Waterhammer. A pressure wave phenomenon for which liquid compressibility plays
a role.
Surging. An unsteady phenomenon governed solely by inertia. Often termed mass
oscillation or referred to as either rigid column or inelastic effect.
Liquid column separation. The formation of vapor cavities and their subsequent collapse and associated waterhammer on rejoining.
Entrapped air. Free air located in a pipeline as a result of incomplete filling, inadequate venting, leaks under vacuum, air entrained from pump intake vortexing, and
other sources.
Acoustic velocity. The speed of a waterhammer or pressure wave in a pipeline.
Joukowsky equation. Fundamental relationship relating waterhammer pressure
change with velocity change and acoustic velocity. Strictly speaking, this equation
only valid for sudden flow changes.
12.2.2 Acoustic Velocity
For wave propagation in liquid-filled pipes the acoustic (sonic) velocity is modified by the
pipe wall elasticity by varying degrees, depending upon the elastic properties of the wall
material and the relative wall thickness. The expression for the wave speed is
VK/pJ_

a0O

/ 1 /^ 1 \

- TT^f- T^IT
V
e E
v
e E
where E is the elastic modulus of the pipe wall, D is the inside diameter of the pipe, e is
the wall thickness, and a0 is the acoustic velocity in the liquid medium. In a very rigid pipe
or in a tank, or in large water bodies, the acoustic velocity a reduces to the well=known
relationship a = a0 = V(/p). For water K = 2.19 GPa (318,000 psi) and p = 998 kg/m3
(1.936 slug/ft3), yielding a value of a0 = 1483 m/sec (4865 ft/sec), a value many times that
of any liquid velocity V.
12.2.3 Joukowsky (Waterhammer) Equation
There is always a pressure change A/? associated with the rapid velocity change AV across
a waterhammer (pressure) wave. The relationship between Ap and AV from the basic
physics of linear momentum yields the well-known Joukowsky equation

Ap = -paAV

(12.2)

where p is the liquid mass density, and a is the sonic velocity of the pressure wave in the
fluid medium in the conduit. Conveniently using the concept of head, the Joukowsky head
rise for instantaneous valve closure is
AH=Ap = _poAV=oK
P
98
g
The compliance of a conduit or pipe wall can have a significant effect on modification
of (1) the acoustic velocity, and (2) any resultant waterhammer, as can be shown from Eq.
(12.1) and Eq. (12.2), respectively. For simple waterhammer waves for which only radial
pipe motion (hoop stress) effects are considered, the germane physical pipe properties are
Young's elastic modulus (E) and Poisson ratio (|n). Table 12.1 summarizes appropriate values of these two physical properties for some common pipe materials.
The effect of the elastic modulus (E) on the acoustic velocity in water-filled circular
pipes for a range of the ratio of internal pipe diameter to wall thickness (Die) is shown in
Fig. 12.1 for various pipe materials.

12.3

HYDRAULIC CHARACTERISTICS OF VALVES

Valves are integral elements of any piping system used for the handling and transport of
liquids. Their primary purposes are flow control, energy dissipation, and isolation of portions of the piping system for maintenance. It is important for the purposes of design and
final operation to understand the hydraulic characteristics of valves under both steady and
unsteady flow conditions. Examples of dynamic conditions are direct opening or closing
of valves by a motor, the response of a swing check valve under unsteady conditions, and
the action of hydraulic servovalves. The hydraulic characteristics of valves under either
noncavitating or cavitating conditions vary considerably from one type of valve design to
another. Moreover, valve characteristics also depend upon particular valve design for a
special function, upon absolute size, on manufacturer as well as the type of pipe fitting
employed. In this section the fundamentals of valve hydraulics are presented in terms of
pressure drop (headloss) characteristics. Typical flow characteristics of selected valve
types of controlgate, ball, and butterfly, are presented.
TABLE 12.1 Physical Properties of Common Pipe Materials
Material

Young's Modulus
E (GPa)

Asbestos cement
Cast iron
Concrete
Concrete (reinforced)
Ductile iron
Polyethylene
PVC (polyvinyl chloride)
Steel

23-24
80-170
14-30
30-60
172
0.7-0.8
2.4-3.5
200-207

Poisson's Ratio
|Ll
0.25-0.27
0.10-0.15
0.30
0.46
0.46
0.30

Elastic Modulus (GPa)

Speed of Sound in m/sec

Steel
207
!Al
uminum
: 68.9
Concrete:
27:6 ;

PVC
2.8
Polyethlene
! 0.69 i

Diameter to Wall Thickness Ratio (Die)

Elastic Modulus (psi)

Speed of Sound in ft/sec

Steel
30(10s)
Copper6
18(10 )
Aluminum
10(1O8)
Concrete
4(106)
PVC
0.4(106V

Polyethlene
: 0.1 no8) i

FIGURE 12.1 Effect of wall thickness of various pipe materials on acoustic velocity
in water pipes.

12.3.1 Descriptions of Various Types of Valves

Valves used for the control of liquid flow vary widely in size, shape, and overall design
due to vast differences in application. They can vary in size from a few millimeters in
small tubing to many meters in hydroelectric installations, for which spherical and butterfly valves of very special design are built. The hydraulic characteristics of all types of
valves, albeit different in design and size, can always be reduced to the same basic
coefficients, notwithstanding fluid effects such as viscosity and cavitation. Figure 12.2
shows cross sections of some valve types to be discussed with relation to hydraulic
performance.

b.) Globe valve

a.) Gate valve
(circular gate)

c.) Needle valve

e.) Butterfly valve

d.) Gate valve

(square gate)

f.) Ball valve

FIGURE 12.2 Cross sections of selected control valves: (From Wood and Jones, 1973).

12.3.2 Definition of Geometric Characteristics of Valves

The valve geometry, expressed in terms of cross-sectional area at any opening, sharpness
of edges, type of passage, and valve shape, has a considerable influence on the eventual
hydraulic characteristics. To understand the hydraulic characteristics of valves it is useful,
however, to express the projected area of the valve in terms of geometric quantities. With
reference to Fig. 12.2 the ratio of the projected open area of the valve Av to the full open
valve Avo can be related to the valve opening, either a linear measure for a gate valve, or
an angular one for rotary valves such as ball, cone, plug, and butterfly types. It should be
noted that this geometric feature of the valve clearly has a bearing on the valve hydraulic
performance, but should not be used directly for prediction of hydraulic performance
either steady state or transient. The actual hydraulic performance to be used in transient
calculations should originate from experiment.

12.3.3 Definition of Hydraulic Performance of Valves

The hydraulic performance of a valve depends upon the flow passage through the valve
opening and the subsequent recovery of pressure. The hydraulic characteristics of a valve
under partial to fully opened conditions typically relate the volumetric flow rate to a characteristic valve area and the head loss AH across the valve. The principal fluid properties
that can affect the flow characteristics are fluid density p, fluid viscosity |i, and liquid
vapor pressure pv if cavitation occurs. Except for small valves and/or viscous liquids or
both, Reynolds number effects are usually not important, and will be neglected with reference to water. A valve in a pipeline acts as an obstruction, disturbs the flow, and in general causes a loss in energy as well as affecting the pressure distribution both upstream
and downstream. The characteristics are expressed either in terms of (1) flow capacity as
a function of a defined pressure drop or (2) energy dissipation (headloss) as a function of
pipe velocity. In both instances the pressure or head drop is usually the difference in total
head caused by the presence of the valve itself, minus any loss caused by regular pipe friction between measuring stations.
The proper manner in determining A/f experimentally is to measure the hydraulic
grade line (HGL) far enough both upstream and downstream of the valve so that uniform
flow sections to the left of and to the right of the valve can be established, allowing for
the extrapolation of the energy grade lines (EGL) to the plane of the valve. Otherwise, the
valve headloss is not properly defined. It is common to express the hydraulic characteristics either in terms of a headloss coefficient K1 or as a discharge coefficient Cf where Av
is the area of the valve at any opening, and A/f is the headloss defined for the valve.
Frequently a discharge coefficient is defined in terms of the fully open valve area. The
hydraulic coefficients embody not only the geometric features of the valve through Av but
also the flow characteristics.
Unless uniform flow is established far upstream and downstream of a valve in a
pipeline the value of any of the coefficients can be affected by effects of nonuniform flow.
It is not unusual for investigators to use only two pressure tapsone upstream and one
downstream, frequently 1 and 10 diameters, respectively. The flow characteristics of
valves in terms of pressure drop or headloss have been determined for numerous valves
by many investigators and countless manufacturers. Only a few sets of data and typical
curves will be presented here for ball, butterfly, and gate, ball, butterfly, and gate valves
C0. For a valve located in the interior of a long continuous pipe, as shown in Fig. 12.3,
the presence of the valve disturbs the flow both upstream and downstream of the obstruction as reflected by the velocity distribution, and the pressure variation, which will be
nonhydrostatic in the regions of nonuniform flow. Accounting for the pipe friction

FIGURE 12.3 Definition of headloss characteristics of a valve.

between upstream and downstream uniform flow sections, the headloss across the valve
is expressed in terms of the pipe velocity and a headloss coefficient K1
*H = KL

(12.4)

Often manufacturers represent the hydraulic characteristics in terms of discharge

coefficients
Q = CfAvoV2^H = CFAVOV2^H ,

(12.5)

where
H = A# + -J(12-6)
2g
Both discharge coefficients are defined in terms of the nominal full-open valve area
Avo and a representative head, A/f for Cf and H for CQJ the latter definition generally
reserved for large valves employed in the hydroelectric industry. The interrelationship
b
e
t
w
e
e
n
C, Cp, and KL is
1
1 C 2
1
^ = i = C^
(12'7>
F
/
Frequently valve characteristics are expressed in terms of a dimensional flow coefficient Cv from the valve industry
Q = CvVZp

(12.8)

where Q is in American flow units of gallons per minute (gpm) and A/? is the pressure loss
in pounds per square inch (psi). In transient analysis it is convenient to relate either the
loss coefficient or the discharge coefficient to the corresponding value at the fully open
valve position, for which Cf = Cfo. Hence,
<L = SL JM-= v fM.
Q0 CfjAH0
^AH0

((i2.9)}

Traditionally the dimensionless valve discharge coefficient is termed i and defined by

c c c
a?
T= C
(12 10)
L
c/o c
c
Jl?
^v0
fo
V AL
12.3.4 Typical Geometric and Hydraulic Valve Characteristics
The geometric projected area of valves shown in Fig. 12.2 can be calculated for ball, butterfly, and gate valves using simple expressions. The dimensionless hydraulic flow coefficient T is plotted in Fig. 12.4 for various valve openings for the three selected valves
along with the area ratio for comparison. The lower diagram, which is based on hydraulic
measurements, should be used for transient calculations rather than the upper one, which
is strictly geometric.
12.3.5 Valve Operation
The instantaneous closure of a valve at the end of a pipe will yield a pressure rise satisfying Joukowsky's equationEq. (12.2) or Eq. (12.3). In this case the velocity difference

Area Characteristics of Valves

Gate
Area Ratio AyA^1

l
D/DBal
b=0.629
Butterfly

Relative Opening y/D (%)

Hydraulic (Tau) Characteristics of Valves

Gate

l
D/DBal
b=0.623
Butterfly

Relative Opening y/D (%)

FIGURE 12.4 Geometric and hydraulic characteristics of typical control valves

AV = O V0 , where V0 is the initial velocity of liquid in the pipe. Although Eq. (12.2)
applies across every wavelet, the effect of complete valve closure over a period of time
greater than 2LIa, where L is the distance along the pipe from the point of wave creation
to the location of the first pipe area change, can be beneficial. Actually, for a simple
pipeline the maximum head rise remains that from Eq. (12.3) for times of valve closure
tc ^ 2LIa, where L is the length of pipe. If the value of tc > 2LIa, then there can be a considerable reduction of the peak pressure resulting from beneficial effects of negative wave
reflections from the open end or reservoir considered in the analysis. The phenomenon
can still be classified as waterhammer until the time of closure tc 2LIa, beyond which
time there are only inertial or incompressible deceleration effects, referred to as surging,
also known as rigid column analysis. Table 12.2 classifies four types of valve closure,
independent of type of valve.

TABLE 12.2 Classification of Valve Closure

Time of Closure tc

Type of Closure

O
< 2LIa
> 2LIa
2L/a

Instantaneous
Rapid
Gradual
Slow

Maximum Head
&Hm
a

K/8
aVJg
< aV0/g
aV0/g

Phenomenon
Waterhammer
Waterhammer
Waterhammer
Surging

Using standard waterhammer programs, parametric analyses can be conducted for the
preparation of charts to demonstrate the effect of time of closure, type of valve, and an
indication of the physical processwaterhammer or simply inertia effects of deceleration.
The charts are based on analysis of valve closure for a simple reservoir-pipe-valve
arrangement. For simplicity fluid friction is often neglected, a reasonable assumption for
pipes on the order of hundreds of feet in length.
72.4

HYDRAULIC CHARACTERISTICS OF PUMPS

Transient analyses of piping systems involving centrifugal, mixed-flow, and axial-flow

pumps require detailed information describing the characteristics of the respective turbomachine, which may pass through unusual, indeed abnormal, flow regimes. Since little if
any information is available regarding the dynamic behavior of the pump in question, invariably the decision must be made to use the steady-flow characteristics of the machine gathered from laboratory tests. Moreover, complete steady-flow characteristics of the machine
may not be available for all possible modes of operation that may be encountered in practice.
In this section steady-flow characteristics of pumps in all possible zones of operation
are defined. The importance of geometric and dynamic similitude is first discussed with
respect to both (1) homologous relationships for steady flow and (2) the importance of the
assumption of similarity for transient analysis. The significance of the eight zones of operation within each of the four quadrants is presented in detail with reference to three possible modes of data representation. The steady-flow characteristics of pumps are discussed
in detail with regard to the complete range of possible operation. The loss of driving power
to a pump is usually the most critical transient case to consider for pumps, because of the
possibility of low pipeline pressures which may lead to (1) pipe collapse due to buckling,
or (2) the formation of a vapor cavity and its subsequent collapse. Other waterhammer
problems may occur due to slam of a swing check valve, or from a discharge valve closing
either too quickly (column separation), or too slowly (surging from reverse flow). For radial-flow pumps for which the reverse flow reaches a maximum just subsequent to passing
through zero speed (locked rotor point), and then is decelerated as the shaft runs faster in
the turbine zone, the head will usually rise above the nominal operating value. As reported
by Donsky (1961) mixed-flow and axial-flow pumps may not even experience an upsurge
in the turbine zone because the maximum flow tends to occur closer to runaway conditions.
12.4.1 Definition of Pump Characteristics
The essential parameters for definition of hydraulic performance of pumps are defined as
Impeller diameter. Exit diameter of pump rotor D1.
Rotational speed. The angular velocity (rad/s) is co, while N =2 7CCO/60 is in rpm.

 Flow rate. Capacity Q at operating point in chosen units.

Total dynamic head (TDH). The total energy gain (or loss) H across pump, defined as
(P

\ (P

\ V2

V2

"-(11H-(H+S-S
where subscripts s and d refer to suction and discharge sides of the pump, respectively,
12.4.2 Homologous (Affinity) Laws
Dynamic similitude, or dimensionless representation of test results, has been applied with
perhaps more success in the area of hydraulic machinery than in any other field involving
fluid mechanics. Due to the sheer magnitude of the problem of data handling it is imperative that dimensionless parameters be employed for transient analysis of hydraulic
machines that are continually experiencing changes in speed as well as passing through
several zones of normal and abnormal operation. For liquids for which thermal effects
may be neglected, the remaining fluid-related forces are pressure (head), fluid inertia,
resistance, phase change (cavitation), surface tension, compressibility, and gravity. If the
discussion is limited to single-phase liquid flow, three of the above fluid effectscavitation, surface tension, and gravity (no interfaces within machine)can be eliminated, leaving the forces of pressure, inertia, viscous resistance, and compressibility. For the steady
or even transient behavior of hydraulic machinery conducting liquids the effect of compressibility may be neglected.
In terms of dimensionless ratios the three forces yield an Euler number (ratio of inertia
force to pressure force), which is dependent upon geometry, and a Reynolds number.
For all flowing situations, the viscous force, as represented by the Reynolds number,
is definitely present. If water is the fluid medium, the effect of the Reynolds number
on the characteristics of hydraulic machinery can usually be neglected, the major
exception being the prediction of the performance of a large hydraulic turbine on the
basis of model data. For the transient behavior of a given machine the actual change in
the value of the Reynolds number is usually inconsequential anyway. The elimination
of the viscous force from the original list reduces the number of fluid-type forces from
seven to two-pressure (head) and inertia, as exemplified by the Euler number. The
appellation geometry in the functional relationship in the above equation embodies primarily, first, the shape of the rotating impeller, the entrance and exit flow passages,
including effects of vanes, diffusers, and so on; second, the effect of surface roughness;
and lastly the geometry of the streamline pattern, better known as kinematic similitude
in contrast to the first two, which are related to geometric similarity. Kinematic similarity is invoked on the assumption that similar flow patterns can be specified by congruent velocity triangles composed of peripheral speed U and absolute fluid velocity V
at inlet or exit to the vanes. This allows for the definition of a flow coefficient,
expressed in terms of impeller diameter D1 and angular velocity co:
C

= J?

(12 12)

The reciprocal of the Euler number (ratio of pressure force to inertia force) is the head
coefficient, defined as

Q = J^

(12.13)

A power coefficient can be defined

C =

"

(1114)

WD?

For transient analysis, the desired parameter for the continuous prediction of pump speed
is the unbalanced torque T. Since T = P/co, the torque coefficient becomes
C

(12 15)

T = WD?

'

Traditionally in hydraulic transient analysis to refer pump characteristics to so-called

rated conditionswhich preferably should be the optimum or best efficiency point (BEP),
but sometimes defined as the duty, nameplate, or design point. Nevertheless, in terms of
rated conditions, for which the subscript R is employed, the following ratios are defined;
Flow: v = -^speed: a = = N
head: h = H
torque: (3 =T
QR
R
R
R
R
Next, for a given pump undergoing a transient, for which D1 is a constant, Eqs.
(12.12-12.15) can be written in terms of the above ratios
= ^o_=Q_to_
a
CQR Q co*

JL = ^L = JL??*.
a2 CHR HR co2

p _ cr _ y CQ/
a2 CTR TR co2

12.4.3 Abnormal Pump (Four-Quadrant) Characteristics

The performance characteristics discussed up to this point correspond to pumps operating
normally. During a transient, however, the machine may experience either a reversal in
flow, or rotational speed, or both, depending on the situation. It is also possible that the
torque and head may reverse in sign during passage of the machine through abnormal
zones of performance. The need for characteristics of a pump in abnormal zones of operation can best be described with reference to Fig. 12.5, which is a simulated pump power
failure transient. A centrifugal pump is delivering water at a constant rate when there is a
sudden loss of power from the prime moverin this case an electric motor. For the postulated case of no discharge valves, or other means of controlling the flow, the loss of driving torque leads to an immediate deceleration of the shaft speed, and in turn the flow.
The three curves are dimensionless head (h), flow (v), and speed (a). With no additional
means of controlling the flow, the higher head at the final delivery point (another reservoir) will eventually cause the flow to reverse (v < O) while the inertia of the rotating
parts has maintained positive rotation (a > O). Up until the time of flow reversal the pump
has been operating in the normal zone, albeit at a number of off-peak flows.
To predict system performance in regions of negative rotation and/or negative flow the
analyst requires characteristics in these regions for the machine in question. Indeed, any
peculiar characteristic of the pump in these regions could be expected to have an influence on the hydraulic transients. It is important to stress that the results of such analyses
are critically governed by the following three factors: (1) availability of complete pump
characteristics in zones the pump will operate, (2) complete reliance on dynamic similitude (homologous) laws during transients, and (3) assumption that steady-flow derived
pumpcharacteristics are valid for transient analysis.

Head (h), Flow (v), and Speed (a)

Pumping

Turbine (III)
Dissipation (IV)

FIGURE 12.5 Simulated pump trip without valves in a single-pipeline system.

In vestigations by Kittredge (1956) and Knapp (1937) facilitated the understanding of

abnormal operation, as well as served to reinforce the need for test data. Following the
work by Knapp (1941) and Swanson (1953), and a summary of their results by Donsky
(1961), eight possible zones of operation, four normal and four abnormal, will be discussed here with reference to Fig. 12.6, developed by Martin (1983). In Fig. 12.6 the
head H is shown as the difference in the two reservoir elevations to simplify the illustration. The effect of pipe friction may be ignored for this discussion by assuming that the
pipe is short and of relatively large diameter. The regions referred to on Fig. 12.6 are
termed zones and quadrants, the latter definition originating from plots of lines of constant head and constant torque on a flow-speed plane (v a axes). Quadrants I (v > O,
a > O) and III (v < O, a < O) are defined in general as regions of pump or turbine operation, respectively. It will be seen, however, that abnormal operation (neither pump nor
turbine mode) may occur in either of these two quadrants. A very detailed description of
each of the eight zones of operation is in order. It should be noted that all of the conditions shown schematically in Fig. 12.6 can be contrived in a laboratory test loop using
an additional pump (or two) as the master and the test pump as a slave. Most, if not all,
of the zones shown can also be experienced by a pump during a transient under the
appropriate set of circumstances.
Quadrant L Zone A (normal pumping) in Fig. 12.6 depicts a pump under normal operation for which all four quantities Q, N, H, and T are regarded as positive. In this case
Q > O, indicating useful application of energy. Zone B (energy dissipation) is a condition
of positive flow, positive rotation, and positive torque, but negative headquite an abnormal condition. A machine could operate in Zone B by (1) being overpowered by another

Zone A. Normal Pumping (I)

Zone C. Reverse Turbine (I)

Zone E. Reverse Rotation Pumping

Radial-Flow Machine (II)

Zone F. Energy Dissipation (III)

Zone B. Energy Dissipation (I)

Zone D. Energy Dissipation (II)

Zone E. Reverse Rotation Pumping

. Mixed-or Axial-Flow Machine (III)

Zone G. Normal Turbine (III)

Zone H. Energy Dissipation (IV)

FIGURE 12.6 Four quadrants and eight zones of possible pump operation. (From Martin,
1983)
pump or by a reservoir during steady operation, or (2) by a sudden drop in head during a
transient caused by power failure. It is possible, but not desirable, for a pump to generate
power with both the flow and rotation in the normal positive direction for a pump, Zone
C (reverse turbine), whichis caused by a negative head, resulting in a positive efficiency
because of the negative torque. The maximum efficiency would be quite low due to the
bad entrance flow condition and unusual exit velocity triangle.
Quadrant IV Zone H, labeled energy dissipation, is often encountered shortly after a
tripout or power failure of a pump, as illustrated in Fig. 12.5. In this instance the combined inertia of all the rotating elementsmotor, pump and its entrained liquid, and
shafthas maintained pump rotation positive but at a reduced value at the time of flow

reversal caused by the positive head on the machine. This purely dissipative mode results
in a negative or zero efficiency. It is important to note that both the head and fluid torque
are positive in Zone H, the only zone in Quadrant IV.
Quadrant III. A machine that passes through Zone H during a pump power failure will
then enter Zone G (normal turbining) provided that reverse shaft rotation is not precluded by a mechanical ratchet. Although a runaway machine rotating freely is not generating
power, Zone G is the precise mode of operation for a hydraulic turbine. Note that the head
and torque are positive, as for a pump but that the flow and speed are negative, opposite
to that for a pump under normal operation (Zone A).
Subsequent to the tripout or load rejection of a hydraulic turbine or the continual operation of a machine that failed earlier as a pump, Zone F (energy dissipation) can be
encountered. The difference between Zones F and G is that the torque has changed sign
for Zone F, resulting in a braking effect, which tends to slow the free=wheeling machine
down. In fact the real runaway condition is attained at the boundary of the two zones, for
which torque J = O .
Quadrant IL The two remaining zonesD and Eare very unusual and infrequently encountered in operation, with the exception of pump/turbines entering Zone E during
transient operation. Again it should be emphasized that both zones can be experienced by
a pump in a test loop, or in practice in the event a machine is inadvertently rotated in the
wrong direction by improper wiring of an electric motor. Zone D is a purely dissipative
mode that normally would not occur in practice unless a pump, which was designed to
increase the flow from a higher to lower reservoir, was rotated in reverse, but did not have
the capacity to reverse the flow (Zone E, mixed or axial flow), resulting in Q > O, TV < O,
T < O, for H < O. Zone E, for which the pump efficiency > O, could occur in practice
under steady flow if the preferred rotation as a pump was reversed. There is always the
question regarding the eventual direction of the flow. A radial-flow machine will produce
positive flow at a much reduced capacity and efficiency compared to TV > O (normal
pumping), yielding of course H > O. On the other hand, mixed and axial-flow machines
create flow in the opposite direction (Quadrant III), and H < O, which corresponds still to
an increase in head across the machine in the direction of flow.

12.4.4 Representation of Pump Data for Numerical Analysis

It is conventional in transient analyses to represent /z/oc2 and (3/oc2 as functions of v/cc, as
shown in Fig. 12.7 and 12.8 for a radial-flow pump. The curves on Fig. 12.7 are only for
positive rotation (a > O), and constitute pump Zones A, B, and C for v > O and the region
of energy dissipation subsequent to pump power failure (Zone H), for which v < O. The
remainder of the pump characteris-tics are plotted in Fig. 12.8 for a < O. The complete
characteristics of the pump plotted in Figs. 12.7 and 12.8 can also be correlated on what
is known as a Karman-Knapp circle diagram, a plot of lines of constant head (h) and
torque (P) on the coordinates of dimensionless flow (v) and speed (a). Fig. 12.9 is such a
correlation for the same pump. The complete characteristics of the pump require six
curves, three each for head and torque. For example, the hia? curves from Figs. 12.7 and
12.8 can be represented by continuous lines for h = 1 and h = 1, and two straight lines
through the origin for h = O. A similar pattern exists for the torque (p) lines. In addition
to the eight zones A-H illustrated in Fig. 12.6, the four Karman-Knapp quadrants in terms
of v and, are well defined. Radial lines in Fig. 12.9 correspond to constant values for v/oc
in Figs. 12.7 and 12.8, allowing for relatively easy transformation from one form of presentation to the other.

Karman-Knapp Quadrant

Homologous Head and Torque Characteristics for Radial-Flow Pump

(ft, - 0.465 in Univenai Units) for Postive Rotation (>o)
FIGURE 12.7 Complete head and torque characteristics of a radial-flow pump
for positive rotation. (From Martin, 1983)

Karman-Knapp Quadrant

Homologous Head and Torque Characteristics for Radial-Flow Pump

(fts 0.465 in Univenai Units) for Negative Rotation (a<o)
FIGURE 12.8 Complete head and torque characteristics of a radialflow pump for negative rotation. (From Martin, 1983)

In computer analysis of pump transients, Figs. 12.7 and 12.8, while meaningful from
the standpoint of physical understanding, are fraught with the difficulty of Iv/ocl becoming
infinite as the unit passes through, or remains at, zero speed (a = O). Some have solved
that problem by switching from /i/oc2 versus via to h/v2 versus o/v, and likewise for (3, for
Iv/ocl > 1. This technique doubles the number of curves on Figs. 12.7 and 12.8, and thereby creates discontinuities in the slopes of the lines at Iv/ocl = 1, in addition to complicating the storing and interpolation of data. Marchal et al. (1965) devised a useful transformation which allowed the complete pump characteris-tics to be represented by two single
curves, as shown for the same pump in Fig. 12.10. The difficulty of via becoming infinite
was eliminated by utilizing the function tan^1 (v/oc) as the abscissa. The eight zones, or
four quadrants can then be connected by the continuous functions. Although some of the
physical interpretation of pump data has been lost in the transformation, Fig. 12.10 is now
a preferred correlation for transient analysis using a digital computer because of function
continuity and ease of numerical interpolation. The singularities in Figs. 12.7 and 12.8 and
the asymptotes in Fig. 12.9 have now been avoided.
12.4.5 Critical Data Required for Hydraulic Analysis
of Systems with Pumps
Regarding data from manufacturers such as pump curves (normal and abnormal), pump and
motor inertia, motor torque-speed curves, and valve curves, probably the most critical for
pumping stations are pump-motor inertia and valve closure time. Normal pump curves are

Dimensionless Flow V=Q/O _

KARNIAN-KNAPP
QUADRANT
I

Dimensionless Speed <* = N/NR

Karman-Knapp Circle Diagram for Radial-Flow Pump (ft, = 0.465 in Universal Units)
FIGURE 12.9 Complete four-quadrant head and torque characteristics
of radial-flow pump. (From Martin, 1983)

Zone

Karman - Knapp
Quadrant

FIGURE 12.10 Complete head and torque characteristics of a radial-flow

pump in Suter diagram. (From Martin, 1983)
usually available and adequate. Motor torque-speed curves are only needed when evaluating pump startup. For pump trip the inertia of the combined pump and motor is important.

12.5

SURGE PROTECTION AND SURGE CONTROL DEVICES

There are numerous techniques for controlling transients and waterhammer, some
involving design considerations and others the consideration of surge protection
devices. There must be a complete design and operational strategy devised to combat
potential waterhammer in a system. The transient event may either initiate a low-pressure event (downsurge) as in the case of a pump power failure, or a high pressure event
(upsurge) caused by the closure of a downstream valve. It is well known that a downsurge can lead to the undesirable occurrence of water-column separation, which itself
can result in severe pressure rises following the collapse of a vapor cavity. In some systems negative pressures are not even allowed because of (1) possible pipe collapse or
(2) ingress of outside water or air.
The means of controlling the transient will in general vary, depending upon whether
the initiating event results in an upsurge or downsurge. For pumping plants the major
cause of unwanted transients is typically the complete outage of pumps due to loss of electricity to the motor. For full pipelines, pump startup, usually against a closed pump discharge valve for centrifugal pumps, does not normally result in significant pressure transients. The majority of transient problems in pumping installations are associated with the
potential (or realized) occurrence of water-column separation and vapor-pocket collapse,
resulting from the tripout of one or more pumps, with or without valve action. The pumpdischarge valve, if actuated too suddenly, can even aggravate the downsurge problem. To

Steady-State HGL

Compressed Gas

Air Chamber
Steady-State HGL

Simple Surge Tank

Steady-State HGL

Surge Relief Valve or

Surge Anticipator

Dump
Steady-State HGL

One-Way Surge Tank

FIGURE 12.11 Schematic of various surge protection devices for pumping installations

combat the downsurge problem there are a number of options, mostly involving the design
and installation of one or more surge protection devices. In this section various surge protection techniques will be discussed, followed by an assessment of the virtue of each with
respect to pumping systems in general. The lift systems shown in Fig. 12.11 depict various surge protection schemes.
12.5.1 Critical Parameters for Transients
Before discussing surge protection devices, some comments will be made regarding the
various pipeline, pump and motor, control valve, flow rate, and other parameters that
affect the magnitude of the transient. For a pumping system the four main parameters are
(1) pump flow rate, (2) pump and motor WR2, (3) any valve motion, and (4) pipeline characteristics. The pipeline characteristics include piping layoutboth plan and profile
pipe size and material, and the acoustic velocity. So-called short systems respond differently than long systems. Likewise, valve motion and its effect, whether controlled valves
or check valves, will have different effects on the two types of systems.
The pipeline characteristicsitem number (4)relate to the response of the system to
a transient such as pump power failure. Clearly, the response will be altered by the addition of one or more surge protection device or the change of (1) the flow rate, or (2) the
WR2, or (3) the valve motion. Obviously, for a given pipe network and flow distribution
there are limited means of controlling transients by (2) WR2 and (3) valve actuation. If
these two parameters can not alleviate the problem than the pipeline response needs to be
altered by means of surge protection devices.

HGL

Orifice

Simple

Simple

One-Way

Check Valve
Check
valve

Orifice

Air Chamber

Accumulator

Vacuum Breaker

FIGURE 12.12 Cross sectional view of surge tanks and gas-n related surge protection devices

a Vacuum Breaker Valve

a. Air Release Valve

c. Surge Reliefer Surge Anticipator Valve

FIGURE 12.13 Cross sections of vacuum breaker, air release and surge relief valves.

12.5.2 Critique of Surge Protection

For pumping systems, downsurge problems have been solved by various combinations of
the procedures and devices mentioned above. Details of typical surge protection devices
are illustrated in Figs. 12.12 and 12.13. In many instances local conditions and preferences of engineers have dictated the choice of methods and/or devices. Online devices
such as accumulators and simple surge tanks are quite effective, albeit expensive, solutions. One-way surge tanks can also be effective when judiciously sized and sited. Surge
anticipation valves should not be used when there is already a negative pressure problem.
Indeed, there are installations where surge anticipation functions of such valves have been
deactivated, leaving only the surge relief feature. Moreover, there have been occasions for
which the surge anticipation feature aggravated the low pressure situation by an additional downsurge caused by premature opening of the valve.
Regarding the consideration and ultimate choice of surge protection devices, subsequent to calibration of analysis with test results, evaluation should be given to simple
surge tanks or standpipes, one-way surge tanks, and hydropneumatic tanks or air chambers. A combination of devices may prove to be the most desirable and most economical.
The admittance of air into a piping system can be effective, but the design of air vacuum-valve location and size is critical. If air may be permitted into pipelines careful
analysis would have to be done to ensure effective results. The consideration of air-vacuum breakers is a moot point if specifications such as the Ten State Standards limit the
pressures to positive values.

12.5.3 Surge Protection Control and Devices

Pump discharge valve operation. In gravity systems the upsurge transient can be controlled by an optimum valve closureperhaps two stage, as mentioned by Wylie and
Streeter (1993). As shown by Fleming (1990), an optimized closing can solve a waterhammer problem caused by pump power failure if coupled with the selection of a surge
protection device. For pump power failure a control valve on the pump discharge can
often be of only limited value in controlling the downsurge, as mentioned by Sanks
(1989). Indeed, the valve closure can be too sudden, aggravating the downsurge and
potentially causing column separation, or too slow, allowing a substantial reverse flow
through the pump. It should also be emphasized that an optimum controlled motion for
single-pump power failure is most likely not optimum for multiple-pump failure.
The use of microprocessors and servomechanisms with feedback systems can be a general solution to optimum control of valves in conjunction with the pump and pipe
system. For pump discharge valves the closure should not be too quick to exacerbate
downsurge, nor too slow to create a substantial flow back through the valve and pump
before closure.
Check valves. Swing check valves or other designs are frequently employed in pump
discharge lines, often in conjunction with slow acting control valves. As indicated by
Tullis (1989), a check valve should open easily, have a low head loss for normal positive
flow, and create no undesirable transients by its own action. For short systems, a
slow=responding check valve can lead to waterhammer because of the high reverse flow
generated before closure. A spring or counterweight loaded valve with a dashpot can (1)
give the initial fast response followed by (2) slow closure to alleviate the unwanted transient. The proper selection of the load and the degree of damping is important, however,
for proper performance.

Check valve slam is also a possibility from stoppage or failure of one pump of several in a parallel system, or resulting from the action of an air chamber close to a pump
undergoing power failure. Check valve slam can be reduced by the proper selection of a
dashpot.
Surge anticipator valves and surge relief valves. A surge anticipation valve, Fig.
12.13c frequently installed at the manifold of the pump station, is designed to open initially under (1) pump power failure, or (2) the sensing of underpressure, or (3) the sensing of overpressure, as described by Lescovitch (1967). On the other hand, the usual type
of surge relief valve opens quickly on sensing an overpressure, then closes slowly, as controlled by pilot valves. The surge anticipation valve is more complicated than a surge
relief valve in that it not only embodies the relief function at the end of the cycle, but also
has the element of anticipation. For systems for which water-column separation will not
occur, the surge anticipation valve can solve the problem of upsurge at the pump due to
reverse flow or wave reflection, as reported in an example by White (1942). An example
of a surge relief valve only is provided by Weaver (1972). For systems for which watercolumn separation will not occur, Lundgren (1961) provides charts for simple pipeline
systems.
As reported by Parmakian (1968,1982a-b) surge anticipation valves can exacerbate the
downsurge problem inasmuch as the opening of the relief valve aggravates the negative
pressure problem. Incidents have occurred involving the malfunctioning of a surge anticipation valve, leading to extreme pressures because the relief valve did not open.
Pump bypass. In shorter low-head systems a pump bypass line (Fig. 12.11) can be
installed in order to allow water to be drawn into the pump discharge line following power
failure and a downsurge. As explained by Wylie and Streeter (1993), there are two possible bypass configurations. The first involves a control valve on the discharge line and a
check valve on the bypass line between the pump suction or wet well and the main line.
The check valve is designed to open subsequent to the downsurge, possibly alleviating
column separation down the main line. The second geometry would reverse the valve
locations, having a control valve in the bypass and a check valve in the main line downstream of the pump. The control valve would open on power failure, again allowing water
to bypass the pump into the main line.
Open (simple) surge tank. A simple on-line surge tank or standpipe (Fig. 12.11) can
be an excellent solution to both upsurge and downsurge problems, These devices are
quite common in hydroelectric systems where suitable topography usually exists. They
are practically maintenance free, available for immediate response as they are on line.
For pumping installations open simple surge tanks are rare because of height considerations and the absence of high points near most pumping stations. As mentioned by
Parmakian (1968) simple surge tanks are the most dependable of all surge protection
devices. One disadvantage is the additional height to allow for pump shutoff head.
Overflowing and spilling must be considered, as well as the inclusion of some damping
to reduce oscillations. As stated by Kroon et al. (1984) the major drawback to simple
surge tanks is their capital expense.
One-way surge tank. The purpose of a one-way surge tank is to prevent initial low
pressures and potential water-column separation by admitting water into the pipeline subsequent to a downsurge. The tank is normally isolated from the pipeline by one or more
lateral pipes in which there ok one or more check valves to allow flow into the pipe if the
HGL is lower in the pipe than the elevation of the water in the open tank. Under normal
operating conditions the higher pressure in the pipeline keeps the check valve closed. The

major advantage of a one-way surge tank over a simple surge tank is that it does not have
to be at the HGL elevation as required by the latter. It has the disadvantage, however, on
only combatting initial downsurges, and not initial upsurges. One-way surge tanks have
been employed extensively by the U.S. Bureau of Reclamation in pump discharge lines,
principally by the instigation of Parmakian (1968), the originator of the concept. Another
example of the effective application of one-way surge tanks in a pumping system was
reported by Martin (1992), to be discussed in section 12.9.1.
Considerations for design are: (1) location of high points or knees of the piping, (2)
check valve and lateral piping redundancy, (3) float control refilling valves and water supply, and other appurtenances. Maintenance is critical to ensure the operation of the check
valve(s) and tank when needed.
Air chamber (hydropneumatic surge tank). If properly designed and maintained, an
air chamber can alleviate both negative and positive pressure problems in pumping systems. They are normally located within or near the pumping station where they would
have the greatest effect. As stated by Fox (1977) and others, an air chamber solution may
be extremely effective in solving the transient problem, but highly expensive. Air chambers have the advantage that the tank-sometimes multiple-can be mounted either vertically or horizontally. The principal criteria are available water volume and air volume for
the task at hand.
For design, consideration must be given to compressed air supply, water level sensing,
sight glass, drains, pressure regulators, and possible freezing. Frequently, a check valve is
installed between the pump and the air chamber. Since the line length between the pump
and air chamber is usually quite short, check valve slamming may occur, necessitating the
consideration of a dashpot on the check valve to cushion closure.
The assurance of the maintenance of air in the tank is essentialusually 50 percent of
tank volume, otherwise the air chamber can be ineffective. An incident occurred at a raw
water pumping plant where an air chamber became waterlogged due to the malfunctioning of the compressed air system. Unfortunately, pump power failure occurred at the same
time, causing water column separation and waterhammer, leading to pipe rupture.
Air vacuum and air release valves. Another method for preventing subatmospheric
pressures and vapor cavity formation is the admittance of air from airvacuum valves
(vacuum breakers) at selected points along the piping system. Proper location and size
of airvacuum valves can prevent water-column separation and reduce waterhammer
effects, as calculated and measured by Martin (1980). The sizing and location of the
valves are critical, as stated by Kroon et al. (1984). In fact, as reported by Parmakian
(1982a,b) the inclusion of air-vacuum valves in a pipeline did not eliminate failures.
Unless the air-vacuum system is properly chosen, substantial pressures can still occur
due to the compression of the air during resurge, especially if the air is at extremely low
pressures within the pipeline when admitted. Moreover, the air must be admitted quickly enough to be effective. Typical designs are shown in Fig. 12.13
As shown by Fleming (1990) vacuum breakers can be a viable solution. The advantage of an air-vacuum breaker system, which is typically less expensive than other measures such as air chambers, must be weighed against the disadvantages of air accumulation along the pipeline and its subsequent removal. Maintenance and operation of valves
is critical in order for assurance of valve opening when needed. Air removal is often
accomplished with a combined airrelease air-vacuum valve. For finished water systems
the admittance of air is not a normal solution and must be evaluated carefully. Moreover,
air must be carefully released so that no additional transient is created.

Flywheel Theoretically, a substantial increase in the rotating inertia (WT?2) of a pumpmotor unit can greatly reduce the downsurge inasmuch as the machine will not decelerate
as rapidly. Typically, the motor may constitute from 75 to 90 percent of the total WR2.
Additional WR2 by the attachment of a flywheel will reduce the downsurge. As stated by
Parmakian (1968), a 100 percent increase in WR2 by the addition of a flywheel may add
up to 20 percent to the motor cost. He further states that a flywheel solution is only economical in some marginal cases. Flywheels are usually an expensive solution, mainly useful only for short systems. A flywheel has the advantage of practically no maintenance,
but the increased torque requirements for starting must be considered.
Uninterrupted power supply (UPS). The availability of large uninterrupted power supply systems are of potential value in preventing the primary source of waterhammer in
pumping; that is, the generation of low pressures due to pump power failure. For pumping stations with multiple parallel pumps, a UPS system could be devised to maintain one
or more motors while allowing the rest to fail, inasmuch as there is a possibility of maintaining sufficient pressure with the remaining operating pump(s). The solution usually
is expensive, however, with few systems installed.

12.6

DESIGNCONSIDERATIONS

Any surge or hydraulic transient analysis is subject to inaccuracies due to incomplete

information regarding the systems and its components. This is particularly true for a water
distribution system with its complexity, presence of pumps, valves, tanks, and so forth,
and some uncertainty with respect to initial flow distribution. The ultimate question is
how all of the uncertainties combine in the analysis to yield the final solution. There will
be offsetting effects and a variation in accuracy in terms of percentage error throughout
the system. Some of the uncertainties are as follows.
The simplification of a pipe system, in particular a complex network, by the exclusion
of pipes below a certain size and the generation of equivalent pipes surely introduces
some error, as well as the accuracy of the steady-state solution. However, if the major flow
rates are reasonably well known, then deviation for the smaller pipes is probably not too
critical. As mentioned above incomplete pump characteristics, especially during reverse
flow and reverse rotation, introduce calculation errors. Valve characteristics that must be
assumed rather than actual are sources of errors, in particular the response of swing check
valves and pressure reducing valves. The analysis is enhanced if the response of valves
and pumps from recordings can be put in the computer model.
For complex pipe network systems it is difficult to assess uncertainties until much of
the available information is known. Under more ideal conditions that occur with simpler
systems and laboratory experiments, one can expect accuracies when compared to measurement on the order of 5 to 10 percent, sometimes even better. The element of judgment
does enter into accuracy. Indeed, two analyses could even differ by this range because of
different assumptions with respect to wave speeds, pump characteristics, valve motions,
system schematization, and so forth. It is possible to have good analysis and poorer analysis, depending upon experience and expertise of the user of the computer code. This element is quite critical in hydraulic transients. Indeed, there can be quite different results
using the same code.
Computer codes, which are normally based on the method of characteristics (MOC),
are invaluable tools for assessing the response based of systems to changes in surge protection devices and their characteristics. Obviously, the efficacy of such an approach is

enhanced if the input data and network schematization is improved via calibration.
Computer codes have the advantage of investigating a number of options as well as optimizing the sizing of surge protection devices. The ability to calibrate a numerical analysis code to a system certainly improves the determination of the proper surge protection.
Otherwise, if the code does not reasonably well represent a system, surge protection
devices can either be inappropriate or under- or oversized.
Computer codes that do not properly model the formation of vapor pockets and subsequent collapse can cause considerable errors. Moreover, there is also uncertainty regarding any free or evolved gas coming out of solution. The effect on wave speed is known,
but this influence can not be easily addressed in an analysis of the system. It is simply
another possible uncertainty.
Even for complicated systems such as water distribution networks, hydraulic transient calculations can yield reasonable results when compared to actual measurements
provided that the entire system can be properly characterized. In addition to the pump,
motor, and valve characteristics there has to be sufficient knowledge regarding the piping and flow demands. An especially critical factor for a network is the schematization
of the network; that is, how is a network of thousands of pipes simplified to one suitable for computer analysis, say hundreds of pipes, some actual and some equivalent.
According to Thorley (1991), a network with loops tends to be more forgiving regarding waterhammer because of the dispersive effect of many pipes and the associated
reflections. On the other hand, Karney and Mclnnis (1990) show by a simple example
that wave superposition can cause amplification of transients. Since water distribution
networks themselves have not been known to be prone to waterhammer as a rule, there
is meager information as to simplification and means of establishing equivalent pipes
for analysis purposes. Large municipal pipe networks are good examples wherein the
schematization and the selection of pipes characterizing the networks need to be
improved in orde to represent the system better.
72.7 NEGATIVEPRESSURESANDWATER
COLUMN SEPARATION IN NETWORKS
For finished water transmission and distribution systems the application of 138 kPa (20
psig) as a minimum pressure to be maintained under all conditions should prevent column
separation from occurring provided analytical models have sufficient accuracy. Although
water column separation and collapse is not common in large networks, it does not mean
that the event is not possible. The modeling of water column separation is clearly difficult
for a complicated network system. Water column separation has been analytically modeled with moderate success for numerous operating pipelines. Clearly, not only negative
pressures, but also water column separation, are unwanted in pipeline systems, and should
be eliminated by installation of properly designed surge protection devices.
If the criterion of a minimum pressure of 138 kPa (20 psig) is imposed then the issue
of column separation and air-vacuum breakers are irrelevant, except for prediction by
computer codes. Aside from research considerations, column separation is simulated for
engineering situations mainly to assess the potential consequences. If the consequences
are serious, as they often are in general, either operational changes or more likely surge
protection devices are designed to alleviate column separation. For marginal cases of column separation the accuracy of pressure prediction becomes difficult. If column separation is not to be allowed and the occurrence of vapor pressure can be adequately predicted, then the simulation of column separation itself is not necessary.

Some codes do not simulate water column separation, but instead only maintain the
pressure at cavity location at vapor pressure. The results of such an analysis are invalid,
if indeed an actual cavity occurred, at some time subsequent to cavity formation. This
technique is only useful to know if a cavity could have occurred, as there can be no assessment of the consequences of column separation. The inability of any code to model water
column separation has the following implications: (1) the seriousness of any column separation event, if any, can not be determined, and (2) once vapor pressure is attained, the
computation model loses its ability to predict adequately system transients. If negative
pressures below 138 kPa (20 psig) are not to be allowed the inability of a code to assess
the consequences of column separation and its attendant collapse is admittedly not so serious. The code need only flag pressures below 138 kPa (20) psig and negative pressures,
indicating if there is a need for surge protection devices.
The ability of any model to properly simulate water column separation depends upon
a number of factors. The principal ones are
Accurate knowledge of initial flow rates
Proper representation of pumps, valves, and piping system
A vapor pocket allowed to form, grow, and collapse
Maintenance of vapor pressure within cavity while it exists
Determination of volume of cavity at each time step
Collapse of cavity at the instant the cavity volume is reduced to zero

12.8

TIME CONSTANTS FOR HYDRAULIC SYSTEMS

Elastic time constant

te =

(12.16)

'/=5f.

<12-17>

Flow time constant

Pump and motor inertia time constant

_ /COR _
/|
~~R~ piaftiw

tm

(1118)

Surge tank oscillation inelastic time constant

'.= 2 ^ JWT

(m9)

12.9

CASESTUDIES

For three large water pumping systems with various surge protection devices waterhammer analyses and site measurements have been conducted. The surge protection systems in question are (1) one-way and simple surge tanks, (2) an air chamber, and (3) airvacuum breakers.
12.9.1

Case Study with One-way and Simple Surge Tanks

A very large pumping station has been installed and commissioned to deliver water
over a distance of over 30 kilometers. Three three-stage centrifugal pumps run at a synchronous speed of 720 rpm, with individual rated capacities of 1.14 mVsec, rated heads of
165 m, and rated power of 2090 kw. Initial surge analysis indicated potential water-column separation. The surge protection system was then designed with one-way and simple
surge tanks as well as air-vacuum valves strategically located.
The efficacy of these various surge protection devices was assessed from site measurements. Measurements of pump speed, discharge valve position, pump flow rate, and
pressure at seven locations were conducted under various transient test conditions. The
site measurements under three-pump operation allowed for improvement of hydraulic
transient calculations for future expansion to four and five pumps. Figure 12.14 illustrates
the profile of the ground and the location of the three pairs of surge tanks. The first and
second pair of surge tanks are of the one-way (feed tank) variety, while the third pair are
simple open on-line tanks.

P r o f i l e of Ground and HGL along Pipe

Simple T

E l e v a t i o n in m

Qne-Way Tanks
Ohe-Way Tanks

D i s t a n c e along Pipe in km
FIGURE 12.14 Case study of pump power failure at pumping station with three pair of surge tanks
two pair one way and one pair simple surge tanks Martin(1992).

Pressure in meters of Water Column

Recorded and Predicted Pressures at Pump Manifold

Three Pump Trip
Predicted

Recorded

Time in seconds

Shaft Speed in rpm

Recorded and Predicted Pump Speed

Three Pump Trip

Predicted

Recorded

Time in seconds
Figure 12.15. The test program and transient analysis clearly indicated that the piping system was adequately protected by the array of surge tanks inasmuch as there were no negative pressures.

Initial Hydraulic Grade Line

Minimum HGL Envelope

Ground and Pipe Profile

38 ft Sphere
Air Chamber
Distance in Feet

Pressure at Pump Manifold in psi

Power Failure of Three Pumps with Air Chamber on Line

Measured
Calculated

Time in Seconds
FIGURE 12.16 Case study of air chamber performance for raw water supply.

Pump trip tests were conducted for three-pump operation with cone valves actuated by
the loss of motor power. For numerical analysis a standard computer program applying
the method of characteristics was employed to simulate the transient events. Figure 12.15
shows the transient pressures for three pump power failure. The transient pressures agree
reasonably well for the first 80 seconds. The minimum HGL's in Fig. 12.14 also show
good agreement, as well as the comparison of measured and calculated pump speeds in
12.9.2

Case Study with Air chamber

Hydraulic transients caused by simultaneous tripping of pumps at the pumping station

depicted on Fig. 12.16 were evaluated to assess the necessity of surge protection. Without
the presence of any protective devices such as accumulators, vacuum breakers, or surge
suppressors, water hammer with serious consequences was shown to occur due to depressurization caused by the loss of pumping pressure following sudden electrical outage. In
the case of no protection a large vapor cavity would occur at the first high point above
the pumping station, subsequently collapsing after the water column between it and the
reservoir stops and reverses. This phenomenon, called water-column separation, can be
mitigated by maintaining the pressures above vapor pressure.
The efficacy of the 11.6 m (38 ft) diameter air chamber shown in Fig. 12.16 was investigated analytically and validated by site measurements for three-pump operation. The
envelope of the minimum HGL drawn on Fig. 12.16 shows that all pressures remained
positive. The lower graph compares the site measurement with the calculated pressures
obtained by a standard waterhammer program utilizing MOC.
12.9.3

Case Study with Air-vacuum Breaker

Air-inlet valves or air-vacuum breakers are frequently installed on liquid piping systems
and cooling water circuits for the purpose of (1) eliminating the potential of water-column
separation and any associated waterhammer subsequent to vapor pocket collapse; (2) protecting the piping from an external pressure of nearly a complete vacuum; and (3) providing an elastic cushion to absorb the transient pressures.
A schematic of the pumping and piping system subject to the field test program is
shown in Fig. 12.17. This system provides the cooling water to a power plant by pumping water from the lower level to the upper reservoir level. There are five identical vertical pumps in parallel connected to a steel discharge pipe 1524 mm (60 in) in diameter. On
the discharge piping of each pump there are 460 mm (18 in) diameter swing check valves.
Mounted on top of the 1524 mm (60 in) diameter discharge manifold is a 200 mm (8 in)
diameter pipe, in which is installed a swing check valve with a counter weight. Air enters
the vacuum breaker through the tall riser, which extends to the outside of the pump house.
Transient pressures were measured in the discharge header for simultaneous tripout of
three, four, and five pumps. The initial prediction of the downsurge caused by pump
power failure was based on the method of characteristics with a left end boundary condition at the pumps, junction boundary condition at the change in diameter of the piping,
and a constant pressure boundary condition at the right end of the system.
The predicted pressure head variation in the pump discharge line is shown in Fig.
12.17 for a simulated five pump tripout. The predicted peak pressure for the five pump

ELEVATION IN METERS

INITIAL HYDRAULIC GRADE LINE

FREE
DISCHARGE

VACUM
DR BAKER
MVT
PUMPS

DISTANCE IN METERS
Figure 12.17. Case Study of Vacuum Breaker Performance for River Water System of Nuclear Plant,
Martin (1980).
tripout compares favorably with the corresponding measured peak, but the time of occurrence of the peaks and the subsequent phasing vary considerably. Analysis without a vacuum breaker or other protective device in the system predicted waterhammer pressure
caused by collapse of a vapor pocket to exceed 2450 kPa (355 psi). The vacuum breaker
effectively reduced the peak pressure by 60 per cent. Peak pressures can be adequately
predicted by a simplified liquid column, orifice, and air spring system. Water-column separation can be eliminated by air-vacuum breakers of adequate size.

REFERENCES
Chaudhry, M. H., Applied Hydraulic Transients, 2d ed., Van Nostrand Reinhold, New Yorle 1987.
Donsky, B., "Complete Pump Characteristics and the Effects of Specific Speeds on Hydraulic
Transients," Journal of Basic Engineering, Transactions, American Society of Mechanical
Engineers, 83: 685-699, 1961.
Fleming, A. J., "Cost-Effective Solution to a Waterhammer Problem," Public Works, 4244, 1990.
Fox, J. A., Hydraulic Analysis of Unsteady Flow in Pipe Networks, John Wiley & Sons, New Yorle
1977.
Karney, B. W., and Mclnnis, D., "Transient Analysis of Water Distribution Systems," Journal
American Water Works Assoriation, 82: 62-70, 1990.
Kittredge, C. P., "Hydraulic Transients in Centrifugal Pump Systems," Transactions, American
Society of Mechanical Engineers, 78: 1307-1322, 1956.
Knapp, R. T, "Complete Characteristics of Centrifugal Pumps and Their Use in Prediction of
Transient Behavior," Transactions, American Society of Mechanical Engineers , 59:683-689,
1937.
Knapp, R. T, "Centrifugal-Pump Performance Affected by Design Features," Transactions,
American Societiy of Mecharnal Engineers, 63:251-260, 1941.
Kroon, J. R., Stoner, M. A., and Hunt, W. A., "Water Hammer: Causes and Effects," Journal
American Water Works Assoriation, 76:39-45, 1984.
Lescovitch, J. E., "Surge Control of Waterhammer by Automatic Valves," Journal American Water
Works Assoriation, 59:632-644, 1967.

Lundgren, C. W., "Charts for Determining Size of Surge Suppressor for Pump-Discharge Lines,"
Journal of Engineering for Power, Transactions, American Society of Mechanical Engineers,
93:43^7, 1961.
Marchal, M., Flesh, G., and Suter, P., "The Calculation of Waterhammer Problems by Means of the
Digital Computer," Proceedings, International Symposium on Waterhammer in Pumped Storage
Projects, American Society of Mechanical Engineers (ASME), Chicago, 1965.
Martin, C. S., "Entrapped Air in Pipelines," Paper F2 Second BHRA International Conference on
Pressure Surges, The City University, London, September 22-24, 1976.
Martin, C. S., "Transient Performance of Air Vacuum Breakers," Fourth International Conference
on Water Column Separation, Cagliari, November 11-13, 1979. "Transient Performance Air
Vacuum Breakere," L'Energia Elettrica, Proceedings No. 382, 1980, pp. 174-184.
Martin, C. S., "Representation of Pump Characteristics for Transient Analysis," ASME
Symposium on Performance Characteristics of Hydraulic Turbines and Pumps, Winter Annual
Meeting, Boston, November 13-18, pp. 1-13,1983.
Martin, C. S., "Experience with Surge Protection Devices," BHr Group International Conference
on Pipelines, Manchester, England, March pp. 171-178 24-26, 1992.
Martin, C. S., "Hydraulics of Valves," in J. A. Schetz and A. E. Fuhs, eds. Handbook of Fluid
Dynamics and Fluid Machinery, Vol. Ill, McGraw-Hill, pp. 2043-2064. 1996,
Parmakian, J., Water Hammer Analysis, Prentice-Hall, New York, 1955.
Parmakian, J., "Unusual Aspects of Hydraulic Transients in Pumping Plants," Journal of the Boston
Society of Civil Engineers, 55:30^7, 1968.
Parmakian, J., "Surge Control," in M. H. Chaudhry, ed., Proceedings, Unsteady Flow in Conduits,
Colorado State University, pp. 193-207 1982,
Parmakian, J., "Incidents, Accidents and Failures Due to Pressure Surges," in M. H. Chaudhry ed.,
Proceedings, Unsteady Flow in Conduits, Colorado State University, pp. 301-311 1982.
Sanks, R. L., Pumping Station Design, Butterworths, Bestar, 1989.
Stepanoff, A. L, Centrifugal and Axial Flow Pumps, John Wiley & Sons, New York, 1957.
Swanson, WM., "Complete Characteristic Circle Diagrams for Turbomachinery," Transactions,
American Society of Mechanical Engineers, 75:819-826, 1953.
Thorley, A. R. D., Fluid Transients in Pipeline Systems, D. & L. George Ltd., 1991.
Tullis, J. P., Hydraulics of Pipelines, John Wiley & Sons, New Yorhl989.
Watters, G. Z., Modern Analysis and Control of Unsteady Flow in Pipelines, Ann Arbor Science,
Ann Arbor, MI, 1980.
Weaver, D. L., "Surge Control," Journal American Water Works Association, 64: 462-466, 1972.
White, I. M., "Application of the Surge Suppressor in Water Systems," Water Works Engineering,
45,304-306, 1942.
Wood, D.J., and Jones, S.E., "Waterhammer Charts for Various Types of Valves", ASCE, Journal of
Hydraulics Division, HYl, 99:167-178, 1973.
Wylie, E. B., and Streeter, V. L., Fluid Transients in Systems, Prentice-Hall, 1993.

CHAPTER 13
HYDRAULIC DESIGN
OF DRAINAGE FOR
HIGHWAYS

G. Kenneth Young, JR.

Stuart M. Stein
GKY and Associates, Inc.
Springfield, VA

13.1

INTRODUCTION

The objective of this chapter is to provide design guidance for highway hydraulic elements: gutters, roadside conveyance, inlets, and bridge scuppers. Proper hydraulic design
of highway drainage is essential to avoid disruption of the highways transportation function, maintain safe travel conditions, and sustain infrastructure.
When rain falls on a sloped pavement surface, it forms a thin film of water that increases in thickness as it flows to the edge of the pavement and concentrates in gutters or roadside ditches; overflow from gutters and ditches spreads out onto the pavement. Factors
that influence the depth of water on the pavement and its spread in gutters are the length
of the flow path, the texture and slope of the surface, and rainfall intensity. As the depth
of water on the pavement or the gutter spread increases, the potential for vehicular
hydroplaning or disruption of the highway's transportation function increases. With
design methods and guidance, the surface drainage elements are sized to function at predetermined reliability thresholds.
The design of highway surface drainage is a critical component of the highway system
(Johnson and Chang, 1984). The hydraulics of cross drainage handled by culverts and
bridges is not considered in this chapter, and subsurface storm drains that accept surface
water removed at inlets also are excluded. This chapter provides information on the following aspects of highway drainage design: highway geometries that influence hydraulic
design, design event selection, design flow estimation, gutter design, ditch hydraulic and
stability design, inlet design, and bridge deck hydraulics.
The reader also is directed to documents of the Federal Highway Administration,
including Hydraulic Engineering Circular No. 12 (HEC-12), "Drainage of Highway
Pavement" (Johnson and Chang, 1984), HEC-15, "Design of Roadside Channels with
Flexible Linings" (Chen and Cotton, 1986), HEC-21, "Design of Bridge Deck
Drainage" (Young et al., 1993), and HEC-22, "Urban Drainage Design Manual"
(Brown et al., 1996). Information on the physics of surface drainage can be found in
Anderson et al. (1995).

73.2 GENERAL GEOMETRIC AND PAVEMENT GUIDELINES

THAT INFLUENCE DRAINAGE
Types of pavement materials and highway geometries can influence drainage. Placement
of drainage structures can have a secondary adverse effect with respect to safety.
13.2.1

Pavement

Sheet flow on the pavement can cause hydroplaning, and pavement materials can influence sheet flow. Smooth pavements can cause water to flow faster and reduce the thickness of the sheet flow film; however, smooth pavements can decrease the coefficient of
friction of pavement and tires and lead to longer skids as well.
As the depth of water flowing over a roadway's surface increases, the potential for
hydroplaning increases. When a rolling tire encounters a film of water on the roadway, the
water is channeled through the tire tread pattern and through the surface roughness of the
pavement. Hydroplaning occurs when the drainage capacity of the tire tread pattern and
the pavement surface is exceeded and the water begins to build up in front of the tire. As
the water builds up, a water wedge is created, producing a hydrodynamic force that can
lift the tire off the pavement surface. This is considered full dynamic hydroplaning, and,
because the water offers little shear resistance, the tire loses its tractive ability and the driver loses control of the vehicle. Hydroplaning is a function of the water depth, roadway
geometries, vehicle speed, treat depth, tire inflation pressure, and conditions of the pavement surface.
The following are several pavement design factors:
1. Design the highway to reduce the drainage path lengths of the water flowing over the
pavement. Crown sections split flow to both sides and are superior to long sloping runs of
pavement from one side to the other. This will prevent build-up of sheet flow thickness.
2. Increase the pavement surface texture depth by such methods as grooving of Portland
cement concrete. An increase of pavement surface texture will increase the drainage
capacity at the tire-pavement interface and reduce hydroplaning.
3. Use open-graded pavements, which have been shown to reduce greatly the
hydroplaning potential of the roadway surface. This reduction is the result of the
water's ability to be forced through the pavement under the tire and to enter the surbase rather than run off. Surface texture also increases the coefficient of tire-to-pavement friction. Such open grading also involves provision of subsurface drainage
details to the pavement design.
4. Use intercepting drainage structures along the roadway to capture the sheet of water
over the pavement, which will reduce the thickness of the film of water and reduce the
hydroplaning potential of the roadway surface. Long slotted inlets perpendicular to the
flow and open expansion joints on bridges can accomplish this.
13.2.2 Grade
The recommended minimum values of roadway longitudinal slope given in the Policy on
Geometric Design of the American Association of State Highway and Transportation
Officials (AASHTO, 1990) provide safe, acceptable pavement drainage. In addition, the
following general guidelines are presented:

1. A minimum longitudinal gradient is more important for a curbed pavement than for an
uncurved pavement because the water is constrained by the curb on the one hand, and
runs off the shoulder, on the other, with crowned cross sections.
2. Desirable gutter grades should not be less than 0.5 percent for curbed pavements, with
an absolute minimum of 0.3 percent. Minimum grades can be maintained in flat terrain by using of a rolling profile or by warping the cross slope to achieve rolling gutter profiles.
3. In sag vertical curves, a minimum grade of 0.3 percent should be maintained within 15
m (50 ft) of the low point of the curve. This is accomplished where the length of the
curve in meters divided by the algebraic difference in grades in percent (K) is equal to
or greater than 50 (156 in English units). This is represented as
K=

G 2 -rG1

(13.1)

where K = vertical curve constant m/percent (ft/percent), L = horizontal length of curve

m (ft), G1 = grade of roadway on one side of point of vertical intersection, percent, G2 =
grade of roadway on other side of point of vertical intersection, percent.

13.2.3 Cross Slope

Table 13.1 indicates an acceptable range of cross slopes (AASHTO, 1990). These cross slopes
are a compromise between the need for reasonably steep cross slopes for drainage and relatively flat cross slopes for driver comfort and safety. These cross slopes represent standard
practice. AASHTO (1990) should be consulted before deviating from these values.
Cross slopes of 2 percent have little effect on driver effort in steering or on friction
demand for vehicle stability (Gallaway et al., 1979). Use of a cross slope steeper than
2 percent on pavements with a central crown line is not desirable; however, in areas of
intense rainfall, a somewhat steeper cross slope (2.5 percent) can be used to facilitate
drainage.
On multilane highways where three or more lanes are sloped in the same direction, it
is desirable to counter the resulting increase in sheet-flow depth by increasing the cross

TABLE 13.1 Normal Pavement Cross Slopes

Surface Type
High-type surface
2 lanes
3 or more lanes, each direction
Intermediate surface
Low-type surface
Shoulders
Bituminous or concrete
With curbs

Range in Rate or Surface Slope

0.015 - 0.020
0.015 minimum, increase 0.005 to 0.010 per
lane, 0.040 maximum
0.015 - 0.030
0.020 - 0.060
0.020 - 0.060
> 0.040

slope of the outermost lanes. The two lanes adjacent to the crown line should be pitched
at the normal slope, and successive lane pairs, or portions thereof outward, should be
increased by approximately 0.5 to 1 percent. The maximum pavement cross slope should
be limited to 4 percent. The following are additional guidelines related to cross slope:
1. Inside lanes can be sloped toward the median if conditions warrant. This is not widely encouraged.
2. Median areas should not be drained across travel lanes.
3. The number and length of flat pavement sections in cross-slope transition areas should
be minimized. Consideration should be given to increasing cross slopes in sag vertical
curves and crest vertical curves and in sections of flat grade.
4. Shoulders should be sloped to drain away from the pavement, except with raised, narrow medians and superelevations.
13.2.4 Safety
The placement of drainage structures in the traveled way can constitute an obstacle to
moving traffic. Median inlets, storm drain inlets, and junctions need to have top elevations that do not project significantly above the surface. Culvert entrances and end walls
need to be situated away from likely trajectories of errant vehicles that have departed
traffic lanes because a driver must maneuver to avoid an accident or is weary.
As a general rule, drainage devices should provide a low silhouette to oncoming traffic.

13.3

DESIGN FREQUENCY AND SPREAD

The most significant design decisions considered in sizing pavement drainage facilities are
the frequency of the design runoff event and the allowable spread of water on the pavement. A related consideration is the use of an event of lesser frequency (greater storm
magnitude) to check the drainage design (AASHTO, 1991; Johnson and Chang, 1984).
Spread and design frequency are not independent. The implications of the use of a criteria for spread of half a traffic lane is considerably different for one design frequency
than for a lesser frequency. Spread also has different implications for a low-traffic, lowspeed highway than for a higher classification highway. Balancing risks and flooding are
central to the issue of highway pavement drainage and are important to highway safety.
13.3.1 Risk Balancing
The objective of highway storm-drainage design is to provide safe passage for vehicles
during the storm event. The design of a drainage system for a section of highway pavement is to collect runoff in the gutter or roadside ditch and convey it to inlets or culverts
in a manner that provides a reasonable degree of safety for traffic and pedestrians at a reasonable cost. As spread from the curb or roadside ditch increases, the risks of traffic accidents and delays and the nuisance and possible hazard to pedestrian traffic increase.
The recurrence interval and allowable spread for design implies acceptable risks of accidents and traffic delays and acceptable costs for the drainage system. Risks associated with
water on traffic lanes are greater with high-volume traffic, high speeds, and higher highway
classifications than with lower-volumes traffic, speeds, and highway classifications.

The major considerations that enter into the selection of design frequency and design
spread are as follows:
1. Functional classification of the highway drives risks: higher functions imply lower
acceptable risks, which translate to low intervals of recurrence with a typical limit of
1 in 100 years. Because ponding on traffic lanes of high-speed, high-volume highways
is contrary to the public's expectations, the risks of accidents and the costs of traffic
delays are high.
2. Design speed is important to the selection of design criteria. (Also, at speeds greater
than 70 km/h (44 mi/h), water on the pavement can cause hydroplaning; in this case,
rain intensity, is more significant than spread.)
3. Projected traffic volumes are an indicator of the economic importance of keeping the
highway open to traffic. The opportunity costs of lost driver and passenger times associated with traffic delays increase rapidly with increasing volumes of traffic.
4. The intensity of rainfall may significantly affect the selection of design frequency and
spread. Risks associated with the spread of water on pavements may be less in arid
areas that are subject to high-intensity thunderstorms than in areas that have frequent
but less intense storms.
5. Capital costs are the other side of the equation. Cost considerations make it necessary to balance the approach to the selection of design criteria to be essentially risk
based. "Tradeoffs" between desirable and practicable criteria are necessary because
of costs.
Other considerations include inconvenience, hazards, and nuisances to pedestrian traffic. These considerations should not be minimized; in some locations, such as commercial
areas, they may assume major importance. Local design practice also may be a major consideration because it can affect the feasibility of designing to higher standards and it influences the public's perception of acceptable practice.
The relative elevation of the highway and surrounding terrain is an additional consideration where water can be drained only through a storm drainage system, as in underpasses and depressed sections. The potential for ponding to hazardous depths should be
considered when selecting the frequency and spread criteria and when checking the design
against storm runoff events of lesser frequency than in the design event.
Spread on traffic lanes can be tolerated to greater widths when traffic volumes and
speeds are low. Spreads of one-half of a traffic lane or more are usually considered to be
a minimum-type design for low-volume local roads.
The selection of design criteria for intermediate types of facilities can be difficult. For
example, some arterials with relatively high traffic volumes and speeds may not have
shoulders that will convey the design runoff without encroaching on the traffic lanes.
13.3.2 Design Guidance Regarding Frequency
and Spread
Table 13.2 provides suggested minimum design frequencies and spread based on the type
of highway and the traffic speed. The recommended design frequency for depressed sections and underpasses where ponded water can be removed only through the storm
drainage system is a 50-yr frequency event. The use of a lesser frequency event, such as
a 100-yr storm, to assess hazards at critical locations where water can pond to appreciable depths is commonly referred to as a check storm or check event.

TABLE 13.2 Suggested Minimum Design Frequency and Spread

Road Classified tion
High volume
Divided
Bi-directional
Collector

Local streets

< 70 km/h (45 mph)

> 70 km/h (45 mph)
Sag point
< 70 km/h (45 mph)
> 70 km/h (45 mph)
Sag point
Low ADT
High ADT
Sag point

Design Frequency

Design Spread

10-yr
10-yr
50-yr
10-yr
10-yr
10-yr
5-yr
10-yr
10-yr

Shoulder + 1 m (3 ft)
Shoulder
Shoulder + 1 m (3 ft)
V2 driving lane
Shoulder
V2 driving lane
V2 driving lane
V2 driving lane
V2 driving lane

13.3.3 Selection of Check Storm and Spread

A check storm should be used any time that runoff could cause unacceptable flooding during less frequent events. In addition, inlets should always be evaluated for a check storm
when a series of inlets terminates at a sag vertical curve where ponding to hazardous
depths could occur.
The frequency selected for the check storm should be based on the same considerations used to select the design storm: i.e., the consequences of spread exceeding that chosen for design and the potential for ponding. Where no significant ponding can occur,
check storms are normally unnecessary. Criteria for spread during the check event are one
lane open to traffic during the check storm event or one lane free of spread during the
check storm event.

13.4

SELECTION OF DESIGN HYDROLOGY

The typical highway drainage area is small, and design flows are linked to design rainfall intensity. Three methods are discussed for the selection of design rainfall intensity:
(1) the rational method that pegs rainfall intensity to a design frequency, (2) the avoidance of the hydroplaning method, and (3) driver vision-impairment method. The first
method uses established drainage policy (such as that implied in Table 13.2), to select
a return period and calculate time of concentration to avoid ponding. The second and
third methods consider vehicle safety directly, either from the standpoint of avoidance
of hydroplaning films or from driver vision being impaired because of heavy rain. All
methods lead to designs for which selected spread is a design requirement. Method 1
involves rainfall return period. Methods 2 and 3 select rainfall intensity on the basis of
physical limits of the vehicle (skid avoidance) or the driver's vision; in these two cases,
the design frequency is imputed by the physical limits rather than selected by the analyst or set by the prevailing policy.
13.4.1 Design Flow Calculation
The commonly used equation for the calculation of peak flow from the small areas associated with highway drainage is a simple mass balance which is considered to be valid for
areas less than 80 ha (200 ac) (ASCE, 1992):

Q = A^
c

where

(13.2)

Q = flow, mVs (ft3/s), C = dimensionless runoff coefficient; outflow volume/rainfall

volume, / = rainfall intensity, mm/h (in/h); to be selected by rational method, avoidance
of hydroplaning method, or driver vision impairment method, A = drainage area ha (ac),
and Kc = units conversion equal to 360 (1 in English units).
Assumptions inherent in Eq. (13.2) are as follows (McCuen, et al., 1996):
Peak flow occurs as the entire area contributes to flow.
Rainfall intensity is uniform over a time duration equal to the time of concentration Tc.
The time of concentration is the time required for water to travel from the most remote
point of the basin to the design element.
Peak flow frequency is the same as that of the rainfall intensity: i.e., the 10-yr intensity of rainfall is assumed to produce the 10-yr peak flow.
The coefficient of runoff is independent of intensity and duration.
Selection of the runoff coefficient C is a function of the ground cover (i.e., land
use). Typical values for C are given in Table 13.3. If the area contains varying types of
land, a composite coefficient is calculated through areal weighting (McCuen et al.,
1996):
I1C1A.
Weighted C = -

(13.3)

?A'

where / is the subscript designating values for different types of land. Highway rights-ofway typically have C > 0.5 because of high pavement coverage.
TABLE 13.3 Runoff Coefficients
Type of Drainage Area
Business
Downtown areas
Neighborhood areas
Residential
Single-family areas
Multi-units, detached
Multi-units, attached
Suburban
Apartment dwelling areas
Industrial
Light areas
Heavy areas
Parks, cemeteries
Playgrounds

Runoff Coefficient C*
0.70 - 0.95
0.50 - 0.70
0.30 - 0.50
0.40 - 0.60
0.60 - 0.75
0.25 - 0.40
0.50 - 0.70
0.50 - 0.80
0.60 - 0.90
0.10 - 0.25
0.20 - 0.40

TABLE 13.3 (Continued)

Type of Drainage Area

Runoff Coefficient C*

Railroad yard areas

Unimproved areas
Lawns
Sandy soil, flat, 2%
Sandy soil, average, 2-1 %
Sandy soil, steep, 7%
Heavy soil, flat, 2%
Heavy soil, average, 2 - 7 %
Heavy soil, steep, 7%
Streets
Asphaltic
Concrete
Brick
Drives and walks
Roofs

0.20 - 0.40
0.10-0.30
0.05 - 0.10
0.10 - 0.15
0.15 - 0.20
0.13 - 0.17
0.18 - 0.22
0.25 - 0.35
0.70 - 0.95
0.80 - 0.95
0.70 - 0.85
0.75 - 0.85
0.75 - 0.95

Source: ASCE, 1960.

*Higher values are usually appropriate for steeply sloped areas and longer return periods because
infiltration and other losses have a proportionally smaller effect on runoff in these cases.
13.4.2 Rainfall Intensity by the Rational Method
The intensity, duration, and frequency (IDF) curves of rainfall are necessary to select the
intensity of rainfall by the rational method. Regional IDF curves are available in most
state highway agency manuals and also are available from the National Oceanic and
Atmospheric Administration and the FHWA HYDRAIN computer program. If IDF curves
are not available, they need to be developed.
The assumption in the rational method is that the time of concentration for a drop of
water to move from the high point to the low point in an area equals the rainfall duration
to realize the maximum flow for a given intensity of rainfall. Thus, the strategy is to calculate the time of concentration as the sum of the travel times of the droplet as it moves
along its path: overland or sheet flow, shallow concentrated flow, gutter or ditch flow, and
pipe flow. The sum of times is accumulated in a downhill direction until the droplet reaches the drainage element to be sized. Thus, time of concentration is calculated as the sum
of the travel times within the various consecutive flow segments. This is called the segment method.
13.4.2.1 Sheet flow travel time. Sheet flow is the shallow mass of runoff on a planar surface
with a uniform depth across the sloping surface. This usually occurs at the headwater of
streams over relatively short distances, rarely more than about 90 m (300 ft) and possibly less
than 25 m (80 ft). Sheet flow is commonly estimated with a version of the kinematic wave
equation, a derivation of Manning's equation, as follows (McCuen et al., 1996):
r=_^_/L>
TS

1"(Vs)

a,4.
(
}

where Ts = sheet flow travel time, min; n = roughness coefficient as given in Table 13.4;
L = flow length, m (ft); 7 = rainfall intensity, mm/h (in/h); S = surface slope, m/m (ft/ft);
and Kc = empirical coefficient equal to 6.943 (0.933 in English units).
Because / depends on Ts and Ts is not initially known, the computation of Ts is an iterative process. An initial estimate of Ts is assumed and used to obtain / from the IDF curve
for the locality. The Ts is then computed from Eq. (13.4) and used to check the initial value
of /. If they are not the same, the process is repeated until two successive Ts estimates are
the same to within a reasonable tolerance (McCuen et al., 1996).
For the following types of segments downhill from the sheet flow, the velocity is calculated and the segment travel time computed. To arrive at the travel time, the segment
length is divided by the velocity.
13.4.2.2 Shallow concentrated flow velocity. After short distances of, at most, 90 m (300
ft), sheet flow tends to concentrate in rills and then gullies of increasing proportions.
Such flow is usually referred to as shallow concentrated flow. The velocity of such flow
can be estimated using a relationship between velocity and slope as follows (McCuen et
al., 1996):
Vc = KckS*

(13.5)

TABLE 13.4 Manning's Roughness Coefficient (n) for Overland Sheet Flow
Surface Description

Smooth asphalt
0.011
Smooth concrete
0.012
Ordinary concrete lining
0.013
Good wood
0.014
Brick with cement mortar
0.014
Vitrified clay
0.015
Cast iron
0.015
Corrugated metal pipe
0.024
Cement rubble surface
0.024
Fallow (no residue)
0.05
Cultivated soils
Residue cover < 20%
0.06
Residue cover > 20%
0.17
Range (natural)
0.13
Grass
Short grass, prairie
0.15
Dense grasses
0.24
Bermuda grass
0.41
Woods*
Light underbrush
0.40
Dense underbrush
0.80
Source Mc Cuen et al 1996*When selecting n, consider cover to a height of about 30 mm. This is the only part of
the plant cover that will obstruct sheet flow.

where Vc = velocity, m/s (ft/s); k = intercept coefficient, as given in Table 13.5; Sp =

slope, percent; and K0 = units conversion equal to 1 (0.3048 in English units).
13.4.2.3 Gutter flow velocity. To find the time of flow in the gutter-flow component of
the time of concentration, a method for estimating the average velocity in a reach of gutter is needed. The time of flow in a triangular channel with uniform inflow per unit of
length can be estimated accurately by using of an average velocity of flow in the gutter.
Integration of the Manning equation for a right triangular channel with respect to time
and distance yields an average velocity for the channel length at the point where spread
is equal to 65 percent of the maximum spread for channels with zero flow at the
upstream end, as discussed in HEC-12, "Drainage of Highway Pavement" (Johnson and
Chang, 1984).
The velocity in a gutter with a curb (shallow cross slope, 1 to 10 percent) is
Vg = G0-5 SQX61 P-67

(13.6)

where Vg = velocity, m/s (ft/s); Kc units conversion equal to 0.752 (1.12 in English
units); G = grade, m/m (ft/ft); Sx = cross slope, m/m (ft/ft); T = spread, m (ft) and, n =
Manning's friction factor (typically taken to be 0.016 for highway gutters).
Thus, if the allowable spread is 5 m, then 0.65 X 5 = 3.25 m of spread is used in Eq. (13.6).
13.4.2.4 Open channel and pipe flow velocity. Flow in gullies empties into channels or
pipes. Open channels are assumed to begin where either the blue streamline shows on the
U.S. Geological Survey quadrangle sheets or the channel is visible on aerial photographs.
Cross-section al geometry and roughness should be obtained for all channel reaches in the
watershed. Manning's equation can be used to estimate average flow velocities in pipes
and open channels as follows:
Vp = #2/3S1/2

(13.7)

where

TABLE 13.5 Intercept Coefficients for Velocity Versus Slope Relationship of Eq. (13.5)
Land Cover/Flow Regime
Forest with heavy ground litter; hay meadow (overland flow)
Trash fallow or minimum tillage cultivation; contour or strip cropped;
woodland (overland flow)
Short grass pasture (overland
flow)
Cultivated straight row (overland
flow)
Nearly bare and untilled (overland flow); alluvial fans in western
mountain regions
Grassed waterway (shallow concentrated
flow)
Unpaved (shallow concentrated
flow)
Paved area (shallow concentrated flow); small upland gullies
Source: McCuen et al., 1996

k
0.076
0.152
0.213
0.274
0.305
0.457
0.491
0.619

n = roughness coefficient as given in Table 13.6; VP = velocity, m/s (ft/s); R = hydraulic

radius (defined as the flow area divided by the wetted perimeter), m (ft); S = slope, m/m
(ft/ft); and Kc = units conversion factor equal to 1 (1.49 in English units).
For a circular pipe flowing full, the hydraulic radius is one-half the radius. For a wide
rectangular channel (w > 10 d), the hydraulic radius is approximately equal to the depth.
13.4.2.5 Combined shallow, gutter, open-channel, and pipe travel time. The travel time
Tc is calculated as follows:

TABLE 13.6 Values of Manning Coefficient (n) for Channels and Pipes
Conduit Material

Mannings n*

Closed conduits
Asbestos-cement pipe
0.0110.015
Brick
0.013-0.017
Cast iron pipe: Cement lined & seal coated
0.011-0.015
Concrete (monolithic)
0.012-0.014
Concrete pipe
0.011-0.015
Corrugated-metal pipe (13 mm X 65 mm
[1/2-in X 2 1/2-in corrugations])
Plain
0.022-0.026
Paved invert
0.018-0.022
Spun asphalt lined
0.011-0.015
Plastic pipe (smooth)
0.011-0.015
Vitrified clay
Pipes
0.011-0.015
Liner plates
0.013-0.017
Open Channels
Lined channels
Asphalt
0.013-0.017
Brick
0.012-0.018
Concrete
0.011-0.020
Rubble or riprap
0.020-0.035
Vegetal
0.030-0.40
Excavated or dredged channels
Earth, straight and uniform
0.020-0.030
Earth, winding, fairly uniform
0.025-0.040
Rock
0.030-0.045
Unmaintained
0.050-0.14
Natural channels (minor streams, top width at flood stage
< 30m [100 ft])
Fairly regular section
0.030.07
Irregular section with pools
0.030.10
Source: Jennings et al., 1994.
*Lower values are usually for well-constructed and maintained (smoother) pipes and channels.

r. = T. +2^

(13.8)

where Tc = time of concentration, min; Ts = sheet flow travel time, min; L1 = flow length
for segment /, m (ft); and V1 = velocity for segment /, m/s (ft/s) (V0, Vg, or Vp.
13.4.2.6 Rainfall intensity as a function of duration and return period. Once Tc is
determined, it is assumed to be equal to the rainfall's duration. With the rational method,
the selected return period, as tabulated in the Table 13.2 guidance, has an associated rainfall intensity-versus-duration curve. This curve is used to establish the rational method
rainfall intensity / for use with Eq. (13.2).
13.4.3 Rainfall Intensity by Avoidance of the
Hydroplaning Method
HEC-21 (Young et al., 1993) developed an alternative method for selecting rainfall intensity that is not dependent on rainfall frequency. (Note that once an intensity is selected,
a frequency is implied.) The method assumes that if drivers have a hydroplaning slip sensation at the design speed (plus a reasonable allowance), they will typically slow down to
90 km/h (55 mph) or less and this behavior governs highway function more than traffic
reaction to gutter flooding.
The method selects values of vehicle speed, tire tread depth, pavement texture, and tire
pressure and calculates the thickness of the sheet flow film at incipient hydroplaning
(assumed at 10 percent spindown: e.g., the tire rolls 1.1 times the circumference to move
one circumference).
The empirical equation for the vehicle speed which initiates hydroplaning is
V = K1SEPM(Kf)M

[(K3TD) + K3]

006

(13.9)

where A = Texas Transportation Institute empirical curve fitting relationship which is the
greater of:
A =
Al

10.409
(K4Cf)OM

+ 3 50?
+5

'^

or
A

> = [I^ ~7'817] K**14

(13 10)

'

where V = vehicle speed in km/h (mph); TD = tire tread depth in mm (1/32 in); TXD =
pavement texture depth, mm (in); d = water film depth, mm (in); P = tire pressure, kPa
(psi); K1 = unit conversion equal to 0.3048 (1 in English units); K2 unit conversion
equal to 6.894 (1 in English units); K3 = unit conversion equal to 0.794 (1 in English
units) K4 =unit conversion equal to 25.4 (1 in English units) and SD = spindown (percent); hydroplaning is assumed to begin at 10 percent spin down.
This occurs when the tire rolls 1.1 times the circumference to achieve a forward
progress distance equal to one circumference.
The method determines a film depth, d, associated with selected values for V, TD,
TXD, P, and with SD =10 percent, by solving the above equation. An estimate of design

d for V = 88 km/h, TD = 5.55 mm (median tire tread), TXD = 0.97 mm (mean pavement
texture), P = 186 kPA (median tire pressure), and SD =10 percent (by definition) is d =
1.87 mm. This is suggested as a sound design value because it represents the combination
of the mean or median of all the above parameters.
Manipulation of Eq. (13.9), using typical values, gives the following sensitivity information. An increase of 1 percent in pavement texture increases the hydroplaning depth by
1.6 percent, an increase of 1 percent in tread depth increases the hydroplaning depth
by 0.8 percent, and an increase of 1 percent in tire pressure increases the hydroplaning
depth by 2.4 percent.
Study of Eq. (13.9) indicates that 90 km/h (55 mph) is the speed value of concern for
practical control of hydroplaning. At this speed, a decrease of 1 percent in speed increases the hydroplaning depth by 25 percent. Speeds below 90 km/h (55 mph) tend to be safe
from the threat of hydroplaning because heavy rainfall is insufficient to generate
hydroplaning depth. An increase in speed of 1 percent decreases the hydroplaning depth
by 25 percent. Above 90 km/h (55 mph), hydroplaning can occur on extremely thin surface films associated with light rainfall intensities-intensities of 25 mm/h (1 in/h) and
less, which are rainfalls that are usually smaller than those used to design gutters, inlets,
and storm sewers.
Once a design d is determined, it is assumed that the thickness of the water film on the
pavement should be less than d. Water flows in a sheet across the surface to the edge of
the gutter flow. The width of sheet flow is the width W of the deck area minus the design
spread T, or (W-T). At the edge of the gutter flow, the design sheet flow depth d is
obtained from Eq. (13.9), with a suggested default value of d 1.867 mm (0.0735 in).
Using Manning's equation and continuity and solving for / gives the hydroplaning design
rainfall intensity in mm/h (in/h), as
K
s
*}\
*
}\ dl'61
[Cn\ [(S2X + G2)0-25 J [(W - T)

1-\

mm
'

where / = rainfall intensity, mm/h (in/h); C = dimensionless runoff coefficient; n =

Manning's roughness coefficient; Sx = cross slope, m/m (ft/ft); G = grade, m/m (ft/ft); d
= thickness of water film, mm (in);c = constant equal to 2289.4612 (64,904.4 in English
units); W = width of deck area, m (ft); and T = design spread, m (ft).
This hydroplaning design rainfall is used in Eq. (13.2) and is determined without
respect to rainfall frequency. However, once established, a frequency is imputed from the
associated time of concentration.
13.4.4 Rainfall Intensity by the Driver Vision-Impairment Method
HEC-21 developed another alternative method for selection of rain intensity (Young et
al., 1993). The method is also not dependent on rainfall frequency. The method assumes
that if drivers cannot see, then they will slow down or stop, and this behavior governs
highway function more than reaction to gutter flooding:
S, = -^y

(13.12)

where Sv = driver visibility, m; / = rainfall intensity, mm; V = vehicle speed, km/h; and
K0 = constant equal to 143,587.88 (338,721.26 in English units).

This empirical relationship was developed based on the Texas A&M test. At 90 km/h
(55 mph), the nonpassing minimum stopping sight distance is 15Om (500 ft) (this is the
lower value of a range given by AASHTO). Substituting 90 km/h and 15Om into Eq.
(13.12) gives a rainfall intensity of 125 mm/h. Note that cars in a travel corridor generate
splash and spray that increase the density of water droplets. Therefore, a lower rainfall of
100 mm/h may be more realistic as a threshold value that will cause slight impairment
because spray will increase the density of rain particles at roadway eye level. That is,
design intensities / above 100 mm/h will probably obscure drivers' visibility in heavy traffic and decrease sight distances to less than minimum stopping sight distance recommended by AASHTO.
Therefore, 100 to 150 mm/h is a suggested threshold design rain fall intensity range
for avoiding drivers' vision impairment; this intensity range is determined without respect
to frequency, but, depending on time of concentration, it imputes a frequency. Rainfall
intensities below this range should not obscure a drivers view through a windshield with
functioning windshield wipers. However, night driving in the rain is highly dependent on
vision, and data supporting the predictive relationship were obtained in daylight.

13.5

GUTTERDESIGN

A pavement gutter is adjacent to the roadway and conveys storm water. Gutter spread
(top flow width) may include a portion or all of a travel lane. Gutter sections can be categorized as conventional or as the shallow swale type illustrated in Fig. 13.1. The most representative conventional gutters have a uniform cross slope (Fig. 13.1 Al) or a composite
cross slope, where the gutter slope varies from the pavement cross slope (Fig. 13.1 A2).
To compute triangular, shallow depth, gutter flow, the Manning equation is integrated for an increment of width across the section (Izzard, 1946). The resulting equation is
Q = ^Sx1-61'Go-5P-67

1. Uniform Section

1. *V-Shope Gutter

2. Composite Section

2. "V-Shop* Medion

(13.13)

3. Curved Section

3. Circular

FIGURE 13.1 Typical gutter sections (Source: Brown, Stein and Warner, 1996)

where K0 - empirical coefficient equal to 0.376 (0.56 in English units); n = Manning's

coefficient, as given in Table 13.7; Q = flow rate, m3/s (ft3/s); T = width of flow (spread),
m (ft); Sx = cross slope, m/m ft/ft; and Sx = cross slope, m/m (ft/ft). G1 = grade, m/m
(ft/ft).Equation (13.13) neglects the resistance of the curb face.
The design of composite gutter sections requires additional consideration of flow in the
depressed segment of the gutter Qw. The depressed flow Qw. relates to the total flow as
GH, = Q ~ Qs

(13.14)

where Qw = flow rate in the depressed section of the gutter, m3/s (ftVs ); Q = gutter flow
rate, m3/s (ft3/s ); and Qs = flow capacity of the gutter section above the depressed section,
calculated using Eq. (13.13), m3/s (ftVs).
The cross-sectional area above the depressed section can be divided into Q8 to get the
gutter flow velocity Vg. Then Qw. is equal to Vg times the cross-sectional area of the
depressed section. The use of depressed sections can put more depth of water above inlet
entrances and increase their efficiency.
When the pavement cross section is curved, gutter capacity varies with the configuration
of the pavement. For this reason, discharge-spread- or discharge-depth-at-the-curb relationships developed for straight cross slopes are not applicable unless approximations are made.
When curbs are not needed for traffic control, a small swale section with a V-shape or
circular shape can be used to convey runoff from the pavement. Also, the control of pavement runoff on fills may be needed to protect the embankment from erosion. Small swale
sections are sized to have sufficient capacity to convey the flow to a location suitable for
interception. Shallow symmetric V-ditches, with side slopes of 10 percent or less, can be
evaluated using Eq. (13.13) and doubling the result.
Flow in shallow, circular gutter sections can be represented by the relationship
d

On
=

o.488
(13 15)

*<[s^H

'

TABLE 13.7 Manning's n for Street and Pavement Gutters

Type of Gutter or Pavement
Concrete gutter, troweled finish
Asphalt pavement
Smooth texture
Rough texture
Concrete gutter, asphalt pavement
Smooth
Rough
Concrete pavement
Float finish
Broom finish
For gutters with a small slope, where sediment may accumulate,
increase above values of n by
Source: Federal Highway Administration, 1977.

Manning's n
0.012
0.013
0.016
0.013
0.015
0.014
0.016
0.02

where d = depth of flow in circular gutter, m (ft); D = diameter of circular gutter, m (ft);
and Kc = empirical coefficient equal to 1.179 (0.972 in English units). The width of circular gutter section Tw is represented by the chord of the arc which can be computed
using Eq. (13.16):
Tw = 2(r* - (2 - </)2)o.5

(13.16)

where Tw = width of circular gutter section, m (ft), and r = radius of flow in circular gutter, m (ft).
As gutter flow approaches the low point in a sag vertical curve, the flow can exceed
the allowable design spread values as a result of the continually decreasing gutter slope.
The spread in these areas should be checked to insure it remains within allowable limits.
If the computed spread exceeds design values, additional flanking inlets should be provided to reduce the flow as it approaches the low point. Sag vertical curves and inlets are
discussed further in a later section. The effects of spread on gutter capacity are greater
than the effects of cross slope and grade, as would be expected because of the larger exponent of the spread term in Eq. (13.13). The magnitude of the effect is demonstrated by the
gutter capacity of a 3-m (9.8-ft) spread being 18.8 times greater than with a 1-m (3.3-ft)
spread, and three times greater than with a spread of 2 m (6.6 ft).
The effects of cross slope also are relatively great, as illustrated by a comparison of
gutter capacities with different cross slopes. At a cross slope of 4 percent, a gutter has 10
times the capacity of a gutter of 1 percent cross slope. A gutter at 4 percent cross slope has
3.2 times the capacity of a gutter at 2 percent cross slope.
Little latitude is generally available to vary grade to increase the gutter capacity, but
grade changes that change the gutter capacity are frequent. A change from G1 = 0.04
to 0.02 will reduce the gutter capacity to 71 percent of the capacity at G1 = 0.04.
However, grade is not typically within the hydraulic designer's set of responsibilities,
whereas the design problem is: for a given grade, size the gutter and space the inlets to
control spread.
The flow time in gutters is an important component of the time of concentration
for the contributing drainage area to an inlet. To find the gutter flow component of the
time of concentration, Eq. (13.6) for estimating the average velocity in a reach of gutter is used.

13.6 ROADSIDEDITCHDESIGN
Roadside and median channels are open-channel systems that collect and convey storm
water from the pavement surface, roadside, and median areas. These channels may outlet
to a storm drain piping system via a drop inlet, to a detention or retention basin or other
storage component, or to an outfall channel. Roadside and median channels are normally
trapezoidal in cross section and are lined with grass or other protective linings.
The design or analysis of roadside and median channels follows the basic principles of
open channel flow. A more complete coverage of open channel flow concepts can be
found in other chapters in this handbook and in Chow (1959) and Richardson et al.,
(1990). Open channel flow is classified using the following characteristics: (1) steady
(discharge passing a cross section is constant) or unsteady, (2) uniform (flow rate and
depth remain constant along a reach) or varied, and (3) subcritical (Froude number less
than 1) or supercritical-sometimes called tranquil or rapid.
Most natural flow conditions are neither steady nor uniform. However, in most cases
of small element drainage design, it can be assumed that the flow will vary gradually in

time and space and can be described as steady, uniform flow for short periods and
distances. Gradually varied flows are nonuniform flows in which the depth and velocity change gradually enough in the direction of flow that vertical accelerations can be
neglected.
Subcritical flow is distinguished from supercritical flow by a dimensionless number
called the Froude number (Fr), which represents the ratio of inertial forces to gravitational forces and is defined for rectangular channels by the following equation:
Fr

=<^

(m7)

where V = mean velocity, m/s (ft/s); g = acceleration of gravity, 9.8 m/s2 (32.2 ft/s2)' and
y = flow depth, m (ft) Critical flow occurs when the Fr has a value of one (1.0). The flow
depth at critical flow is referred to as critical depth.
Subcritical flow occurs when the F1. is less than one (Fr > 1). Subcritical flow is characterized by slower velocities, deeper depths, and mild slopes, whereas supercritical flow
is represented by faster velocities, shallower depths, and steeper slopes. Supercritical flow
occurs when the Fr is greater than one (Fr > 1). Supercritical flows are apt to be erosive
and may require energy dissipation measures to slow them down to manageable velocities.
Most small-element open-channel flows are Subcritical. However, supercritical flows are
not uncommon for smooth-lined ditches on steep grades; if such elements transition to flat
grades or channels with higher friction, a hydraulic jump may result. It is important to evaluate the Fr in open channel flows to determine how close a particular flow is to the critical
condition. In other words, a hydraulic jump occurs as an abrupt transition from supercritical to subcritical flow. Significant changes in depth and velocity occur in the jump and
energy is dissipated. The potential for a hydraulic jump to occur should be considered in
all cases where the Fr is close to one (1.0) or where the slope of the channel bottom changes
abruptly from steep to mild. The characteristics and analysis of hydraulic jumps are covered in detail in Chow (1959) and in HEC-14 (Federal Highway Administration, 1983).
13.6.1 Steady Uniform Flow Design
Small element design is based on the depth of steady uniform flow. This flow is called
the normal flow and is computed with Manning's equation. The general form of
Manning's equation is:
K A #0.67 C0.5
Q=KnAK^ ^
(13.18)
where Kn = empirical conversion equal to 1.0 (1.486 in English units); Q - discharge rate,
mVs (ftVs); A = cross sectional flow area, m2 (ft2); R = hydraulic radius, m (ft); = A/P, m
(ft); P = wetted perimeter, m (ft); S0 = energy grade line slope, m/m (ft/ft); and n =
Manning's roughness coefficient.
The selection of an appropriate Manning's n value for design purposes is often based
on observation and experience. Manning's n values also vary with normal flow depth
since n is a friction factor and friction is a boundary effect that is averaged into the
water column. Table 13.8 provides a tabulation of Manning's n values for various lining materials for open channels Manning's roughness coefficient for vegetative and
other linings vary significantly, depending on the amount of submergence. The classification of vegetal covers by degree of retardance is provided in Table 13.9; Table 13.10
provides a list of Manning's n relationships for five classes of vegetation defined by
their degree of retardance.

13.6.2 Water Surface Superelevation in Bends

Flow around a bend in an open channel induces centrifugal forces because of the change
in the direction of flow (Chow, 1959). This results in a superelevation of the water surface at the outside of bends and can cause the flow to splash over the side of the channel
if adequate freeboard is not provided. This superelevation can be estimated by the following equation:
V2T
Ad = Af
(13.19)
KC
TABLE 13.8 Manning's Roughness Coefficients for Open Channels
Lining Category

Lining Type

n Values for Given Depth Ranges

0.15-0.6Om
>0.60 m
0-015 m
(0.5 -2.0 ft)
(> 2.0 ft)
(0.05 ft)

0.015
Concrete
0.013
0.013
0.030
0.040
Grouted riprap
0.028
0.032
0.042
Stone masonry
0.030
0.022
0.025
Soil cement
0.020
0.016
0.016
0.018
Asphalt
0.020
0.020
0.023
Bare soil
Unlined
0.025
0.035
0.045
Rock cut
Woven paper net
Temporary*
0.015
0.015
0.016
0.022
Jute net
0.019
0.028
0.021
0.028
Fiberglass roving
0.019
Straw with net
0.033
0.065
0.025
0.028
0.035
0.066
Curled wood mat
0.021
0.025
0.036
Synthetic mat
0.044
25 mm (1 in), D50
0.030
0.033
Gravel riprap
0.034
0.041
50 mm (2 in), D50
0.066
0.104
150 mm (6 in), Z)50
Rock riprap
0.069
0.035
300 mm (12 in), D50
0.040
0.078
Source: Chen and Cotton, (1986).
Note: Values listed are representative values for the respective depth ranges. Manning's roughness coefficients,(n)
vary with the flow depth.
*Some "temporary" linings become permanent when buried.
Rigid

TABLE 13.9 Classification of Vegetal Covers According to Degree of Retardance.

Retardance
Class
Cover

A
B

Weeping lovegrass
Yellow bluestem schaemum
Kudzu
Bermuda grass
Native grass mixture
(little bluestem, bluestem,
blue gamma, and other long
and short midwest grasses)

Condition
Excellent stand, tall, average 0.76 m (2.5 ft)
Excellent stand, tall, average 0.91 (3.0 ft)
Very dense growth, uncut
Good stand, tall, average 0.30 m (1.0 ft)
Good stand, unmowed

TABLE 13.9

(Continued)

Retardance
Cover
Class
Weeping lovegrass
Lespedeza sericea
Alfalfa
Weeping lovegrass
Kudzu
Blue gamma
Crabgrass
Bermuda grass
Common lespedeza
Grass-legume mixture:
summer (orchard grass,
redtop Italian ryegrass, and
common lespedeza)
Centipedegrass
Kentucky bluegrass
Bermuda grass
Common lespedeza
Buffalo grass

Condition
Good stand, tall, average 0.61 m (2.0 ft)
Good stand, not woody, tall, average 0.48 m (1.6 ft)
Good stand, uncut, average 0.28 m (0.91 ft)
Good stand, unmowed, average 0.33 m (1.1 ft)
Dense growth, uncut
Good stand, uncut, average 0.33 m (1.1 ft)
Fair stand, uncut, avg. 0.25-1.20 m (0.9-4.0 ft)
Good stand, mowed, average 0.15 m (0.5 ft)
Good stand, uncut, average 0.28 m (0.91 ft)
Good stand, uncut average 0.15-0.20 m (0.5-1.5 ft)

Very Dense cover, average 0.15 m (0.5 ft)

Good stand, headed, avg. 0.15 to 0.30 m (0.5 to 1.0 ft)
Good stand, cut to .06 m (0.2 ft)
Excellent stand, uncut, average
Good stand, uncut, average 0.08 tp 0.15 m (0.3 to 0.5

ft)

Good stand, uncut, 0.10 to 0.13 (0.3 to 0.4 ft)

Grass-legume mixture: fall,
spring (orchard grass, redtop
Italian ryegrass, and
common lespedeza)
After cutting to 0.05 m (0.2 ft) height. Good stand
Lespedeza sericea
before cutting
Good stand, cut to average 0.04 m (0.1 ft)
Bermuda grass
E
Burned stubble
Bermuda grass
Source: Reproduced from HEC-15 (Chen and Cotton, 1986)
Note: Covers classified have been tested in experimental channels. Covers were green and generally uniform.

TABLE 13.10 Manning's n Relationships for Vegetal Degree of Retardance

Retardance Class
A
B
C
D
E

Manning's n Equation*
1.22 R^ I [30.2 + 19.97 log (R^ S00-4)]
1.22 R^ I [37.4 + 19.97 log (R S004)]
1.22 R^ I [44.6 + 19.97 log (R^ S00-4)]
1.22 R^ I [49.0 + 19.97 log (R S00-4)]
1.22 R^ I [52.1 + 19.97 log (R1-4 S00-4)]

* Equations are valid for flows less than 1.42 m3/s (50 ft3/s). Nomograph solutions for these equations are contained
in HEC-15 (Chen and Cotton, 1986).

where Ad = difference in water surface elevation between the inner and outer banks of the
channel in the bend, m (ft); V= average velocity, m/s (ft/s); T= surface width of the channel, m (ft); g = gravitational acceleration, 9.8 m/s2 (32.2 ft/s2); and Rc = radius to the center line of the channel, m (ft).
Equation (13.19) is valid for subcritical flow conditions. The elevation of the water
surface at the outer channel bank will be Ad/2 higher than the center line water surface
elevation (the average water surface elevation immediately before the bend) and the elevation of the water surface at the inner channel bank will be Ad/2 lower than the elevation
of the water surface at the center line. Flow around a channel bend imposes higher shear
stress on the channel's bottom and banks and is discussed in more detail in the next section. The increased stress requires additional design considerations within and downstream of the bend.
13.6.3 Shear Stresses in Open Channels
Stable channel design concepts presented in HEC-15 (Chen and Cotton, 1986) provide a
means of evaluating and defining channel configurations that will perform within acceptable limits of stability. Stability is achieved when the material forming the channel boundary effectively resists the erosive forces of the flow. Principles of rigid boundary
hydraulics can be applied to evaluate this type of system.
A pragmatic approach is to limit channel velocities to the range of 0.3 to 0.6 m/s (1 to
2 ft/s) to avoid erosion. Better estimates are possible by making tractive force calculations
that consider actual physical processes occurring at the channel boundary.
In uniform flow, the shear stress exerted on the bed is equal to the effective component
of the gravitational force acting on the body of water parallel to the channel bottom. The
average shear stress is equal to
T = yxS

(13.20)

where T = average shear stress, Pa (lb/ft ); y = unit weight of water, 9810 N/m3 (62.4
lb/ft2) at 15.6 0C (60 0F); x = height of water column, m (ft); and S = average bed slope
(or energy slope in varied flow conditions) m/m (ft/ft).The maximum shear stress on the
channel bed is computed as follows (Chow, 1959):
id = ydS

(13.21)

where id = maximum shear stress, Pa (lb/ft ) and d = maximum depth of flow, m (ft).
Shear stress in channels is not distributed uniformly along the wetted perimeter of a
channel. A typical distribution of shear stress in a trapezoidal channel tends toward zero at
the corners with a maximum on the bed of the channel at its center line, and the maximum
for the side slopes occurs around the lower third of the slope, as illustrated in Fig. 13.2.
Flow around bends creates higher shear stresses on the channel sides and bottom compared to straight reaches. The maximum shear stress in a bend is a function of the ratio of
the channel's curvature to its bottom width. This ratio increases as the bend becomes
sharper and the maximum shear stress in the bend increases. The bend shear stress can be
computed using the following relationship:
1 = Kfii
2

(13.22)

where ib = bend shear stress, Pa (lb/ft ); Kb = function of RJB as shown in Figure 13.3;
Rc = radius to the centerline of the channel, m (ft); B = bottom width of channel, m (ft);

FIGURE 13.2 Distribution of shear stress. (Source: Brown, Stein, Warner, 1996)

FIGURE 13.3 Kb Factor for maximun shear stress on chanel bends.

(Source: Brown, Stein, Warner, 1996)

ld = maximum channel shear stress, Pa (lb/ft2) calculated with Eq. (13.21). The increased
shear stress produced by the bend persists downstream of the bend a distance Lp9 which is
computed using the following relationship:
L, = ^-

(13.23)

where Lp = length of protection (length of increased shear stress due to the bend) from
the point of tangency downstream, m (ft); nb = Manning's roughness in the channel bend;
R = hydraulic radius, m (ft); and Kc = coefficient equal to 0.897 (0.736 in English units).
13.6.4 Parameters for Stable Channel Design
Parameters required for the evaluation and design of stable roadside and median channels
include discharge frequency, channel geometry, safety, channel slope, vegetation type,
freeboard, and shear stress. This section provides criteria relative to the selection or computation of these design elements. The applied shear stress is calculated and compared
with the permissible alternative shear stresses in the next section.
Discharge frequency. Roadside and median drainage channels are designed to accommodate the hydrologic design flow. However, when designing temporary channel linings,
a 2-yr return period is appropriate.
Channel geometry. Highway drainage channels are typically trapezoidal in shape.
Several typical shapes with equations for determining channel properties are illustrated in
Fig. 13.4. The depth, bottom width, and top width of the channel must be selected to provide the necessary flow area. The side slopes for triangular and trapezoidal channels
should not exceed the angle of repose of the soil, the lining material, or both and should
generally be 3:1 or flatter (Chen and Cotton, 1986) without consideration of the stability
of the side slopes.
Safety. Design of roadside and median channels should be integrated with the highway's geometric and pavement-design to insure proper consideration of safety and pavement drainage needs. In areas where traffic safety may be a concern, the side slopes of the
channel should be 4:1 or flatter.
Channel slope. Channel bottom slopes are generally dictated by the road profile.
However, if channel stability conditions warrant, it may be feasible to adjust the channel
gradient slightly to achieve a more stable condition. Channel gradients greater than 2 percent may require the use of flexible linings to maintain stability. Most flexible lining materials are suitable for protecting channel gradients of up to 10 percent, with the exception
of some grasses. Linings, such as riprap and wire-enclosed riprap, are more suitable for
protecting steep channels with gradients in excess of 10 percent. Rigid linings, such as
concrete paving, are stable to erosion but can be susceptible to failure from other causes,
such as overtopping, freeze-thaw cycles, swelling, and excessive soil pore water pressure.
Freeboard. The freeboard of a channel is the vertical distance from the water's surface
to the top of the channel. The importance of this factor depends on the consequence of
overflow of the channel bank. At a minimum, the freeboard should be sufficient to prevent waves, changes in superelevation, or fluctuations in water surface that result in the
sides overflowing. In a permanent roadside or median channel, about 150 mm (0.5 ft) of
freeboard is generally considered to be adequate. For temporary channels, no freeboard is

FIGURE 13.4 Channel geometries. (Source: Brown, Stein, Warner, 1996)

necessary. However, Fr's greater than one warrant freeboard height equal to the flow
depth to compensate for the large variations in flow caused by waves, splashing, and surging. Transitions from steep to mild also should have increased freeboard.
Permissible shear stress. The exerted shear stress is calculated with Eq. (13.21) or
(13.22) using the design parameters. Then, the permissible shear stress is used to find
an acceptable design alternative. The permissible or critical shear stress in a channel
defines the force required to initiate movement of the channel bed or lining material.
Table 13.11 presents permissible values of shear stress for manufactured, vegetative,
and riprap channel lining. The permissible shear stress values for noncohesive soils is a
function of the mean diameter of the channel material (Fig. 13.5). For larger sized
stones not shown in Fig. 13.5 and rock riprap, the permissible shear stress is given by
the following equation:
\ = K1P50

(13.24)

Particle Diameter, D (mm)

FIGURE 13.5 Permissible shear for non-cohesive soils. (Source: Brown,
Stein, Warner, 1996, English units preserved from source document)
TABLE 13.11 Permissible Shear Stresses for Lining Materials.
Lining Category

Lining Type

Temporary*

Woven paper net

Jute net
Fiberglass roving:
Single
Double
Straw with net
Curled wood mat
Synthetic mat
Class A
Class B
Class C
Class D
Class E
25 mm (1 in)
50 mm (2 in)
150 mm (6 in)
300 mm (12 in)
Noncohesive
Cohesive

Vegetative

Gravel riprap
Rock riprap
Bare soil

*Some "temporary" linings become permanent when buried.

Source: From HEC-15 (Chen and Cotton, 1986).

Permissible Unit Shear Stress

Pa
lb/ft2
7.2
21.6

0.15
0.45

28.7
40.7
69.5
74.3
85.7
177.2
100.6
47.9
228.7
16.8
15.7
31.4
95.7
191.5
See Fig. 13.5
See Fig. 13.6

0.60
0.85
1.45
1.55
2.00
3.70
2.10
1.00
0.60
0.35
0.33
0.67
2.00
4.00

Explonolion
"N" Volue
Loose
4-10
Medium Compoct 1 0 - 3 0
Compoct
30 - 50
NEffect
Number
Of Blows OfRequired
12" Penetration
The 2" To
Split-Spoon Sampler Seated To A
Depth Of 6" Driven With A 140 Lb.
Weight Foiling 30".

Plasticity Index - P.I.

FIGURE 13.6 Permissible shear stress for cohesive soils. (Source: Brown,
Stein, Warner, 1996, English units preserved from source document)

where ip = permissible shear stress, Pa (lb/ft2); D50 = mean riprap size, m (ft); and Kp =
empirical constant equal to 628 (4.0). For cohesive materials, the plasticity index provides
a good guide for determining the permissible shear stress illustrated in Fig. 13.6.
For trapezoidal channels protected with gravel or riprap having side slopes steeper
than 3:1, stability of the side slopes must be considered. This analysis is performed by
comparing the ratio of the tractive force between side slopes and channel bottom with the
ratio of shear stresses exerted on the channel sides and bottom. The ratio of shear stresses on the sides and bottom of a trapezoidal channel K1 is given in Fig. 13.7, and the tractive force ratio K2 is given in Fig. 13.8. The angle of repose 0 for different rock shapes
and sizes is provided in Fig. 13.9. The required rock size for the side slopes is found using
the following equation:

Angle Of Side Slope - 0 Deg.

FIGURE 13.7 Channel side stress to botton shear stress ratio, K1. (Source: Brown,
Stein, Warner, 1996)

FIGURE 13.8 Tractive force ratio, K2. (Source: Brown, Stein, Warner, 1996)

Angle Of Repose 9 - Deg.

Mean Stone Size, D50, Ft

Mean Stone Size, D50, mm

FIGURE 13.9 Angle of repose of riprap in terms of mean size and shape
of stone. (Source, Brown, Stein, Warner, 1996, English units preserved from source
document)

(D50)sUes = ^(D
50)bomm
A
2

(13.25)

where D50 = the mean riprap size, m (ft.); K1 = ratio of shear stresses on the sides and
bottom of a trapezoidal channel as shown in Fig. 13.7; and K2 = ratio of tractive force on
the sides and bottom of a trapezoidal channel shown in Fig. 13.8.

73.7

DRAINAGEINLETDESIGN

As water accumulates and flows in the gutters and ditches, spread increases. At the point
where spread reaches the design value, an inlet is provided to capture some or all of the

flow to control the spread. Letting some gutter flow pass the gutter ("flow by") for subsequent capture can be efficient because it is uneconomical to make the inlets as wide as
the design spread.
The hydraulic capacity of a storm drain inlet depends on its geometry as well as on the
characteristics of the gutter flow. Inlet capacity governs both the rate of water removal
from the gutter and the amount of water that can enter the storm drainage system.
Inadequate inlet capacity or poor inlet location may cause flooding on the roadway, resulting in a hazard to the traveling public.
13.7.1 Inlets
Storm drain inlets are used to collect runoff and discharge it to an underground storm
drainage system. Inlets are typically located in gutter sections, paved medians, and roadside and median ditches. Inlets used for draining of highway surfaces can be divided into
the following four classes: (1) Grate inlets (grate-covered openings); (2) Curb-opening
inlets (vertical curb openings); (3) Combination inlets (grate plus curb opening), and (4)
Slotted inlets (under surface pipe with slotted crown).
Grate inlets, as a class, perform satisfactorily over a wide range of gutter grades but
lose capacity with increase in grade. A disadvantage is clogging. In addition, where bicycle traffic occurs, grates should be bicycle safe (narrow opening widths or transverse grating with the latter being hydraulically inefficient).
Curb-opening inlets are most effective on flatter slopes (less than 3 percent grade), in
sags, and with flows which typically carry significant amounts of floating debris. The interception capacity of curb-opening inlets significantly decreases as the gutter grade steepens.
Combination inlets provide a high-capacity inlet that offers the advantages of both
grate and curb-opening inlets. The curb opening precedes the grate in a "sweeper" configuration, and it acts as a trash interceptor. Used in a sag configuration, the sweeper inlet
can have a curb opening on both sides of the grate.
Slotted inlets can be used in areas where it is desirable to intercept sheet flow before
it crosses onto a section of roadway. Their principal advantage is their ability to intercept sheet flow over a wide section of roadway. Slotted inlets are highly susceptible to
clogging.
Inlet interception capacity Q1 is the flow intercepted by an inlet. The efficiency (E) of
an inlet is the fraction of total flow that the inlet will intercept. The efficiency changes
with changes in cross slope, longitudinal slope, total gutter flow, and, to a lesser extent,
pavement roughness. Efficiency is defined by
E=

(13-26)

where E = inlet efficiency; Q = total gutter flow, mVs (ftVs); and Q1 = intercepted flow,
m3/s (ftVs). Flow that is not intercepted by an inlet is termed carryover, bypass, or flow by
and is equal to
Q1, = Q - Q,

(13.27)

when Qh = bypass flow, mVs (ftVs).

The interception capacity of all inlet configurations increases and the efficiency
decreases as flow rates increase. The depth of water next to the curb is the major factor in
the interception capacity of both grate inlets and curb-opening inlets. The interception

capacity of a grate inlet depends on the amount of water flowing over the gate, the size
and configuration of the grate, and the velocity of flow in the gutter. The interception
capacity of a curb-opening inlet is largely dependent on flow depth at the curb and curbopening length. Flow depth at the curb and, consequently, the interception capacity and
efficiency of the curb-opening inlet is increased by the use of a local gutter depression at
the curb-opening or a continuously depressed gutter to increase the proportion of the head
and flow adjacent to the curb. For curb inlets, top slab supports placed flush with the curb
line can substantially reduce the interception capacity of curb openings and therefore
should be avoided or be recessed and rounded in shape.
Slotted inlets function in essentially the same manner as curb opening inlets: i.e., as
weirs with flow entering from the side. Interception capacity depends on the depth of
flow and the length of the inlet. Efficiency depends on the depth of flow, the inlet length
of the inlet, and the total flow in the gutter.
A combination inlet consisting of a curb-opening inlet placed upstream of a grate inlet
has a capacity equal to that of the curb-opening length upstream of the grate plus that of
the grate, taking into account the reduced spread and depth of flow over the grate because
of the interception by the curb opening. This inlet configuration has the added advantage
of intercepting debris that might otherwise clog the grate and deflect water away from
the inlet.
Grate inlets in sag vertical curves operate as weirs for shallow ponding depths and as
orifices at greater depths. Between weir and orifice flow depths, a transition from weir to
orifice flow occurs. The capacity at a given depth can be affected severely if debris collects on the grate.
Curb-opening inlets operate as weirs in sag vertical curve locations up to a ponding
depth equal to the opening height. At depths above 1.4 times the opening height, the inlet
operates as an orifice and the transition between weir and orifice flow occurs between
these depths. The curb-opening height and length and the water depth at the curb affect
inlet capacity. The effective water depth at the curb can be increased by using a continuously depressed gutter, a locally depressed curb opening, or an increased cross slope, thus
decreasing the width of spread at the inlet. Slotted drains are not recommended in sag
locations because they are susceptible to clogging by debris.
13.7.2 Grate Inlet Design
Grates are effective highway pavement-drainage inlets when clogging by debris is not a
problem. Typical grate configurations are shown in Fig. 13.10. If clogging may be a problem, see Table 13.12, where grates are ranked for their susceptibility to clogging based on
laboratory tests using simulated leaves.
When the velocity approaching the grate is less than the "splash-over" velocity, the
grate will intercept essentially all the frontal flow. Conversely, when the gutter flow velocity exceeds the "splash-over" velocity for the grate, only part of the flow will be intercepted. A part of the flow along the side of the grate will be intercepted, depending on the
cross slope of the pavement, the length of the grate, and the velocity of flow.
The ratio of frontal flow to total gutter flow E0 for a uniform cross slope is expressed by
*. - ^ - 1 - fl - "I

(13.28)

where Q = total gutter flow, mVs (ftVs); Qw = flow in width, W, m3/s (ftVs); W = width
of depressed gutter or grade, m (ft); and T = total spread of water, m (ft). The ratio of side
flow Qs to total gutter flow is

Grate Type
Parallel bars with
1-7/8* spacing
(not bicycle safe)

Description

Comments
Adding tranaverae
) bars
4' o. c. to

1/4 x 4* bars'
on center

Parallel bars with

1-1/8* spacing
(not bicycle safe)
Curved vane

that of 30 ' van*.

3/8 x 4* bars
1-1/8* bar spacing
on center

(See fig. 9, HEC-12)

4-1/2* vane spacing (See fig 10, HEC-12)

1/2 x 2* flat
longitudinal bars
3-7/32* bar spacing
on center
4* vane spacing
1/2 x 2* flat
longitudinal bars
2-1/4* bar spacing
on center

45 tilt bar

4* vane spacing
1/2 x 2* flat
(See fig. 12, HEC-12)
longitudinal bars
3-7/32* bar spacing
on center

30 tilt bar

1/4 x 4* flat
tranaverae membera
2-9/16* o. c.
3/16 x 2* reticuline performance.
bars

Reticuline

Not*: L*ttra in circles refer to Chart 10.

Letter@ refera to 1-7/8 parallel bara with
Source: Johnson and Chang, 1984.
FIGURE 13.10 Typical grates. Source: (HEC-12: Johnson and Chang, 1984,
English units preserved from source document)
TABLE 13.12 Average Debris Handling Efficiencies of Grates Tested
Rank
1
2
3
4
5
6
7
8

Grate
Curved vane
30-85 tilt bar
45-85 tilt bar
P-50
P-50 X 100
45-60 tilt bar
Reticuline
P-30

Longitudinal Slope
0.005
0.040
46
44
43
32
18
16
12
9

61
55
48
32
28
23
16
20

^ = 1 - ^ = 1 - E0

(13.29)

The ratio of frontal flow intercepted to total frontal flow Rf is expressed by

Rf= I-K0(V-Vc)

(13.30)

where K0 = constant equal to 0.295 (0.09 in English units); V = velocity of flow in the gutter, m/s; and V0 = gutter velocity where splash-over first occurs, m/s. (Note: /^cannot exceed
1.0.) The splash-over velocity and /^ factor are solved by the chart in Fig. 13.11 as a function
of the type and length of the grate and average gutter velocity approaching the inlet.
The ratio of side flow intercepted to total side flow Rs, or side flow interception efficiency, is expressed by
*-r^|
(
SJU")
where Kc = a constant equal to 0.0828 (0.15 in English units).
The overall efficiency E of a grate is expressed by the following equation:
E = R1E0 + R1(I- E0)

(13.32)

Parallel bar* wild

Parallel
ban with
1-1/8* apaolng
Curved vane
46' till bar
30* lilt bar
Retloullne
Parallel bars with

Flow

EXAMPLE
Given:
W 12 In.; L - 20 In.; T 10 ft.
SX 0.03; Su 4%; n 0.010
(aee Chart 6)
Then:
V 8.2 fpe from Chart 9
E0 0.26 from Chart 7
R, - 0.88 from chart at left
Orate Efficiency E R E
E 0.8S x 0.28 0.21 1 0
Note: Side Inflow to email
rectangular Inlet* Ie negligible.
FIGURE 13.11 Grate inlet frontal flow interception efficiency. Source: HEC-12 (Johnson and
Chang, 1984, English units preserved from source document)

The interception capacity of a grade inlet on a grade is equal to the efficiency of the
grate multiplied by the total gutter flow:
Qi = EQ = Q[RfE0 + Rs(l- E0)]

(13.33)

13.7.3 Curb-Opening Inlet Design

Curb opening heights vary in dimension, however, a typical maximum height is approximately 100 to 150 mm (4 to 6 in). The length of the curb-opening inlet required for total
interception of gutter flow on a pavement section with a uniform cross slope is expressed
by the following equation:
Lr = ^ go.42 50.3f i V
r
(!3.34)
r**J
where Kc = constant equal to 0.817 (0.6 in English units); LT = curb opening length
required to intercept 100 percent of the gutter flow, m (ft); SL = grade; and Q = gutter
flow, m-Vs (ft-Vs).
The efficiency of curb-opening inlets shorter than the length required for total interception is expressed by
=l-l-ir
L
I
)

(13.35)

where L = curb-opening length, m (ft);

The length of inlet required for total interception by depressed curb-opening inlets or
curb openings in depressed gutter sections can be found by the use of an equivalent cross
slope Se in Equation (13.34) in place of Sx. S6 can be computed as
S6 = Sx + S^0

(13.36)

where S'w = cross slope of the depressed gutter and E0 is the same ratio used to compute
the frontal-flow interception of a grate inlet (Eq. (13.28)). The length of curb opening
required for total interception can be reduced significantly by increasing the cross slope
or the equivalent cross slope. Increasing the equivalent cross slope can be accomplished
by using of a continuously depressed gutter section or locally depressed gutter sections.
To compute efficiency, E for curb inlets shorter than the length needed for 100 percent
capture of flow, Eq. (13.35) is applicable with either straight cross slopes or composite
cross slopes.
13.7.4 Slotted Inlet Design
Placed at right angles to the flow, the slot acts as a short grate. Assuming a splash-over
velocity of 0.3 m/s (1 ft/s) and no side flow, the grate-inlet efficiency equations can
be used. Placed parallel to the flow, in the gutter notch, the flow interception by slotted
inlets and curb-opening inlets is similar. Analysis of data from the Federal Highway
Administration's tests of slotted inlets with slot widths > 45 mm (1.75 in) indicates that
the length of slotted inlets required for total interception can be computed with the curb-

opening equations. It should be noted that it is less expensive to add length to a slotted
inlet to increase its interception capacity than it is to add length to a curb-opening inlet.
Wide experience with the debris-handling capabilities of slotted inlets is not available. Deposition of debris in the pipe is a commonly encountered problem. However,
the configuration of slotted inlets make them accessible for cleaning with a high-pressure water jet.
13.7.5 Combination Inlet Design
Because the interception capacity of a combination inlet consisting of a curb opening
and a grate placed side by side is no greater than that of the grate alone, capacity is computed by neglecting the curb opening. Typical designs place the curb opening upstream
of the grate. The curb opening intercepts debris that might clog the grate and is called
a "sweeper" inlet. A sweeper combination inlet has an interception capacity equal to the
sum of the curb opening upstream of the grate plus the grates capacity, except that the
frontal flow, and thus the interception capacity of the grate, is reduced by interception
by the curb opening.
13.7.6 Design Adjustments for Sag Locations
Inlets in sag locations operate as weirs under low head conditions and as orifices at greater
depths and are more susceptible to clogging. Orifice flow begins at depths dependent on
the grate size, the height of the curb opening, or the slot width of the inlet. At depths
between those at which weir flow definitely prevails and those at which orifice flow prevails, flow is in a transition stage. Grate inlets alone are not recommended for use in sag
locations because of the tendencies of grates to become clogged. Combination inlets or
curb-opening inlets are recommended for use in these locations.
The capacity of grate inlets operating as weirs is
G1- = CJ>d^

(13.37)

where P = perimeter of the grate in m (ft) disregarding the side against the curb, Cw =
weir coefficient = 1.66 (3.0 English units) and d = flow depth, m (ft). The capacity of a
grate inlet operating as an orifice is
Q1 = C^g (2gdy*

(13.38)

where C0 - orifice coefficient = 0.67; Ag = clear opening area of the grate, m2 (ft2); g =
9.90 m/s2 (32.16 ft/s2); and d = flow depth, m (ft).
Use of Eq. (13.38) requires the clear area of opening of the grate. Tests of three grates
for the Federal Highway Administration (Burgi, 1978) showed that for flat bar gates, the
clear opening is equal to the total area of the grate less the area occupied by longitudinal
and lateral bars. However, curved vane grate performed about 10 percent better than one
would expect by calculating a net opening equal to the total area less the area of the bars
projected on a horizontal plane. Tilt bar and curved vane grates are not recommended for
sump locations.
The capacity of a depressed curb-opening inlet operating as a weir (curb water depths
less than opening height) is
G1- = Cw (L + 1.8W)J15

(13.39)

where Cw weir coefficient = 1.25 (2.3 in English units); L = length of curb opening,
m (ft); W = lateral width of depression, m (ft); and d = depth at curb measured from the
normal cross slope, m (ft): i.e., d = TSx. For curb-opening inlets with a continuously
depressed gutter, it is reasonable to expect that the effective weir length would be as great
as that for an inlet in a local depression. The weir equation for curb-opening inlets without any depression simplifies to
Qf = CwLd1-5

(13.40)

Without depression of the gutter section, the weir coefficient Cw becomes 1.60 (3.0,
English system).
At curb-opening lengths greater than 3.6 m (12 ft), Eq. (13.40) for nondepressed inlets
produces captured flows that exceed the values for depressed inlets computed using Eq.
(13.39). Since depressed inlets will perform at least as well as nondepressed inlets of the
same length, Eq. (13.40) should be used for all curb-opening inlets longer than 3.6 m
(12 ft).
Curb-opening inlets operate as orifices at curb depths greater than approximately
1.4 times the opening height. The capacity of curb-opening inlets operating as
orifices is
Qi= C0hL(2gd)*

(13.41)

r (
}a5
Q1 = C0AJ 2 ^ U - I

(13.42)

or

where C0 = orifice coefficient (0.67); d0 = effective head on the center of the orifice
throat, m (ft); L = length of orifice opening, m (ft); A^ = clear area of opening, m2 (ft2);
di depth of lip of curb opening, m (ft); and h = height of curb-opening orifice, m (ft).
The above variables are detailed in Fig. 13.12 as a function of typical variations in geometric configuration. For curb-opening inlets with other than vertical faces (see Fig.
13.12), Eq. (13.41) can be used with h = orifice throat width, m (ft) and d0 = effective
head on the center of the orifice throat, m (ft).
Slotted inlets in sag locations perform as weirs to depths of above 0.06 m (0.2 ft),
depending on slot width. At depths greater than approximately 0.12 m (0.4 ft), they perform as orifices. Between these depths, flow is in a transition stage. The capacity of a slotted inlet operating as a weir can be computed by
Q1 = CJLW*)

(13.43)

where Cw = weir coefficient =1.4 (2.2248 for English units); L = length of slot, m (ft);
and d = depth at curb measured from the normal cross slope, m (ft). The capacity of a
slotted inlet operating as an orifice can be computed by
Q1 = 0.8 L W(2gd)-5

(13.44)

where W = width of slot, m (ft); L = length of slot, m (ft); d = depth of water at slot for
d ^0.12 m (0.4 ft), m (ft); and g = 9.81 m/s2 (32.16 ft/s2).
Combination inlets consisting of a grate and a curb opening are considered advisable
for use in sags. Equal-length inlets refer to a grate inlet placed alongside a curb-opening

a. Horizontal Throat

b. Inclined Throat

c. Vertical Throat
FIGURE 13.12 Curb-opening inlets.
Source: (Brown, Stein, and Warner, 1996)

inlet, both of which have the same length. A sweeper inlet refers to a grate inlet placed at
the downstream end of a curb-opening inlet; the curb-opening inlet is longer than the grate
inlet and intercepts the flow before the flow reaches the grate. The sweeper is more efficient than the equal-length combination inlet and the curb opening and has the ability to
intercept any debris which may clog the grate inlet. The capture capacity of the equallength combination inlet in weir flow is essentially equal to that of a grate alone. In orifice flow, the capacity of the equal-length combination inlet is equal to the capacity of the
grate plus the capacity of the curb opening. A typical assumption is to size sweeper inlets
at the sag, assuming that the grate in between is 100 percent clogged.
13.7.7 Inlet Locations
The location of inlets is determined by geometric controls that require inlets at specific
locations, the use and location of flanking inlets in sag vertical curves, and the criterion
of spread on the pavement at the design rainfall intensity. To design the location of the
inlets adequately for a given project, the following information is needed:
A layout or plan sheet suitable for outlining drainage areas.
Road profiles.
Typical cross sections.

Grading cross sections.

Superelevation diagrams.
Contour maps.
Intensity-duration-frequency rainfall curves.

Because the typical inlet is located at a gutter location having sufficient drainage area
to generate the design spread, full or partial removal of spread is accomplished. Bypass of
some of the gutter flow, although leading to more frequent spacing, can result in much
smaller inlets at considerable cost savings.
There are a number of locations where inlets may be necessary with little regard to
contributing drainage area. These locations should be marked on the plans before any
computations regarding discharge, water spread, inlet capacity, or flow bypass. Examples
of such locations follow:
at all sag points in the gutter grade
immediately upstream of median breaks, entrance/exit ramp gores, crosswalks, and
street intersections: i.e., at any location where water could flow onto the travelway
immediately upgrade of bridges (to prevent pavement drainage from flowing onto
bridge decks)
immediately downstream of bridges (to intercept bridge deck drainage)
immediately upgrade of cross-slope reversals
at the end of channels in cut sections
on side streets immediately upgrade from intersections
behind curbs, shoulders, or sidewalks to drain low areas
In addition to the areas identified above, runoff from areas draining toward the highway
pavement should be intercepted by roadside channels or inlets before it reaches the roadway. This applies to drainage from cut slopes, side streets, and other areas alongside the
pavement.
13.8

BRIDGE-DECK DRAINAGE DESIGN

From a hydraulic standpoint, the ideal bridge has no sag point and bridge deck drainage
is not needed (Young et al., 1993). Such a bridge has off-bridge inlets at the upper end to
remove gutter flow before it reaches to the bridge and at the lower end to collect bridge
deck drainage. If the bridge is too long, spread on the bridge deck may necessitate inlets.
On a long bridge, typical inlets are small and are sometimes called scuppers: a 3-m (10ft) shoulder is desirable for maintenance of scuppers.
The off-bridge inlets are typically curb inlets and are designed using methods presented in Sec. 13.7. The on-bridge inlets, if rectangular, also can be designed using
equations in Sec. 13.7; the dimensions typically are less than 1 m to a side. If the inlets
are circular, their diameter is typically low. Figure 13.13 gives the capture efficiency for
circular inlets.
The dimensions of inlets are relatively constrained for placement and integration into
bridge decks. Many decks are pre- or post-tensioned structural slabs, and inlets are
detailed that may interfere with structural continuity. Thus, the surface grate and the
recessed collection chamber must be considered.

Frontal Flow Area

Note: One cross

bar did not
significantly
reduce efficiency
for D 4 inches.

Empiric curves based on laboratory data by Anderson, 1973

100% of frontal flow area of width, D

EXAMPLE
Given:
D * 6 in.; T 4 ft.
Su = 2%
Then:
D/TS.125
E * 0.33

FIGURE 13.13 Efficiency curves for circular scuppers. (Source: Young, Walker and Chang, 1993, English units preserved from source
document)

The constraints lead to small inlets and are associated with reinforcing bar schedules
or post-tensioned cable spacings. Typical dimensions need to be less than 0.5 m (18 in);
these details need to be structurally designed to transfer loads into the slab. Large inlet
spans cause a need for special designs and reinforcing details.
The hydraulic problem is that spreads of 2.5 to 3.0 m (8 to 10 ft) of water in gutters
are not reduced effectively with small inlets having capture efficiencies of 5 to 10 percent.
Numerous closely spaced small inlets are necessary to control spread.
Ideally, a bridge should have 3-m (10-ft) shoulders and the inlet boxes should be
placed at the outside edge of the shoulder. In this position, the maintenance crew can
park on the shoulder and work on the side away from traffic in reasonable safety. These
drains have a good chance of being maintained regularly. Unfortunately, accidents may
happen when lanes must be blocked to service the drains or when the maintenance crew
must work on the edge of the stream of traffic. This can result in poorly placed inlets
receiving inadequate maintenance. The larger the inlet, the fewer inlets to maintain.
Hydraulically, larger inlets handle larger volumes of flow and are more apt to clean
themselves. The larger the inlet, the easier it is to clean with a shovel. Inlets should be
sized as large as possible: Practically speaking, 1 m (36 in) is probably an upper limit
for inlets placed within deck slabs.
Round vertical scuppers should not be less than 155 mm (6 in) in diameter, with a
diameter of 200 m (8 in) preferred. Scuppers with a diameter of 100 mm (4-in) are not
uncommon in practice. However, their limited hydraulic capacity, coupled with their tendency to plug with debris, mitigates recommending their use. Although such features are
easy to place, they are relatively ineffective with capture efficiencies on the order of 5 percent. Nonetheless, they may be convenient when drainage can fall directly to underlying
surfaces without under-deck pipe collection and downspouts. With direct fall of water,
their small size is an advantage.
The water collected at inlets either falls directly to surfaces beneath bridges or is collected. Collected storm water is conveyed to bridge support columns and downspouts
that are affixed to vertical bridge members. The collectors are typically cast iron. The
collector pipes should be sloped at least 2 percent (20 mm/m [1/4 in/ft] or more-preferably 8 percent-to provide sufficient velocities at low flow to move silt and small debris
to avoid clogging.
13.8.1 Inlet Design for Constant-Grade Bridges
The general procedure is to start at the high end of the bridge and work downslope from
inlet to inlet. First, the designer selects a rainfall intensity and design spread. General
bridges dimensions, bridge grade, roughness and runoff coefficients, and inlet specifications are assumed to be known. The procedure is as follows:
1. Find the appropriate rainfall intensity. If the rational method is used, assume a time of
concentration of 5 min.
2. Find the flow on the deck Q at design spread T using Eq. (13.13).
3. Start at the high end of the bridge and compute the inlet spacing using Eqs. (13.45) and
(13.46), the derivations of which are given by Young et al. (1993):
L0 = ^j-for the first inlet
and

(13.45)

Lc = -^?- E, for the general case

^i Wp

(13.46)

where / = design rainfall intensity, mm/h (in/h) (Step V)Q = gutter flow, m3/s Lc = constant distance between inlets, m (ft). L0 distance to first inlet, m (ft). C = Rational
runoff coefficient. Wp = width of pavement contributing to gutter flow, m (ft). E = constant, which is equal to 1 for first inlets in all cases and is equal to capture efficiency for
subsequent inlets of constant-slope bridges and Kc = unit conversion equal to 79,662.5
(43,560 in English units).
Because the first inlet receives virtually no bypass from upslope inlets, the constant E
can be assumed to be equal to 1. If the computed distance L0 is greater than the length of the
bridge, then inlets are not needed and only bridge-end treatment design need be considered.
4. If inlets are required, the designer should proceed to calculate the constant inlet spacing Lc for the subsequent inlets.
5. If the bridge slope is nearly flat (less than about 0.003 m/m [0.003 ft/ft]), then the procedures for flat bridges should also be followed as a check.
13.8.2

Inlet Design for Flat Bridges

Four steps are involved when designing inlets for flat bridges:
1. Rainfall intensity, design spread (T), pavement width (Wp), bridge length (LB),
Manning's n (n), rational runoff coefficient (C), and gutter cross slope (Sx) are assumed
to be known. If the rational method is used, assume a time of concentration of 5 min
and determined the intensity from the intensity-duration-frequency (IDF) curves.
2. Constant inlet spacing Lc can then be computed using Equ. (13.47), the derivation of
which is given by Young, et al. (1993):
^(^c^8*-4472'11

(13 47)

where Kc = constant equal to 63,415 (1312 in English units).

3. The computed spacing is then compared with the known bridge length, If Lc is greater
than the length of the bridge, then there is no need for inlets and the designer need be
concerned only with the design of bridge end treatments. If Lc is less than the bridge
length, then the total needed inlet perimeter (P) can be computed using Eq.(13.48),
which is based on critical depth along the perimeter of the inlet (weir flow)
P _ (C
\^jWWP) "10.33 1T 0-61
^
KcSx"-<*n->

H348)
(
;

where Kc = 320.1 (102.5 in English units).

4. Adapt spacing to accommodate structural and aesthetic constraints.
Bridges with vertical curves, having either sags or high points, are nearly flat at their
low-or high-point stations. Bridges with grades less than about 0.3 percent are nearly flat.

For nearly flat stations on vertical curves or constant grades, the designer should set spacing assuming tthat he bridge is flat.

13.9

SAMPLEPROBLEMS

Example 1. (Brown et al, 1996)

Given: The following characteristics of the flow path
Flow segment

Length (m)

Slope (m/m)

1
2
3
4

25
43
79
146

0.005
0.005
0.006
0.008

Segment description
Short-grass prairie
Short-grass pasture
Grassed waterway
380-mm concrete pipe

Find: Rainfall intensity I = 60 mm/hr the time of concentration Tc for the area.
Solution
Step 1. Calculate time of concentration for each segment.
Segment 1: Obtain Manning's n roughness coefficient from Table 13.4: n = 0.15.
then determine the sheet flow travel time using Eq. (13.4):
T+1 = (6.943/104) (/z/S-5)0-6 = [6.943/(6O)04] [(0.15) (25)7(0.005)5]06 - 14.6 min
Segment 2: Obtain intercept coefficient k from Table 13.5: k = 0.213. Then
determine the concentrated flow velocity from Eq. 13.5:
Vc = K1IcSn0-5 = (1)(0.213)(0.5)0.5 - 0.15 m/s
Segment 3: Obtain intercept coefficient k from Table 13.5: k = 0.457. Then
determine the concentrated flow velocity from Eq. (13.5):
Vc= KckSp-5 = (1)(0.457)(0.6)05 - 0.35 m/s
Segment 4: Obtain Manning's n roughness coefficient from Table 13.6: n =
0.011. Then determine the pipe flow velocity from Eq. (13.7):
V= (1.0/0.011)(0.38/4)067 (0.008)05 = 1.7 m/s
Step 2. Determine the total travel time from Eq. 13.8:
Tc = T+1 + T+2 + T+3 + T+4
Tc = 4.6 + 43/[(6O)(0.15)] + 79/[(60)(0.35)] + 146/[(6O)(1.7)] - 24.5 min.
use 25 min.

Example 2. (Brown et al., 1996)

Given: The following existing and proposed land uses.
Existing conditions (unimproved):
Land use
Unimproved Grass
Grass
Total

Area, ha

Runoff coefficient C

8.95
8.60
17.55

0.25
0.22

Proposed conditions (improved):

Land use

Area, ha

Paved
Lawn
Unimproved Grass
Grass
Total

Runoff coefficient C

2.20
0.66
7.52
7.17
17.55

0.90
0.15
0.25
0.22

Times of concentration:
Existing condition (unimproved):
Proposed condition (improved):

Time of concentration Tc (min)

88
66

Partial 10-yr IDF information:

Duration (min)

Rainfall intensity (mm/hr)

60
65
70
75
80
85
90

60
57
55
53
51
49
47

Find: The 10-yr peak flow using the rational formula.

Solution
Step 1. Determine weighted C for existing (unimproved) conditions using Eq. 13.3:
Weighted C = S(CxA^M = [(8.95)(0.25) + (8.60)(0.22)]/(17.55) = 0.235
Step 2. Determine weighted C for proposed (improved) condition using Eq. 13.3:
Weighted C = [(2.2) (0.90) + (0.66)(0.15) + (7.52) (0.25) +
(1.17)(0.22)]/(17.55) = 0.315
Step 3. Determine rainfall intensity, I, from the lOyr IDF curve for each time of concentration.

Rainfall intensity I
Existing condition (unimproved):
48 mm/hr
Proposed condition (improved):
58 mm/hr
Step 4. Determine peak flow rate Q using equation 13.2:
Existing condition (unimproved):
Q = CIAIKc
= (0.235)(48)(17.55)/360 = 0.55 mVs
Proposed condition (improved):
Q = CIAlKc
= (0.315)(58)(17.55/360 = 0.88 mVs
Example 3
Given: Gutter flow with a spread of 3 m, n = 0.016, Sx = 0.02, and 1 percent grade.
Find: Flow
Solution
Determine the flow using Eq. 13.13:
Q = (0.376/0.016X0.02)167(0.01)05 (3)2'67 = 0.064 mVs
Example 4
Given: A curved vane grate inlet which is 0.3 m wide and 0.5 m long with a full flow of
0.064 mVs and gutter characteristics from Example 3.
Find: The inlet capture efficiency.
Solution
Step 1. Determine the ratio of frontal flow to gutter flow using Eq. 13.28:
E0 = 1 - [1 - (0.3/3)]267 - 0.245
Step 2. Determine the velocity using Eq. 13.6:
Vg = (0.752/0.016) (O.Ol)0'5 (0.02)067 (3)67 = 0.71 m/s
Step 3. Determine the frontal flow interception ratio using Fig. 13.11: For a 0.5 m (1.6 ft)
long grate, the splashover velocity is .4 m/s (4.6 ft/s), which is greater than the
gutter velocity. This implies an Rfof 1.0.
Step 4. Determine the side flow interception ratio using Eq. 13.31:
R5 = 1/(1 + ([(0.0828)(0.71)18]/[(0.02)(0.5)2/3])) - 0.22
Step 5. Calculate the overall efficiency using Eq. 13.32:
E = (1.0X0.245) + (0.22) (1 - 0.245) - 0.411
Example 5
Given: the configuration described in Example 4 and pavement width of 11 m, a rainfall intensity of 75 mm/hr, and a rational runoff coefficient of 0.7 (pavement and median).

Find: The inlet spacing.

Step 1. Calculate the distance to the first inlet using Eq. 13.45:
L0 = (3 600 000) (0.064)/[(0.7) (7S)(Il)] = 400 m
Step 2. Calculate general inlet spacing using Eq. 13.46:
Lc = L0E = (40)(0.411) = 165 m

REFERENCES
AASHTO, "Storm Drainage Systems", Model Drainage Manual Chapter 13. American Association
of State Highway and Transportation Officials, Washington, DC, 1991.
AASHTO, Policy on Geometric Design of Highways and Streets, American Association of State
Highway and Transportation Officials, Washington, DC, 1990.
Anderson, D. A., J. R., Reed, R. S., Huebner, J. J., Henry, W. R, Kilareski, and J. C., Warner,
Improved Surface Drainage of Pavements, Hydraulic Engineering Circular No. 12, Federal
Highway Administration, U.S. Department of Transportation, Washington, DC, 1995.
ASCE, "Design and Construction of Urban Stormwater Management Systems ", ASCE Manuals
and Reports of Engineering Practice No. 77, WEF Manual of Practice RD-20, American Society
of Civil Engineers, New York, 1992.
ASCE, Design Manual for Storm Drainage, American Society of Civil Engineers, New York, 1960.
Brown, S. A., S. M., Stein, and J. C., Warner, Urban Drainage Design Manual, Hydraulic
Engineering Circular No. 22, FHWA-SA-96-078, Federal Highway Administration, U.S.
Department of Transportation, Washington, DC, 1996.
Burgi, P. H., D. E. Gober, June 1977. Bicycle-Safe Grates Inlets Study, Volume 1 Hydraulic and
Safety Characteristics of Three Selected Grate Inlets on Continous Grades. Report No. FHWARD-77-24, Federal Highway Administration.
Burgi, P. H., May 1978. Bicycle-Safe Grate Inlets Study, Volume 2 Hydraulic Characteristics of
three Selected Grate Inlets on Continous Grades. Report No. FHWA-RD-78-4, Federal Highway
Administration.
Burgi, P. H., Bicycle-Safe Grate Inlets Study: Vol. 3 - Hydraulic Characteristics of Three Selected
Grate Inlets in a Sump Condition, Report No. FHWA-RD-78-70, Federal Highway
Administration, Washington, DC, September 1978.
Chen, Y. H., and G. K.,Cotton, Design of Roadside Channels with Flexible Linings, Hydraulic
Engineering Series No. 15, Publication No. FHWA-IP-87-7, prepared for the Federal Highway
Administration, U.S. Department of Transportation, Washington, DC, 1986.
Chow, V. T, Open-Channel Hydraulics, McGraw-Hill, New York, 1959.
Federal Highway Administration, Design Charts for Open-Channel Flow, Hydraulic Design Series
No. 3, U.S. Department of Transportation, Washington, DC, 1977 (reprint).
Federal Highway Administration, Hydraulic Design of Energy Dissipators, Hydraulic Engineering
Circular No. 14, U.S. Department of Transportation, Washington, DC, 1983.
Gallaway, B. C., et al, December 1979.Pavement and Geometric Design Criteria for Minimizing
Hidroplaning. Texas Transportation Institute, Texas A&M University, Federal highway
Administratuion, Report No. FHWA-RD-79-30, A Technical Summary.
Izzard, C. R, "Hydraulics of Runoff from Developed Surfaces," Proc. Highway Research Board,
Vol. 26, p. 129-150, Highway Research Board, Washington, DC, 1946.
Jennings, M. E., W. O., Thomas, Jr., and H. C., Riggs, Nationwide Summary of U.S. Geological
Survey Regional Regression Equations for Estimating Magnitude and Frequency of Floods for
Ungaged Sites, 1993, U.S. Geological Survey, Water-Resources Investigations Report No. 944002, prepared in cooperation with the Federal Highway Administration and the Federal
Emergency Management Agency, Reston, VA, 1994.

Johnson, F. L., and F. M., Chang, Drainage of Highway Pavements, Hydraulic Engineering Circular
No. 12, FHWA-TS-84-202, Federal Highway Administration, Washington, DC, 1984.
McCuen, R. H., P. A., Johnson, and R. M., Ragan, Hydrology, Hydraulic Design Series No. 2,
FHWA-SA-96-067, Federal Highway Administration, U.S. Department of Transportation,
Washington, DC, 1996.
M. E. Jennings, W. O. Thomas, Jr., and H. C. Riggs, 1994. Nationwide Summary of U.S.
Geological Survey Regional Regression Equations for Estimating Magnitude and Frequency of
Floods for Ungaged Sites, 1993. US Geological Survey, Water-Resources Investigations Report
94-4002, prepared in cooperation with the federal Highway Administration and the Federal
Emergency Managment Agency, Reston, Virginia.
Richardson, E. V., D. B., Simons, and P. Y., Julien, Highways in the River Environment, prepared
for the Federal Highway Administration, Washington, D.C, by the Department of Civil
Engineering, Colorado State University, Fort Collins, Colorado, 1990.
Young, G. K., S. M., Walker, and R, Chang, Design of Bridge Deck Drainage, Hydraulic
Engineering Circular No. 21, FHWA-SA-92-010, Federal Highway Administration, U.S.
Department of Transportation, Washington, D.C, 1993.
V. B. Sauer, W O. Thomas Jr., V. A. Strieker, and K. V. Wilson, 1983. Flood Characteristics of
Urban Watersheds in the United States. Prepared in cooperation with U. S. Department of
Transportation, Federal Highway Administration, U.S. Geological Survey Water-Supply Paper
2207.
V. B. Sauer, 1989. Dimensionless Hydrograph Method of Simulating Flood Hydro graphs. Preprint,
68th Annual Meeting of the Transportation Research Board, WEashington, D.C., (January) pp.
22-26.

CHAPTER 14
HYDRAULIC DESIGN OF
URBAN DRAINAGE SYSTEMS

Ben Chie Yen

Department of Civil & Environmental Engineering
University of Illinois at Urbana-Champaign
Urbana, Illinois
A.Osman Akan
Department of Civil and Enviromental Engineering
Old Dominion University
Norfolk, Virginia

U.I.

INTRODUCTION

Generally speaking urban drainage systems consist of three parts: the overland surface
flow system, the sewer network, and the underground porous media drainage system.
Some elements of these components are shown schematically in Fig. 14.1. Traditionally
no design is considered for the urban porous media drainage part. Recently porous media
drainage facilities such as infiltration trenches have been designed for flood reduction or
pollution control in cities with high land costs. For example, preliminary work on this
aspect of urban porous media drainage design can be found in Fujita (1987), Morita et al.
(1996), Takaaki and Fujita (1984) and Yen and Akan (1983). Much has yet to be developed to refine and standardize on such designs; no further discussion on this underground
subject will be given in this chapter.
From a hydraulic engineering viewpoint, urban drainage problems can be classified
into two types: (1) design and (2) prediction for forecasting or operation. The required
hydraulic level of the latter is often higher than the former. In design, a drainage facility
is to be built to serve all future events not exceeding a specified design hydrologic level.
Implicitly the size of the apparatus is so determined that all rainstorms equal to and
smaller than the design storm are presumably considered and accounted for. Sewers,
ditches, and channels in a drainage network each has its own time of concentration and
hence its own design storm. In the design of a network all these different rainstorms
should be considered. On the other hand, in runoff prediction the drainage apparatus has
already been built or predetermined, its dimensions known, and simulation of flow from
a particular single rainstorm event is made for the purpose of real-time forecasting to be
used for operation and runoff control, or sometimes for the determination of the flow of
a past event for legal purposes. The hydrologic requirements for these two types of problems are different. In the case of prediction, a given rainstorm with its specific temporal
and spatial distributions is considered. For design purposes, hypothetical rainstorms with
assigned design return period or acceptable risk level and assumed temporal and spatial

FIGURE 14.1 Schematic of components of urban catchment. (From Metcalf & Eddy, Inc. et al., 1971).
distributions of the rainfall are used. Table 14.1 lists some of these two types of design
and prediction problems.
In the case of sanitary sewers, for design purposes the problem becomes the estimation of the critical runoffs in both quantity and quality, from domestic, commercial, and
TABLE 14.1 Types of Urban Drainage Problems (a) Design Problems
Type

Design Purpose

Hydro Information
Sought

Required Hydraulic
Level

Sewers

Pipe size (and slope)

determination
Channel dimensions

Peak discharge, Qp for

design return period
Peak discharge, Qp for
design return period
Design hydrograph, Q(t)

Low

Design hydrograph, Q(t)

Low to moderate

Design peak discharge,

G,
Design peak discharge,
G,,
Design hydrograph
Design hydrograph

Low to moderate

Drainage channels
Detention/retention
storage ponds
Manholes and
junctions
Roadside gutters

Geometric dimensions
(and outlet design)
Geometric dimensions

Inlet catch basins

Geometric dimensions

Pumps
Control gates
or valves

Capacity
Capacity

Geometric dimensions

Low to moderate
Low to moderate

Low

Moderate to high
Moderate to high

TABLE 14.1 (continued) Types of Urban Drainage Problems (b) Prediction Problems
Type

Purpose

Hydro Input

Real-time
operation

Real-time
regulation
of flow
Simulation for
evaluation of
a system
Determination
of runoff at
specific locations
for particular past
or specified events
Determination
of the extent
of flooding
Reduce and
control of water
pollution due to
runoff from
rainstorms
Long-term, usually
large spatial scale
planning for
stormwater
management

Predicted and/or just Hydrographs, Q(t, jc() High

measured rainfall,
network data
Specific storm
Hydrographs, Q(t, x) High
event, network data

Performance
evaluation
Storm event
simulation

Flood level
determination
Storm runoff
quality
control
Storm runoff
master
planning

Hydro Information
Sought

Required Hydraulic
Level

Given past storm

event or specified
input hyetographs,
network data

Hydrographs, Q(t, jc.)

Moderate-high

Specific storm
hyetographs,
netwark data
Event or continuous
rain and pollutant
data, network data

Hydrographs and
stages

High

Long-term data

Hydrographs Q(t, Jt1) Moderate to high

Pollutographs, c(t, X1)

Runoff volume
Pollutant volume

Low

industrial sources over the service period in the future. For real-time control problems it
involves simulation and prediction of the sanitary runoff in conjunction with the control
measures.
The basic hydraulic principles useful for urban drainage have been presented in
Chapter 3 for free surface flows, Chapters 2 and 12 for pipe flows, and Chapter 10 for
pump systems. In the following, more specific applications of the hydraulics to urban
drainage components will be described. However, the hydraulic design for drainage of
highway and street surfaces, roadside gutters, and inlets has been described in Chapter 13,
design of stable erodible open channels in Chapter 16, and certain flow measurement
structures adaptable to urban drainage in Chapter 21; therefore they are not included in
this chapter.

74.2

HYDRAULICS OF DRAINAGE CHANNELS

Flows in urban drainage channels usually are open-channel flows with a free water surface.
However, sewer pipes, culverts, and similar conduits under high flow conditions could
become surcharged, and pressurized conduit flows do occur. Strictly speaking, the flow is

always unsteady, that is, changing with time. Nevertheless, in a number of situations, such
as in most cases of flow in sanitary sewers and for some rainstorm runoffs, change of flow
with time is slow enough that the flow can be regarded as approximately steady.
14.2.1 Open-Channel Flow
Open-channel flow occurs on overland, ditches, channels, and sewers in urban areas.
Unsteady flow in open channels can be described by a momentum equation given below
in both discharge (conservative) and velocity (nonconservative) forms together with its
various simplified approximate models:
_ L M + i afPfi!]+ i f Uj11*,+ -*. +s,f = o.
gA dt
gAdx(A ) gA Ja ^
dx

(i4.i)

dynamic wave
quasi-steady dynamic wave
noninertia
kinematic wave
lf

(2p-l)^ + ( p - l ) ^ + M + f - 5 . + ^a

(14.2)

where x = flow longitudinal direction measured horizontally (Fig. 14.2); A = flow crosssectional area normal to x\ y = vertical direction; Y = depth of flow of the cross section,
measured vertically; Q = discharge through A; V = QIA, cross-sectional average velocity along x direction; S0 = channel slope, equal to tan 6, 9 = angle between channel bed
and horizontal plane; Sf = friction slope; a = perimeter bounding the cross section A; ql
= lateral flow rate (e.g., rain or infiltration) per unit length of channel and unit length of
perimeter a, being positive for inflow; Ux = ^-component velocity of lateral flow when

FIGURE 14.2 Schematic of open, channel flow.

joining the main flow; g = gravitational acceleration; t = time; M = CgA)-1 I (Ux

V)q1 da and (3 = Boussinesq momentum flux correction coefficient for velocity distribution:
P = ^l u2dA

(14.3)

u = ^-component of local (point) velocity averaged over turbulence.

The continuity equation is
(14 4)
+ f = i*
If the channel is prismatic or very wide, such as the case of overland flow, Eq. (14.4) can
be written as

f + s (vy) = U-*

(14 5)

where b is the water surface width of the cross section.

In practice, it is more convenient to set the x and y coordinates along the horizontal
longitudinal direction and gravitational vertical direction, respectively, when applied to
flow on overland surface and natural channels for which S0 = tan 0. For human-made
straight prismatic channels, sewers, pipes, and culverts, it is more convenient to set the xy directions along and perpendicular to the longitudinal channel bottom. In this case, the
flow depth h is measured along the y direction normal to the bed and it is related to Y by
Y= h cos 6, whereas the channel slope S0 = sin 0.
The friction slope Sf is usually estimated by using a semiempirical formula such as
Manning's formula
riiy\-y\
n2Q\Q\
Sf = ^P R -^ = -^ R ^
(14.6)
or the Darcy-Weisbach formula
^-ife-M-j^
where n = Manning's roughness factor, Kn = 1.486 for English units and 1.0 for SI units;
/ = the Weisbach resistance coefficient; and R = the hydraulic radius, which is equal to
A divided by the wetted perimeter. The absolute sign is used to account for the occurrence
of flow reversal.
Theoretically, the values of n and / for unsteady nonuniform open-channel and pressurized conduit flows have not been established. They depend on the pipe surface roughness and bed form if sediment is transported, Reynolds number, Froude number, and
unsteadiness and nonuniformity of the flow (Yen, 1991). One should be careful that for
unsteady nonuniform flow, the friction slope is different from either the pipe slope, the dissipated energy gradient, the total-head gradient, or the hydraulic gradient. Only for steady
uniform flow without lateral flow are these different gradients equal to one another.
At present, we can only use the steady uniform flow values of n and/given in the literature as approximations. The advantage of/is its theoretical basis from fluid mechanics and its being nondimensional. Its values for steady uniform flow can be found from
the Moody diagram or the Colebrook-White formula given in Chap. 2, as well as in stan-

dard hydraulics and fluid mechanics references. Its major disadvantage is that for a given
pipe and surface roughness, the value of/varies not merely with the Reynolds number but
also with the flow depth. In other words, as the flow depth in the sewer changes during a
storm runoff,/must be recomputed repeatedly.
Manning's n was originally derived empirically. Its major disadvantage is its troublesome dimension of length to one-sixth power that is often misunderstood. Its main advantage is that for flows with sufficiently high Reynolds number over a rigid boundary with
a given surface roughness in a prismatic channel, the value of n is nearly constant over a
wide range of depth (Yen, 1991). Values of n can be found in Chow (1959) or Chap. 3.
Other resistance coefficients and formulas, such as Chezy's or Hazen-Williams's, have
also been used. They possess neither the direct fluid mechanics justification as/nor independence of depth as n. Therefore, they are not recommended here. In fact, HazenWilliams' may be considered as a special situation of Darcy-Weisbach's formula. A discussion of the preference of the resistance coefficients can be found in Yen (1991).
Equations (14.6) and (14.7) are applicable to both surcharged and open-channel
flows. For the open-channel case, the pipe is flowing partially filled and the geometric
parameters of the flow cross section are computed from the geometry equations given
in Fig. 14.3.
The pair of momentum and continuity equations [Eqs. (14.1) and (14.4) or Eqs. (14.2)
and (14.5)] with ( 3 = 1 and no lateral flow is often referred to as the Saint-Venant equations or full dynamic wave equations. Actually, they are not an exact representation of the
unsteady flow because they involve at least the following assumptions: hydrostatic
pressure distribution over A, uniform velocity distribution over A (hence (3 = 1), and negligible spatial gradient of the force due to internal stresses.
Those interested in the more exact form of the unsteady flow equations should refer to
Yen (1973b, 1975,1996). Conversely, simplified forms of the momentum equation, namely, the noninertia (misnomer diffusion wave) and kinematic wave approximations of the
full momentum equation [Eq. (14.1)] are often used for the analysis of urban drainage
flow problems.
Among the approximations shown in Eqs. (14.1) or (14.2), the quasi-steady dynamic
wave equation is usually less accurate and more costly in computation than the noninertia equation, and hence, is not recommended for sewer flows. Akan and Yen (1981),
among others, compared the application of the dynamic wave, noninertia, and kinematic
wave equations for flow routing in networks and found the noninertia approximation generally agrees well with the dynamic wave solutions, whereas the solution of the kinematic wave approximate is clearly different from the dynamic wave solution, especially when
the downstream backwater effect is important. Table 25.2 of Yen (1996) gives the proper
form of the equations to be used for different flow conditions.

Flow Area A= -^-($-sin$)

Hydraulic Radius R * -^-(I--~^-)
Depth h = Y11"008"!"5
Water Surface Width 8 = Dsin-|4, in Radians
FIGURE 14.3 Sewer pipe flow geometry. (From Yen, 1986a)

TABLE 14.2 Theoretical Comparison of Approximations to Dynamic Wave Equation

Boundary conditions required

Account for downstream
backwater effect and flow reversal
Damping of flood peak
Account for flow acceleration

Kinematic
wave

Noninertia

Quasi-steady
dynamic wave

Dynamic
wave

1
No

2
Yes

2
Yes

2
Yes

No
No

Yes
No

Yes
Only
convective
acceleration

Yes
Yes

Analytical solutions do not exist for Eqs. (14.1) and (14.2) or their simplified forms
except for very simple cases of the kinematic wave and noninertia approximations.
Solutions are usually sought numerically as described in Chap. 12. In solving the differential equations, in addition to the initial condition, boundary conditions should also be
properly specified. Table 14.2 shows the boundary conditions required for the different
levels of approximations of the momentum equation. It also shows the abilities of the
approximations in accounting for downstream backwater effects, flood peak attenuation,
and flow acceleration.
For flows that can be considered as invariant with time the steady flow momentum
equations which are simplified from Eq. (14.2) for different conditions are given in Table
14.3. The lateral flow contribution, mq, can be from rainfall (positive) or infiltration (negative) or both. Instead of these equations, the following Bernoulli total head equation is
often used for flow profile computations:
TABLE 14.3 Cross-Section-Averaged One-Dimensional Momentum Equations for Steady Flow of
Incompressible Homogeneous Fluid
Prismatic channel
Constant piezometric
pressure distribution
K = K' = 1
P = constant
K=K' = I
Prismatic or wide
channel
Definitions:
Dh=A/ water surface width;

K and K* = piezometric pressure distribution cor

rection factors for main an lateral flows;

oc V2
a V2
-^f + Y2 +yb2 = -^+Yl+ybl + he + hq,

(14.8)

where the subscripts 1 and 2 = the cross sections at the two ends of the computational reach,
Ax, of the channel, Y + yb the stage of the water surface where the channel bed elevation
at section 1 is ybl and that at section 2 is yb2 = ybl + S0 Ax; hq is the energy head from the
lateral flow, if any; the energy head loss he = Se Ax where Se is the slope of the energy line;
and a = the Coriolis convective kinetic energy flux correction coefficient due to nonuniform velocity distribution over the cross section (Chow, 1959; Yen, 1973). If there are other
energy losses, they should be added to the right-hand side of the equation. Methods of backwater surface profile computation using these equations are discussed in Chap. 3.
If the flow is steady and uniform, Eqs. (14.1) and (14.4) or Eqs. (14.2) and (14.5)
reduce to S0 = Sf and Q = AV. Hence, for steady uniform flow using Manning's formula,
Q = 0.0496 ^ S0"2 D ^ ~(t)S2^(|))5/3

(14.9)

where <|) is in radians (Fig. 14.3). Correspondingly, the Darcy-Weisbach formula yields
Q = l^ D 5 f l (^n^
Figure 14.4 is a plot of these two equations that can be used to find $.

(141Q)

14.2.2 Surcharge Flow

Sewers, culverts, and other drainage pipes sometimes flow full with water under pressure,
often known as surcharge flow (Fig. 14.5). Such pressurized conduit flow occurs under
extreme heavy rainstorms or under designed pipes. There are two ways to simulate
unsteady surcharge flow in urban drainage: (1) The standard transient pipe flow approach
and (2) the hypothetical piezometric open slot approach.
14.2.2.1 Standard transient pipe flow approach. In this approach, the flow is considered as it is physically, that is, pressurized transient pipe flow. For a uniform size
pipe, the flow cross-sectional area is constant, being equal to the full pipe area A^
hence 3A/3.X = O. The continuity and momentum equations [Eqs. (14.4) and (14.2)
with ql = O] can be rewritten as
Q=AfV

(14.11)

S /
-^
f
g 8r + ^{
dx ( g + -}=YJ

<14-12>

where P0 = the piezometric pressure of the flow and y = the specfic weight of the fluid.
If the pipe has a constant cross section and is flowing full with an incompressible fluid
throughout its length, then 3V/3jt = O. By further neglecting the spatial variation of P,
integration of Eq. (14.12) over the entire length, L, of the sewer pipe yields
P

exit

-f entrance

V2
=

yi

i av"i

v 5/+ ^
"-*i-^-*'IH

'

(1413)

(Degrees)

< (Radians)
FIGURE 14.4 Central angle cj> of water surface
in circular pipe (from Yen, 1986a).

FIGURE 14.5 Surcharge flow in a sewer. (After

Pansic, 1980).
or
(14 14)
Jr?-*--*'-^ + *^-^
'
where H11 = the total head at the entrance of the pipe, Hd = the water surface outside the
pipe exit, and Ku and Kd = the entrance and exit loss coefficients, respectively (Fig. 14.5).
Equations (14.11) and (14.12) can also be derived as a special case of the commonly used
general, basic, closed conduit transient flow continuity equation for waterhammer and
pressure surge analysis, see, e.g., Chaudhry, 1979; Stephenson, 1984; Wood, 1980; Wylie
and Streeter, 1983.

or

+
T^
A Jr -
p Jr ?
a* =

<14-15)

i a# a# c2 av
----1 + ^-+- -- + ^n 9 = 0
F a/
dx gV dx

(14.16)

and the momentum equation

i av v av a//
-^- + -^- + - ^ - + &f = 0
S ar
g dx
dx

(14.17)

where p = the bulk density of the fluid, H = PJj = the piezometric head above the reference datum, and c = the celerity of the pressure surge. The fact that Eqs. (14.11) and
(14.12) can be derived from Eqs. (14.2) and (14.5) is the theoretical basis of the
Preissmann hypothetical slot concept, which will be discussed below.
14.2.2.2 Hypothetical slot approach. This approach introduces hypothetically a continuous, narrow, piezometric slot attached to the pipe crown and over the entire length of the
pipe as shown in Fig. 14.6. The idea is to transform the pressurized conduit flow situation
into a conceptual open-channel flow situation by introducing a virtual free surface to the
flow. The idea was suggested by Preissmann (Cunge and Wegner, 1964). The hypothetical open-top slot should be narrow so that it would not introduce appreciable error in the
volume of water. Conversely, the slot cannot be too narrow, with the aim of avoiding the
numerical problem associated with a rapidly moving pressure surge.
A theoretical basis for the determination of the width of the slot is to size the width
such that the wave celerity in the slotted sewer is the same as the surge celerity of the compressible water in the actual elastic pipe. The celerity C1 of the slot pipe is
C1 = ^A/b

(14.18)

where b = the slot width and A = the flow cross-sectional area. Neglecting the area contribution of the slot and hence A = Af = nD2/4 for a circular pipe, and equating C1 to the
pressure wave speed c in the elastic pipe without the hypothetical slot, the theoretical slot
width is

FIGURE 14.6 Preissmann hypothetical piezometric open slot.

b = ngD2/4c2

(14.19)

The surge speed in a pipe usually ranged from a few hundred feet per second to a few
thousand feet per second. For an elastic pipe with a wall thickness e and Young's modulus of
elasticity EpJ assuming no pressure force from the soil acting on the pipe, the surge speed c is
Hf)('+(^)]
where Ef is the bulk modulus of elasticity and p/ is the bulk density, respectively, of the
flowing water (Wylie and Streeter, 1983). Special conditions of pipe anchoring against
longitudinal expansion or contraction and elasticity relevant to the surge speed c are given
in Table 14.4 where co = Poisson's ratio for the pipe wall material, that is, -co is the ratio
of the lateral unit strain to axial unit strain, and a is a constant to account for the rigidity
with respect to axial expansion of the pipe. For small pipes, Eq. (14.19) may give too
small a slot width, which would cause numerical problems. Cunge et al. (1980) recommend a width of 1 cm or larger.
The transition between part-full pipe flow and slot flow is by no means computationally smooth and easy, and assumptions are necessary (Cunge and Mazadou, 1984). One
approach is to assume a gradual width transition from the pipe to the slot. Sjoberg (1982)
suggested two alternatives for the slot width based on two different values of the wave
speed c in Eq. (14.19). For the alternative applicable to h/D > 0.9999, his suggested slot
width b can be expressed as
L = 10-6 + 0.05423 exp[-(/i/>)24]

(14.21)

He further proposed to compute the flow area A and hydraulic radius R when the depth h
is greater than the pipe diameter D as
TABLE 14.4 Special Conditions of Surge Speed in Full Pipe, Eq. (14.20)
Factor
Pipe Anchor

Condition
T| = ^(l + (D) + a D D+ e

Freedom of pipe
longitudinal
expansion

Only one
Entirely free
(expansion joints at both ends) end anchored

Entire length
anchored

Value of axial
expansion factor a

1 -co 2

Rigid pipe
, = 00

Air entrainment
P7 = pwVw + pfla

c> = Ef/Pf

Ef

1 - 0.5 co

Elasticity E

1 + VJi(EJEJ - 1

Subscript w denotes water (liquid); Subscript a = air; subscript/ = fluid mixture; Y = volume.

No air

P/ = Pw
E

f=

E
W

A = (nD2/4) + (h- D)b

(14.22)

R = D/4

(14.23)

A slight improvement to Sjoberg's suggestion to provide a smoother computational

transition is to use
A = A9999 + b(h - 0.9999Z))
(14.24)
for hiD > 0.9999 and assume that the transition starts at hiD = 0.91. Between hiD = 0.91
and 0.9999, real pipe area A and surface width B are used. However, for h/D > 0.91, R is
computed from Manning's formula using pipe slope and a discharge equal to the steady
uniform flow at h = 0.9ID, Q91; thus, for h/D > 0.91
R = (A91IA)R91

(14.25)

Because of the lack of reliable data, neither the standard surcharge sewer solution
method nor the Preissmann hypothetical open-slot approach has been verified for a single
pipe or a network of pipes. Past experiences with waterhammer and pressure surge problems in closed conduits may provide some indirect verification of the applicability of the
basic flow equations to unsteady sewer flows. Nevertheless, direct verification is highly
desirable.
Jun and Yen (1985) performed a numerical testing and found there is no clear superiority of one approach over the other. Nevertheless, specific comparison between them is
given in Table 14.5. They suggested that if the sewers in a network are each divided into
many computational reaches and a significant part of the flow duration is under surcharge,
the standard approach saves computer time. Conversely, if transition between open-channel and pressurized conduit flows occurs frequently and the transitional stability problem
is important, the slot model would be preferred.

TABLE 14.5 Comparison Between Standard Surcharge Approach and Slot Approach
Item

Standard Surcharge Approach

Hypothetical Slot Approach

Concept
Flow equations

Direct physical
Two different sets, one equation
for surcharge flow, two equations
for open-channel flow

Discretization for solution

Whole pipe length for

surcharge flow
Constant

Conceptual
Same set of two equations
(continuity and momentum)
for surcharge and openchannel flows
Divide into AJC' s

Water volume within pipe

Discharge in pipe at
given time

Same

Transition between
open channel flow and
surcharge flow

Specific criteria

Varies slightly with slot

size, inaccurate if slot is too
wide, stability problems if
slot is too narrow
Varies slightly with AJC, thus
allows transition to progress
within pipe
Slot width transition to avoid
numerical instability

TABLE 14.5 (Continued)

Item

Standard Surcharge Approach

Part full over pipe length

Assume entire pipe length full

or free
Time accounting for transition Yes, specific inventory of
surcharged pipes at different times
Programming efforts
More complicated because of two
sets of equations and time accountirig
and computer storage for transition

Computational effort

74.3

Depending mainly on accounting

for transition times

Hypothetical Slot Approach

Assume full or free Ax by Ar
No, implicit
Relatively simple because of
one equation set and no
specific accounting and
storage for transition
between open-channel and
full-pipe flows
Depending mainly on space
discretization Ax

FLOWINASEWER

14.3.1 Flow in a Single Sewer

Open-channel flow in sewers and other drainage conduits are usually unsteady, nonuniform, and turbulent. Subcritical flows occur more often than supercritical. For slowly time
varying flow such as the case of the flow traveling time through the entire length of the
sewer much smaller than the rising time of the flow hydrograph, the flow can often be
treated approximately as stepwise steady without significant error.
The flow in a sewer can be divided into three regions: the entrance, the pipe flow, and
the exit. Figure 14.7 shows a classification of 10 different cases of nonuniform pipe flow

subcritical

supercritical to surcharge

supercritical

subcritical to surcharge

supercritical to subcritical

surcnarge to supercritical

subcritical to supercritical

surcharge to subcritical

supercritical jump to surcharge

surcharge

FIGURE 14.7 Classification of flow in a sewer pip

(After Yen, 1986a).

easel

(b) case II

(c) case HI

(d) case IV
FIGURE 14.8 Types of sewer entrance flow.
(After Yen, 1986a).
based on whether the flow at a given instant is subcritical, supercritical, or surcharge.
There are four cases of pipe entrance condition, as shown in Fig. 14.8 and below:
Case

Pipe entrance hydraulic condition

I
II
III
IV

Nonsubmerged entrance, subcritical flow

Nonsubmerged entrance, supercritical flow
Submerged entrance, air pocket
Submerged entrance, water pocket

Case I is associated with downstream control of the pipe flow. Case II is associated
with upstream control. In Case III, the pipe flow under the air pocket may be subcritical,
supercritical, or transitional. In Case IV, the sewer flow is often controlled by both the
upstream and downstream conditions.
Pipe exit conditions also can be grouped into four cases as shown in Fig. 14.9 and
below:
Case

Pipe exit hydraulic condition

A
B
C
D

Nonsubmerged, free fall

Nonsubmerged, continuous
Nonsubmerged, hydraulic jump
Submerged

case A

case B

case C

case D
FIGURE 14.9 Types of sewer exit flow. (After
Yen, 1986a).
In Case A, the pipe flow is under exit control. In Case B, the flow is under upstream
control if it is supercritical and downstream control if subcritical. In Case C, the pipe flow
is under upstream control while the junction water surface is under downstream control.
In Case D, the pipe flow is often under downstream control, but it can also be under both
upstream and downstream control.
The possible combinations of the 10 cases of pipe flow with the entrance and exit conditions are shown in Table 14.6 for unsteady nonuniform flow. Some of these 27 possible
combinations are rather rare for unsteady flow and nonexistent for steady flow, for example, Case 6. For steady flow in a single sewer, by considering the different mild-slope M
and steep-slope 5 backwater curves (Chow, 1959) as different cases, there are 27 possible
cases in addition to the uniform flow, of which six types were reported by Bodhaine (1968).
TABLE 14.6 Pipe Flow Conditions
Pipe Flow

Possible
Entrance Conditions

Possible Exit
Conditions

Subcritical
Supercritical
Subcritical > hydraulic drop - supercritical
Supercritical > hydraulic jump > subcritical
Supercritical > hydraulic jump > surcharge
Supercritical surcharge
Subcritical > surcharge
Surcharge -> supercritical
Surcharge > subcritical
Surcharge

I, III
II, III
I, III
II, III
II, III
II, III
I, III
IV
IV
IV

A, B
B, C
B, C
A, B
D
D
D
B, C
A, B
D

Case
1
2
3
4
5
6
7
8
9
10

Source: From Yen (1986a).

The nonuniform pipe flows shown in Fig. 14.7 are classified without considering the
different modes of air entrainment. The types of the water surface profile, equivalent to
the M, S, and A (adverse slope) types of backwater curves for steady flow, are also not
taken into account. Additional subcases of the 10 pipe flow cases can also be classified
according to rising, falling, or stationary water surface profiles. For the cases with a
hydraulic jump or drop, subcases can be grouped according to the movement of the jump
or drop, be it moving upstream or downstream or stationary. Furthermore, flow with
adverse sewer slope also exists because of flow reversal.
During runoff, the change in magnitude of the flow in a sewer can range from only a few
times dry weather low flow in a sanitary sewer to as much as manyfold for a heavy rainstorm runoff in a storm sewer. The time variation of storm sewer flow is usually much more
rapid than that of sanitary sewers. Therefore, the approximation of assuming steady flow is
more acceptable for sanitary sewers than for storm and combined sewers.
In the case of a heavy storm runoff entering an initially dry sewer, as the flow enters the
sewer, both the depth and discharge start to increase as illustrated in Fig. 14.10 at times J1,
t2, and t3 for the open-channel phase. As the flow continues to rise, the sewer pipe becomes
completely filled and surcharges as shown at t4 and ^5 in Fig. 14.10. Surcharge flow occurs
when the sewer is underdesigned, when the flood exceeds that of the design return period,
when the sewer is not properly maintained, or when storage and pumping occur.
Under surcharge conditions, the flow-cross-sectional area and depth can no longer
increase because of the sewer pipe boundary. However, as the flood inflow continues
to increase, the discharge in the sewer also increases due to the increasing difference
in head between the upstream and downstream ends of the sewer, as sketched in the
discharge hydrograph in Fig. 14.10. Even under surcharge conditions while the sewer

FIGURE 14.10 Time variation of flow in a sewer. (After

Yen, 1986a).

diameter remains constant, the flow is usually nonuniform. This is due to the effects of
the entrance and exit on the flow inside the sewer, and hence, the streamlines are
not parallel.
As the flood starts to recede, the aforementioned flow process is reversed. The sewer
will return from surcharged pipe flow to open-channel flow, shown at t6 and I1 in Figure
14.1. Since the recession is usuallybut not alwaysmore gradual than the rising of the
flood, the water surface profile in the sewer is usually more gradual during flow recession
than during rising.
The differences in the gradient of the water surface profiles during the rising and recession of the flood bear importance in the self-cleaning and pollutant-transport abilities of
the sewer. During the rising period, with relatively steep gradient, the flow can carry not
only the sediment it brings into the sewer but also erodes the deposit at the sewer bottom
from previous storms. For a given discharge and gradient, the amount of erosion increases with the antecedent duration of wetting and softening of the deposit. During the recession, with a flatter water surface gradient and deceleration of the flow, the sediment being
carried into the sewer by the flow tends to settle onto the sewer bottom.
If the storm is not heavy and the flood is not severe, the rising flow will not reach surcharge state. The flood may rise, for example, to the stage at r3 shown in Fig. 14.10 and
then starts to recede. The sewer remains under open-channel flow throughout the storm
runoff. For such frequent small storms, the flow in the sewer is so small that it is unable
to transport out the sediment it carries into the sewer, resulting in deposition to be cleaned
up by later heavy storms or through artificial means.
For a single-peak flood entering a long circular sewer having a diameter D and pipe
surface roughness k, Yen (1973a) reported that for open-channel flow, the attenuation of
the flood peak, Qpx, at a distance x downstream from the pipe entrance (x = O) and the corresponding occurrence time of this peak, tpx, can be described dimensionlessly as
n
(
( V / H17 (K Vo.42 ( Qn v-16
|-= exp
^
MH=
p -0.0771*- A
Qp0
l
(D][DJ (Dj
(D^ V^)
x

r (f
j?132 ~M
p *-^iM \

<14-26)

v
r
\(o ^
i
}(j V0-11 (R >66
(^-W^={6.03.oglo[{|)-0,8]-52o}(A)
[]
f
x

O4-4 l 0 -^

a0.82~|0.5

(1427)
[slM F ^ - ^ i M '
where Qp0 and tp0 = the peak discharge and its time of occurrence at x = O, respectively;
Qb is the steady base flow rate and Rb = hydraulic radius of the base flow; tg = the time
to the centroid of the inflow hydrograph at x = O above the base flow; g = the gravitational acceleration; and Vw = (QJA1) + (gAJB1)112 = the wave celerity of the base flow,
where Ab = the base flow cross = sectional area and Bb = the corresponding water-surface width. In both equations, the second nondimensional parameter in the right-hand side
klD is a pipe property parameter; the third parameter R1JD is a base flow parameter; the
fourth nondimensional parameter represents the influence of the flood discharge; whereas the fifth and last nondimensional parameter reflects the shape of the inflow hydrograph.
The single - peak hydrograph shown in Fig. 14.10 is an ideal case for the purpose of
illustration. In reality, because the phase shift of the peak flows in upstream sewers and
the time-varying nature of rainfall and inflow, usually the real hydrographs are multipeak.

Piezometric Gradient

Because the flow is nonuniform and unsteady, the depth-discharge relationship, also
known as the rating curve in hydrology, is nonunique. Even if we are willing to consider
the flow to be steady uniform as an approximation, the depth-discharge relation is nonlinear, and within a certain range, nonunique, as shown nondimensionally and ideally in Fig.
14.11 for a circular pipe. The nonunique depth-discharge relationship for nonuniform flow,
aided by the poor quality of the water and restricted access to the sewer, makes it difficult
to measure reliably the time-vary ing flow in sewers. Among the many simple and sophisticated mechanical or electronic measurement devices that have been attempted on sewers
and reported in the literature, the simple, mechanical Venturi-type meter, which has side
constriction instead of bottom constriction to minimize the effect of sediment clogging,
still appears to be the most practical measurement means, that is, if it is properly designed,
constructed, and calibrated and if it is located at a sufficient distance from the entrance and
exit of the sewer. On the other hand, the hydraulic performance graph described in Sec.
14.6.1 can be used to establish the rating curve for a steady nonuniform flow.

Depth-Diameter Ratio,

Area

Discharge

Hidraulic Radius

Velocity

Hydraulic Elements Relative to Full Cross Section,

FIGURE 14.11 Rating curve for steady uniform flow in
circular pipe.

Flow in sewers is perhaps one of the most complicated hydraulic phenomena. Even for
a single sewer, there are a number of transitional flow instability problems. One of them
is the surge instability of the flow in pipes of a network. The other four types of instabilities that could occur in a single sewer pipe are the following: The instability at the
transition between open-channel flow and full conduit flow, the transitional instability
between supercritical flow and subcritical flow in the open-channel phase, the water-surface roll-wave instability of supercritical open-channel flow, and a near dry-bed flow
instability. Further discussion on these instabilities can be found in Yen (1978b, 1986a). It
is important to realize the existence of these instabilities in flow modeling.
14.3.2. Discretization of Space-Time Domain of
a Sewer for Simulation
No analytical solutions are known for the SaintVenant equations or the surcharged
sewer flow equation. Therefore, these equations for sewer flows are solved numerically
with appropriate initial and boundary conditions. The differential terms in the partial differential equations are approximated by finite differences of selected grid points on a
space and time domain, a process often known as discretization. Substitution of the finite
differences into a partial differential equation transforms it into an algebraic equation.
Thus, the original set of differential equations can be transformed into a set of finite difference algebraic equations for numerical solution.
Theoretically, the computational grid of space and time need not be rectangular.
Neither need the space and time differences AJC and Af be kept constant. Nonetheless, it is
usually easier for computer coding to keep AJC and At constant throughout a computation.
For surcharge flow, Eq. (14.14) dictates the application of the equation to the entire length
of the sewer, and the discretization applies only to the time domain. In an open-channel
flow, it is normally advisable to subdivide the length of a sewer into two or three computational reaches of Ax, unless the sewer is unusually long or short. One computational
reach tends to carry significant inaccuracy due to the entrance and exit of the sewer and
is usually incapable of sufficiently reflecting the flow inside the sewer. Conversely, too
many computational reaches would increase the computational complexity and costs
without significant improvement in accuracy.
The selection of the time difference Ar is often an unhappy compromise of three criteria. The first criterion is the physically significant time required for the flow to pass
through the computational reach. Consider a typical range of sewer length between 100
and 1000 ft and divide it into two or three AJC, and a high flow velocity of 5-10 ft/s, a suitable computational time interval would be approximately 0.2-2 min. For a slowly varying unsteady flow, this criterion is not important and larger computational Ar will suffice.
For a rapidly varying unsteady flow, this criterion should be taken into account to ensure
the computation is physically meaningful.
The second criterion is a sufficiently small Ar to ensure numerical stability. An often-used
guide is the Courant criterion
Ac/Ar > V + VgAIB

(14.28)

In sewers, which usually have small AJC compared to rivers and estuaries, this criterion
often requires a Ar less than half a minute and sometimes 1 or 2 s.
The third criterion is the time interval of the available input data. It is rare to have rainfall or corresponding inflow hydrograph data with a time resolution as short as 2, 5, or
even 10 min. Values for Ar smaller than the data time resolution can only be interpolated.
This criterion becomes important if the in-between values cannot be reliably interpolated.

In a realistic application, all three criteria should be considered. Unfortunately, in many

computations only the second numerical stability is considered.
There are many, many numerical schemes that can be adopted for the solution of the
Saint-Venant equations or their approximate forms [Eqs. (14.1)-(14.5)]. They can be classified as explicit schemes, implicit schemes, and the method of characteristics. Many of
these methods are described in Chap. 12, as well as in Abbott and Basco (1990), Cunge
et al. (1980), Lai (1986), and Yen (1986a).
14.3.3 Initial and Boundary Conditions
As discussed previously and indicated in Table 14.1, boundary conditions, in addition to
initial conditions, must be specified to obtain a unique solution of the Saint-Venant equations or their approximate simplified equations.
The initial condition is, of course, the flow condition in the sewer pipe when computations start, t = O, that is, either the discharge Q(X, O), or the velocity V(x, O), paired with
the depth h(x, O). For a combined sewer, this is usually the dry-weather flow or base flow.
For a storm sewer, theoretically, this initial condition is a dry bed with zero depth, zero
velocity, and zero discharge. However, this zero initial condition imposes a singularity in
the numerical computation. To avoid this singularity problem, either a small depth or a
small discharge is assumed so that the computation can start. This assumption is justifiable because physically there is dry-bed film flow instability, and the flow, in fact, does
not start gradually and smoothly from dry bed. Based on dry-bed stability consideration,
an initial depth on the order of 0.25 in., or less than 5 mm, appears reasonable.
However, in sewers, this small initial depth usually is unsatisfactory because negative
depth is obtained at the end of the initial time step of the computation. The reason is that
the continuity equation of the reach often requires a water volume much bigger than the
amount of water in the sewer reach with a small depth. Hence, an initial discharge, or base
flow, that permits the computation to start is assumed. For a storm sewer, the magnitude
of the base flow depends on the characteristics of the inflow hydrograph, the sewer pipe,
the numerical scheme, and the size of Ar and AJC used. For small Ax and Ar, a relatively
large base flow is required, but may cause a significant error in the solution. In either case,
it is not uncommon that in the first few time steps of the computation, the calculated depth
and discharge decrease as the flood propagates, a result that contradicts the actual physical process of rising depth and discharge. Nonetheless, if the base flow is reasonably
selected and the numerical scheme is stable, this anomaly would soon disappear as the
computation progresses. An alternative to this assumed base flow approach to avoid the
numerical problem is to use an inverted Priessmann hypothetical slot throughout the pipe
bottom and assigning a small initial depth, discharge or velocity to start the computation.
Currey (1998) reported satisfactory use of slot width between 0.001 and 0.01 ft.
As to boundary conditions, when the Saint-Venant equations are applied to an interior
reach of a sewer not connected to its entrance or exit, the upstream condition is simply the
depth and discharge (or velocity) at the downstream end of the preceding reach, which are
identical with the depth and discharge at the upstream of the present reach. Likewise, the
downstream condition of the reach is the shared values of depth and discharge (or velocity) with the following reach. Therefore, the boundary conditions for an interior reach
need not be explicitly specified because they are implicitly accounted for in the flow equations of the adjacent reaches.
For the exterior reaches containing either the sewer entrance or the exit, the upstream
boundary conditions required depend on whether the flow is subcritical or supercritical as
indicated in Table 14.7.

TABLE 14.7 Some Types of Specified Boundary Conditions for Simulation of Exterior
Reaches of Sewers
Location

Upstream End of Sewer

Entrance Reach (x = O)

Downstream End of Sewer

Exit Reach (x = L)

Subcritical flow

One of
G(O, O
/z(0, f}
V(O, O

One of
/z(L, t)', e.g. ocean tides, lakes
Q(L, t)', release hydrograph
<2(/i); rating curve
V(h); storage- velocity relation
for all t to be simulated
None

Supercritical flow

for all t to be simulated

Two of the above

For a sewer that is divided into M computational reaches and M + 1 stations, there is
a continuity equation and a momentum equation written in finite difference algebraic form
for each reach. There are 2(M + 1) unknowns, namely, the depth and discharge (or velocity) at each station. The 2(M + 1 ) equations required to solve for the unknowns come
from M continuity equations and M momentum equations for the M reaches, plus the two
boundary conditions. If the flow is subcritical, one boundary condition is at the sewer
entrance (x = O) and the other is at the sewer exit (x = L). If the flow is supercritical, both
boundary conditions are at the upstream end, the entrance, one of them often is a critical
depth criterion. If at one instant a hydraulic jump occurs in an interior reach inside the
sewer, two upstream boundary conditions at the sewer entrance and one downstream
boundary condition at the sewer exit should be specified. If a hydraulic drop occurs inside
the sewer, one boundary condition each at the entrance and exit of the sewer is needed;
the drop is described with a critical depth relation as an interior boundary condition.
Handling the moving surface discontinuity, shown schematically in Fig. 14.12, is not a
simple matter. The moving front may travel from reach to reach slowly in different Ar, or
it may move through the entire sewer in one Ar. If, for any reason, it is desired to compute
the velocity of the moving front Vw between two computational stations / and / + 1 in a
sewer, the following equation can be used as an approximation;

Supercriticol to Subcriticol

Subcritical to Supercritical

Supercritical to Subcritical

Subcritical to Supercritical

FIGURE 14.12 Moving water surface discontinuity in a

sewer. (After Yen, 1986a).

AV-A V
Vn = ' ' . '^V + '
A- + 1 A

(14-29)

14.3.4 Storm Sewer Design with Rational Method

The most important components of an urban storm drainage system are storm sewers.
A number of methods exist for designing the size of such sewers. Some are highly sophisticated, using the Saint-Venant equations, whereas others are relatively simple. In contrast
to storm runoff prediction/simulation models, sophisticated storm sewer design methods
do not necessarily provide a better design than the simpler methods, mainly because of the
discrete sizes of commercially available sewer pipes.
If the peak design discharge Qp for a sewer is known, the required sewer dimensions
can be computed by using Manning's formula such that
ARW = ^e(14.30)
K
n VS0
which can be obtained from Eq. (14.6) by assuming the friction slope Sf is equal to the
sewer slope S0. All other symbols in the equation have been defined previously. For a circular sewer pipe, the minimum required diameter dr is
f
n Q T/8
d,= 3.208
(14.3Ia)
K fM
L
n ^o J
where kn = 1 for SI units and 1.486 for English units. If the Darcy-Wesibach formula (Eq.
14.7) is used,
r
/
11/5
dr= 0.811 -^- G,2
(14.3Ib)
L
OO J
These two equations are plotted in Fig. 14.13 for design applications. The assumption S0
= Sf essentially implies that around the time of peak discharge, the flow can well be
regarded approximately as steady uniform flow for the design, despite the fact that the
actual spatial and temporal variations of the flow are far more complicated as described
in Sec. 14.3.1.
In sewer designs, there are a number of constraints and assumptions that are commonly
used in engineering practice. Those pertinent to sewer hydraulic design are as follows:

1. Free surface flow exists for the design discharge, that is, the sewer is under "gravity
flow" or open-channel flow. The design discharge used is the peak discharge of the
total inflow hydrograph of the sewer.
2. The sewers are commercially available circular sizes no smaller than, say, 8 in. or 200
mm in diameter. In the United States, the commercial sizes in inches are usually 8, 10,
12, and from 15 to 30 inches with a 3-in. increment, and from 36 to 120 in. with an
increment of 6 in. In SI units, commercial sizes, depending on location, include most
if not all of the following: 150, 175, 200, 250, 300, 400, 500, 600, 750, 1000, 1250,
1500, 1750, 2000, 2500, and 3000 mm.
3. The design diameter is the smallest commercially available pipe that has a flow capacity equal to or greater than the design discharge and satisfies all the appropriate constraints.
4. To prevent or reduce permanent deposition in the sewers, a nominal minimum permissible flow velocity at design discharge or at nearly full-pipe gravity flow is speci-

FIGURE 14.13 Required sewer diameter, (m or ft)

5.

6.
7.
8.
9.

fied. A minimum full-pipe flow velocity of 2 ft/s or 0.5 m/s at the design discharge is
usually recommended or required.
To prevent the occurrence of scour and other undesirable effects of high velocity =
flow, a maximum permissible flow velocity is also specified. The most commonly
used value is 10 ft/s or 3 m/s. However, recent studies indicate that due to the improved
quality of modern concrete and other sewer pipe materials, the acceptable velocity can
be considerably higher.
Storm sewers must be placed at a depth that will allow sufficient cushioning to prevent
breakage due to ground surface loading and will not be susceptible to frost. Therefore,
minimum cover depths must be specified.
The sewer system is a tree-type network, converging toward downstream.
The sewers are joined at junctions or manholes with specified alignment, for example,
the crowns aligned, the inverts aligned, or the centerlines aligned.
At any junction or manhole, the downstream sewer cannot be smaller than any of the
upstream sewers at that junction, unless the junction has significantly large detention

storage capacity or pumping. There also is evidence that this constraint is unnecessary
for very large sewers.
Various hydrologic and hydraulic methods exist for the determination of the design
discharge Qp. Among them the rational method is perhaps the most widely and simplest
used method for storm sewer design. With this method, each sewer is designed individually and independently, except that the upstream sewer flow time may be used to estimate
the time of concentration. The design peak discharge for a sewer is computed by using the
rational formula
Q11 = I ^ C f J

(14.32)

where i = the intensity of the design rainfall; C = the runoff coefficient (see Chap. 5 for
its values); and a is surface area. The subscript j represents they'th subarea upstream to be
drained. Note that ^a. includes all the subareas upstream of the sewer being designed.
Each sewer has its own design i because each sewer has its own flow time of concentration and design storm. The only information needed from upstream sewers for the design
of a current sewer is the upstream flow time for the determination of the time of concentration.
The rational formula is dimensionally homogenous and is applicable to any consistent
measurement units. The runoff coefficient C is dimensionless. It is a peak discharge coefficient but not a runoff volume fraction coefficient. However, in English units usually the
formula is used with the area aj in acres and rain intensity i in inches per hour. The conversion factor 1.0083 is approximated as unity.
The procedure of the rational method is illustrated in the following in English units for
the design of the sewers of the simple example drainage basin A shown schematically in
Fig. 14.14. The catchment properties are given in Table 14.8. For each catchment, the
length L0 and slope S0 of the longest flow pathor better, the largest LJvS0 should
first be identified. As discussed Sec. 14.7, a number of formulas are available to estimate
the inlet time or time of concentration of the catchment to the inlet. In this example, Eq.
(14.86) is used with K = 0.7 for English units and heavy rain, that is, t0 =
0.7(nL0A/S^)-6. The catchment overland surface texture factor TV is determined from
Table 14.16
The design rainfall intensity is computed from the intensity-duration-frequency relation for this location,
i QQT 0.2
i(in./h) = -^(14.33)
ld T- ZJ
TABLE 14.8 Characteristics of Catchments of Example Drainage Basin A
Catchment

Area
(acres)

I
II
III
IV
V

2
3
3
5
5

Total area =18 acres

Longest Overland Path

Length L0
Slope
Surface Texture
(ft)
N
250
420
400
640
660

0.010
0.0081
0.012
0.010
0.010

0.015
0.016
0.030
0.020
0.021

Inlet Time
t0
(min)

Runoff
Coefficient
C

6.2
9.3
11.7
12.9
13.1

0.8
0.7
0.4
0.6
0.6

Manhole
Ground

(10-YrFlood Level)
Note: All elevations are in feet
(Channel Bed)
FIGURE 14.14 Sewer design example drainage basin A. (a) Layout (b) Profiles.
where td = the rain duration (min) which is assumed equal to the time of concentration,
tc, of the area described, and Tr = the design return period in years. For this example, Tr
= 10 years. Determination of / for the sewers is shown in Table 14.9a. The entries in this
table are explained as follows:
Column L Sewer number identified by the inlet numbers at its two ends.
Column 2. The sewer number immediately upstream, or the number of the catchment
that drains directly through manhole or junction into the sewer being considered.
Column 3. The size of the directly drained catchment.
Column 4. Value of the runoff coefficient for each catchment.
Column 5. Product of C and the corresponding catchment area.
Column 6. Summation of C}a. for all the areas drained by the sewer; it is equal to the
sum of contributing values in Column 5.

Column 7. Values of inlet time to the sewer for the catchments drained, that is, the
overland flow inlet time for directly drained catchments, or the time of
concentration for the immediate upstream connecting sewers.
Column 8. The sewer flow time of the immediate upstream connecting sewer as
given in Column 9 in Table 14.9&.
Column 9. The time of concentration tc for each of the possible critical flow paths,
tc = inlet time (Column 7) + sewer flow time (Column 10) for each
flow path.
Column 10. The design rainfall duration td is assumed equal to the longest of the different times of concentration of different flow paths to arrive at the
entrance of the sewer being considered, for example, for Sewer 31, td is
equal to 13.9 min from Sewer 21, which is longer than that from directly
contributing Catchment V (13.1 min).
Column IL The rainfall intensity i for the duration given in Column 10 is obtained
from the intensity-duration relation for the given location, in this case, Eq.
(14.33) for the 10-year design return period.
Table 14.9/? shows the design of the sewers for which the Manning n = 0.015, minimum soil cover is 4.0 ft, and minimum nominal design velocity is 2.5 ft/s. The exit sewer
of the system (Sewer 31) flows into a creek for which the bottom elevation is 11.90 ft, the
ground elevation of its bank is 21.00 ft, and its 10-year flood water level is 20.00 ft.
Column L
Column 2.
Column 3.
Column 4.

Sewer number identified by its upstream inlet (manhole) number.

Ground elevation at the upstream manhole of the sewer.
Length of the sewer.
Slope of the sewer, usually follows approximately the average ground
slope along the sewer.
Column 5. Design discharge Qp computed according to Eq. (14.32); thus, the product
of Columns 6 and 11 in Table 14.90.
Column 6. Required sewer diameter, as computed by using Eq. (14.31) or Fig. 14.13;
for Manning's formula with n = 0.015 and dr in ft, Eq. (14.3Ia) yields
dr

(
o Y8
= [0.0324 ^-j

in which Qp, in ftVs, is given in Column 5 and S0 is in Column 4.

Column 7. The nearest commercial nominal pipe size that is not smaller than the
required size is adopted.
Column 8. Flow velocity computed as V = Q/A^ that is, it is calculated as Column 5
multiplied by 4/n and divided by the square of Column 7. As discussed in
Yen (1978b), there are several ways to estimate the average velocity of the
flow through the length of the sewer. Since the flow is actually unsteady
and nonuniform, usually the one used here, using full pipe cross section,
is a good approximation.
Column 9. Sewer flow time is computed as equal to LIV, that is, Column 3 divided
by Column 8 and converted into minutes.

TABLE 14.9 Rational Method Design of Sewers of Example Drainage Basin A

(a) Design Rain Intensity
Sewer
(1)

Directly Drained Area

a
Catchment or
i
Contributing
Upstream Sewer (acres)
(2)
(3)

Runoff
Coefficient

Inlet Time
CA

-L CH

Cj
(4)

<5)

(6)

11-21

I
II

2
3

0.8
0.7

1.6
2.1

12-21
21-31

m
IV
ii
12

3
5

0.4
0.6

1.2
3.0
3.7
1.2

V
21

3.7
1.2

(min)
(7)

Upstream
Time of
Design Rain Design Rain
Sewer Flow Concentration Duration
Intensity
Time
i
(min)
(min)
(min)
(in./h)
(10)
(8)
(U)
(9)

6.2
9.3

6.2
9.3

11.7
12.9
9.3
11.7

1.4
0.9

11.7
12.9
10.7
12.6

1.0

13.1
13.9

7.9

31-41

0.6

3.0
7.9

13.1
12.9
10.9

9.3
11.7

4.62
4.32

12.9

4.18

13.9

4.07

TABLE 14.9 (Continued)

(b) Sewer Design
Length

Slope

(1)

Upstream
Manhole
Ground
Elev.
(ft)
(2)

(ft)
(3)

11
12
21
31
(31

35.00
41.50
31.90
28.70
28.70

450
360
400
500
500

Sewer

Design Required Diameter Flow

Discharge Diam.
Used
Velocity
d
V
r
J
QP
n

Sewer
Flow
Time

SL

Upstream
Upstream Downstream Downstream
Crown Elev. Invert Elev. Crown Elev. Invert Elev.

(4)

(ft3/*)
(5)

(ft)
(6)

(ft)
(7)

(ft/*)
(8)

(min)
(9)

(ft)
(10)

(ft)
(U)

(ft)
(12)

(ft)
(13)

(ft)
(14)

0.0081
0.0290
0.0100
0.0144
0.0156

17.1
5.2
33.0
44.4
44.4

1.98
0.99
2.43
2.53
2.50

2.00
1.00
2.50
2.75
2.50

5.4
6.6
6.7

1.4
0.9
1.0

3.65
10.444.00
7.20
7.80

31.00
37.50
27.85
23.85
23.85

29.00
36.50
25.35
21.10
21.35

27.35
27.06
23.85
16.65
16.05

25.35
26.06
21.35
13.90
13.55)

Column 10. Product of Columns 3 and 4; this is the elevation difference between the
two ends of the sewer.
Column 11. The upstream pipe crown elevation of Sewer 11 is computed from the
ground elevation minus the minimum soil cover, 4.0 ft, to save soil excavation cost. In this example, sewers are assumed invert aligned except the
last one (Sewer 31), which is crown aligned at its upstream (23.85 ft for
upstream of Sewer 31 and downstream of Sewer 21) to reduce backwater
influence from the water level at sewer exit.
Column 12. Pipe invert elevation at the upstream end of the sewer, equal to Column
11 minus Column 7.
Column 13. Pipe crown elevation at the downstream end of the sewer, equal to
Column 11 minus Column 10.
Column 14. Pipe invert elevation at the downstream end of the sewer, equal to Column
13 minus Column 7. For the last sewer, the downstream invert elevation
should be higher than the creek bottom elevation, 11.90 ft.
The above example demonstrates that, in the rational method, each sewer is designed
individually and independently, except the computation of sewer flow time for the purpose of rainfall duration determination for the next sewer, that is, the values of tf in
Column 8 of Table 14.9a are taken from those in Column 9 of Table 14.96.
The profile of the example designed sewers are shown as the solid lines in Fig. 14.14.
If the water level of the creek downstream of Sewer 31 is ignored, theoretically a cheaper design could be achieved by putting the exit Sewer 31 on a slightly steeper slope, from
0.0144 to 0.0156 to reduce the pipe diameter from 2.75 to 2.50 ft. The new slope can be
estimated from
(1434)
H^H'*"
This alternative is shown with the parentheses in Table 14.9Z? and as dashed lines in Fig.
14.14b. However, one should be aware that the water level of a 10-year flood in the creek
is 20.00 ft and hence, the last sewer is actually surcharged and its exit is submerged. The
sewer will not achieve the design discharge unless its upstream manhole is surcharged by
almost 4 ft (20.00-16.05). Therefore, the original design of 2.75 ft diameter is a safer and
preferred option in view of the backwater effect from the tailwater level in the creek. In
fact, Sewer 21-31 may also be surcharged due to the downstream backwater effect.
Sometimes, a backwater profile analysis is performed on the sewer network to assess
the degree of surcharge in the sewers and manholes. In such an analysis, energy losses in
the pipes and manholes should be realistically accounted for. However, the intensity-duration-frequency-based design rainfall used in the rational method design is an idealistic,
conceptual, simplistic rain and the probability of its future occurrence is nil. The actual
performance of the sewer system varies with different actual rainstorms, each having different temporal and spatial rain distributions. But it is impossible to know the distributions of these future rainstorms, whereas the ideal rainstorms adopted in the design of the
rational method are used as a consistent measure of protection level. Although designing
sewers using the rational method is a relatively simple and straightforward matter, checking the performance of the sewer system is a far more complex task requiring thorough
understanding of the hydrology and hydraulics of watershed runoff. For instance, checking the network performance by using an unsteady flow simulation model would require
simulation of the unsteady flow in various locations in the network accounting for losses

in sewer pipes as well as in manholes and junctions (the latter will be discussed in the next
section).
Moreover, for a given sewer network layout, by using different sewer slopes, alternative designs of the network sewers can be obtained. A cost analysis should be conducted
to select the most economic feasible design. This can be done with a system optimization
model such as Illinois Least-Cost Sewer System Design Model (ILSD) (Yen et al., 1984).

74.4

HYDRAULICSOFSEWERJUNCTIONS

There are various auxiliary hydraulic structures such as junctions, manholes, weirs,
siphons, pumps, valves, gates, transition structures, outlet controls, and drop shafts in a
sewer network. Information relevant to design of most of these apparatuses are well
described in standard fluid mechanics textbooks and references, particularly in the
German text by Hager (1994) and Federal Highway Administration (FHWA, 1996). In this
section, the most important auxiliary component in modeling the sewer junctions are discussed. For sewers of common size and length, the headloss for the flow through a sewer
is usually two to five times the velocity head. Thus, the head loss through a junction is
comparable to the sewer pipe loss, and is not a minor loss.
14.4.1 Junction Classifications
A sewer junction usually has three or four sewer pipes joined to it. Under normal flow
conditions, one downstream pipe receives the outflow from the junction and other pipes
flow into the junction. However, junctions with only two or more than four joining pipes
are not uncommon. The most upstream junctions of a sewer network are usually one-way
junctions having only one sewer connected to a junction. The horizontal cross section of
the junction can be circular or square or may be another shape. The diameter or horizontal dimension of a junction normally is not smaller than the largest diameter of the joining sewers. To allow the workers room to operate, usually junctions are not smaller than
3 ft (1 m) in diameter. For large sewers, the access to the junction can be smaller than the
diameter of the largest joining sewer.
Sewers may join a junction with different vertical and horizontal alignments, and they
may have different sizes and slopes. Vertically, the pipes may join at the junction with
their centerlines or inverts or crowns aligned, or with any line of alignment in between.
There is no clearly preferred alignment that could simultaneously satisfy the requirements
of good hydraulics at low and high flows without complicating either construction cost or
design. The bottom of the junction is usually at or slightly lower than the lowest invert of
the joining sewers.
In the horizontal alignment, often the outflow sewer is aligned with one (usually the
major) inflow sewer in a straight line with other sewers joining at an angle. For cities with
square blocks, right-angle junctions are most common. Typical sewer benching and flow
guides injunctions are shown in Fig. 14.15.
With the alignment of the joining pipes and the shape and dimensions of junctions not
standardized, the precise, quantitative hydraulic characteristics of the junctions vary considerably. As a result, there are many individual studies of specified junctions, but a general comprehensive quantitative description of junctions is yet to be produced.
For the purpose of hydraulic analysis, junctions can be classified according to the following scheme (Yen, 1986a):

Directly Opposed Lolerol With Deflector

(Heod Losses Are Slill Excessive With
lhis Method. But Are Significontly
Less Thon When No Deflectors Exist)

Bend With Stroight Deflector

Depressed

Flat
Bend With Curved Deflector

Inline Uostreom Moin & 9(J Loterol

With Deflector
Half

Full

FIGURE 14.15 Junction benching of sewers and flow guides.

1. According to the geometry: (a) one-way junction, (b) two-way junction, (c) threeway junctionmerging (two pipes flow into one pipe) or dividing (one pipe flows into
two pipes), and (d) four- or more-way junctionmerging, dividing, or merging and
dividing.
2. According to the flow in the joining pipes: (a) open-channel junction (with openchannel flow in all joining pipes), (b) surcharge junction (with all joining pipes surcharged), and (c) partially surcharged junction (with some, but not all, joining pipes
surcharged).
3. According to the significance of the junction storage on the flow: storage junction or
point junction.
Hydraulically, the most important feature of a junction is that it imposes backwater
effects to the sewers connected to it. A junction provides, in addition to a volumehowever smallof temporal storage, redistribution and dissipation of energy, and mixing and
transfer of momentum of the flow and of the sediments and pollutants it carries. The precise, detailed hydraulic description of the flow in a sewer junction is rather complicated
because of the high degree of mixing, separation, turbulence, and energy losses. However,
correct representation of the junction hydraulics is important in realistic simulation and
reliable computation of the flow in a sewer system (Sevuk and Yen, 1973).

14.4.2 Junction Hydraulic Equations

The continuity equation of the water in a junction is
ds
1,Q1+Qj = ^

(14.35)

where Q1 = the flow into or out from the junction by the /-th joining sewer, being positive for inflow and negative for outflow; Q. = the direct, temporally variable water inflow
into (positive) or the pumpage or overflow or leakage out from (negative) the junction, if
any; s = the storage in the junction; and t = time. For a two-way junction, the index i =
1, 2; for a three-way junction, / = 1, 2, 3, and so on.
The energy equation in a one-dimensional analysis form is

T/2 p
"N
_i + _i + Z j j + f i ^

fly
T/2
,_ + lG^_i

(14.36)

where Z1., P1, V1 = the pipe invert elevation above the reference datum, piezometric pressure above the pipe invert, and velocity of the flow at the end of the section of the /th
sewer where it meets the junction, respectively; Hj = the net energy input per unit volume
of the direct inflow expressed in water head; K1 = the entrance or exit loss coefficient for
the /th sewer; Y the depth of water in the junction; and g = the gravitational acceleration. The first summation term in Eq. (14.36) is the sum of the energy input and output by
the joining pipes. The second term at the left-hand side of the equation is the net energy
brought in by the direct inflow. The first term to the right of the equal sign is the energy
stored in the junction as its water depth rises. The last term is the energy loss.
The momentum equations for the two horizontal orthogonal directions x and z are
1(2^,) = Jg^JA

(14.37)

2(avfe) = JgydA

(14.38)

and

where px and pz = the x and z components of the pressure acting on the junction boundary, respectively, and A = the solid and water boundary surface of the junction. The direct
flow QJ is assumed entering the junction without horizontal velocity component. The
right-hand side term of Eqs. (14.37 and 14.38) is the x or z component force, where the
integration is over the entire junction boundary surface A. The left-hand side term is the
sum of momentum of the inflow and outflow of the joining pipes. Note that for a threeway merging junction, two of the Q. 's are positive and one Q. is negative, whereas for a
three-way dividing junction, two of the Q. 's are negative.
Joliffe (1982), Kanda and Kitada (1977), Taylor (1944), and others suggested the use
of momentum approach to deal with high velocity situations. To illustrate this approach,
consider the three-way junction shown in Fig. 14.16. The control volume of water at the
junction enclosed by the dashed line is regarded as a point, and there is no volume change
associated with a change of depth within it. One of the two merging sewers is along the
direction of the downstream sewer, whereas the branch sewer makes an angle (p with it.
When one assumes that the pressure distribution is hydrostatic and the flow is steady, the
force-momentum relation can be written as

FIGURE 14.16 Control volume of junction

for momentum analysis.

Jh2A2 + Jh3A3 cos 9 - Jh^ sin (p - Y^1A1 + F

(14.39)

= PQ1V1 - PQ2V2 - PQ3V3 cos cp

where A = the flow cross-sectional area in a sewer; h = depth of the centroid of A; y
= the specific weight; p = the density of water; Q = the discharge; V = QIA = the
cross-sectional mean velocity; and F = the sum of other forces that are normally
neglected. Some of these neglected forces are the component of the water weight in the
control volume along the small bottom slope, the shear stresses on the walls and bottom, and the force due to geometry of the junction if the sewers are not invert aligned
or the longitudinal sewers are of different dimensions. The subscripts 1, 2, and 3 identify the sewers shown in Fig. 14.16, and b represents the exposed wall surface of the
branch in the control volume shown as ab in the figure. For the special case of invert
aligned sewers with the branch (pipe 3) joining at right angle, (p = 90, Eq. (14.39) can
be simplified as
A2(gh2 +\q)= A1^h1 + Vf)

(14.40)

or

& JW (SW)+ 1T"

1)
(144
(
Q1 LUJ(sW)+ iJ
'
Based on experimental results of invert-aligned equal-size pipes merging with (p = 90,
Joliffe (1982) observed that the upstream depth /I1 = h2 and proposed that
=
b
(14 42)
h = 7T
7r
cl hcl ^where hcl = the critical depth in the downstream sewer, F3 = the Froude number of the
flow in the branch sewer, and

^ = 0.999 - 0.482 (-^ 1 - 0.381 (^-T

(Qi)
(Qi)

(14.43)

b = 0.514 - 0.067 [^] + 0.197 [^ [ - 0.122 ^T

(Qi)
(Qi)
(Qi)

(14.44)

The equation describing the load of sediment or pollutants, expressed in terms of concentration c, can be derived from the principle of conservation as
^J cds + I Qf1 + Qfj = G
S

(14.45)

where G = a source (positive) or sink (negative) of the sediment or pollutants in the

junction.
Equations (14.35-14.44) are the theoretical basic equations for sewer junctions. They
are applicable to junctions under surcharge as well as open-channel flows in the joining
pipes. However, more specific equations can be written for the point-type and storagetype junctions.
14.4.3 Experiments on Three-Way Sewer Junctions and
Loss Coefficients
Proper handling of flow in sewer networks required information on the loss coefficients at the junctions. Unfortunately, there exists practically no useful quantitative
information on energy and momentum losses of unsteady flow passing through a junction. Therefore, steady flow information on sewer junction losses are commonly used
as an approximation.
Table 14.10 summarizes the experimental conditions of three-way merging, surcharging, top-open sewer junctions conducted by Johnston and Volker (1990), Lindyall (1984),
and Sangster et al. (1958, 1961). Also listed in the table are the experiments by Blaisdell
and Mason (1967), Serre et al. (1994), and Ramamurthy and Zhu (1997); these experiments were not conducted on open-top sewer junctions but on three-way merging closed
pipes. They are listed in the table as an example because these tests were conducted with
different branch and main diameter ratios and with different pipe alignments. Hence, they
may provide helpful information for sewer junctions. There exists considerably
more information on merging or dividing branched closed conduits than on sewer junctions. The reader may look elsewhere (e.g., Fried and Idelchik, 1989; Miller, 1990) for
information about centerline-aligned three-way joining pipes as an approximation to
sewer junctions.
The loss coefficients K21 and ^31 for the merging flow are defined as
Tfv 2
1 fv2
11
+h
+z
+h
z
,tj _
- \\tz
LLfS > \-\tz
) ^ \\ JJ
k

(14 46)

Figure 14.17 shows the experimental resultsfej

of (1987) and Sangster et al. (1958) and
Lindvall for the case of identical pipe size of the main and 90 merging lateral. The corresponding curves suggested by Miller (1990) and Fried and Idelchik (1989) for threeway identical closed pipe junctions are also shown as a reference. The values of the loss
coefficients in a sewer junction that is open to air on its top are expected to be slightly
higher than the enclosed pipe junction cases given by Miller because of the water volume
at the junction above the pipes.
The effect of the relative size of the joining branch pipe is shown in Fig. 14.18. The experimental data of Sangster et al. (1961) have identical upstream pipe sizes, D2 = D3 for four different values of lateral branch to downstream main pipe area ratio, A3M1. The data of Johnston
and Volker (1990) on surcharged circular open-top sewer junction are not plotted in Fig. 14.18
because the mainline pipe area ratio A2M1 = 0.41 instead of unity in the figure. Conversely,
as a comparison, the smoothed curve of ^T21 for A3M1 = 0.5 of the three-way pipe junction of
Serre et al. (1994) with A 1 =A 2 is plotted in Fig. 14.18a, and their experimental curves of ^T31
for A3M1 = 0.21 and 0.118 are plotted in Fig. 14.18b. Also shown in the figure, as reference,
are the three-way pipe junction curves for different values of A3M1 suggested by Fried and
Idelchik (1989) and Miller (1990) for identical size of main pipes, A2 = A 1 . The experiments
of Sangster et al. (1961) indicated that for a given A3M1, the effect OfA 2 M 1 on the loss coefficients is minor. Therefore, their curves should be comparable with those of Fried and Idelchik,

TABLE 14.10 Experimental Studies on Three-way Junction of Merging Surcharged Channels

Reference

Type of Junction

Shape of Channels

Channel
Slope

Sangster et al. Square,

(1958, 1961) rectangular, or
round box

Circular,
D = 3.0 in.3.75 in.
4.75 in. or 5.72 in.

Horizontal

Flushed bottom

Lindvall
(1984, 1987)

Circular,
Dmam = 144 mm,
A>/Anain = 1A
0.686, or 0.389

Horizontal

Round box

Johnston and Square box

Volker (1990)
Blaisdell and Enclosed pipe
Mason (1967) junction
Serre et al.
(1994)
Ramamurthy
and Zhu
(1997)

Enclosed pipe
junction

Circular,
Horizontal
Anain d = 70 ^Hl,
/U up/Anain,= 0.64,
>b/Dmain, = 0.91
Circular,
DJD^ = 0.25 ~ 1.0

Pipe Alignment at Junction

Vertical
Longitudinal

Type of
Flow

Remarks

Straight through and

one 90 merging channel

Steady

Center aligned

Straight through and one

90 merging channel

Steady

Also tests of opposed

lateral pipes;
tests with grate
inflow into junction
Loss coefficient
dependent on
junction diameter,
lateral pipe diameter,
and flow ratio

Flushed bottom

Centerline aligned with

slight deflector for
lateral in junction

Steady

Center or top
aligned

Straight through and one

Steady
mergin channel at 15-165
by 15 increments
Straight through and one
Steady
90 merging channel

Circular,
(Horizontal) Center aligned
^main = 4441^11'
/V/U, = 0.14, 0.23,
0.34, or 0.46
Enclosed
Rectangular,
(Horizontal) Same height
rectangular
4. 14 mm high,
conduit junction main width 91.5 mm,
branch width
20.4 mm,70.5 mm,
or 91.5 mm

Straight through and one

90 emerging branch

Steady

Reynolds number
effect insignificant

Sangsteretal. (1958)
Lindvall (1987)
Lindvall (1984)
Mil er (1990)
Fri(1989)
ed & ldelchick

Marsalek(1985)M1
Marsalek(1985)M3
deGroot&Boyd(1983)
Sangsteretal. (1958)
Lindvall (1984) Type 1
Mil er (1990)
Fri
ed & ldelchik
(1989)

FIGURE 14.17 Experimental headless coefficients for surcharged 3-way sewer junction with identical
pipe sizes and 90 merging lateral, (a) Mainline loss coefficient K21; (b) Branch loss coefficient K31.
Miller, and Serre et al. However, Fig. 14.18 depicts considerable disagreement among the different sources, indicating the need for more reliable investigations.
The joining angle of the lateral branch is a significant factor affecting the loss coefficients, particular on K^ P The values of the loss coefficients decrease if the joining angle
more or less aligns with the flow direction of the main, and increase if the lateral flow is
directed against the main. The references of Fried and ldelchik (1989) and Miller (1990)
provide some idea on the adjustment needed for the K values due to the joining angle.

Mil er (1990)
Serre etal( 1994)

Sangsteretal (1961)
Mil er
lFridelecdhi&k
Serre
etal

Sangsteretal (1961)

FIGURE 14.18 Headless coefficients for surcharged 3-way junction with 90 merging lateral of different sizes, (a) Mainline loss coefficient K21. (b) Branch loss coefficient K31.

Townsend & Prina

(1978)
Marsalek
(1985)

FIGURE 14.19 Headless coefficients for 3-way open-channel sewer junction with identical pipe
sizes and 90 merging lateral, (a) Mainline loss coefficient K21 (B) Branch loss coefficient ^31 (After
Yen, 1987).
Listed in Table 14.11 is a summary of experiments on steady flow in three-way merging open-channel junctions. Most of the studies were done with point-type junctions. The
experimental subcritical flow results of storage-type junctions by Marsalek (1985) and
Townsend and Prins (1978) are plotted in Fig. 14.19 for lateral joining 90 to the same size
mainline pipes. Yevjevich and Barnes (1970) gave the combined main and lateral loss
coefficient but not the separate coefficients, making the result difficult to be used in routing simulation. The points in the figure scatter considerably, but they are generally in the
same range of the loss coefficient values for surcharged three-way 90 merging junction
except K3, for Townsend and Prins' data. It is interesting to note that the most frequently
encountered sewer junctions are three- and four-way box junctions with unsteady subcritical flow in the joining circular sewers. None of the open-channel experiments was
conducted under these conditions. All were tested with steady flow. It is obvious that
existing experimental evidence and theory do not yield reliable quantitative information
on the loss coefficients of three-way sewer junctions. Before more reliable information is
obtained, provincially for design and simulation of three joining identical size sewers, for
K2 j a curve drawn between that of Lindvall and that of Sangster et al. can be used as an
approximation. For ^3}, the curve of Lindvall can be used. For joining pipes of unequal
sizes, the curves of Sangster et al. appear to be tentatively acceptable.
14.4.4 Loss Coefficient for Two-Way Sewer Junctions
Two-way junctions are used for change of pipe slope, pipe alignment, or pipe size.
Experimental studies on two-way, surcharged, top-open sewer junctions are listed in Table

TABLE 14.11 Experimental Studies on Three- Way Junction of Merging Open Channels
Pipe Alignment at Junction
Channel
Slope

Vertical

Longitudinal

Point

Rectangular,
identical width,
B = 4 in.

Horizontal

Flushed
bottom

Straight through Subcritical

and one merging
channel at 45
or 135

Bowers (1950) Point

Trapezoidal,
identical width,
B = 7.2 in.

0.0062, 0.012 Flushed

bottom

Behlke and
Pritchett
(1966)

Rectangular or
trapezoidal
(side slope 1:1)

Each channel Flushed

slope varied bottom
independently

Webber and
Point
Greated (1966)

Rectangular,
B = 5 in.

Horizontal

Flushed
bottom

Yevjevich and
Barnes (1970)

Circular,
Anain = 6'25 inDbT= 1.87 in.

0.00008
0.00054
0.00107

Flushed
bottom
or crown
aligned

Taylor (1944)

Type of
Junction

Type of Flow
Upstream
Downstream Remarks
Pipes
Pipe

Shape of
Channels

References

Point

Square
box

Subcritical

Also theoretical analysis

based on momentum,
good agreement with
45 merging but not
with 135 merging
Straight through Supercritical Supercritical Structure P7, hydraulic
and one merging
jumps formed upstream
channel at 51
of junction, other
structures with lateral
bottom up to 3 ft
above main
Straight through Supercritical Supercritical Use of tapered wall in
and one merging
the junction to diminish
channel at 15,
diagonal wave and
30, or 45
pile-up problems
Straight through Subcritical Subcritical Greater losses associated
and one merging
with increasing merging
channel at 30,
angles of branch channel
60, or 90
Straight through Subcritical Subcritical Greater loss for the case
and one 90
of crown aligned lateral
merging pipe

TABLE 14.11

(Continued)

Kanda and
Kitada (1977)

Point

Rectangular,
Horizontal
* main =100,200,
400mm
Bbr= 100mm;
Circular, different
sizes
Circular,
Horizontal
different sizes

Radojkovic and Point

Maksimovic
(1977)
Townsend and Rectangular Circular,
box
Prins (1978)
Anain = 161^11'
DbT = 102 mm

Flushed
bottom

Flushed
bottom

Straight through Supercritical Supercritical Also theoretical analysis

and one merging
based on momentum
channel at 30
60, or 90

Straight through
and one merging
channel
Less than 0.01 Invert drop Straight through
and one merging
across
junction box channel at 45
or 90
Rectangular,
Lin and Soong Point
Invert drop Straight through
Horizontal
B = 457 mm
(15 or
and one 90
(1979)
merging channel
18mm)
Circular,
Joliffe (1982) Point
Flushed
Straight through
Horizontal,
equal diameter,
and one 90
0.0001, 0.0075, bottom
D = 69mm
merging channel
0.005, or 0.01
Rectangular,
Point
Channel slope Not available Straight through
Best and
and one merging
Reid (1984)
identical width,
adjustable
B = 0.5 ft
channel at 15,
to achieve
45, 70, or 90
equilibrium
water depth
Marsalek
Square box Circular,
Flushed
Horizontal
(1985)
or round box identical diameter
bottom

Supercritical Supercritical Junction zone with

expansion and without
expansion
Subcritical Subcritical Simple junction box
and special junction
box with flow deflector
Subcritical

Subcritical

Subcritical
Subcritical

Sub- or
supercritical
Subcritical

Subcritical

Subcritical

Energy loss coefficient

as a function of lateral
to totalflowrate ratio
Upstream flow depth
depends on critical depth
in downstream pipe
Systems operated for
O.K F < 0.3

TABLE 14.12 Experimental Studies on Straight-Through Two- Way Open-Top Junction of Surcharged Channels
References

Type of
Junction

Shape of Channels

Channel Slope

Pipe Alignment at Junction

Vertical Longitudinal

Type of
Flow

Remarks

Sangster et al.
(1958, 1961)

Square,
rectangular,
or round box
Rectangular
box

Circular,
D = 3.0, 3.75,
4.75, or 5.72 in.
Circular, identical
diameter, D = 6 in.

Horizontal

Flushed
bottom

Steady

Also test of grate inflow

into junction

0.0094-0.0192

Flushed
bottom

Steady

Also studied part-full

supercritical flow

Rectangular
box or round
box
Rectangular
box or round
box
Round box

Circular, identical
diameter,
D = 102 mm
Circular, identical
diameter,
D = 88 mm
Circular, identical
diameter,
D = 144 mm
Circular, identical
diameter,
D = 6 in
Circular, identical
diameter,
D = 88 mm
Circular, identical
diameter,
D = 90 mm

0.002 and 0.010

Flushed
bottom

Horizontal

Flushed
bottom

Horizontal

Center
aligned

Straight through

Steady

Horizontal

Flushed
bottom

Straight through

Steady

Horizontal

Center
aligned

Straight through

Steady

Manhole diameter < 2.26D,

headloss coefficient constant
for given junction geometry
Three types of benching

Horizontal

Center
aligned

Straight through

Steady

Four types of benching

Ackers (1959)

Archer et al.
(1978)
Howarth and
Saul (1984)
Lindvall
(1984)
Marsalek
(1984)

Square box
or round box

Johnston and
Volker (1990)

Square box

Bo Pedersen
and Mark
(1990)

Round box

Straight
through
or 90 bend
Straight through
or 45 bend in
junction or 52
bend downstream
of junction
Straight through,
or 30 or 60
bend in junction
Straight through

Steady
Steady or
unsteady

Loss coefficient increases as

junction diameter increases

14.12. All the experiments were conducted with the same size upstream and downstream
pipes joining the junction. Only Sangster et al. (1958, 1961) tested also the effect of different joining pipe sizes. These experimental results show that for a straight-through, twoway junction, the value of the loss coefficient is usually no higher than 0.2. Alignment of
the joining pipes and benching in the junction are also important factors to determine the
value of the loss coefficient.
Figure 14.2Oa shows the headloss coefficiet of a surcharged two-way open-top junction connecting two pipes of identical diameters aligned centrally given by the experiments of Archer et al. (1978), Howarth and Saul (1984), Johnston and Volker (1990) and
Lindvall (1984). Noticeable is the swirl and instability phenomena when the junction submergence (junction depth to pipe diameter ratio) is close to two and the corresponding
high head loss coefficient. The ranges of loss coefficient given by Ackers (1959),
Marsalek (1984), and Sangster et al. (1958) are also indicated in Fig. 14.2Oa, but the data
on the variation with the pipe-to-junction-size ratio was not given by these investigators.
Sangster et al. (1958) also tested the effect of different sizes of joining pipes for surcharged two-way junction. Some of their results are plotted in Fig. 14.2Ob. They did not
indicate a clear influence of the effect of the size of the junction box. However, Bo
Pedersen and Mark (1990) demonstrated that the loss coefficient of a two-way junction
can be estimated as a combination of the exit headloss due to a submerged discharging jet
and the entrance loss of flow contraction. They suggested that the loss coefficient K
depends mainly on the size ratio between the junction and the joining pipes of identical
size. For an infinitely large storage junction, the theoretical limit of K is 1.5. For the junction-diameter to pipe-diameter ratio, DJD less than 4, they proposed to estimate the K
values according to benching as shown in Fig. 14.21.
14.4.5 Storage Junctions
For a storage- (or reservoir-) type junction, the storage capacity of the junction is relatively large in comparison to the flow volume and hydraulically it behaves like a reservoir. A water surface, and hence, the depth in the junction can be defined without great
difficulty. A significant portion of the energy carried in by the flows from upstream sewers is dissipated in the junction. If the horizontal cross-sectional area of the junction A7
remains constant, independent of the junction depth 7, the storage is s = AjY. Hence, dsldt
= Aj(dY/dt) = Aj(dH/dt), where H = Y + Z = the water surface elevation above the reference datum, and Z = the elevation of the junction bottom. Therefore, from Eq. (14.35),
^Q,+ Qj = Aj ^-

(14.47)

Either the energy equation (Eq. 14.36) or the momentum equations [Eqs. (14.37) and
(14.38)] can be used as the dynamic equation of the junction. If the energy loss coefficient
K1 in Eq. (14.36) can be determined, use of the energy equation is appropriate. On the
other hand, if the pressure on the junction boundary can be determined, the momentum
equation is also applicable. If the junction were truly a large reservoir, both the loss coefficients and the pressure could reasonably be estimated on the basis of information on
steady flow entering or leaving a reservoir.
Customarily for the convenience of computation, instead of Eq. (14.36), the junction
energy relationship is divided for each joining sewer by relating the total head of the sewer
flow to the total head in the junction. Assuming that the energy contribution from the
direct lateral inflow Qj is negligible, the component of Eq. (14.36) for each joining sewer
/ can be written as

Marsalek, half benched

Lindvall
Howarth &
Saul
Archer
etal.
Johnston &
Volker

Sangster et al.
(1958, Eq.7)

FIGURE 14.20 Headless coefficient for surcharged 2way open-top straight-through sewer junction,
(a) Same size sewers upstream and downstream, (b)
Different joining pipe sizes. (After Yen, 1987).

Shape

FIGURE 14.21 Effect of benching on loss coefficient of surcharged two-way sewer junction
according to Bo Pedersen and Mark (1990).
H=(I- K1)(VfVg) + (P1Ii) + Z1

(14.48)

For open-channel flow in the joining pipes, the piezometric term P/y is
P/y=ht

(14.49)

where ht is the open-channel flow depth of the ith pipe at the junction. It should be cautioned that Eq. (14.48) is applicable only when there is no free surface discontinuity
between the junction and the sewer. In other words, they are applicable to Cases B and D
in Fig. 14.9 and all four cases in Fig. 14.8. The flow equations for these pipe exit and
entrance cases are given in Table 14.13.
14.4.6 Point Junctions
A point-type junction is the one whose storage capacity is negligible, and the junction is
treated as a single confluence point. Hence, Eq. (14.35) is reduced to
Za + QJ. = O

(14.50)

For subcritical flow in the sewers emptying into the point junction, the flow can discharge freely into and without the influence of the junction only when a free fall exists
over a nonsubmerged drop at the end of the pipe (Case A in Fig. 14.9). Otherwise, the subcritical flow in the sewer is subject to backwater effect from the junction. Since the junction is treated as a point, the dynamic condition of the junction is usually represented by
a kinematic compatibility condition of common water surface at the junctions for all the
joining pipes without a free fall (Harris, 1968; Larson et al., 1971; Roesner et al., 1984;
Sevuk and Yen, 1973; and Yen, 1986a). Thus, by neglecting the junction storage and for
subcritical sewer flow into the junction,
ht = hic

if Z,.+ /*,,> Z 0 + /*0

(Case A in Fig. 14.9)

(H51)

TABLE 14.13 Storage Junction Flow Equations

Case in
Fig. 14.8
or 14.9

Condition

Equation

Remarks

For reservoir
junction K = I

For sewerflowinginto junction

D

Submerged sewer exit

H>Z + D

H=(I- K)(V2Kg) + (P/y) + Z

Free fall at sewer exit

h = hc

Subcritical flow
in sewer

Supercritical
flow in sewer

H<Z+hc
F< 1
Z + D>H
H>Z + hc
F< 1
Z + D>H
Z + hc>H
F> 1
H>Z+hc
but F > 1
hc> h

Supercritical flow in
sewer, hydraulic jump
in junction
For outflow from junction into sewer
I

Subcritical flow
in sewer

II

Supercritical flow
in sewer
Submerged sewer
entrance, openchannel flow in sewer
Surcharged sewer

III
IV

h>hc
H<Z + D + S(V2 2Kg)
H>Z+hc + (Vc /2g)
H < Z + D + S(Vc2/2)
(h = hc)
H> Z + D + S(V2Kg)
h<D
H> Z + D + 8(V2/2g)
h> D

H=(I- K)(V1IIg) + h + Z

For reservoir
junction K = I

H = h +Z

Upstream control
for sewer
Upstream control for
sewer; H determined
by other sewers

H = (1 - K)(V2SIg) + h + Z

5-1

H = ( I - K)(V2Kg) + hc + Z

8-1

Q = AVVV
Vv = Cv V2g(H - Z)

8-1

H=(I- K)(V2Kg) + (P/y) + Z

8-1

ht + Z. = h0 + Z0

otherwise

(Case B in Fig. 14.9)

where Z0 and h0 = the invert elevation and depth of the flow at the entrance of the downstream sewer taking the outflow from the junction, respectively.
For a supercritical flow in a sewer flowing into a point junction, Case C in Fig. 14.9
would not occur. Only a subcase of Case B in Fig. 14.9 with ht < hic exists where Eq.
(14.51) applies. For Case D of Fig. 14.9 with submerged exit,
Zi + WY) = (PJD + Z0> ZI + D1

(14.52)

The flow in the downstream sewer, which takes water out from the junction, may be
subcritical, supercritical, or submerged, depending on the geometry and flow conditions. The flow equations are the same as the storage junction outflow Cases I-IV given
in Table 14.13.

74.5.

HYDRAULICS OF A SEWER NETWORK

Hydraulically, sewers in a network interact, and the mutual flow interaction must be
accounted for to achieve realistic results. In designing the sewers in a network, the constraints and assumptions on sizing sewers as discussed in Sec. 14.3.4 should be noted.
The rational method is the most commonly and traditionally used method for the
design of sewer sizes. As described in Sec. 14.3.4, each sewer is designed independently
without direct, explicit consideration of the flow in other sewers. This can be done
because to design a sewer, only the peak discharge, not the entire hydrograph, of the
design-storm runoff is required. As previously explained in Sec. 14.3.4, each sewer has its
own design storm. The information needed from upstream sewers is only for the alignment and bury depth of the sewer and the flow time to estimate the time of concentration
for the determination of the rainfall intensity / for the sewer to be designed.
Contrarily, in simulation of flow in an existing or predetermined sewer network for
urban stormwater control and management, often the hydrograph, not merely the peak discharge, is needed, and a higher level of hydraulic analysis that considers the interaction of
the sewers in the network is required. This network system analysis involves combining
the hydraulics of individual sewers as described in Sec. 14.3, together with the hydraulics
of junctions described in Sec. 14.4.
A sewer network can be considered as a number of nodes joined together by a number
of links. The nodes are the manholes, junctions, and network outlets. The links are the
sewer pipes. Depending on their locations in the network, the nodes and links can be classified as exterior or interior. The exterior links are the most upstream sewers or the last
sewer having the network exit at its downstream end. An exterior sewer has only one end
connected to other sewers. Interior links are the sewers inside the network that have both
ends connected to other sewers. Exterior nodes are the junctions or manholes connected
to the upstream end of the most upstream sewers, or the exit node of the network. An exterior node has only one link connected to it. Interior nodes (junctions) inside the network
have more than one link connected to each node.
A systematic numeric representation of the nodes or links is important for computer
simulation of a network. One approach is to number the links (sewers) according to the
branches and the order of the pipes in the branch, similar to Horton's numbering of river
systems. Another approach is to identify the links by the node numbers at the two ends of
the link. Using the node number identification system, a numbering order technique similar to topographic contour lines called the isonodal line method, can be used. This method
Next Page

CHAPTER 15
HYDRAULIC

DESIGN OF

CULVERTS AND

HIGHWAY

STRUCTURES

I. Kaan Tuncok
Stanley Consultants, Inc.
Phoenix, Arizona
Larry W. Mays
Department of Civil and Environmental Engineering
Arizona State University
Tempe, Arizona

15.1

INTRODUCTION

The aim in highway drainage is to prevent on-site water standing on the surface and convey the off-site storm runoff from one side of the roadway to the other. To accomplish the
off-site drainage either a culvert or a bridge can be used. Culverts are closed conduits in
which the top of the structure does not form part of the roadway. Bridges are mainly provided for large streams and the road is practically a part of the span or drainage structure.
National Bridge Inspection Standards (NBIS) define bridges as those structures that have
at least 20 ft of length along the roadway centerline. The main operational differences
between culverts and bridges may be described in terms of:

economics,
hydraulics,
structural aspects,
maintenance attention requirements.

Economically, the initial and operating costs of culverts are considerably less than that
of bridges. Thus, the total investment of public funds for culverts constitutes a substantial
share of the highway budget in relation to the investment for bridges.
The hydraulic properties for both the culverts and bridges can be computed by using
the conservation of mass, energy, and momentum principles defined in Chap. 3. In the
case of bridges the magnitude of energy losses must be carefully computed. One part of
this loss is due to the contraction and expansion that occur in reaches immediately
upstream and downstream from the bridge. The other part is due to the losses at the structure itself and is calculated with either the momentum or energy principles (U.S. Army
Corps of Engineers, 1990). To compute the structural losses, the flow must be identified

as low flow, pressure flow, or weir flow and equations based on the momentum principle,
orifice flow, and a standard weir equation can be used, respectively, to separate the associated losses. The hydraulics of culverts is complicated and the concept of inlet control
and outlet control is commonly used to simplify the analysis. Inlet and outlet control culverts have a barrel capacity larger and smaller than the capacity of the culvert entrance,
respectively. This chapter will focus on the hydraulic design of culverts and stream stability at highway structures.
Structurally, culverts are designed with a heavy dependence on good and proper backfill around the perimeter of the culvert conduit. Often culverts must accommodate significant dead loads (i.e., embankment loads) in addition to the live loads (i.e., vehicles and
pedestrians). The earth load transmitted to a culvert is largely dependent on the type of
installation, and the four common types are:
trench, used in the construction of sewers, drains and water mains,
positive projecting embankment, used in relatively flat streambed or drainage path,
negative projecting embankment, used in relatively narrow and deep streambed or
drainage path,
induced trench, used in the construction of culverts placed under high embankment.
For the details of these different installation types the reader should refer to the
American Concrete Pipe Association (1992). Bridges barely have significant embankment
loads and therefore the live load on bridges is the predominant structural consideration.
Maintenance requirements for culverts are considerable. Constant efforts must be
made to ensure:

clear and open conduits,

protection or repair against corrosion and abrasion,
repair and protection against local and general scour,
structural distress repair,
maintenance of traffic safety devices pertinent to the culvert.

Bridges require some of the same items of maintenance attention, but not to the extent
required by culverts.

75.2

DESIGNPARAMETERS

To better understand the design of culverts, basic design parameters must be carefully studied. The headwater, outlet velocity, and tailwater are factors of significant importance.
15.2.1 Headwater and Tailwater
Headwater is the vertical distance from the culvert invert at the entrance, to the energy
line (depth + velocity head) of the headwater. Because of the low velocities in most
entrance pools and the difficulty in determining the velocity head for all flows, the water
surface and the energy line at the entrance are assumed to be coincident (American Iron
and Steel Institute, 1983). Therefore, headwater is the depth of water at the upstream face
of the culvert. The main reason for the accumulation of water is to build up the energy
required to pass the water through the culvert opening and to overcome the friction,

Control Section
FIGURE 15.1 Typical inlet control flow condition. dc = critical depth; HW = headwater; TW = tailwater; WS = water surface. (From Normann et al., 1985)
entrance, and exit losses along the culvert. By this process, the potential energy accumulated in water is transformed into kinetic energy through the culvert. Figure 15.1 shows a
typical inlet control flow condition with the headwater depicted on it.
The designer should not ignore the design limitations on the maximum or allowable
headwater (AHW). AHW is the level to which the culvert headwater may rise before causing an unwanted inundation or damage under the circumstances of the design flood
(Reagan, 1993). The level of AHW may be based on the fill heights, elevation level of
property, or features which would be affected by ponded water due to culvert headwater
levels. In cases with a heavy load of debris in the runoff, provision for its passage (or
retention) must be made. The designer should avoid the use of very high allowable headwater that may result in unacceptable turbulence and objectionable velocities of flow into
and through the culvert. The criteria in establishing the AHW can be summarized as:
(1) AHW should not give damage to upstream properties,
(2) AHW should be below the traffic lanes of interest or lower than the shoulder,
(3) AHW should be lower than the low point in the road grade,
(4) AHW should not be equal to the elevation where flow diverts around the culvert.
The ponding effect at the upstream face of the culvert may cause an attenuation in the
peak discharges, therefore resulting in smaller culvert size requirements for the design.
The tailwater is the depth of water downstream of the culvert measured from the outlet invert. It can be a very significant factor in culvert hydraulic operation. The depth of
the tailwater can affect the outlet velocity, and depending on the character of the culvert
flow it can also affect the operating headwater on the culvert.
15.2.2 Outlet Velocity
For design purposes, the outlet velocity should be similar in magnitude to the velocity in
the channel to provide appropriate protection at the downstream end (American Concrete
Pipe Association, 1992). But generally due to the constriction inside the culvert opening,
the outlet velocity is higher than the channel velocity. High channel velocities can be mitigated in several different ways such as the following.
(1) Stabilization of the channel.
(2) Construction of energy dissipation structures, such as hydraulic jump basin, drop
structure, stilling basin, riprap, and sill. For a detailed description of the individual
methods, the designer should refer to Corry et., al. (1983).
(3) Changing the size or the roughness of the culvert.

A more detailed discussion of outlet velocity will be given in Sec. 15.3.3. Another
important factor is the minimum velocity that must be maintained within the culvert for an
efficient operation. The ideal minimum velocity should be adjusted such that the sediment
particles transported through the culvert should not be allowed to settle.
The culvert barrel should result in a tractive force, T, greater than the critical T of the
transported stream bed material at low flow rates. In case of unknown stream bed material size, a culvert velocity of 2.5 ft/s should be maintained within the culvert. If clogging is
probable, installation of a sediment trap or sizing the culvert to facilitate the cleaning
should be considered.

75.3

CHARACTERISTICSOFFLOW

A culvert is a type of structure that can transmit water as full or partly full. Full flow is not
common for culverts unless governed by a high downstream or upstream water surface elevation. Full flow can be described by the fundamentals of pipe flow. Partly full flow culverts
follow the rules of open-channel flow and need to be classified as either subcritical or supercritical flow to accomplish the design procedure. The reader should refer to Chap: 3 for the
details of open-channel flow and the related terminology that will be used in this chapter.
An exact theoretical analysis of culvert flow is difficult due to the following factors:
(l)the culvert may have both gradually and rapidly varied flow zones and non-uniform
flow conditions must be considered to analyze the flow,
(2) the location of hydraulic jump, if it occurs, must be identified,
(3) the results of hydraulic model studies must be applied,
(4) the change in the flow type as a function of discharge and hydraulic characteristics of
the culvert must be studied.
In the culvert design procedures of this chapter, the concept of control section, as related to the relationship between the flow rate and the upstream water surface elevation will
be used. A culvert can operate either under inlet or outlet control depending on whether
the barrel or the inlet has a greater hydraulic capacity. Under inlet control, the cross sectional area of the culvert barrel, the inlet geometry, and the amount of headwater at the
entrance are of primary importance. Outlet control involves the additional consideration
of tailwater elevation in the outlet channel, the slope, the roughness and the length of the
culvert barrel.
When a culvert operates under inlet control, the barrel will flow partly full and depending on the headwater, the flow may be divided either by using the weir or orifice equations. For outlet control, the culvert barrel is intended to flow full for design conditions.
A more detailed description of both the inlet and outlet control culverts will be described
in the following sections.
15.3.1 Inlet Control
A culvert operates under inlet control when the barrel hydraulic capacity is higher than
that of the inlet. A typical flow condition is critical depth near the inlet and supercritical
flow in the culvert barrel. Depending on the tailwater, a hydraulic jump may occur downstream of the inlet. The barrel geometry and roughness have no direct influence on the
hydraulic characteristics of the culvert.

Due to the severe constriction of the flow at the culvert entrance, the inlet configuration have a significant effect on the hydraulic performance. To increase the capacity of the
culvert the designer can use beveled inlet edges, side-tapered inlet, or slope-tapered inlet.
The details of different inlet conditions can be found in Sec. 15.7.4.
Typically both the inlet and outlet control culverts can be studied in four different types
and can be summarized as:
(1)
(2)
(3)
(4)

Inlet and outlet unsubmerged,

Inlet unsubmerged, outlet submerged,
Inlet submerged, outlet unsubmerged,
Inlet and outlet submerged.

Culverts operating under steep slope regimes usually operate with inlet control of the
headwater, and as a result supercritical flow is a very common flow condition. In cases
where the specific energy within the culvert barrel is in the vicinity of minimum specific
energy, the flow depth is sensitive to the changes (i.e., roughness) in the culvert. In application, if the normal depth and critical depth are within 5 percent of each other, hydraulic
computations should be performed separately to identify the worst case for the design. In
the case of inlet control, these four cases are depicted in Fig. 15.2. The flow will be governed by weir flow when the entrance is unsubmerged, and by orifice flow in the case of
a submerged entrance. Laboratory investigations showed that for an entrance, either inlet
or outlet control, to be considered as unsubmerged, the headwater must be less than a critical value, //cr,and can be approximated as (1.2-1.5) times the height of the barrel.
Another important factor is the identification of the culvert as hydraulically short or long
(Chow, 1959), which depends on the length, slope, size, entrance geometry, headwater,
entrance, and outlet conditions.
The most typical case of inlet submerged and outlet unsubmerged condition is shown
in Fig. 15.2c. As soon as the flow enters the culvert, critical depth is observed and the freeflow conditions at the downstream end results in supercritical normal depth. Supercritical
flow is maintained along the culvert barrel. A very similar condition with both inlet and
outlet unsubmerged is shown in Fig. 15.2a. The difference from the previous condition is
the low headwater at the culvert entrance, but again the flow passes through critical depth
at the culvert entrance. In the case of inlet unsubmerged and outlet submerged (Fig.
15.2), even though the high downstream water surface depth cannot enforce outlet control conditions, a hydraulic jump is observed within the culvert. Similar to the previous
cases the flow passes through critical depth at the culvert entrance, but the partly full flow
is maintained only upstream of the hydraulic jump (ARMCO, 1950). An uncommon condition shown in Fig. 15.2J depicts submerged inlet and outlet with partly full flow. Similar
to the previous case, a hydraulic jump will be observed within the culvert.
15.3.2 Outlet Control
The culvert will operate under outlet control conditions if the culvert barrel has a smaller
hydraulic capacity than the inlet does. The flow regime is always subcritical; as a result,
the control of the flow is either at the downstream end of the culvert or further downstream of the culvert outlet, depending on the depth of the tailwater. (Portland Cement
Association, 1964). Typical flow conditions include a full or partially full culvert barrel
for all or part of its length.
Similar to the inlet control culverts, four types of control are available for the outlet
control culverts (Fig. 15.3). In the first flow condition in which both the inlet and the out-

.WATER SURFACE

WATER SURFACE

WATER SURFACE

MEDIAN DRAIN

WATER SURFACE
FIGURE 15.2 Types of inlet control: (A) outlet unsubmerged, (B) outlet submerged, inlet unsubmerged, (C) inlet submerged (D) outlet submerged. (From
Normann et al, 1985)
let are unsubmerged, the flow is subcritical and the culvert barrel is partly full over its
entire length. The unsubmerged flow condition results in critical depth at the culvert outlet. For the second flow condition, in which the inlet is unsubmerged and outlet is submerged, headwater is an important factor. Typically, the water surface will drop at the culvert entrance and experience a contraction through the culvert opening. This is depicted
in Fig. 15.3B and is a result of shallow headwater. The unusual case shown in Fig. 15.3C
can be established with high submergence at the culvert inlet. Although there is no tailwater, the submergence at the inlet maintains the pressure flow along the culvert barrel.
The condition where the inlet is submerged and outlet is unsubmerged (Fig. 15.3D), is a
result of tailwater that does not submerge the culvert depth but is greater than the critical
depth at the outlet. Finally, the common condition where both the inlet and the outlet are
submerged (Fig. 15.3A) results in pressure flow through the culvert opening.
In all of these flow types, in addition to the factors that affect the inlet control culverts, the
barrel characteristics such as roughness, area, length, and slope also play an important role.
Hydraulics To evaluate the outlet control hydraulics the condition of full flow in the
culvert barrel will be used. The energy equation, Eq. (15.1), must incorporate the losses

FIGURE 15.3 Types of outlet control. (From Normann et al., 1985)

due to entrance (he), friction (hf), exit (AJ, bend (hb), junctions (A,), and grates (hg) and
can be written as
HL = he + hf+hex + hb + hi + hs

(15.1)

where HL = total energy loss (ft).

The velocity head, (Av), can be expressed as
(YZ yi \
hv = ^ 2g a)
where V = velocity of flow in the culvert barrel (ft/s), and can be computed as

(15.2)

V =-J

(15.3)

where Qb = discharge per barrel (ft3/s), A= cross-sectional area of flow with the barrel
full (ft2), V0 = approach velocity of flow (ft/s), and g = acceleration due to gravity (32.2
ft/s2).
For design purposes, the approach velocity is usually ignored and the velocity head
can be expressed as

V2
hv = ^

(!5-4)

The entrance loss can be computed in terms of velocity head as

(V2I
* = .\fr]

<15'5)

where ke = entrance loss coefficient (Table 15.1).

Similar to the entrance loss, the exit loss can be computed as

V2
* = 2

<15'6)

TABLE 15.1 Entrance Loss Coefficients-Outlet Control, Full or Partly Full

^-[j]
Type of Structure and Design of Entrance

Coefficient ke

Pipe, concrete
Mitered to conform to fill slope
End-section conforming to fill slope*
Projecting from fill, sq. cut end
Headwall or headwall and wingwalls
Square-edge
Rounded (radius = 1/12D)
Socket end of pipe (groove-end)
Projecting from fill, socket end (groove-end)
Beveled edges, 33.7 or 45 bevels
Side- or slope-tapered inlet

0.5
0.2
0.2
0.2
0.2
0.2

Pipe, or pipe-arch, corrugated metal

Projecting from fill (no metal)
Mitered to conform to fill slope, paved or unpaved slope
Headwall or headwall and wingwalls aquare-edge

0.9
0.7
0.5

0.7
0.5
0.5

TABLE 15.1 Entrance Loss Coefficients-Outlet Control, Full or Partly Full

*-*
Type of Structure and Design of Entrance

Coefficient ke

End-section conforming to fill slope*

Beveled edges, 33.7 or 45 bevels
Side-or slope-tapered inlet

0.5
0.2
0.2

Box, reinforced concrete

Wingwalls parallel (extension of sides)
Square-edges at crown
Wingwalls at 10-25 or 30-75 to barrel
Square-edged at crown
Headwall parallel to embankment (no wingwalls)
Square-edged on three edges
Rounded on three edges to radius of 1/12 barrel
dimension, or beveled edges on 3 sides
Wingwalls at 30-75 to barrel
Crown edge rounded to radius of 1/12 barrel
dimension, or beveled top edges
Side- or slope-tapered inlet

0.7
0.5
0.5
0.2
0.2
0.2

Source : From Normann et al., (1995).

*Note:
"End section conforming to fill slope," made of either metal or concrete, are the sections commonly available from manufacturers. From limited hydraulic tests they are equivalent in operation to a headwall in both inlet and
outlet control. Some end sections, incorporating a closed taper in their design have a superior hydraulic performance.
These latter sections can be designed using the information given for the beveled inlet.
The friction loss can be expressed as
hf = Sf L
(15.7)
where Sf = friction slope, and can be computed by a manipulation of Manning's equation
(Brater, et al 1996) and L = length of the culvert barrel (ft).

-\

*f

&

[1.486 A R061I

in American customary units where, n = Manning's roughness coefficient, A = cross-sectional area (ft2), R = hydraulic radius (ft), A and R are based on full-flow conditions.
When the expression for friction slope is inserted into Eq. (15.7), the expression for
friction loss can be written as
HT^)S
The bend losses are also a function of velocity head and can be computed as
(V2}
* = K*]fr)

<15-9>

TABLE 15.2 Loss Coefficients for Bends

Radius of Bend
Equivalent Diameter
(ft)

Angle

90
0.50
0.30
0.25
0.15
0.15

1
2
4
6
8

of

45
0.37
0.22
0.19
0.11
0.11

Bend,0

22.5
0.25
0.15
0.12
0.08
0.08

Source: From Normann, et al. (1985).

where Kb = bend loss coefficient (Table 15.2).
Junction losses, hj9 are significant whenever more than two culverts flow into a single
culvert downstream, as shown in Fig. 15.4. In such cases the energy and momentum principles can be used to calculate hr
The energy equation for the junction can be written as
*,= y + hm~ hvd

(15.10)

where /i; = junction loss in the main culvert (ft) hvu = velocity head in the upstream culvert (ft), hvd = velocity head in the downstream culvert (ft),and y = change in hydraulic
grade line through the junction (ft).
The momentum principle can be used to compute y as

TRIBUTARY STREAM

MAIN
STREAM

PIPE 2

PIPE I

FIGURE 15.4 Culvert junction. (From Normann et al., 1985)

y=

Q2V2-Q1V1-Q^CQsQ
0.5(A^A2)S

(15.11)

where A1 and A2 = cross-sectional areas for culverts 1 and 2, respectively; V1 and V2 =

velocity for culverts 1 and 2, respectively; and 9 = angle between culverts 1 and 3.
If only the entrance, friction, and exit losses are considered, and the respective equations are inserted into Eq. (15.1), the total headless can be expressed in American customary units as
(
29n2L\ V2
H = \l + k. + -jjiir| 2J = ha + he + hf

(15.12)

By using the schematic in Fig. 15.5 that depicts the hydraulic and energy grade lines,
the energy equation with the entrance, exit, and friction loss terms for a full flow, can be
written between the upstream and downstream ends of a culvert system as
HW0 + ^ = TW + ^ + H

(15.13)

where HW0 = headwater depth above the outlet invert, TW = tailwater depth above the
outlet invert, and the subscripts u and d denote the upstream and downstream conditions,
respectively.
Because of the smaller values of upstream and downstream velocities, compared to the
culvert velocity, they can be neglected. Then the energy equation can be expressed as
HW= TW + H-S0L

(15.14)

where
S0L= drop in elevation from the inlet invert to the outlet invert
A practical way to compute the headloss H is to use the outlet control nomographs
(Figs. 15.6 and 15.7). Outlet control nomographs presented for full flow can also be used

SECTION
FIGURE 15.5 Full-flow energy and hydraulic grade lines. (From Normann et al.,
1985)

SECTION

for the partly full flow for the computation of H. Details on the use of outlet control nomographs will be given in Sec. 15.4.2.2.
Equations (15.1-15.14) were developed for full barrel flow in which TW is greater
than or equal to the culvert diameter, D. However in the case of partly full flow, backwater calculations may be required starting at the downstream end of the culvert and proceeding upstream. During the backwater computations if the hydraulic grade line cuts the
top of the barrel, full flow will be observed upstream of this intersection point. FHWA has
also developed an approximate method to overcome the tedious backwater computations
for partly full flow. They found out that the hydraulic grade line at the culvert outlet is at
a point between the critical depth and the culvert diameter, and can be computed as (dc +
D)/2. It was oncluded that the observed TW should only be used if it is higher in magnitude than (dc + D)/2. In such a case the following equation should be used:
HW=h0 + H-S0L
where h0 = max [TW, (dc + D) 12].

(15.15)

The use of these two equations gives reasonable results for HW depths greater than
0.75Z).

HEAD (H) IN FEET

DIAMETER (D) IN INCHES

DISCHARGE (O) IN CFS

SUBMERGCO OUTLET CULVCMT FLOWING FULL

FIGURE 15.6 Head for concrete pipe culverts, (n = 0.012) (From Normann
et al., 1985)

HEAD (H) IN FEET

AREA OF RECTANGULAR BOX IN SQUARE FEET

DIMENSION OF SQUARE BOX IN FEET

UISUMANUt IQJ IN CFS

SUBMERGED OUTLET CULVERT FLOWING FULL

For
owtlcrdMcri
crownbidnotin Mithebmidctlon
rd, com
putt HW by
nuthodi
proctdurt

FIGURE 15.7 Head for concrete box culverts flowing full (n = 0.012)
(From Normann, et al., 1985)
15.3.3 Outlet Velocity
Outlet velocity is an important factor to define the type of outlet protection. Because of
the decreased area of flow in the culvert, the velocity increases through the culvert barrel
and results in higher velocities than that of the natural stream. Because of the high velocities, outlets must be protected by using riprap or an energy dissipator. To accurately compute the outlet velocity, the type of control must be defined.
For inlet control culverts, either an exact or approximate method can be used. In the
exact method, water surface profile computations start at the upstream and proceed downstream and the velocity is computed by using the cross-sectional area at the exit. For the
approximate method, normal depth is assumed to occur at the culvert outlet as depicted in
Fig. 15.8. This method is more commonly used and results in more conservative estimates
of the outlet velocity. Normal depth can be computed by using Manning's equation.
For outlet control culverts, the outlet velocity depends on the outlet geometry and the
magnitude of tailwater depth with respect to either the critical depth or the barrel diameter. The following procedure can be used to define the tailwater depth and to compute the
outlet velocity with the given geometry.

If
If
If

TW < dc
dc < TW < D
TW > D

Depth at culvert outlet = dc.

Depth at culvert outlet = TW.
Depth at culvert outlet = D.

The schematic in Fig. 15.9 also summarizes these three flow conditions.
15.3.4 Roadway Overtopping
Roadway overtopping starts soon after the headwater elevation reaches the top of roadway elevation. To compute the magnitude of the overtopping flow, the following broadcrested equation can be used:
Qr=CjL(HW,)1*

(15.16)

where Qr = overtopping flow rate (ftVs), Cd = overtopping discharge coefficient = Ic1 Cr,
the discharge coefficient Cr, and the submergence factor Ic1, are presented in Fig. 15.10; L
= length of roadway crest, (ft), HWr = the upstream depth, measured above the roadway
crest, (ft)

AREA
OF FLOW PRISM BASED
BARREL
DEPTHON
EQUAL TOGEOMETRYANO
NORMAL DEPTH
FIGURE 15.8 Inlet control outlet velocity. (From Normann et al., 1985)

AREA OF FLOW PRISM BASED ON BARREL

GEOMETRY AND d
FIGURE 15.9 Outlet control outlet velocity. (From Normann et al., 1985)

The main reason for this type of equation is the similarity of this flow to the
broad-crested weir flow. One of the most important factors in the computation of Qr is a
good estimate of the length of roadway crest. The length can be established in two different ways. The first method, although time consuming, results in more accurate estimates
of the length, especially if the elevation of the roadway crest varies. It is based on an
approximation of the vertical curve as a series of horizontal segments and can be applied
to cases where the crest is defined by a roadway sag vertical curve (for definition see Sec.
15.7.3) and is depicted in Fig. 15.11. Then the flow over each segment is calculated for a
given headwater and finally the flows for each segment are added together to determine
the total flow. In the second method, which is more practical, the length can be represented by a single horizontal line at a constant roadway elevation. For correct estimates
of Q1, this horizontal line must span the roadway profile accurately. The depth of flow
used in this method is the average depth of upstream pool above the roadway.
Given the headwater, length and the coefficient, Qr can be computed by using Eq.

(A) Discharge Coefficient for

(C) Submergence Factor

(B) Discharge Coefficient for

FIGURE 15.10 Discharge coefficient and submergence factor for roadway overtopping.
(From Flood Control District of Maricopa County, 1996)

MOAOWAY VERTICAL CVWVE

ELEVATION Of CREST

FIGURE 15.11 Weir crest length determinations for

roadway overtopping. (A) Method 1 subdivision in
to to segments. (B) Method 2use of a single segment. (From Normann et al., 1985)
(15.16). The difficult task is to determine the magnitude of total flow which equals roadway
overflow plus the culvert flow and can be determined by using a trial and error procedure.
Performance curves, as defined in Sec. 15.5, can also be used to compute the total flow.

15.4

METHODOFCULVERTDESIGN

The culvert design procedures for both the inlet and outlet control culverts must consider
some important factors (AASHTO, 1990) namely,

Establishment of hydrology
Design of downstream channel
Assumption of a trial configuration
Computation of inlet control headwater
Computation of outlet control headwater at inlet
Evaluation of the controlling headwater
Computation of discharge over the roadway and then the total discharge
Computation of outlet velocity and normal depth
Comparison of headwater and velocity to limiting values
Adjustment of configuration (if necessary)
Recomputation of hydraulic characteristics (if necessary)

15.4.1 Inlet Control

The computation of headwater for inlet control culverts is based on either the design equations or the nomographs.

15.4.1.1 Design equations. The design equations to be used depend on the condition
of the inlet control. A culvert performs as an orifice when the inlet is submerged and
as a weir when it is unsubmerged. The submerged (orifice) equation can be written as
[^] =c [lfc] 2 + r + z

for

[A*H-

<15-17)

where
HWf = headwater depth above the inlet control section invert (ft); D = interior height of
the culvert barrel (ft), Q = discharge (ftVs), A = full cross-sectional area of the culvert
barrel (ft2), c, Y = constants from Table 15.3, and Z = term for culvert barrel slope (ft/ft).
For mitered inlets, Z = 0.7S. For all other conditions, Z = -0.5S.
The unsubmerged (weir) condition can be presented in two forms. The first one is
based on the specific head at critical depth, and can be written as
[^]-BHAN

'[J-H

where Hc = specific head at critical depth (Hc = dc + V*c/ 2g) (ft), and K, M = constants
from Table 15.3.
The second form is similar to a weir equation and has a simpler form:
[H^F

for

[^H5

(15 19)

'

The latter form is the only documented form for some of the design inlet control nomographs in Normann, et al., (1985).
Nomographs As an alternative procedure, the inlet control nomographs presented in
Figs. 15.12 and 15.13 can be used as described in the following:
(1) Identify the culvert size and flow rate to be used. It is important to note that for box
culverts, the flow rate per foot of barrel width is used.
(2) Connect the culvert size and discharge and extend a straight line to cut the HWID axis.
In case HWID is required in another scale, extend the value at the original intersection
horizontally to cut the appropriate scale.
(3) Once the HWID ratio is computed either by the design equations or the nomograph,
compute the inlet control headwater depth, HW1, by multiplying the barrel diameter by
the ratio HWID.
15.4.2 Outlet Control
Similar to the inlet control culverts, the headwater can be determined either by the design
equations or the outlet control nomographs.
15.4.2.1 Design equations. Unlike the inlet control culverts, the headwater required for
an outlet culvert cannot be computed by using a single equation, but Eq. (15.12) can be
used to compute the energy losses. This concept is illustrated in the Example in Sec.
15.4.2.2.

15.4.2.2 Nomographs.

The nomographs can be used as described in the following.

(1) Identify the culvert size, D, and length, L,

(2) Use the appropriate scale for ke to connect D and L with a straight line and identify the
intersection of this line with the turning line (i.e., point X),
TABLE 15.3 Constants for Inlet Control Design Equations
Shape and
Material

Unsubmerged
Inlet Edge Description Equation
No.
K

Circular concrete Square edge w/headwall

15.18
Groove end w/headwall
Groove end projecting
Circular CMP
Headwall
15.18
Mitered to slope
Projecting
Circular ring Beveled
ring,
15.18
45 bevels
Beveled
ring
33.7 bevels
Rectangular box 30-75
15.18
wingwall flares
90 and 15
winwall flares
0 wingwall
flares
Rectangular box 45 wingwall
flare
15.19
18-33.7
wingwall flare
Rectangular box 90 headwall w/
15.19
3/4 in chamfers
90 headwall w/
45 bevels
90 headwall w/
33.7 bevels
Rectangular box 3/4 in chamfers,
15.19
45 skewed headwall
3/4 in chamfers,
30 skewed headwall
3/4 in chamfers,
15 skewed headwall
45bevels,1045 skewed wall
Rectangular box, 45 non offset
15.19
3/4 in. chamfers wingwall flares
18.4 non offset
wingwall flares
18.4 non offset
wingwall flares, 30 skewed barrel

Submerged
M

0.0098 2.000
0.0078 2.000
0.0045 2.000
0.0078 2.000
0.0210 1.330
0.0340 1.500
0.0018 2.500

0.0398
0.0292
0.0317
0.0379
0.0463
0.0553
0.0300

0.67
0.74
0.69
0.69
0.75
0.54
0.74

0.0018

2.500

0.0243

0.83

0.0260

1.000

0.0385

0.81

0.0610

0.750

0.0400

0.80

0.0610 0.750
0.5100 0.667
0.4860 0.667

0.0423
0.0309
0.0249

0.82
0.80
0.83

0.5150

0.667

0.0375

0.79

0.7950

0.667

0.0314

0.82

0.4860

0.667

0.0252

0.87

0.5220 0.667

0.0402

0.73

0.5330

0.667

0.0425

0.71

0.5450

0.667

0.0451

0.68

0.4980

0.667

0.0327

0.75

0.4970 0.667

0.0339

0.80

0.4930 0.667

0.0361

0.81

0.4930 0.667

0.0386

0.71

TABLE 15.3

(Continue)

Shape and
Material
Rectangular box,
top bevels
Corrugated metal
boxes

Inlet Edge Descrition

Unsubmerged
Equation
No.
K
M

45 wingwall flaresoffset 15.19

33.7 wingwall flaresoffset
18.4 wingwall flaresoffset
90headwall
15.18
Thick wall projecting
Thin wall projecting

0.4970 0.667
0.4950 0.667
0.4930 0.667
0.0083 2.000
0.0145 1.750
0.0340 1.500

Submerged
c

0.0302
0.0252
0.0227
0.0379
0.0419
0.0496

0.84
0.88
0.89
0.69
0.64
0.57

Source: From Normann et al. (1985).

Angl*
Wlngwolofl
Flor*

SCALE WINGWALL
FLARE

HEADWATER DEPTH IN TERMS OF HEIGHT (HW/D

RATIO OF DISCHARGE TO WIDTH (Q/B) IN CFS PER FOOT

HEIGHT OF BOX (D) IN FEET

EXAMPLE

To otosontolMOllyOto(2)tcalo
or (3)(I),pro|thonet
Korl
HM ilroloht lnclinod lino throvoh
OlllMtlrotoO.
ontf O OCoUt1Or rovorto ot

FIGURE 15.12 Headwater depth for box culverts with inlet control. (From
Normann et al., 1985)

INLET FACE-ALL EDGES

HEADWATER DEPTH IN TERMS OF HEIGHT (HWD)

TNOTES ON BEVELSI
FACC
DIMENSI
N of ALL
SlSHALL
OC AMD
TOPBEOKVELS
NOT
LESS
THAN
SHOWN.
TO
OBTAI
TEHMI
NATION IN NONEBEVEL
PLANE
IEINTAHERRECTANGULAR
BOX,
4 OAANCLE.
6. OR
DCCMCASEINCREASE
THE BEVEL
Bvl Angle

DISCHARGE PER FOOT OF BARREL WJDTH(CVNB)IN CFS PE-J FOOT

HEIGHT OF BARREL(D) IN FEET

EXAMPLE
TFT. o-4rr. Q.aoocFs Q/NB .TI.S
HW HW

HM(II O m FMI
Bi AMK
FACE DlMARE
CNSCNS
RELATED
ANO Of TO
BEVELS
NG EACH
DITOMENSI
N AT
RITHEGHTOPENI
ANGLES
THE OEDOC
FIGURE 15.13 Headwater depth for inlet control rectangular box culverts, 90 headwall chamfered or beveled inlet edges. (From Normann et
al, 1985)
(3) Draw a new line that connects the flow rate Q and headless H and goes through point X,
(4) The value read from this nomograph for //, includes the entrance, friction, and exit
losses.
Some of the important steps for design of outlet control culverts can be summarized as
(1) Compute the tailwater depth
(2) Compute the critical depth, and remember the following restrictions;
dc cannot exceed the diameter of the culvert, D.
if dc < 0.9A use Fig. 15.14
if dc > 0.9D, (use the guidelines in Chapter 3 for a more accurate estimate)
(3) Compute (dc + D) / 2
(4) Compute h0

h0 = max [TW, (dc + D) 12]

(15.20)

(5) Determine the entrance loss coefficient, ke

(6) Determine H either by using the design equations or the nomograph
For Maning's n value different from that of the outlet nomograph, a modified length L1
is used as the length scale, given by

CAXNOTEXCEED
CRITICAL DEPTH
RECTANGULAR SECTION

CANNOT EXCEED O

CRITICAL DEPTH

FIGURE 15.14 Critical depth rectangular section. (From Normann et al,

1985)

L1-L^]

(15.21)

where L1 = adjusted culvert length (ft), L = actual culvert length (ft), H1 = desired Manning
n value, and n = Manning n value from the chart,
(8) Compute the outlet control headwater by using the following equation
HW0111 = H + H0-S0L

(15.22)

Once the inlet control headwater, HW1 and the outlet control headwater, HWout are
computed, the controlling headwater is determined by comparing HW1 and HWout,
if HWout > HW1, the culvert is inlet control;
if HWout > HW1, the culvert is outlet control.
Then the discharge over the roadway is computed as given in Sec. 15.3.4. After computing the total discharge, Qt, the outlet velocity, V09 and the normal depth, dn, the results
are evaluated so that the barrels have adequate cover, the headwalls and wingwalls fit to
the site conditions, and the allowable headwater and overtopping flood frequency are not
exceeded.
Example. Design a reinforced concrete box culvert for a roadway crossing to pass a
50-year discharge of 400 ft3/s. The site conditions are as follows:
Shoulder elevation = 155 ft
Streambed elevation at culvert face = 140 ft
Natural stream slope =1.5%
Tail water depth = 3.0 ft
Approximate culvert length = 200 ft
Downstream channel is a 10 X 10 ft rectangular channel
The inlet is not depressed.
Solution. The design will be performed by using the procedures given in Sec. 15.4
Step L The 50-year design discharge is given as 400 ftVs.
Step 2. The geometry for the downstream channel is given (10 ft X 10 ft rectangular).
Step 3. Select a 7 ft X 5 ft reinforced concrete box culvert with 45 beveled edges in a
headwall.
Step 4. Determine inlet control headwater HW1
1. Either by using the inlet control nomograph (Fig. 15.13)
a) D = 5 ft
b) QIB = 400/7 = 57 ft3 / s / ft
c) HWID = (1) 1.80 for 3/4 in chamfer
(2) 1.64 for 45 bevel
d) HW1 = (HWID) D = 1.64 X 5 = 8.2 ft
Neglect the approach velocity.

2. Or the design equations (Eqs. 15.17, 15.18, or 15.19).

To identify whether the inlet is submerged or unsubmerged,
Compute[^55] = [(35)4^)05] = 5.11 > 4.0
Therefore the inlet is submerged; use Eq. 15.17

F]-W"-
HW^
f 400 I2
-ggij = 0.0314 [(35y(5)o.5J + 0.82 - 0.5(0.015)
HWi = 8.20 ft
Step 5. Determine the outlet control headwater depth at inlet,
1. The tailwater depth is specified as 3.0 ft which is obtained either from a backwater computation or normal depth calculation.
2. Compute critical depth, dc
a. Either by using the following equation (Chaudry, 1993)

A - I^
'-VT
total discharge
where q (ft3/s/ft), unit2 discharge = cm*vert width * anc* ^
acceleration, 32.2 ft/s ; then
/(4OO/7) 2
^ = V 32.2

3.
4.
5.
6.

ravitatina^

=4 7ft

'

b. Or by using Fig. 15.14, dc = 4.7 ft

(dc + D) I 2 = (4.7 + 5.0) / 2 = 4.85 ft
h0 = max [TW, (dc +D) I 2] = max [3.0, 4.85] = 4.85 ft
ke = 0.2 from Table 15.1
Determine H
a. Either by using Eq. (15.12)
(
29n2L] V2
" = [ l + k e + -*HrJ^
where A = 7 X 5 = 35 ft2, V = 400/35 = 11.4 ft/s, and/? = AIP = 30/ (7 +
7 + 5 + 5) = 1.25ft, then

4-.2 + ^~)^ = 3,ft

b. Or by using Fig. 15.7

where ke scale = 0.2, culvert length, L = 200 ft (n = 0.012, same as on chart),
Area = 35ft2, H = 3.7 ft.
7. Compute the outlet control headwater by using Eq. (15.22)
HWout = H + h0 - S0L = 3.7 + 5.0 - (0.015) (200) = 5.7 ft
Step 6. Determine the controlling headwater, HW0:
HW1 = 8.2 ft > HWout 5.7 ft. therefore HWc = 8.2 ft and the culvert is in inlet
control.
Step 7. Compute the discharge over the roadway, Qr
1. Calculate the depth over the roadway, HWr:
HWr = HWC - hc where hc (shoulder height) = 155 - 140 = 15 ft, then
HWr = 8.2 - 15 = -6.8 ft
2. As HWrr < O, Qr = O
Step 8. Compute the total discharge.
Qt = Qd + Qr =40 + = 40 ft3 / s

Step 9. Compute the outlet velocity, V0 and the normal depth, dn

Normal depth for the culvert can be computed using Manning's equation
Q= IA9 RS1'2
3
400
4 9 0012
iZ^r_JZ^_r
015V/2
4uu = 1i.4v
^7 + ^J (0
(urns)

By trial and error, dn = 2.8 ft.

Normal depth in the culvert is less than the critical depth, therefore, the type of
flow in the culvert is supercritical, a typical case for inlet control culverts.
400
Compute the velocity at the culvert outlet, V0 ,~ ^ g^ = 20.4 ft/s
400
Compute the velocity at the downstream channel, V = ,~ ,^ = 13.3 ft/s

Since V= 1.5 \ an armoring riprap, as defined in Sec. 15.8.1 can be appropriate for erosion protection at culvert outlet.
Step 10. Review the results: the culvert barrel has 5 ft of cover, which is adequate and
the allowable headwater (10 ft) is greater than HW1 = 8.2 ft, which is appropriate for design.

75.5 PERFORMANCECURVES
A performance curve is a plot of flow rate versus headwater depth or elevation for a culvert. Because a culvert has several possible control sections (inlet, outlet, throat), a given
installation will have a performance curve for each control section and one for roadway
overtopping, as shown in Fig. 15.15. In addition to these individual performance curves,
an overall culvert performance curve can be constructed by the controlling portions of the
individual performance curves for inlet, outlet, and overtopping.
Inlet control. The curves for the inlet control culverts can be plotted either by using the
design equations or the inlet control nomographs described in Sec. 15.4.1.
Outlet control. For the outlet control culverts there are a number of ways to plot the
performance curves; the use of Eqs. (15.1-15.14), outlet control nomographs, or backwater calculations. In the first two cases, the flow rates that will be used for the design of the
culvert are chosen, then the corresponding total losses are computed and are added to the
elevation of the hydraulic grade line at the culvert outlet to obtain the headwater. In the
backwater method the headwater elevations are computed for the corresponding flow
rates by adding the inlet losses to the energy equation. The energy grade line at the culvert inlet should reflect this adjustment.
Roadway overtopping. An overall performance curve is a useful tool to separate the
culvert flow from the roadway overtopping flow and to compute their respective magnitudes. In order to develop a typical overall performance curve as shown in Fig. 15.15, the
step-by-step procedure as defined in the following can be used:
1. A set of discharge values that will be used in the design of the culvert must be selected. Then the corresponding inlet and outlet headwater values must be computed.
2. The inlet and outlet control performance curves must be combined to develop a single
performance curve for the culvert.
3. To compute the flow rates of the overtopping flow; calculate the depth of upstream
water surface above the roadway for each one of the flow rates selected in Step 1, and
use Eq. (15.16) and the depth calculated in Step 3.
4. Then the overall performance curve will be generated by adding the overtopping and
culvert flow at the corresponding backwater elevations.
Example. Develop a performance curve for two 48-in corrugated metal pipe culverts
(Manning's n = 0.024) with metal end sections. The culvert is 150 ft long and on a 0.01
percent slope. The roadway is a 30-ft-wide paved crossing that can be approximated as a
broad-crested weir 130 ft long. The roadway centerline elevation is 109 ft. The culvert
invert elevations are 100 ft. at the inlet, and 99.98 ft at the outlet. Tail water discharges and
depths are listed below:

Q (ft3 /s)

80

120

160

200

240

280

320

360

TW (ft)

102.4

103.0

103.3

103.5

103.7

104.1

104.3

104.5

Step 1. Compute inlet and outlet control headwater elevations as tabulated below.
Step 2. Develop the rating curve for the overtopping flow by using Eq. (15.16)
Qr=CdL(HWr)"

H E A D W A T E R ELEVATION ( f t )

CULVERT FLUS
ROADWAY
OVERTOPPING
ROADWAY CREST

OUTLET
CONTROL

INLET CONTROL

TOP OF
CULVERT
OVERALL
PERFORMACE
CURVE

FLOW RATE (It3Xs)

FIGURE 15.15 Culvert performance curve with roadway overtopping.
(From Normann et al., 1985)

HWr

C1

kt

Q,
(Overtopping Flow)

O
(Culvert Flow)

Qt
(Total Flow)

0.30
0.50

3.00
3.05

1
1

130
130

64
184

235
240

299
424

Step 3. Draw the performance curve for the inlet control, outlet control, and overtopping
flow, shown in Fig. 15.17, by using the headwater calculations and Step 2.
Total flow Row per barrel
Q^
(1)
80
40
120
60
160
80
200
100
240
120

Inlet Control
HW1TDl H W ~ I ELh1
(2)
(3)
0.68 2.72 102.72
0.88 3.52 103.52
1.08 4.32 104.32
1.30 5.20 105.20
1.58 6.32 106.32

Outlet Control
Tw I dc I (dc+D)/2 I h0 I ke I H I ELh0
ELh
(4)
(5) (6) (7) (8)
(9)
2.40 1.84 2.92 2.92 0.50 0.65 103.55 103.55
3.00 2.31 3.16 3.16 0.50 1.50 104.64 104.64
3.30 2.73 3.37 3.37 0.50 2.60 105.95 105.95
3.50 3.11 3.56 3.56 0.50 4.20 107.74 107.74
3.70 3.47 3.74 3.74 0.50 5.80 109.52 109.52

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
75.6

Use Q/NB for box culverts

Use Figure 15.16, Scale 1
ELh1 = HW1 + 100
dc = 0.325 (Q/ND) A 0.66 + 0.083
Use Equation 15.1
Use Table 15.1
Use Figure 15.17
ELh0 = 99.98 + H +h0
EIh = max (EIh1, ELh0)
MATERIALS AND CULVERT GEOMETRY

Culverts are required to serve under various conditions including wide temperature variations, heavy abrasion, erosion and sedimentation within the culvert barrel. Because of
these factors there are a variety of culvert materials used to maintain the long life and
durability of culverts. The three main types used in application are the precast concrete,
monolithic or cast-in-place concrete, and corrugated aluminum and steel. Some other
options are listed in Table 15.4. The culvert must be able to carry loads without large
deflection and efficiently convey the water through the culvert opening. Therefore, the
selection of culvert material should consider the desired hydraulic efficiency, structural,
and operational characteristics.
One of the factors in the physical strength of the culverts is the type of the material used. The two main applications are with rigid and flexible culverts. Rigid culverts
are usually of reinforced concrete pipe and their primary strength is built into the pipe
itself. Special designs are possible that permit the pipe to support any magnitude of
overload. Their durability against freezing, thawing and erosion are major advantages,
combined with good hydraulic efficiency. In design with smaller culvert size requirements and lighter loads, nonreinforced concrete pipe can be used. Flexible culverts,
generally corrugated metal pipe or plastic pipe, provide their strength by the interaction of the pipe and the surrounding backfill material. Corrosion is a concern with the
corrugated metal pipes.
In many applications the operational characteristics of a culvert are as important as
the hydraulic and structural elements. The culvert designer must select the material
with the consideration of hydraulics, structures, maintenance, and geotechnical characteristics of the culvert site. The proper culvert design should optimize the hydraulic
and structural characteristics of the culvert while minimizing the total cost. The culvert
material which directly affects the service life, is a key parameter in the total cost of
the culvert.
Another important factor for the efficient design is the shape of the culvert conduit.
The shape has an impact on the hydraulics and structural characteristics. The type of the
maintenance and accessibility for deterioration and deformation repairs are also affected by the shape. The most common culvert conduit shapes are circular, rectangular,
pipe-arch, and arch.

15.7

LOCATION AND ALIGNMENT OF CULVERTS

The location and alignment has an important effect on the economics of culvert design. As
a general criteria, culvert location should not alter the natural drainage, and be placed at the

ENTRANCE
TYPE
Hatf*aH
Mttr*d conform
t ilop
?r)!ti4

HEAOWATER DEPTH IN DIAMETERS (HW /D)

DISCHARGE {Q) IN CFS

STRUCTURAL PLATE C. M.
DIAMETER OF CULVERT (O) IN INCHES

&~ SCALE

T9 ui teat* (Z) Or (5) proj.t!

kotlzntlfy t icU (I), thtft
M Tr;M ti*tlifttf IiM through
D < Q ffcol**, or rvtrtt *
l!ittrt4.

FIGURE 15.16 Headwater depth for CM pipe culverts with inlet control. (From Normann
et al., 1985)
bottom of the ravine. It is almost impossible to find a unique rule that could be applied to
every culvert. The final location and alignment of the culverts depend on the designer's
own experience in hydrology, hydraulics and structural aspects of culvert design. This will
help the designer to achieve maximum economy, utility, and safety (Hendrickson, 1957).
There are different options to locate the culvert; namely, bottom location placement,
and top location placement.

TABLE 15.4 Culvert Materials

Shape

Material

Protective Coating

Circular
Rectangular
Arch
Pipe-arch
Elliptical

Steel
Aluminium
Al/Fe Alloy
Concrete
Plastic

Galvanizing
Aluminized
Bituminous
Polymer
Concrete

Source: From Reagan (1993).

HEAD (H) IN FEET

DIAMETER (D) IN INCHES

DISCHARGE (Q) IN CFS

SUBMERGED OUTLET CULVENT FLOWING 'ULL

FIGURE 15.17 Head for standard CM pipe culverts flowing full, (n = 0.024). (From
Normann et al., 1985)

Headwater Elevation (ft)

Wifo Overtopping

Flow Rate (cfs)

FIGURE 15.18 Example performance curve.

15.7.1 Bottom Location Placement

The invert of the culvert with bottom location placement follows the line and grade of the
natural stream channel as closely as possible (Fig. 15.19). Although this type of installation may result in some curvature and possible changes in grade to obtain satisfactory bedding, it has a good hydraulic performance. However, there are several disadvantages to
this type of location. In applications with steep natural gradients, the velocity at the culvert outlet can be high and may require erosion and scour protection. In cases where the
bottom of the stream channel lies in a ravine, the culverts can be under high fills resulting
in long culverts and heavy loads. Cost comparison with other locations must include consideration of headwalls, spillways, energy dissipation structures, channel changes, and
extra maintenance.
In cases of steep channels, it is possible to reduce the length and required strength of
culvert by modifying the natural slope and alignment. This can be done by keeping the
culvert inlet at the bottom of the channel, but shifting the culvert outlet horizontally and
vertically and placing it in the downstream face of the embankment as shown in Fig.
15.20. Although this configuration can result in a more economical solution, an appropriate design must be provided to return the flow at the culvert outlet to the natural
drainage channel. It may be necessary to protect the downstream slope against erosion,
depending on the type of the material in the fill. In some instances, a form of spillway
may be required, particularly if the return channel is steep. In other cases it may be feasible to project the culvert beyond the toe of the embankment and allow the water to drop
into a pool.

ROADWAV GRADE
FILL
BOTTOM
OF RAVINE
SECT/ON A-A
FIGURE 15.19 Bottom location culverts. (From Hendrickson, 1957)
15.7.2 Top Location Placement
In some applications the culvert can be placed near the top of the embankment fill just
beneath the roadway grade. The main advantage over the bottom installation is its shorter length and considerably reduced load that the culvert has to carry. This type of installation forces the embankment to serve as a dam and results in a rise in the level of the
upstream pool until water flows through the culvert. Therefore the embankment material
must ensure stability during flood events. One other concern is the possible accumulation
of sediment and debris due to stagnant ponding on the upstream side when the water level
drops below the level of the culvert invert. Similar to the second solution of the bottom
location culverts, spillways or auxiliary channels might be required to carry water from
the outlet without undercutting or eroding the embankment.
15.7.3

Siphons

The site conditions might require the modification in the horizontal and vertical orientation, and under certain conditions a culvert may become a siphon. A siphon is defined as
a tube by which liquid can be transferred by means of atmospheric pressure from a higher to a lower level over an intermediate elevation. This is the case for a culvert laid on a
broken grade line as shown in Fig. 15.2IA. Due to the change in grade, the hydraulic
grade line drops below the crown of the culvert when the inlet and outlet are submerged.
The negative pressure or vacuum in the center section of the culvert results in a siphon.
In certain cases the designer may have to use the inverted siphon alternative, which is
also named as sag culvert. The inverted siphon is a consequence of a hydraulic grade line
above the crown of the culvert, thus causing the culvert to flow under pressure. These culverts can be constructed under low fills where it is necessary to excavate down to get the

BOTTOM
OF RAVIHE

FIGURE 15.20 Modified bottom location culverts. (From Hendrickson, 1957)

required cross-sectional area for hydraulic capacity (Fig. 15.21B), or they can result from
the gradual accumulation of sediment and debris at the upstream and downstream ends.
In irrigated areas, inverted siphons are used to carry the water from open ditches and
canals below highways, railroads, and other obstructions. This type of culvert must be
periodically maintained as a protection against sediment and debris accumulation. Also
they are not suggested for intermittent streams, because stagnant water can lay in them for
a long period of time and can result significant health hazards.
15.7.4 End treatments
End treatments are applied to culverts for many purposes, including; retention of roadway embankment, improvement of hydraulic efficiency, protection of highway embankment from discharge momentum, protection against piping, debris control traffic safety.
Headwall end treatments are very common structures standardized with construction
details and are economical and practical. In some instances, hydraulic efficiency can be
improved by minor details such as flared wingwalls on the headwall. Prefabricated*^
sections are very popular due to their convenience of installation and economy, lneir
application is limited to smaller culverts.
The flowline length of a mitered end culvert (Fig. 15.22) corresponds to the distance
between the upstream and downstream roadway toe-of-slope locations. The top length is
measured between the points of interception with the roadway side slopes and the barrel
top. Mitered end treatments are economical and relatively simple in construction.
Projecting end treatments are similar to mitered end treatments except that the upper
portion of the culvert barrel protrudes from the side slopes of the roadway embankment
as shown in Fig. 15.22. Projecting end treatments are convenient and economical.
However, they have serious problems regarding, hydraulics (greater entrance losses),
structural (inadequate perimeter containment), operational (potential for collision with
maintenance operations).
Another alternative is the use of different improved inlet conditions. Since a culvert
usually represents a severe constriction to the stream flow, the physical characteristics of

f/yafr0uJ/c
'/s>e

//ye/r0t///c
/s?e

Grade

Grat/e

FIGURE 15.21 A. Culvert siphons: culvert siphon, B. inverted siphon culvert. (From
Hendrickson, 1957)

the entrance to the culvert have a significant effect on the operational capacity. This
effect is reflected by the headloss coefficient, fce, associated with the entrance type used.
With reference to the table of entrance loss coefficients in Table 15.1, note that some ke
are much lower than others depending on the type of entrance. Often, improvement in
the hydraulic capacity of the culvert can be realized by simple and inexpensive expedients such as, beveling or rounding of the inside perimeter of the entrance to the conduit
and ensuring installation of reinforced concrete pipe with the grooved end upstream.
Beveled edges are commonly used with box culverts and headwall structures. As given
in Normann, et al., (1985), design charts are available for two bevel angles, 45 and 33.7
(Fig. 15.23). Although the 33.7 bevels result in a better inlet performance than the 45
bevel, it requires additional structural modifications. As a result, the 45 bevel is the preferred alternative. Groove ends provide similar hydraulic performance as the bevels. Figs.
15.24 and 15.25 demonstrate bevels with other inlet improvements. Side-tapered inlets
provide enlarged culvert entrance with a transition to the original barrel dimensions. The

PROJECTING BARREL

PRECAST END SECTION

CAST-IN-PLACECONCRETE
HEADWALL a WINGWALLS

END MITERED TO THE SLOPE

FIGURE 15.22 Four standard inlet types. (From Norman et al., 1985)

test results of FHWA (Harrison et al., 1972) resulted in a section geometry as shown in
Fig. 15.24. The throat section as identified in this figure is used to control the capacity and
maximize the efficiency of side-tapered inlets. Slope-tapered inlets have steeper slopes at
the entrance as compared to the culvert barrel. The steeper slope is provided to increase
the head on the throat section and create additional fall. Depending on the magnitude of
the available fall, the inlet capacities can be significantly higher than the conventional culvert with square edges.
Culvert inlet performance also can be improved cost effectively, by the incorporation
of special entrance geometries. Effectively, such improvements are used to reduce the
headlosses due to flow contractions and to increase capacity of culverts operating under
steep slope regimes.
Expanded entrances have a "funneling" effect on the flow. These flared entrances
allow more discharge into the conduit, thus significantly improving the efficiency of the
conduit. Their main advantage is in the reduction of constriction loss coefficients. While
there may be a minor loss reduction in outlet control culvert operations, they are mostly
effective in situations of steep slope regime. Culverts operating in a supercritical flow
regime often exhibit excessive outlet velocities. The usual effect of an improved entrance
on such a configuration results in more efficient use of the culvert barrel with the further
result of a lower outlet velocity.
Transitions of flow from the relatively wide channel approach to the constricted culvert opening may be useful in some instances. Both expanded entrances and transitions
have the potential of clogging due to debris. Larger debris is allowed to enter the culvert
partially, and subsequently becomes lodged in the smaller opening provided by the main
conduit.
Improved inlets are useful in some situations but the following considerations should
be made in design:

1.5 : I TOP
BEVEL
W/33.70AN GLE
OF
BARREL < Lj
HEIGHT
f

Cl TOP BEVEL
W/45ANGLE .
I in/ft OF
BARREL
HEIGHT ;
1/2 BARREL
in/ft (
OF
HEIGHT

33.7 TOP
AND SIDE
BEVELS

l.5!l SIDE
BEVEL
W/33.7 0 ANGLE

I in/ft OF
BARREL
WIDTH

450TOP
AND SIDE
BEVELS

1/2
in/ft
OF
BARREL
WIDTH

FIGURE 15.23 Beveled edges. (From Normann et al., 1985)

III SIDE BEVEL

W/45-ANGLE

Face Section
Bevel
Throat Section

ELEVATION

Wingwalls

Taper
PLAN
FIGURE 15.24 Typical side-tapered inlet detail. (From State of
Florida Department of Transportation, 1987)

The improved inlet often is a relatively expensive item, sometimes costing more than
the rest of the culvert.
Improved inlets are effective only in supercritical flow situations.
Debris, erosion, and other maintenance problems can represent significant operational
costs when improved inlets are used.
The design charts and methods for improved inlet design are applicable to one or two
barrels only of reinforced concrete box culvert configurations.
The design charts require a culvert face which is perpendicular to the stream flow. This
can be a construction and maintenance problem when the culvert is skewed to the
roadway centerline.

Face Section
Bevel

Throat Sectic

ELEVATION

Wingwalls

Taper

PLAN
FIGURE 15.25 Typical slope-tapered inlet detail. (From State of
Florida Department of Transportation, 1987)

15.8 SPECIALCONSIDERATIONS
15.8.1 Erosion
The high outlet velocities observed at the culvert outlets may result in excessive scour of
the channel in the vicinity of the outlet. The variety in the soil type of natural channels and
varying flow characteristics at the culvert outlet enforces the use of different methods to
Next Page

CHAPTER 16
HYDRAULIC DESIGN OF
FLOOD CONTROL CHANNELS

George K. Cotton
Simons & Associates, Inc.
Fort Collins Colorado

16.1

INTRODUCTION

Flood protection is essential to the economic and environmental integrity of most civil
engineering projects. The reliability of transportation systems, the livability of urban and
suburban developments, the long-term stability of landfills, the operation of mining excavations, the reclamation of disturbed lands, and the management of forest and agricultural lands require careful planning for flood hazards. In nearly all these cases, design of a
flood-control system will include a variety of conveyance channels referred to as floodcontrol channels.
To perform reliably, flood-control channels must behave in a stable, predictable
manner. This ensures that a known flow capacity will be available for a planned flood
event. In most cases, the design goal is a noneroding channel boundary, although, in
certain cases, a dynamic channel is sought. Since most soils erode under a concentrated flow, channel linings are needed either temporarily or permanently to achieve channel stability.
Channel linings can be classified in two broad categories: rigid or flexible. Rigid lining include channel pavements of concrete or asphaltic concrete and a variety of precast
interlocking blocks and articulated mats. Flexible linings include such materials as loose
stone (riprap), vegetation, manufactured mats of light-weight materials, fabrics, or combinations of these materials. The selection of a particular lining is a function of the design
context, involving issues related to the consequences of flooding, the availability of land,
and environmental needs.
A rigid lining is capable of high conveyance and high-velocity flow. Flood-control
channels with rigid linings are often used to reduce the amount of land required for a surface drainage system. When land is costly or unavailable because of restrictions, use of
rigid channel linings is preferred.
Flexible channel linings are distinguished from rigid linings because they can respond
to a change in channel shape. They can sustain some adjustment of the channel's shape
and still maintain their integrity. This would not be the case for a rigid lining, where local
damage to the lining may lead to a general unraveling. Damage to a rigid lining can result
from secondary forces, such as frost heave, piping, or slumping, although uplift and shear
forces are often causes of failure.
Flexible linings are used as temporary channel linings for control of erosion during
construction or reclamation of disturbed areas. Also, when environmental requirements

are part of the design, flexible lining materials have several advantages. Flexible linings
are inexpensive, permit infiltration and exfiltration, and allow growth of vegetation.
Hydraulically, flow conditions in the channel lined with flexible materials generally can
be made to conform to conditions found in a natural channel. This provides better habitat
opportunities for local flora and fauna. By permitting the growth of vegetation in the channel, flexible linings can provide a buffering effect for runoff contaminant's and sediment.
We are fortunate that a large number of innovative and traditional products are currently available for channel linings. In most cases, these products have received extensive
testing, including laboratory flume studies, large-scale prototype modeling, and documentation of field case histories. From this effort has come a better understanding of complicated flow conditions associated with each type of lining and a better understanding of
product performance and of methods needed for design. Construction experience has
resulted in better specifications for channel lining materials and their installation.
The presentation in this chapter covering flexible lining materials is based on work in
preparing Design of Roadside Channels "with Flexible Linings (Hydraulic Engineering
Circular No. 15) for the Federal Highway Administration. Chen and Cotton (1988), Since
1988, when that manual was published, the commercial market for channel-lining products has expanded. Because product testing and performance monitoring. also have
increased, this chapter also serves to update the reader on the status of current practice.
The so-called tractive force (or boundary shear stress) that acts on the channel's
perimeter describes the basic mechanics of channel stability. Design and field methods
based on tractive force are well suited for the evaluation of the stability of smaller channels where the grade of the channel dominates. In these cases, the calculation of shear
stress is simple, the determination of channel slope is the most difficult estimation. For
field observations on a channel of known grade, depth alone needs to be measured to estimate the maximum shear stress. Field estimates of flow velocity require some type of current meter. In design, the performance criteria are simple to recall for a specific type of
lining because it is represented by a single permissible shear stress value. This permissible shear stress value is applicable over a wide range of channel slopes and shapes.
Permissible velocity criteria, on the other hand, are a function of channel slope, lining
roughness, and channel shape.

76.2 DESIGNCONCEPTS
The design methods are based on the concept of maximum permissible tractive force, coupled with the hydraulic resistance of the particular lining material. The method includes
two parts: computation of the flow conditions for a given design discharge and determination of the degree of erosion protection required. The flow conditions are a function of
the channel geometry, design discharge, channel roughness, and channel slope. The erosion protection required can be determined by computing the shear stress on the channel
at the design discharge and by comparing the calculated shear stress to the permissible
value for the type of channel lining used.
16.2.1 Flood-Control Channel Design
The methods given in this chapter are for small (less than 1 m3/s) to medium-sized floodcontrol channels (less than 10 m3/s). The design of larger flood-control channels is
increasingly complex in terms of the physical environment, the nature of the flood risk,
and the number of issues and constraints that a design team must address. Small channels

can often be designed on the basis of simplified hydrologic analysis (i.e., the rational
method), without detailed geotechnical data, and by assuming uniform flow. A mediumsized flood-control channel typically requires hydrologic routing, information on soil
properties, and a hydraulic analysis based on nonuniform flow.
The level of effort for design increases steadily as the size of the channel increases.
It is expected that it may require approximately four times the hours for technical staff
to accomplish a medium-sized flood channel compared with a small channel. Since
costs will increase by a factor of about 10, there is a definite economy of scale in the
design of larger channels. However, the larger a channel becomes, the more tributary
and appurtenant features will be needed for a complete design. Therefore, the design of
a medium-sized channel often includes the design of several small channels and other
features.
16.2.2 Open-Channel Flow
16.2.2.1 Types of flow. Open-channel flow can be classified according to three general
conditions: (1) uniform or nonuniform flow, (2) steady or unsteady flow, and (3) subcritical or supercritical flow. In uniform flow, the depth and discharge remain constant along
the channel. In steady flow, no change in discharge occurs over time. Most natural flows
are unsteady and are described by a runoff hydrograph. One can assume, in most cases,
that the flow will vary gradually and can be described as steady, uniform flow for short
periods. Subcritical flow is distinguished from supercritical flow by a dimensionless number called the Froude number (Fr), which is defined as the ratio of inertial forces to gravitational forces in the flow. Subcritical flow (Fr < 1.0) is characterized as tranquil and has
deep, slower velocity flow. Supercritical flow (Fr > 1.0) is characterized as rapid and
has shallow, high-velocity flow.
For design purposes, uniform flow conditions also are considered to be steady. The
channel grade S09 water-surface grade Sw, and the energy grade Sj are assumed to be equal.
This allows the development of a flow equation based on a friction loss formula (such as
Manning's equation) and channel grade.
Computation of uniform flow is suitable for small channels and a useful approximation for medium-sized channels. In medium channels, steady uniform flow is rare, and a
more complete analysis using gradually varied flow methods is required. In solving gradually varied flow, the designer should rely on a current computer program, the accuracy
of which is well documented.
Design of steep channels with supercritical flow presents a number of special concerns.
Waves can form in a channel bed with a flexible lining (beginning near a Fr of 0.8) that
ultimately approach the depth of flow. In extremely steep channels, the flow may splash
and surge in a violent manner and additional freeboard is required. Ultimately, at Fr's near
2.0, the flow becomes unstable. Channel designs in this range should be avoided.
16.2.2.2 Resistance to flow and boundary shear stress. Flow resistance is the result of
the drag of moving water against the channel boundary. For practical purposes, the flow
in most channels will be fully turbulent and the velocity, V, will be proportional to the
square root of the shear stress, T, on the channel boundary. This gives the following simple formula for estimating channel velocity:
V=CfViTp
where Cf = the conveyance factor and p = the density of water.

(16.1)

FIGURE 16.1 Definition sketch for boundary shear stress

For uniform flow, the average shear stress on the channel boundary acts opposite to the
weight component of the flow, as shown in Fig. 16.1. For channel grades of less than 10,
the sine of the grade can approximated by the tangent and, hence, the slope of the bed.
This leads to the equation for mean boundary shear:
lALP = jAALSf
x = jRSf

(16.2)

where AL = the incremental reach length (Fig. 16.1), P = the wetted perimeter, y = the
specific weight of water, A = the cross sectional area of flow, and Sf = the friction slope
(energy grade).
The conveyance factor in Eq. 16.1 is not constant but is a function of both boundary
and flow conditions. If this were not the case, there would be little difference between a
velocity-based approach and a tractive shear-based approach to channel design. In
Manning's formulation, the conveyance factor varies as the one-sixth power of the
hydraulic radius, R, giving
C7f = R'
n

(16.3)

Note: Because Cf is dimensionless, Manning's coefficient n has the dimension of length

to the one-sixth power.
Combining Eqs. 16.1, 16.2, and 16.3 gives the standard form of the Manning's resistance equation:
V = RS

(16.4)

Not all channel linings behave according to the Manning's formulation for the conveyance factor. Most rigid channel linings have a Manning's n that is approximately constant. For shallow flows, the n value increases in rigid channels, but this effect is often

neglected even for small flood-control channels. However, for flexible types of channel
linings, the conveyance factor is difficult to describe using Eq. 16.3.
A channel lined with a good stand of vegetation cannot be described by a single n
value. The flow resistance is complicated by the mechanics of the vegetation since the
stem of the plant will bend because of the drag force, changing the height of the plant
relative to the flow depth. The Soil Conservation Service (SCS) (USDA, 1954), through
the work of Ree and Palmer (1949), developed standard classifications for vegetative
flow retardance. Grasses are classified into five broad categories of flow retardance:
Class A identifies grasses with the highest flow retardance and Class E identifies the
grasses with the lowest flow retardance. In general, high flow-retardance species of grass
form a dense cover, are tall, and have stiff stems. Sparse cover or short, flexible grasses
have lower flow retardance.
Temple (1980; Temple et al., 1987) and Kouwen, (1988; Kouwen and Li, 1980;
Kouwen and Unny, 1969) have developed modern approaches to the flow resistance of
vegetation. Temples approach is based on a mathematical fit to the SCS retardance charts
and a generalized retardance index that is a function of two properties of gress: stem
length and cover density. Temple compiled standard values of these properties for selected species for reference stem densities representing "good" cover conditions. Adjusting
these reference properties for local conditions and expected growth allows the designer to
determine an appropriate index for specific design conditions.
Kouwen analyzed the biomechanics of vegetation and developed conveyance factors
that are a function of the density, height, and stiffness of the grass. He showed that the
resulting method gives results that are consistent with the SCS retardance curves. Since
Kouwen's method is physically based, it is a useful tool for extending retardance curves
to channel conditions that were not studied by the SCS, including stiff vegetation and mild
channel grades.
16.2.3

Flood-Control Channel Components

A basic flood-control channel consists of a minimum of five components. There are specific design issues associated with the local flow conditions at each component. More
complex flood-control systems also may include additional components for flood storage
(detention basins) and grade control (low dams).
Channel inlet. At the point where water enters a flood-control channel, local shear
stresses can develop as a result of local acceleration of the flow. For small flood-control channels, the inlet may consist of small berms or dikes that collect runoff and direct
this diffuse flow to the channel. A medium-sized flood-control channel will typically be
the result of the combination of several tributaries or the extension of an existing channel. Collection and control of incoming flow is essential to the proper operation of a
larger channel. In most cases, a flood-control channel will discharge as a tributary to a
larger channel.
Reach. The reach component is the length of channel, with only minor variation in the
properties of grade, discharge, cross section, and lining material. A reach may have a
straight or curving alignment. In larger flood-control channels, a reach also may include
a bridge or culvert.
Flow around a bend in an open channel induces centrifugal forces created by the
change in the direction of flow. This results in a superelevation of the water surface, with
its surface being higher at the outside of the bend than at the inside of the bend. Flow
around a channel bend also imposes higher shear stress on the channel bottom and banks.

The increased shear stress requires additional design considerations within and downstream of the bend.
Confluence. A confluence is the site where two or more flows merge without a significant grade difference between each channel. The process of merging flow involve turbulent mixing and a local increase in energy loss and boundary shear stress. Flow separation, wave formation, and bank impingement can occur in a confluence that creates much
higher local boundary shear stresses and a potential for erosion. Control of the velocity
and direction of tributary inflows often is required to prevent damage to the channel within a confluence. Where a large tributary joins the main channel, performance of the confluence can be improved if the direction of the two flows is as nearly parallel as possible.
Side inlet. A side inlet allows flow to enter the channel over the bank. Example of the
need for a side inlet to a channel are flow from a field or a street. Often, the flow enters
the channel from a chute constructed on the channel bank. When the bank channel is at a
mild grade, vegetation may be a sufficient lining for the chute. For steeper bank slopes,
however, riprap or gabion linings are common.
Channel crossing. A channel crossing is required where a private or public road passes over the channel. Structures used for this purpose are stream fords, culverts, and
bridges. Stream fords are best suited for low-traffic-volume roads crossing channels with
an intermittent type of flow. Fords can be hazardous if flow depths of more than 0.30 to
0.45 m. are expected. Culverts of concrete or metal pipe provide an economical crossing
for small channels. Since the shape of the culvert is typically circular, elliptical, or arched,
the culvert will obstruct a portion of the channel section. Bridges of concrete, steel, or timber are used when the channel must remain largely unobstructed.
Transition. A transition is a gradual expansion or contraction between two channel
sections. Transitions can occur between one reach and another or between a reach and the
channel inlet or outlet.
Channel Outlet. At the point where water exits a flood control channel, local shear
stresses can develop as aresult of deceleration of the flow. At the outlet, the flow changes
to match the local velocity and depth of the receiving channel.
16.2.4 Stable Channels
16.2.4.1 Stable channel modes. A stable channel can be either static or dynamic, but
over time the net effect of scour and deposition must be zero. In a static stable channel,
scour and deposition occur within the limits of a channel boundary that effectively resists
the erosive force of the flow. Although the transport of sediment can be significant in a
static channel, the effective change in channel section at any given time is small. This
implies equilibrium between the incoming supply of sediment and the transport of sediment within a channel reach. When the supply of sediment is small, the principles of rigid
boundary hydraulics can be applied to evaluate the channel's capacity and the lining's stability. When the supply of sediment is large, then the determination of the channel's
capacity must consider the transport of both water and sediment. The transport of sediment affects the flow resistance and therefore the shear forces and the stability of the
channel lining.
In a dynamic stable channel, some change in the channel bed and banks are expected.
The channel is considered to be stable if the morphology (i.e., shape) of the channel
remains unchanged over time. Important measures of channel shape include the channel
width, depth, slope, pattern (straight, meandering, or braided), hydraulic variables (flow

velocity and discharge), and sediment variables (particle size, suspended concentration of
sediment). A dynamic channel can be regarded as stable as long as local changes in channel shape do not have adverse consequences. For most small- or moderate-capacity floodcontrol systems, design of a dynamic channel is often too complex or it presents too many
uncertainties. Consequently, development of static stable channels is usually preferable to
using dynamic approaches to the design.
Several empirical methods are commonly applied to determine whether a channel is
stable. These methods are defined as permissible velocity methods. Permissible velocity
approaches were first developed in the early 1920s (Lacey, 1920, Lindley, 1919) and are
related to the successful development of regime theory for large irrigation canals beginning in 1885 with Kennedy's formula. Regime theory continues to be a basic tool for evaluating large rivers and irrigation canals (Blench, 1969; Simons and Albertson, 1963);
however, its empirical nature limits its application in the case of flood control channels.
In the 1950s, under the direction of Lane at the U.S. Bureau of Reclamation, research
was conducted that refined the permissible tractive force method (Glover and Florey,
1951; Lane, 1955). This work and subsequent research clarified the actual physical
processes occurring for flow in stable channels. This method extended the range of stable
channel design beyond the empirical limits of regime theory. In most cases, a more realistic model of channel stability is based on permissible tractive force.
16.2.4.2 Tractive force. The flow resistance of water moving against a channel boundary results in shear force along the channel boundary referred to as the tractive force. As
was discussed in Sec. 16.1.2, there is a close relationship between the tractive force and
the flow velocity. In a uniform flow with a normal (semilog) velocity distribution, the
tractive force is equal to the effective component of the gravitational force acting on the
body of water parallel to the channel bottom (see derivation of Eq. 16.2).
However, shear stress is not uniformly distributed along the wetted perimeter in a
channel, and Eq. 16.2 describes only the average shear force on the channel. For design,
it is necessary to determine the maximum shear along the channel perimeter so that the
lining material is suitable for resisting the shear. At the same time, the stability of a soil
particle decreases as the lateral slope of the channel increases. The relative combination
of particle stability and boundary shear is needed to attain a stable channel. Simply stated, the principle of stable channel design using tractive force is to have the shear strength
of the lining material ip, exceed the boundary shear force ib at every point in the channel's wetted perimeter:
\> \

(16-5)

The permissible shear stress of the lining material lp is the maximum shear stress that
the lining can safely withstand. Using the critical shear stress Tc for a soil particle as a reference value, the permissible shear stress can be defined as
1

P=CA*

(!6-6)

where Ca is the critical shear-stress adjustment factor. The design shear stress is affected
by a number of factors, including: the density of the soil particles, the submerged soil friction coefficient (\JL = tan <]), where is the angle of repose), the lateral (side slope) and
down-slope components of soil particle weight, and the direction of flow.
The boundary shear stress ib varies within a channel section primarily as a function of
depth and also in response to the lateral diffusion of momentum in the channel section.
Using the mean shear stress T0 as a reference value, the local boundary shear stress can be
defined as

Velocity Distribution

Bank Shfrar
Bed Shear
Shear Distribution
FIGURE 16.2 Boundary shear distribution in channel section

High Shear Stress Zone

FIGURE 16.3 Boundary shear distribution in a bend

\=&a\

(16.7)

where K0 is the boundary shear-stress adjustment factor. Channel alignment, the relative
channel width and depth (aspect ratio), and the channel side slope affect the boundary
shear stress.
Figure 16.2 shows a typical distribution in a parabolic channel. Boundary shear is
greatest at the middle of the channel (maximum flow depth) and tends toward zero
along the channel bank. The distribution shear of in a channel bend is shown in
Fig. 16.3.
16.2.5 Design Parameters
Flood frequency. The most important question facing the designer of a flood-control
channel is - What is the probability of failure? Because flood-control channels are built to
protect valuable facilities, damage should occur only rarely. The probability of flood damage, although never zero, can be low enough to make the risk to the facility acceptable.

TABLE 16.1 Commonly Used Flood Frequencies

Type of Facility
Collector road
Roadside or median ditch
Urban collector streets
Rural flood control
Urban flood control

Potential Flood Damage

Delay costs
Impeded traffic
Delay costs
Impeded emergency access
Detour costs
Crop damage
Road damage
Property damage
Infrastructure damage

Flood Frequency
(Return Period, years)
2 to 5 years
10 to 25 years
25 to 50 years
100 years

When flows are small and potential damages are small, such as in the case of a roadway drainage system, more frequent flooding is allowed. For larger flood-control channels, where flood damage is severe, a low frequency of flooding is desired. Although economic analysis can be used to determine the optimal range of flood frequency for a floodcontrol facility, only larger projects warrant such an effort. In most cases, standard flood
frequencies are used. Table 16.1 provides a brief summary of common flood frequencies
and their use in the design of various types of facilities.
Duration offloading.
Small flood-control channels with intermittent flooding tend
to have short times of flooding, typically on the order of several hours. Larger flood-control channels, or channels with sustained flows, require a design that considers the persistence of boundary stresses. Well-maintained channels that sustain only short periods of
boundary stress often can sustain higher stresses without damage.
Channel profile. Major project features, such as roads generally dictate the slope of
smaller channels. If channel stability cannot be maintained for these conditions, it may be
feasible to reduce the channel gradient slightly relative to other grading. This typically
results in short, steep reaches that are constructed as rigid chutes or low drops.
Channel slope is among the major parameters affecting boundary shear stress. Thus
at a given discharge, the shear stress is less for a mild (subcritical regime) gradient
compared with a steep (supercritical regime) gradient. It is not uncommon for smaller
flood-control channels to operate mainly in supercritical regime. However, extremely
steep channels (above 10 percent) and larger channels require additional design considerations.
Channel section. The most common shapes of drainage channel are trapezoidal or triangular. These shapes are a close approximation of the most efficient hydraulic section, a
semicircle. For smaller channels with flexible linings, these shapes tend to be constructed
with slightly rounded corners, which gives the as-built section a slightly parabolic shape.
The banks are the least stable portions of the channel section. Bank stability increases
as the side slope flattens. For smaller channels, a 3:1 side slope is sufficient to prevent erosion in excess of the channel bed. For larger channels, or for small channels that require
steep side slopes, the stability of the side slope should be checked.
In larger flood-control channels with persistent low flows, a low-flow channel often is
included within the channel section. It is advisable to situate the low-flow channel near

the center of the main channel to avoid creating increased shear near the channel bank.
Low-flow channels are often designed to meander within the larger section. The meander
reduces the gradient of the channel, allowing the enhancement of the channel section for
habitat or appearance. The potential effect at meander bends on the main channel bank
line should be considered.
Although a channel section can be widened to reduce boundary shear, channels with
large width to depth ratios typically are not stable. Over time, a high stress area will develop within the section and cause erosion that will form a narrower, deeper channel, the
channel eventually fails.

16.3

CHANNELLININGS

Considerable research and development has produced a variety of rigid and flexible lining materials. Before the late 1960s, channel linings were constructed primarily of natural materials, such as rock riprap, stone masonry, concrete and vegetation. Material manufactured or fabricated into rolls offered several advantages for erosion control, particularly during construction and during the establishment of vegetation. Use of rolled erosioncontrol products for temporary erosion control assured improved long-term performance
of vegetative linings.
16.3.1 Lining Types
Because of the large number of channel stabilization materials currently available, it is
useful to classify materials according to their performance characteristics. Lining types
are classified as either rigid or flexible. Flexible linings are grouped further as either permanent or temporary.
Rigid linings
Cast-in-place concrete
Cast-in-place asphaltic concrete
Stone masonry
Soil cement
Fabric formed concrete
Grouted riprap
Flexible linings (Long-term
nondegradable)
Riprap
Wire-enclosed stone
Vegetation
Gravel
Synthetic mat
16.3.2

Flexible linings (temporary

degradable)
Straw with net
Curled wood mat
Jute net
Woven paper net
Fiberglass roving

Performance Data

16.3.2.1 Rigid linings. Rigid linings are useful in applications where high shear-stress or
nonuniform flow conditions exist, such as at transitions in channel shape or at an energy
dissipation structure. In areas where loss of water or seepage from the channel is undesir-

able, a rigid lining can provide an impermeable barrier. Because rigid linings are
nonerodible, the designer can use any channel shape that provides adequate conveyance.
Rigid linings may be the best option if right-of-way limitations restrict the channel size.
Despite the nonerodible nature of rigid linings, they are highly susceptible to failure from structural instability. For example, cast-in-place or masonry linings often
break up and deteriorate if the foundation is poor. Once a rigid lining deteriorates, it is
highly susceptible to erosion because it forms large, flat, broken slabs that are easily
moved by flow.
The major causes of structural instability and failure of rigid linings are freeze-thaw,
soil swelling, and excessive soil pore-water pressure. Freeze-thaw and swelling soils exert
upward forces against the lining, and the cyclic nature of these conditions eventually causes failure. Excessive soil-pore pressure may occur when the flow levels in the channel
drop quickly but the soil behind the lining remains saturated. This can result in instability of the bank's slope caused by the high water-table gradients within the channel bank.
Construction of rigid linings requires specialized equipment using relatively costly
materials. As a result, the cost of rigid channel linings is high. Prefabricated linings can
be a less expensive alternative if shipping distances are not excessive.
There often are significant environmental issues associated with rigid channel linings. In
environmentally sensitive areas, replacement of concrete linings with articulated block mats
is a partial solution, because openings between the blocks allows vegetation to take hold.
16.3.2.2 Flexible linings. Riprap and gabion are suitable linings for hydraulic conditions
similar to those requiring rigid linings. Because flexible linings are permeable, they may
require protection of the underlying soil to prevent washout. For example, filter cloth is
often used with riprap to inhibit soil piping.
Vegetative and temporary linings are suited to hydraulic conditions where uniform
flow exists and shear stresses are moderate. Most grass linings cannot survive sustained
flow conditions or long periods of submergence. Grass-lined channels with sustained low
flow and intermittent high flows often are designed with a composite lining of a riprap or
concrete low-flow section.
The primary use of temporary linings is to provide protection from erosion until vegetation is established. In most cases, the lining will deteriorate over the period of one
growing season, which means that successful revegetation is essential to the overall channel stabilization effort. Temporary channel linings can be used without vegetation to control erosion on construction sites temporarily.
16.3.3

Information About Flexible Linings

The erosion control industry has grown in response to continued infrastructure development and increased awareness of water-quality problems. Traditional methods of erosion
control are well documented, and a variety of specifications exist for these types of linings. In 1997, the Erosion Control Technology Council (ECTC) established standard terminology and testing methods for the variety of manufactured products used for erosion
control. The ECTC terminology for rolled erosion control products (RECPs) is used here
for clarity,
16.3.3.1 Long-term, Nondegradable flexible linings. Vegetation. Vegetative linings
consist of planted or sodded grasses placed in and along the drainage (Fig. 16.4). If planted, grasses are seeded and fertilized according to the requirements for the particular variety or mixture and to soil conditions. Sod is laid parallel to the direction of flow and can
be secured with staples or stakes.

FIGURE 16.4 Vegetative channel lining

FIGURE 16.5 Rock channel lining

Rock riprap. Rock riprap is dumped or hand placed on prepared ground with a filter
blanket or a prepared bedding material interface (Fig. 16.5). The stone layer is placed to
form a well-graded mass with a minimum of voids. Stones should be hard, durable, prefer-

FIGURE 16.6 Wire-enclosed stone channel lining

ably angular in shape, and free from overburden, shale, and organic material. Resistance
to disintegration from channel erosion should be determined from service records or from
specified field and laboratory tests.

FIGURE 16.7 Gravel channel lining

Wire-enclosed stone. Wire-enclosed stone consists of a wire basket or tube filled

with stone (Fig. 16.6). The wire basket is made of steel wire woven in a uniform pattern and reinforced on corners and edges with heavier wire. Common forms of wireenclosed stone include boxlike baskets, thin mattresses, and tubes. The containers are
filled with stone, connected together, and anchored to the channel side slope. Stones
are graded fairly uniformly, with the smallest size larger than the wire mesh opening.
The stones should be hard, durable, and free from overburden, shale, and organic material. Wire-enclosed stone typically is used when rock riprap is either not available or
not large enough to be stable

FIGURE 16.8 Turf-reinforcement mat channel lining

Gravel. Gravel consists of coarse gravel or crushed stone placed on filter fabric or
prepared bedding material to form a well-graded lining with a minimum of voids (Fig.
16.7). The stones should be hard, durable, and free from overburden, shale, and organic material.

FIGURE 16.9 Woven-paper net channel lining

Turf reinforcement mat. Turf reinforcement mat is composed of ultraviolet-stabilized, nondegradable, synthetic fibers, netting, or filaments processed into a three-dimensional reinforcement matrix ranging in thickness from 6 to 20 mm (Fig. 16.8). The mat
provides sufficient thickness, strength, and void space to permit soil filling and development of vegetation within the matrix. The mat is laid parallel to the direction of flow is
stapled or staked to the channel surface, and is anchored into cutoff trenches at regular
intervals along the channel. The mat is top-dressed with fine soil at a depth equal to the
mat's thickness and is seeded and fertilized. Turf reinforcement occurs as root growth penetrates and entangles with the mat.

FIGURE 16.10 Jute net channel lining

16.3.3.2 Temporary degradable flexible linings. Woven paper net. Woven paper net is
erosion-control net that consists of knotted plastic netting interwoven with paper strips
(Fig. 16.9). The net is applied evenly on the channel slopes, with the fabric running parallel to the channel's direction of flow. The net is stapled to the ground and placed into
cutoff trenches at regular intervals along the channel. Woven paper net is usually installed
immediately after seeding operations.

FIGURE 16.11 Curled-wood mat channel lining

FIGURE 16.12 Straw with net channel lining

Jute net. Jute net is an erosion control mat that consists of jute yarn approximately 6
mm in diameter that is woven into a net with openings approximately 10 to 20 mm (Fig.
16.10). The jute net is loosely laid in the channel parallel to the direction of flow. The net
is secured with staples and is placed int cutoff trenches at regular intervals along the channel. Jute net is usually installed immediately after seeding operations.
Curled wood mat. Curled wood mat is an erosion control blanket that consists of
wood fibers, 80 percent of which are 150 mm or longer, with a consistent thickness and
an even distribution of fiber over the entire mat (Fig. 16.11). The topside of the mat is covered with biodegradable plastic mesh. The mat is placed in the channel parallel to the
direction of flow and is secured with staples and cutoff trenches.
Straw with net. Straw with net can be manufactured as an erosion control blanket
or can be constructed from straw mulch secured with erosion control net. The constructed version consists of plastic mesh with 20-mm2 square openings overlaying
straw mulch (Fig. 16.12). Straw is spread uniformly over the channel surface at a rate
of approximately 4.5 metric tons/hectare and can be crimped into the soil. Plastic mesh
is placed over the straw and stapled to the soil. The manufactured version consists of
straw, either sewn to a layer of erosion control net or sandwiched between to two layers of erosion control net. Straw weight is approximately 2.5 metric tons/hectare. The
mat is placed parallel to the direction of flow and is secured with staples and cutoff
trenches.
Fiberglass roving. Fiberglass roving consists of continuous fibers drawn from molten
glass that is coated and lightly bound together into roving (Fig. 16.13). The roving is ejected by compressed air, forming a random mat of continuos glass fibers. The material is
spread uniformly over the channel and anchored with asphaltic materials.

FIGURE 16.13 Fiberglass roving channel lining

76.4 MILD-GRADIENT CHANNEL DESIGN
(SYMMETRIC SECTION)
This section outlines a method of stable channel design for channel gradients of less than
10 percent. Most of the lining types presented will perform acceptably up to grades of

Next Page

CHAPTER 17
HYDRAULIC DESIGN
OF SPILLWAYS

H. Wayne Coleman
C. Y. Wei
James E. Lindell
Harza Engineering Company
Chicago, Illinois

17.1

INTRODUCTION

The spillway is among the most important structures of a dam project. It provides the project with the ability to release excess or flood water in a controlled or uncontrolled manner to ensure the safety of the project. It is of paramount importance for the spillway facilities to be designed with sufficient capacity to avoid overtopping of the dam, especially
when an earthfill or rockfill type of dam is selected for the project. In cases where safety
of the inhabitants downstream is a key consideration during development of the project,
the spillway should be designed to accommodate the probable maximum flood. Many
types of spillways can be considered with respect to cost, topographic conditions, dam
height, foundation geology, and hydrology. The spillways discussed in this chapter
include overflow, overfall, side-channel, orifice, morning-glory, labyrinth, siphon, tunnel,
and chute spillways A section on design of spillways that considers cavitation and aeration also is included.

77.2 OVERFLOWSPILLWAY
An overflow spillway can be gated or ungated, and it normally provides for flow over a
gravity dam section. The flow remains in contact with the spillways surface (except for
possible aeration ramps) from the crest of the dam to the vicinity of its base. The hydraulic characteristics are defined as follows:
1. Determine design head H0. Normally, H0 is 75 to 80% of maximum head #max.
2. Use the depth from the crest to ground surface P to find the basic discharge coefficient C0 from Fig. 17.1.
3. Find discharge coefficient C for the full range of heads from Fig. 17.2.

4. Correct discharge coefficient C for the sloping upstream face from Fig. 17.3. The
sloping upstream face is normally for structural stability, not hydraulic efficiency.
5. Correct discharge coefficient C for the downstream apron from Fig. 17.4.
6. Correct discharge coefficient C for tail water submergence from Fig. 17.5
7. Define the shape of the pier nose. Normally, use Type 3 or 3A from Fig. 17.6.
8. Check the minimum crest pressure from Fig. 17.7 or 17.8. If the minimum pressure
is below -1/2 atmosphere, increase H0 and start over.
9. Define the crest shape from Fig. 17.9(a) and (b).
10. Determine the effective crest length for the full range of heads from
L = L -2(NKp +K0)H6

(17.1)

where: L = effective length of crest, L = net length of crest, N = number of piers,

K = pier contraction coefficient, Ka = abutment contraction coefficient, and He =
total head on crest.
The following are values of the pier contraction coefficient

For square-nosed piers with corners

rounded on a radius equal to approximately
0.1 of the pier thickness
for round-nosed piers
for pointed-nose piers

K
,
0.02

0.01
0.00

The following are values of the abutment contraction coefficient

*.
for square abutments with headwall
0.20
at 90 to the direction of flown
for rounded abutments with headwall
0.10
at 90 to the direction of flow
when0.5# 0 >r>0.15# 0
for rounded abutments, where
0.00
r > 0.5H0, and the headwall is placed
no more than 45 to the direction of flow,
where r = radius of the abutment rounding.
11. Determine the discharge rating curve from
Q = CLH*2

(17.2)

Exhibits 17.1 and 17.2 illustrate overflow spillways for hydroelectric projects.

VALUES OF COEFFICIENT CQ

VALUES OF-H0

RATIO OF COE FFl C IE N TS -i-

FIGURE 17.1 Discharge coefficients for a vertical-faced ogee crest. (USBR, 1987).

RATIO OF HEAD ON CREST TO DESIGN HEAD=

H0
FIGURE 17.2 Coefficient of discharge for other than the design head. (USER, 1987).

Slope

Angle
with
tht vrticol

FREE DISCHARGE GOEFF ICIE N T *

FIGURE 17.3 Coefficient of discharge for an ogee-shaped crest with a sloping

upstream face. (USER, 1987)

hrifd
POSITION OF DOWNSTREAM APRON -^-

FREE DISCHARGE COEFFICIENT i

FIGURE 17.4 Ratio of discharge coefficients associated with the apron effect.
(USER, 1987).

DEGREE OF SUBMERGENCE ^f
FIGURE 17.5 Ratio of discharge coefficients associated with the tailwater effect.
(USER, 1987).

CREST AXIS
TYPE 3

TYPE I

CREST AXIS
TYPES 3 ANO 3A
TYPE 4
PIER NOSE SHAPES
NOTE: PIER
NOSE LOCATED
IN SAME
PLANE
AS UPSTREAM
FACE OF
SPILLWAY.
DIMENSIONS
IN
PARENTHESES
ARE
FOR
TYPE 3A.

COEFFICIENT OF PIER CONTRACTION-K 0

FIGURE 17.6 Pier-contraction coefficients for high-gated overflow crests. (USAGE, 1988)

HIGH GATED OVERFLOW CRESTS

PIER CONTRACTION COEFFICIENTS
EFFECT OF NOSE SHAPE
HYDRAULIC DESIGN CHART UI-S

PRESSURE jhp\

HORIZONTAL
DISTANCE
DESIGN HEAD
AXIS BOTH QUADRANTS

CCMART MI-S)

CREST

NOTE DATA BASED ON F.SOOI T^STS.

HIGH OVERFLOW DAMS

CREST PRESSURES
ALONG PIERS
HYDRAULIC DESIGN CHART III-I6/2

FIGURE 17.7 Crest pressures along piers (Type 3A) for high-overflow dams. (USAGE, 1988)

PRESSURC
CESlGN HEAD

HORIZONTAL DISTANCE
DtSiCN HKAO
AXIS BOTH QUADRANTS

TYPE 3A PIER
(CHART in-4)

CREST

NOTE DATA OASED ON esaoi TESTS.

HIGH OVERFLOW DAMS

CREST PRESSURES
CENTER LINE OF PIER BAY
HYDRAULIC DESIGN CHART III-I6/I

FIGURE 17.8 Crest pressures along the centerline of a pier (Type 3A) bay. (USACE, 1988)

Water surface upstream from weir drawdown

.Origin and open of crest

upstream face

(A) ELEMENTS OF NAPPE-SHAPED CREST PROFILES

VALUES OF K

FIGURE 17.9 (a) Factors for the definition of nappe-shaped crest profiles (USER, 1987)

LOCATlOW OF CENTER FOR R,

FIGURE 17.9(b) Factors (xc, yc, R1, and R2) for definition of nappe-shaped crest
profiles.(USBR, 1987)

Exhibit 17.1 Macagua hydroelectric project, Venezuela (Courtesy CVG-EDELCA, Caracas.)

(a) General view of the spillway in operation.
(b) Free flow condition at the ogee crest.
(c) Flow condition at the ogee crest with gate partially open.
(d) Profile of the spillway and stilling basin.

El 31.50:

El 34.00

CREST STA. S.S. 7 + 17.00

FLIP BUCKET LIP
STA. S.S. 36 + 83.8
EL. 1202.00

BEGINNING OF 1% SLOPE
1% SLOPE

PROTECTION APRON

Exhibit 17.2 Tarbela Hydroelectric Project, Pakistan (Courtesy Water and Power
Development Authority, Pakistan)
(a) General view of the spillway in operation.
(b) Profile of the spillway including the flip bucket.

OVERFALLSPILLWAY

77.3

An overfall spillway can be gated or ungated and provide for flow over an arch or archbuttress dam, wherein the flow free-falls some distance before entering a plunge-pool
energy dissipator in the tailrace. The hydraulic characteristics are defined as follows:
1.

The crest structure and discharge rating are similar to those for the overflow spillway.
2. The flow normally leaves this structure shortly below the crest. The exit structure
is normally some variation of a flip-bucket.
3. The flip-bucket radius for an overfall spillway is normally smaller than the ideal,
which is at least 5d, where d is the flow depth at the bottom of the bucket. The
radius is usually undersized to minimize the size of the overhang, which can destabilize the top of a thin-arch dam. However, the radius should be sufficient to
fullydeflect a significant flood say, the 100-year event.
4. The bucket exit angle is selected to throw the jet to a suitable location in the tailrace. The trajectory can be estimated by
y = * * n g - 36f/;os2,

(17.3)

where: y = vertical distance from the bucket lip, x = horizontal distance from the
bucket lip, 6 = bucket exit angle, and H = depth + velocity head at the bucket lip.
5.

6.
7.

The trajectory can be estimated from Step 4 as long as the bucket radius exceeds
5d. For larger depths, the flow overrides the bucket and will fall short of the maximum trajectory. The trajectory in this range is determined best by a physical
model.
Pressure load on the bucket can be estimated from Fig. 17.10. For larger floods,
where d > R/5, this load is determined best by a physical model.
The energy from an overfall spillway is normally dissipated by a plunge pool,
which can be lined or unlined. If unlined, the scour and the scour rate will be
based on both flow and geology. The scour hole development is usually indeterminate. However, the terminal scour depth for a uniformly credible material can
be estimated from the following empirical formula and from Fig. 17.11
(Coleman, 1982; USER, 1987)
ys = ds 5woc = terminal vertical scour depth

(17.4)

ds = C8 225 q-54

(17.5)

and

where Cs = 1.32 (for units in ft and cfs/ft) 1.90 (for units in m and cms/m), H =
effective head at tailwater level, q = unit discharge = Q/B, B = width of the bucket, and a = average jet entry angle.
The extent of the scour hole is based on judgment of stable slope of material surrounding the deepest hole. A physical model is normally used where topography is
complex and where scour can endanger project structures.

8.

9.

As the jet plunges into the pool, it diffuses almost linearly and entrains air at the surface of the pool and the water from the pool at the boundary of the jet. The behavior
of the plunging jet, including dynamic pressures, can be approximated (Hinchliff and
Houston, 1984) using Fig. 17.12 and Table 17.1 for both rectangular and circular jets.
If the plunge pool is lined because the scour would be unpredictable and unacceptable, the lining must be designed for pressure pulsations from the jet's impact.

O./O.T-'.0(FL/P BUCKETS)

NOTE: NUMBERS ARE VALUES

Or
* 'O2 FOR
DATA(O./O.T)
POINTS.

A
X
A
a

DEFiNITION SKETCH

LEGEND
PINE TLAT MODEL
HARTWELL MODEL
PINC TLAT PROTOTYPE
ES aoi (TOE CURVE)

HIGH OVERFLOW DAMS

ENERGY DISSIPATORS
FLIP BUCKETANDTOE CURVE PRESSURES
HYDRAULIC DESICNCHART 112-7

FIGURE 17.10 Pressures for flip-buckets and toe curves of an overflow spillway. (USAGE, 1988).

TRAJECTORY

ORIGINAL GROUND

SCOUR

FIGURE 17.11 Definition sketch of free-jet trajectory and scour depth of an overflow spillway.
(Coleman, 1982)

jet core

velocity head
dynamic pressure
velocity

FIGURE 17.12 Schematic diagram of a diffusing plunging jet.

(Vischer and Hager, 1995).

TABLE 17.1 Plunging Jet Characteristics (Whittaker and Schleiss, 1984)

Rectangular jet
5
v
EL

Circular jet

1 + 0.414 ylyk

1 + 0.507 ylyk + 0.500 (ylytf-

1 - 0.184 y/yk

1 - 0.550 ylyk + 0.2IKyJy1)2

g-n/8(l + x/Bu'y/y/yk-yk/y)2

-1/2(1 + r/R^Yfi-yfip

y<yk

Q
jE

^u
Z.
vz
P_
Pz
y
:r
V
u
%
Pu

e-nl\6(xlBufi

e-l/2(r/Ru)2

^7)7

yJy

yJy

Wyf

^-

1.414 V^

-j-

0.816 V^

Ji

e-rtS(ABu'yk/yP

e-l/2(r/Ru'yk/yfi

JL
e-nl6(x/Eu-y]<!yfi
z
Source: Whittaker and Schleiss (1984).

-l/4(r//eu-y^/y)2

y>yk

Exhibit 17.3 Boyd's Corner Dam Project, New York,

(a) General View of the spillway.

2y/yt
0.667 y/y

Flow
Existing masonry
lining (typ.)

Existing masonry
fining <typ>
Existing dam
concrete

Downstream face
of existing dam

(Bond length)

T/rock
varies

(C)
Exhibit 17.3 Boyd's corner dam project, New York (Continued).
(b) A view of the overfall spillway in operation.
(c) Profile of the spillway including the flip bucket.

The design pressures can be determined best from physical model studies.
Exhibit 17.3 illustrates an overfall spillway in operation.

77.4

SIDE-CHANNELSPILLWAY

A side-channel spillway can be gated or ungated and provides for flow into a chute or tunnel at right angles because the abutment topography is not favorable for a normal crest
alignment. Figure 17.13 shows a typical arrangement. Exhibit 17.4 illustrates a side-channel spillway in operation The hydraulic features are defined as follows (USBR, 1987):
1. The crest is often ungated since a long crest may be suitable because of favorable
topography.
2. The crest shape is based on the same criteria as the criteria for an overflow spillway.
3. The trough is sized by trial and error to prevent the maximum discharge water surface
from encroaching on the crest's free-discharge capacity. The trough should be as nearly V-shaped as possible to promote efficient dissipation of energy.
4. The chute crest is proportioned to produce subcritical flow in the trough for all discharges
to dissipate the overflow energy and produce uniform flow into the chute.
5. The trough geometry for the first trial is proportioned to produce an approximately
uniform reduction in trough velocity from downstream to upstream. This will usually
minimize the trough size.
6. The water surface profile in the trough is estimated by the following:

-ff^h-^^]
where AF = change in water level between two sections AX apart, Q2,V2 = discharge
and velocity at the downstream section, and Q1^V1 = discharge and velocity at the
upstream section. The computation begins with a known water level at the downstream
end produced by critical depth control at the chute crest, and proceeds upstream in
increments of AX by trial and error, calculating the water level along the length of the
side channel. If the calculated water profile encroaches on the side channel's capacity,
the geometry of the trough or chute crest or both should be adjusted.
SlOECHANNCl.
CBEST

CHUTE
CNUTE BLOCKS
OENTflTEO SILLS

FIGURE 17.13 Typical arrangement of a side-channel spillway.

(USER, 1987).

Exhibit 17.4 Ullum Dam project, Argentina.

(a) General view of the side-channel spillway in operation
(b) Layout of the spillway including the stilling basin.

7. Once an acceptable geometry for the trough and chute crest has been determined, the
chute should be sloped to provide supercritical flow away from the chute crest.
8. Large side-channel spillways are almost always model-tested because of the complex
flow conditions in the trough area.

77.5

ORIFICESPILLWAY

An orifice spillway is normally gated and is used when substantial discharge capacity is
needed at low reservoir levels, as illustrated in Exhibit 17.5. For instance, it is useful when
sediment sluicing is required. It also is useful for diverting flow during construction. Gate
sizes are normally smaller for these spillways, but higher head and sealing details can
make them expensive.
Orifice spillways can be found in gravity dams, adjacent to embankment dams, and in
arch dams. Figs. 17.14 and 17.15 show a typical layout. The proportions of the orifice are
defined as follows:
1. Gates are sized on the basis of discharge requirements at maximum and minimum
headwater levels. The discharge coefficient can be assumed to be 0.90.
2. The roof curve must be shaped carefully to prevent cavitation if the head is high. For
instance, the design in Fig. 17.15 has a roof curve composed of a circular curve with
a large radius, followed by a 1:15 roof slope to the gates top seal. This shape was based
on a physical model test (Fig. 17.14).

Exhibit 17.5 Mangla Hydroelectric Project, Pakistan. A three-dimensional rendition of

the spillway gate structure^Courtesy Water and Power Development
Authority, Pakistan).

Groyne Well

Colt Control Buildirx)

Upper
St.llmg
Bonn

Upocr Ch!t

Chute

PLAN
Scolt of Fttl

SECTION
FIGURE 17.14 Typical arrangement of an orifice spillway. (Institute of Civil Engineers, (1968).

Toilroct
Cltonntl

FIGURE 17.15 Typical arrangement of an orifice spillway control structure.

(Institute of Civil Engineers, 1968)

3. The floor curve is less critical; however, for the design in Figs. 17.14 and 17.15 the
curve radius was 150 ft.
4. Potential formation of vortex is a special concern for orifice spillways. A vortex suppressor will generally be required and can be determined by a physical model.
5. The spillway downstream of the gate structure changes to a chute that may lead to a
stilling basin (see Figs. 17.14 and 17.15) or a flip-bucket.

77.6

MORNINGGLORYSPILLWAY

The morning glory spillway is normally used in conjunction with a tunnel spillway when
the intake is a vertical shaft. Also, because of flow entry from the entire periphery, the
crest capacity is relatively high. Crest gates are not normally used because of access and
cost considerations as well as poor hydraulic conditions in the shaft with partially open
gates Exhibit 17.6 illustrates a morning glory spillway. Figures 17.16 and 17.17 show the
range of possible flow conditions in the crest area for a morning-glory spillway
(USER, 1987). The design procedure is as follows:
1. By trial and error, determine the required design head H0 and the crest radius from Fig.
17.18; H0IRs = 0.3 is recommended. Note that in Fig. 17.18, the discharge coefficient
C0 is for English units. For metric units, the coefficient should be multiplied by a conversion factor of 0.552.
2. Determine the discharge rating curve for the full range of heads from Fig. 17.19.
3. Determine the lower nappe profile from Figs. 17.17 and 17.20 and Tables 17.2, 17.3,
and 17.4.

4. Check for throat control in the shaft using the following:

gl/2
R^

R =C

(17-7)

where: CR = 0.275 for units in m and mVsec, = 0.204 for units in ft and ftVsec, R =
Crest
Transition tub*
Throot of tronsition tub*
inclined shaft
AcceUrotmg

Outltt 1*0, of conduit

Dc*l*roting flow
CONDITION I. CREST CONTROL

Transition tub*
Oific* control at throat of transition

CONDITION 2. TUBE OR ORIFICE CONTROL

Hydraulic gradient
Toilwoter.
CONDITION 3. FULL PIPE FLOW

Design head

Pip* control, Q '0(H,- hv.) , condition 3-

Point
controlof chong*
to pip* from
flow orifice
Tub* or orifice control, O F(H9'4), condition 2
H*ad
at which tunnel
at downstream
end. flows 0.79 full

Erratic
rqnge flow

Crett contra),0 f (H, ^condition |

DISCHARGE - SCCONO-FEtT

FIGURE 17.16 Possible flow conditions for a morning-glory spillway. (USER, 1987).

Noppe-shoped crest
Crotch,where onnulor flow
becomes jet flow.

COEFFICIENT C0

FIGURE 17.17 Elements of nappe-shaped profile for a morning-glory

spillway. (USBR, 1987)

NOTE Dotted lines are

Dosed on extrapolation
of data.

FIGURE 17.18 Circular crest coefficients for a morning-glory spillway with aerated nappe.
(USER, 1987)

FIGURE 17.19 Circular crest coefficients of discharge for other than design head. (USER, 1987)

TABLE 17.2 Coordinates of Lower Nappe Surface for Different Values of HJR When PIR = O.15
^
0.20
0.2
0.30
0.35
0.40
0.45
0.50
0.60
0.80
X
Y For portion of the profile above the weir crest
"s
"j
0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000 0.0000
.010
.0120 .0120 .0115 .0115 .0110 .0110 .0105 .0100 .0090
.020
.0210 .0200
.0195 .0190 .0185 .0180 .0170 .0160 .0140
.030
.0285
.0270
.0265
.0260
.0250
.0235
.0225
.0200 .0165
.040
.0345
.0335
.0325
.0310 .0300
.0285
.0265
.0230 .0170
.050
.0405
.0385
.0375
.0360
.0345
.0320
.0300
.0250 .0170
.060
.0450
.0430
.0420
.0400
.0380
.0355
.0330
.0365 .0165
.070
.0495
.0470
.0455
.0430
.0410 .0380
.0350
.0270 .0150
.080
.0525
.0500
.0485
.0460
.0435
.0400
.0365
.0270 .0130
.090
.0560
.0530
.0510 .0480
.0455
.0420
.0370
.0265 .0100
.100
.0590
.0560
.0535
.0500
.0465
.0425
.0375
.0255
.0065
.120
.0630
.0600
.0570
.0520
.0480
.0435
.0365
.0220
.140
.0660
.0620
.0585
.0525
.0475
.0425
.0345 .0175
.160
.0670
.0635
.0590
.0520
.0460
.0400
.0305 .0110
.180
.0675
.0635
.0580
.0500
.0435
.0365
.0260
.0040
.200
.0670
.0625
.0560
.0465
.0395
.0320
.0200
.250
.0615 .0560
.0470
.0360
.0265
.0160 .0015
.300
.0520
.0440
.0330
.0210 .0100
.350
.0380
.0285
.0165 .0030
.400
.0210 .0090
.450 .0015
.500
.550
I
-tH1
0.000
.020
.040
.060
.080
.100
.150
.200
.250
.300
.400
.500
.600
.800
1.000
1.200
1.400
1.600
1.800
2.000
2.500
3.000
3.500
4.000
-4.500
5.000
5.500
6.000
^j-K

0.454
.499
.540
.579
.615
.650
.726
.795
.862
.922
1.029
1.128
1.220
1.380
1.525
1.659
1.780
1.897
2.003
2.104
2.340
2.550
2.740
2.904
3.048
3.169
3.286
3.396
0.20

Source: USBR (1987).

0.422
.467
.509
.547
.583
.616
.691
.760
.827
.883
.988
1.086
1.177
1.337
1.481
1.610
1.731
1.843
1.947
2.042
2.251
2.414
2.530
2.609
2.671
2.727
2.769
2.800
0.25

0.392
.437
.478
.516
.550
.584
.660
.729
.790
.843
.947
1.040
1.129
1.285
1.420
1.537
1.639
1.729
1.809
1.879
2.017
2.105
2.153
2.180
2.198
2.207
2.210

0.30

-Jj-H For portion of the profile below the weir crest

s
0.358
0.325
0.288
0.253
0.189 I 0.116
.404
.369
.330
.292
.228
.149
.444
.407
.368
.328
.259
.174
.482
.443
.402
.358
.286
.195
.516
.476
.434
.386
.310
.213
.547
.506
.462
.412
.331
.228
.620
.577
.526
.468
.376
.263
.685
.639
.580
.516
.413
.293
.743
.692
.627
.557
.445
.319
.797
.741
.671
.594
.474
.342
.893
.828
.749
.656
.523
.381
.980
.902
.816
.710
.567
.413
1.061
.967
.869
.753
.601
.439
1.202
1.080
.953
.827
.655
.473
1.317
1.164
1.014
.878
.696
.498
1.411
1.228
1.059
.917
.725
.517
1.480
1.276
1.096
.949
.750
.531
1.533
1.316
1.123
.973
.770
.544
1.580
1.347
1.147
.997
.787
.553
1.619
1.372
1.167
1.013
.801
.560
1.690
1.423
1.210
1.049 .827
1.738
1.457
1.240
1.073 .840
1.768
1.475
1.252 1.088
1.780
1.487 1.263
1.790 1.491
1.793

0.35

0.40

0.45

0.50

0.60

0.80

TABLE 17.3 Coordinates of Lower Nappe Surface for Different Values of HJR When PIR = 0.30
^-

-jj"s
0000
.010
.020
.030
.040
.050
.060
.070
.080
.090
.100
.120
.140
.160
.180
.200
.250
.300
.350
.400
.450
.500
.550
fj-Y
"s
0.000
.020
.040
.060
.080
.100
.150
.200
.250
.300
.400
.500
.600
.800
1.000
1.200
1.400
1.600
1.800
2.000
2.500
3.000
3.500
-4.000
-4.500
5.000
5.500
6.000
^

0.20

0.2

0.30

0.0000 I 00000 I 0.0000

.0130 .0130 .0130
.0245
.0242
.0240
.0340
.0335
.0330
.0415 .0411 .0390
.0495
.0470
.0455
.0560
.0530
.0505
.0610 .0575
.0550
.0660
.0620
.0590
.0705
.0660
.0625
.0740
.0690
.0660
.0800
.0750
.0705
.0840
.0790
.0735
.0870
.0810 .0750
.0885
.0820
.0755
.0885
.0820
.0745
.0855
.0765
.0685
.0780
.0670
.0580
.0660
.0540
.0425
.0495
.0370
.0240
.0300
.0170 .0025
.0090
.0060

OM9
.560
.598
.632
.664
.693
.760
.831
.893
.953
1.060
1.156
1.242
1.403
1.549
1.680
1.800
1.912
2.018
2.120
2.351
2.557
2.748
2.911
3.052
3.173
3.290
3.400
0.20

Source: USER (1987).

0488
.528
.566
.601
.634
.664
.734
.799
.860
.918
1.024
1.119
1.203
1.359
1.498
1.622
1.739
1.849
1.951
2.049
2.261
2.423
2.536
2.617
2.677
2.731
2.773
2.808
0.25

O455
.495
.532
.567
.600
.631
.701
.763
.826
.880
.981
1.072
1.153
1.301
1.430
1.543
1.647
1.740
1.821
1.892
2.027
2.113
2.167
2.200
2.217
2.223
2.228

0.30

0.35

0.40

0.45

0.50

0.60

0.80

-jy^ For portion of the profile above the weir crest

"s
I 0.0000 I 0.0000 I 0.0000 I 0.0000 I 0.0000 I 0.0000
.0125 .0120 .0120 .0115 .0110 .0100
.0235
.0225
.0210 .0195 .0180 .0170
.0320
.0300
.0290
.0270
.0240 .0210
.0380
.0365
.0350
.0320
.0285
.0240
.0440
.0420
.0395
.0370
.0325
.0245
.0490
.0460
.0440
.0405
.0350
.0250
.0530
.0500
.0470
.0440
.0370
.0245
.0565
.0530
.0500
.0460
.0385
.0235
.0595
.0550
.0520
.0480
.0390 .0215
.0620
.0575
.0540
.0500
.0395 .0190
.0650
.0600
.0560
.0510 .0380 0.120
.0670
.0615 .0560
.0515 .0355
.0020
.0675
.0610 .0550
.0500 .0310
.0675
.0600
.0535
.0475
.0250
.0660
.0575
.0505
.0435
.0.180
.0590
.0480
.0390
.0270
.0460
.0340
0.220
.0050
.0295 .0150
.0100

--Y For portion of the profile below the weir crest

"s
O422
O384
O349
O310
O238 I O144
.462
.423
.387
.345
.272 .174
.498
.458
.420
.376
.300
.198
.532
.491
.451
.406
.324
.220
.564
.522
.480
.432
.348 .238
.594
.552
.508
.456
.368
.254
.661
.618
.569
.510
.412 .290
.723
.677
.622
.558
.451 .317
.781
.729
.667
.599
.483 .341
.832
.779
.708
.634
.510 .362
.932
.867
.780
.692
.556
.396
1.020 .938
.841
.745
.595 .424
1.098
1.000
.891
.780
.627 .446
1.227 1.101 .970
.845
.672
.478
1.333 1.180
1.028
.892
.707
.504
1.419
1.240
1.070
.930
.733 .521
1.489
1.287 1.106
.959
.757 .540
1.546
1.323 1.131
.983
.778
.551
1.590
1.353 1.155
1.005 .797 .560
1.627 1.380
1.175
1.022 .810
.569
1.697
1.428 1.218
1.059 .837
1.747
1.464
1.247
1.081 .852
1.778 1.489 1.263 1.099
1.796
1.499 1.274
1.805 1.507
1.810

0.35

0.40

0.45

0.50

0.60

0.80

TABLE 17.4 Coordinates of Lower Nappe Surface for Different Values of HJR When
PIR = 2.00
^

0.00 0.10* 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.60 0.80 1.00 1.20 1.50 2.00

For portion of the profile above the weir crest

"s
s
0.000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
.010 .0150 .0145 .0133 .0130 .0128 .0125 .0122 .0119 .0116 .0112 .0104 .0095 .0086 .0077 .0070
.020 .0280 .0265 .0250 .0243 .0236 .0231 .0225 .0220 .0213 .0202 .0180 .0159 .0140 .0115 .0090
.030 .0395 .0365 .0350 .0337 .0327 .0317 .0308 .0299 .0289 .0270 .0231 .0198 .0168 .0126 .0085
.040 .0490 .0460 .0435 .0417 .0403 .0389 .0377 0.363 .0351 .0324 .0268 .0220 .0176 .0117 .0050
.050 .0575 .0535 .0506 .0487 .0471 .0454 .0436 .0420 .0402 .0368 .0292 .0226 .0168 .0092
.060 .0650 .0605 .0570 .0550 .0531 .0510 .0489 .0470 .0448 .0404 .0305 .0220 .0147 .0053
.070 .0710 .0665 .0627 .0605 .0584 .0560 .0537 .0514 .0487 .0432 .0308 .0201 .0114 .0001
.080 .0765 .0710 .0677 .0655 .0630 .0603 .0578 .0550 .0521 .0455 .0301 .0172 .0070
.090 .0820 .0765 .0722 .0696 .0670 .0640 .0613 .0581 .0549 .0471 .0287 .0135 .0018
.100 .0860 .0180 0.762 .0734 .0705 .0672 .0642 .0606 .0570 .0482 .0264 .0089
.120 .0940 .0880 .0826 .0790 .0758 .0720 .0683 .0640 .0596 .0483 .0195
.140 .1000 .0935 .0872 .0829 .0792 .0750 .0005 .0654 .0599 .0460 .0101
I
I
.160 .1045 .0980 .0905 .0855 .0812 .0760 .0710 .0651 .0585 .0418
.180 .1080 .1010 .0927 .0872 .0820 .0766 .0705 .0637 .0559 .0361
.200 .1105 .1025 .0938 .0877 .0819 .0756 .0688 .0611 .0521 .0292
.250 .1120 .1035 .0926 .0850 .0773 .0683 .0596 .0495 .0380 .0068
.300 .1105 .1000 .0850 .0764 .0668 .0559 .0446 .0327 .0174
.350 .1060 .0930 .0750 .0650 .0540 .0410 .0280 .0125
.400 .0970 .0830 .0620 .0500 .0365 .0220 .0060
.450 .0845 .0700 .0450 .0310 .0170 .000
.500 .0700 .0520 .0250 .0100
.550 .0520 .0320 .0020
.600 .0320 .0080
.650 .0000
y For portion of the profile below the weir crest
^-Y
-^"s
"j
0.000 0.668 0.615 0.554 0.520 0.487 0.450 0.413 0.376 0.334 0.262 0.158 0.116 0.093 0.070 0.048
-.020 .705 .652 .592 .560 .526 .488 .452 .414 .369 .293 .185 .145 .120 .096 .074
-.040 .742 .688 .627 .596 .563 .524 .487 .448 .400 .320 .212 .165 .140 .115 .088
-.0.60 .777 .720 .660 .630 .596 .557 .519 .478 .428 .342 .232 .182 .155 .129 .100
-.080 .808 .752 .692 .662 .628 .589 .549 .506 .454 .363 .250 .197 .169 .140 .110
-.100 .838 .784 .722 .692 .657 .618 .577 .532 .478 .381 .266 .210 .180 .150 .188
-.150 .913 .857 .793 .762 .725 .686 .641 .589 .531 .423 .299 .238 .204 .170 .132
-.200 .978 .925 .860 .826 .790 .745 .698 .640 .575 .459 .326 .260 .224 .184 .144
-.250 1.040 .985 .919 .883 .847 .801 .750 .683 .613 .490 .348 .280 .239 .196 .153
-.300 1.100 1.043 .976 .941 .900 .852 .797 .722 .648 .518 .368 .296 .251 .206 .160
-.400 1.207 1.150 1.079 1.041 1.000 .944 .880 .791 .706 .562 .400 .322 .271 .220 .168
-.500 1.308 1.246 1.172 1.131 1.087 1.027 .951 .849 .753 .598 .427 .342 .287 .232 .173
-.600 1.397 1.335 1.260 1.215 1.167 1.102 1.012 .898 .793 .627 .449 .359 .300 .240 .179
-.800 1.563 1.500 1.422 1.369 1.312 1.231 1.112 .974 .854 .673 .482 .384 .320 .253 .184
-.1.000 1.713 1.646 1.564 1.508 1.440 1.337 1.189 1.030 .899 .710 .508 .402 .332 .260 .188
-.1,200 1.846 1.780 1.691 1.635 1.553 1.422 1.248 1.074 .933 .739 .528 .417 .340 .266
-.1.400 1.970 1.903 1.808 1.748 1.653 1.492 1.293 1.108 .963 .760 .542 .423 .344
-.1.600 2.085 2.020 1.918 1.855 1.742 1.548 1.330 1.133 .988 .780 .553 .430
-.1.800 2.196 2.130 2.024 1.957 1.821 1.591 1.358 1.158 1.008 .797 .563 .433
-.2.000 2.302 2.234 2.126 2.053 1.891 1.630 1.381 1.180 1.025 .810 .572
I
I
-.2.500 2.557 2.475 2.354 2.266 2.027 1.701 1.430 1.221 1.059 .838 .588
-.3.000 2.778 2.700 2.559 2.428 2.119 1.748 1.468 1.252 1.086 .853
-.3.500
2.916 2.749 2.541 2.171 1.777 1.489 1.267 1.102
-.4.000
3.114 2.914 2.620 2.201 1.796 1.500 1.280
-.4.500
3.306 3.053 2.682 2.220 1.806 1.509
-.5.000
3.488 3.178 2.734 2.227 1.811
-.5.500
3.653 3.294 2.779 2.229
-.6.000
3.820 3.405 2.812 2.232
^
K
I 0.00 I 0.10 I 0.20 I 0.25 I 0.30 I 0.35 I 0.40 I 0.45 I 0.50 I 0.60 I 0.80 I 1.00 1.20 1.50 2.00
The tabulation for ^
K = 0.10 was obtained by interpolation between -^
K = O and 0.20.
Source: USER (1987).

Exhibit 17.6 Big Dalton Dam, California (Courtesy Los Angeles Dept. of Public Works)
(a) General view of the modified morning-glory spillway.
(b) Profile of the spillway.

required shaft radius for throat control (m or ft), Ha = head from headwater level to
throat location (m or ft), and Q = discharge under consideration (mVsec or ftVsec). If
R exceeds the radius of the shaft for a given flow, then throat control exists and the discharge is based on the shaft radius.
5. The downstream tunnel is ordinarily sized to flow no more than three-fourths full with
the maximum value of Manning's roughness coefficient n = 0.016 to avoid potentially unstable flow conditions. If the tunnel flows full, with throat control already developed, the capacity is dictated by full flow throughout.
6. An ideal design has crest control throughout the range of discharge.

77.7

LABYRINTHSPILLWAY

The labyrinth spillway is used to concentrate discharge into a narrow chute, where space
does not permit a linear ungated crest. It generally minimizes approach excavation,
whereas the concrete weir is more complicated to construct. Labyrinth spillways have
been built with a wide range of sizes and discharge capacities and are well suited for rehabilitation of existing spillway structures when increased spillway capacity is needed
(Hinchliff and Houston, 1984; Tacail et al., 1990; Tullis el al., 1995). Labyrinth structures
can be built economically if an adequate foundation is available.
Figure 17.21 is a typical layout (Tacail et al., 1990). The most efficient spillway
entrance for most reservoir applications is a curved approach adjacent to each end cycle
of the spillway, with the approach flow parallel to the centerline of the spillway cycles for
more uniform approach flow. The spillway entrance should be placed as far upstream in
the reservoir as possible to reduce localized upstream head losses. To avoid the submergence
effect, the supercritical flow condition should be maintained at the chute crest downstream
from the labyrinth. The number of spillway cycles should be determined on the basis of the
magnitude of the upstream head, the effect of nappe interference, and the economics of the
design. With normal operating conditions, the vertical aspect ratio for each labyrinth cycle,
w/P (Fig. 17.22), should be 2.5 or greater (Tullis et al., 1995).
Subatmospheric pressures under the nappe will cause nappe oscillation and noise and
should be avoided by venting for structural reasons. Splitter piers can be placed along the
spillway's side walls, or crushed stone can be placed along the downstream edge of the crest.
The piers should be located at a distance equal to 8 to 10 percent of the wall length upstream
of the downstream apexes (Tullis et al., 1995). Although the use of crushed stone may
reduce the spillway's capacity, it can be cost effective. Discharge capacity is a function of
the head over the crest, the height of the crest wall, the shape of the crest, the angle of the
labyrinth, the number of cycles, and the length of the side leg (Tullis et al., 1995). An efficient design will result from the following procedure:
1. Knowing the design discharge and the maximum head, determine the required effective length from
2 = |QLV2^f3/2

(178)

where Q = 0.30 is derived from Fig. 17.23, with a = 8, and H/p = 0.8.

Next Page

CHAPTER 18
HYDRAULIC DESIGN OF
STILLING BASINS AND
ENERGY DISSIPATORS

C. Y. Wei and James E. Lindell

Harza Engineering Company
Chicago, Illinois.

18.1

INTRODUCTION

Stilling basins and energy dissipators are usually provided in conjunction with development of spillways, outlet works, and canal structures. It is often necessary to perform
hydraulic model studies of individual structures to be certain that these energy dissipating
devices will operate as anticipated. A relatively large volume of data is available from
many laboratory and field studies performed in the past (Blaisdell, 1948; Bowers and
Toso, 1988; Bowers and Tsai, 1969; Chadwick and Morfett, 1986; Chaudhry, 1993; Chow
1959; French, 1985; George, 1978; Hendreson, 1966; International Commission on Large
Dams (ICOLD), 1987; Novak et al., 1990; Peterka, 1964; Robertson et al., 1988; Senturk,
1994; Toso and Bowers, 1988; U.S. Bureau of Reclamation (USBR), 1974, 1987; Vischer
and Hager, 1995, 1998). Based on the results of many intensive studies, in 1958, A. J.
Peterka (1964) of the U.S. Bureau of Reclamation (USBR) published a summary report
of USBR'S studies entitled Hydraulic Design of Stilling Basins and Energy Dissipators,
Engineering Monograph No. 25. Since then, this publication has been referenced widely
within the hydraulic engineering community and still is one of the best references on this
subject available today. Energy dissipators are used to dissipate excess kinetic energy possessed by flowing water. An effective energy dissipator must be able to retard the flow of
fast moving water without damage to the structure or to the channel below the structure.
Vischer (1995) classified various types of energy dissipators by their features as: (1) by
sudden expansions, (2) by abrupt deflections, (3) by counterflows, (4) by rough walls, (5)
by vortex devices, and (6) by spray inducing devices. The stilling basins and energy dissipators discussed in this chapter are related to energy dissipation by expansion and
deflection.
There are two basic types of energy dissipators. They are hydraulic jump-type dissipators and impact-type dissipators. The hydraulic jump type energy dissipators dissipate
excess energy through formation of highly turbulent rollers within the jump. The impacttype dissipators direct the water into an obstruction that diverts the flow in all directions
and generates high levels of turbulence and in this manner dissipates the energy in the
flow. In other cases, the flow is directed to plunge into a pool of water where the energy

Max design flood

El. 575.0
1,400,000 cts
H. W. EL 570.0
Flow

Radial gate: 50ft. wide

65 ft. high
Max. T.W. El 527.4
1,400,000 cfs

Crest EL 5050
Min. T.W. E1481.4
)
KEYPLAN
EL. 456.0

**
WANAPUM SPILLWAY
(State of Washington, USA )

Exhibit 18.1: Wanapum project, Washington

(a) General view of the spillway in operation.
(b) Layout of the spillway and stilling basin.

is diffused and dissipated. The impact-type energy dissipators include check drops and
vertical drops, baffled outlets, baffled aprons, and vertical stilling wells. Generally, the use
of an impact-type energy dissipator results in smaller and more economical structures.

18.2 STILLINGBASINS
Using six test flumes, USBR conducted model studies for five stilling basin designs. The
results are summarized and presented in the Engineering Monograph No. 25 mentioned
above (Peterka, 1964). In Basin I tests, all test flumes were used and the test data obtained
provides basic hydraulic information concerning hydraulic jumps on a horizontal apron.
The Type II basin was developed for high dam and earth dam spillways and large canal
structures where the approach velocity is high and the corresponding Froude number
exceeds 4.5. Type III stilling basin is suitable for general canal structures, small outlet
works, and small spillways where the approach velocity is moderate or low and does not
exceed 50-60 ft/s (15-18 m/s) and the unit discharge is less than 200 ftVs/ft (18 mVs). For
smaller canal structures, outlet works, and diversion dams where the approach Froude
number is relatively low (between 2.5 and 4.5) and the heads of the structures do not
exceed 50 ft (15 m), Type IV stilling basin may be used. However, the jumps in the basin
may be unstable and alternative design such as the modified Type IV basin may be considered. To achieve greater structure economy for high dam spillways, Type V stilling
basin with sloping apron may be considered. Photos of several stilling basins in operation
are given in Exhibits 17.2, 17.4, 18.1, and 18.2.
18.2.1 General Hydraulic Jump Basin (Basin I)
The basic elements and characteristics of a hydraulic jump on horizontal aprons (Figs.
18.1 and 18.2) is provided to aid designers in selecting more practical basins such as
Basins II, III, IV, V, and VI. Jump occurs on a flat floor with no chute blocks, baffled piers
or end sill in the basin. Usually, it is not a practical basin because of its excessive length.
For a high-velocity flow down a spillway chute with known terminal velocity (Fig. 18.1)
and depth entering the basin, the required tail water depth, the length of jump, and loss of
energy can be determined based on the curves provided in Fig. I8.2a-e.
18.2.2 Stilling Basins for High Dam and Earth Dam Spillways
and Large Canal Structures (Basin II)
This stilling basin was developed for use on high spillways, large canal structures, and
so forth for approach Froude numbers above 4.5. With chute blocks and dentated end sill,
the jump and basin length can be reduced by about 33 percent. The basic design features
of Basin II stilling basin are given in Fig. 18.3a. For preliminary designs, the curves for
estimating required tailwater depth, and length of jump are given in Figures 18.3Z? and c.
The water surface and pressure profiles can be determined based on Fig. 18.3J and e.
The water surface profile in this basin can be closely approximated by a straight line
making an angle a(jump angle) with the horizontal. This line can also be considered as
a pressure profile. The USBR guidelines for designing this type of stilling basins are
given as follows:

7OThOiSt

28V at axis of dam

Axis of dam

Radial gate: 41' wide

40' high

Flow
Crest El. 385.0.

Intermediate
. raining wall

KEYPLAN
End training wall
MAYFIELD SPILLWAY
(State of Washington, USA ),

Exhibit 18.2 Mayfield hydroelectric project, Washinton

(a) Layout of the spillway including flip bucket.
(b) General view of the spillway and the stilling pool
(c) General view of the spillway and stilling pool.

Z=FALU FROM RESERVOIR LEVEL TO STILLING BASIN FLOOR-FEET

PROTOTYPE TESTS
X Shasta Dam
o Grand Coulee Dam

VA (actual)
VT (theoretical)
FIGURE 18.1 Curves for velocity entering stilling basins from 0.8:1 to 0.6:1 steep slopes.(From
Peterka, 1964)

LOSS IN ENERGY IN PERCENT

FROUOE NUMBER
LOSS OF ENERGY IN JUMP
FROUOE NUMBER
RATIO OF TW DEPTH TO O1
F 1.7-2.5, PRE-JUMP

F.-2.5-4.S, TRANSITION

F, 4.5-9.0, RANGE OF GOOD JUMPS

FROUOE NUMBER
LENGTH OF JUMP
F ( 9.0-UPWARD, EFFECTIVE BUT ROUGH
JUMP FORMS

FIGURE 18.2 Basic hydraulic jump basins on horizontal aprons. (Basin I) (From Peterka, 1964)

OC DEGREES

Slooe 2:l

FROUOE NUMBER
FROUOE NUMBER
MINIMUM TW DEPTHS
(SWEEPOUT)

FROUOE NUMBER
LENGTH OF JUMP

Profile for greater than

conjugate depth
Pressure profile for
conjugate depth

WATER SURFACE AND

PRESSURE PROFILES

FIGURE 18.3 Stilling basins for high dam and earth dam spillways and large canal structures. (Basin
H)(From Peterka, 1964)

1. Determine velocity V1 of flow entering the jump. Figure 18.1 may be used. This
chart represents a composite of experience, computation, and a limited amount of
experimental information obtained from prototype tests on Shasta and Grand Coulee
Dams. The chart provides a fair degree of accuracy for chute having slopes of 0.8:1
or steeper, where computation is a difficult and arduous procedure. The asymptotic nature of the terminal velocity curves is also depicted in Fig. 18.1. For a constant
head of 2.5 ft (0.8m) on the spillway crest, the terminal velocity does not increase
significantly (from 51 ft/s or 15.5 m/s to 53 ft/s or 16.2 m/s) as the vertical distance
(fall) from the reservoir level to stilling basin floor increases from 200 to 600 ft
(61-183 m).
2. Set apron elevation to utilize full conjugate tail water depth. Add a factor of safety
if needed. A minimum margin of safety of 5 percent of tailwater depth (D2) is recommended.
3. Exercise caution with effectiveness of the basin at lower values of the Froude number
(V1 / (^D1)172) of 4 or lower. D1 is the depth of the flow entering the basin.
4. Determine the length of basin using the curve shown in Fig. 18.3c.
5. Use the depth of flow entering the basin, D1 as the height of chute blocks. The
width and spacing should be equal to approximately D1 but can be varied to avoid
fractional blocks. A space equal to D/2 is preferable along each side of wall to
reduce spray and maintain desirable pressures.
6. As shown in Fig. 18.3a, set the height of the dentated sill equal to 0.2D2 and the
maximum spacing approximately 0.15D2. For narrow basins, the width and spacing may be reduced but they should remain equal.
7. It is not necessary to stagger the chute blocks with respect to the sill dentates.
8. It is recommended that the sharp intersection between chute and basin apron be
replace with a curve of reasonable radius of at least 4D1 when the chute slope is 1:1
or greater. Chute blocks can be incorporated on the curve surface as readily as on
the plane surfaces. The chute slope (0.6:1-2:1) does not have significant effect on
the stilling basin action unless it is nearly horizontal.
Following the above rules should result in a safe, conservative stilling basin for
spillways up to 200 ft (60 ms) high and for flows up to about 500 (ftVs/ft) [46.5
(mVs/m)] basin width, provided that jet entering the basin is reasonably uniform
both as to velocity and depth. For greater falls, larger unit discharges, or possible
asymmetry, a model study of the specific design is recommended.
18.2.3 Short Stilling Basins for Canal Structures, Small Outlet Works,
and Small Spillways [Basin III and the St. Anthony Falls (SAF) Basin]
For structures carrying relatively small discharges at moderate velocities, a shorter
basin having a simpler end sill may be used if baffled piers are placed downstream from
the chute blocks (Fig. 18.4). In this section, stilling basins for smaller structures in
which velocity at the entrance to the basin are moderate or low ( up to 50-60 ft/s or
15-18 m/s) and discharges of up to 200 ft3/s/ft or 18 m3/s/m are discussed. The stilling
basin action is very stable for this design. It has a large factor of safety against sweepout of the jump and operates equally well for all values of the Froude number above 4.0.

Slope 2:1

MIn. IVt depth

Basin IIi

FROUDE NUMBER
MINIMUM TW DEPTHS
(SWEEPOUT)

FROUOE NUMBER
HEIGHT OF BAFFLE BLOCKS
AND END SILLS
Profile for
conjugate depth

FROUOE NUMBER
LENGTH OF JUMP

WATER SURFACE AND

PRESSURE PROFILES

FIGURE 18.4 Short stilling basins for canal structures, small outlet works and small spillways.(Basin
III) (From Peterka, 1964)

This basin should not be used for velocities above 50 ft/s or 15 m/s to avoid potential cavitation damages aginst baffle piers. Instead, Basin II type stilling basin should be considered or hydraulic model studies should be performed. The following USBR guidelines
pertain to the design of the Basin III type stilling basin:
1. The stilling basin operates best at full conjugate tail water depth, D2. A reasonable
factor of safety is inherent in the conjugate depth for all values of the Froude number and it is recommended that this margin of safety not be reduced.
2. Determine the length of basin using the design curve given in Fig. ISAc. It is less
than one-half the length of the natural jump. It should be noted that an excess of
tail water depth does not substitute for basin length or vice versa.
3. Exercise caution with effectiveness of the basin at lower values of the approach
Froude number [V11 CgD1)172] of 4.5 or lower.
4. Height, width, and spacing of chute blocks should equal the average depth of flow
entering the basin, or D1. Width of blocks may be decreased, provide spacing is
reduced a like amount. Should D1 proved to be less than 8 in or 20 cm, the blocks
should be made 8 in or 20 cm high.
5. The height of the baffle piers (Fig. 18.4a) varies with the Froude number and is
given in Fig. 18.4d. In narrow structures, block width and spacing may be reduced,
provided both are reduced a like amount. A half space is recommended adjacent to
the walls.
6. The upstream face of the baffle piers should be set at a distance of 0.8D2 from the
downstream face of the chute blocks. This dimension is important.
7. The height of the solid end sill is given in Fig. 18.4d. The slope is 2:1 upward in
the direction of flow.
8. It is undesirable to round or streamline the edges of the chute blocks, end sill, or
baffle piers. It reduces the effectiveness of the energy dissipation. However, small
chamfers on the block edges to prevent chipping of the edges and to reduce cavitation erosion may be used.
9. It is recommended that a radius of reasonable length greater than 4D1 be used at the
intersection of the chute and basin apron for slopes of 45 or greater.
10. As a general rule, the slope of the chute has little effect on the stilling basin action
unless long flat slopes are involved.
11. Experience indicates that the Type III basin works well for flow less than 200
ft3/s/ft or 18 m3/s/m based on basin width and approach velocity at the entrance of
up to 50-60 ft/s or 15-18 m/s.
The St. Anthony Falls (SAF) Hydraulic Laboratory of the University of Minnesota had
also developed a similar basin for small spillways, outlet works, and small canal structures
for approach Froude numbers ranging from 1.7 to 17 (Blaisdell, 1948; Chow, 1959). This
basin was developed to achieve about 70 to 90 percent reduction of the jump lengths.
This basin is commonly known as the SAF stilling basin (Fig. 18.5). Since the basin is
relatively short so that a significant amount of residual energy can still exist downstream
from the end sill, the channel reach downstream from the stilling basin should be allowed
to erode until a stable scour depth is reached. Otherwise riprap protection should be provided to minimize scour (Sec. 18.7). The guidelines for designing this basin are summarized as follows (Blaisdell, 1948; Chow, 1959):

RECTANGULAR STILLING BASIN

HALF-PLAN

O0Io 90
45preferred

ODfM
IT)
in
Cl
VH
Il
?*
5-3O
9
<*
VMCM
CD
O
^T
O

TRAPEZOIDAL STILLING BASIN

HALF-PLAN

Chute blocks
Floor blocks

End sill

CENTERLINE SECTION

Vories
Cut-off woll

Side woll
Wing woll
Top slope is1M

TRAPEZOIDAL STILLING BASIN ' RECTANGULAR STILLING BASIN

DOWNSTREAM ELEVATION
FIGURE 18.5

The SAF stilling basin. (From Blaisdell, 1948)

1. The length L of the stilling basin is determined by the following equation.

_ 4.5y2
fO.16
where y2 is the theoretical sequent depth of the jump corresponding to the approach
flow depth J1 and F1 is the approach Froude number.
2. The height of the chute blocks and floor blocks is y19 and the width and spacing are
approximately 0.7Sy1.
3. The distance from the upstream end of the stilling basin to the floor blocks is Z/3.
4. No floor blocks should be placed closer to the side-wall than 3^/8.
5. The floor blocks should be placed downstream from openings between the chute
blocks.
6. The total width of the floor blocks should occupy about 40 to 55 percent of the stilling basin width.
7. The widths and spacings of the floor blocks for diverging stilling basins should
be increased in proportion to the increase in stilling basin width at the floor block
location.
8. The height of end sill is given by c = 0.07V2.
9. The depth of tail water above the stilling basin floor is given by
y2=

'

( U ~ "S)V2

for F1 = 1.7 to 5.5

y'2 = 0.8Sy2

for F1 = 5.5 to 11.0

y 2 = fl-00 - -^-I2
^
oUUJ

for F1 = 11.0 to 17.0

10. The top of the side-wall above the maximum tailwater level to be expected during
the life of the structure is given by z = J2I^11. Wing-walls should be equal in height to the stilling basin side-walls. The top of the
wing-wall should have a slope of IH:IV.
12. The wing-wall should be placed at an angle of 45 to the outlet center line.
13. The stilling basin side-walls may be parallel for a rectangular stilling basin or they
may diverge as an extension of the transition side-walls for a trapezoidal stilling
basin as shown in Fig. 18.5.
14. A cutoff wall of nominal depth should be used at the end of the stilling basin.
15. The effect of entrained air should be neglected in the design of the stilling basin.
18.2.4 Low Froude Number Stilling Basins
(Basin IV and Modified Basin IV)
This stilling basin was developed for canal structures, outlet works, and diversion dams
where the approach Froude number of the basin is relatively low (between 2.5 and 4.5)

MQX. tooth width D1

Spoce-2.5w
Top surface on 5*slope

Slope 2 - i
FIGURE 18.6 Low Froude number (2.5-4.5) stilling basin design (Basin
IV).(From Peterka, 1964)
and the heads of the structures are about 50 ft (or 15 m). In this case, the jump is not fully
developed and unstable and the methods of design discussed previously do not apply.
Alternative design and/or wave suppressors or Basin VI type stilling basin with a hanging baffle for energy dissipation may be considered. Guidelines for developing
a low Froude number stilling basin (Basin IV) as depicted in Fig. 18.6 are given as
follows (Peterka, 1964):
1. A model study of the stilling basin is imperative.
2. Reduction of excessive waves created in the unstable jump is the main problem
concerning the design of the stilling basin.
3. A tailwater depth of 10 percent greater than the conjugate depth is strongly recommended.
4. Place as few appurtenances as possible in the path of the flow, as volume occupied
by appurtenances helps to create a backwater problem, thus requiring higher
training walls.
5. Use Fig. 18.6 to develop the design of the stilling basin. The number of deflector
blocks shown in the figure is a minimum requirement.
6. The length of basin can be obtained from Fig. 18.2c. No baffle piers are needed in
the basin.
7. The recommended maximum width of the deflector blocks is equal to D1 but
0.75D1 is preferable from a hydraulic standpoint. The ratio of block width to spacing should be maintained as 1:2.5.
8. The extreme tops of the deflector blocks are 2D1 above the floor of the stilling
basin.
9. To accommodate the various slopes of chutes and ogee shapes encountered, the
horizontal top length of the deflector blocks should be at least 2D1. The upper surface of each block is sloped at 5 in a downstream direction for better operation
especially at lower discharges.

10. The addition of a small triangular sill placed at the end of the apron for scour control is desirable. An end sill of the type developed for short stilling basins (Basin
III) can be used. The slope of the upstream face of the sill is 2:1 and the height of
the sill can be determined based on Fig. 18.4J.
11. Basin IV stilling basin is applicable to rectangular cross sections only to minimize
potential wave-related problems.
Type IV stilling basin performs effectively in dissipating the energy at low Froude
number flows for small canals and for structures with small unit discharges. It is also
effective in minimizing wave problems. Based on additional model tests, the U.S. Bureau
of Reclamation (USBR) has developed a modified stilling basin for low Froude number
approach flows (George, 1978). This stilling basin is suitable for approach flows with
Froude numbers ranging from 2.5 to 5.0. The basin is relatively short and is provided with
chute blocks, baffle piers, and a dentated end sill as shown in Fig. 18.7a. The guidelines
for designing Modified Type IV stilling basin are given as follows (George, 1978):

PLAN
Baffle piers

Chute blocks

Chute blocks

Baffle piers
2:1 Slope

Toe of chute

ELEVATION
FIGURE 18.7(a) Low Froude number stilling basin (Modified Basin IV).(From George, 1978)

1. A hydraulic model study is recommended to confirm the design. Erosion tests

should be included. Such tests should be made over a full range of discharges to
determine erosion potential downstream from the basin and to determine the potential for the abrasive bed materials to move upstream into the basin.
2. Determine the theoretical D2 based on the known unit discharge and the approach
flow depth D1.

- Test Data
o -Natural Basin

FIGURE 18.7(b) Design curves for modified Basin IV stilling basin.(From

Georges, 1978)

3.
4.
5.
6.
7.
8.
9.
10.

Determine tailwater depth as TW = 1.05 D2.

Set the length of the basin L = 3D2 (approximately).
Use Fig. 18.7'a to develop the basic dimensions of the basin.
Determine the distance X from the chute blocks to the baffle piers. X varies from 1.3
to 0.7 times D2 as the approach Froude number varies from 2.5 to 5.6 as shown in
Fig. 18.7ft.
Determine the distance L1 from the toe of the chute to the upstream face of the end
sill from Fig. 18.7ft.
If (L1 + the length of the end sill) is longer than L then the stilling basin should be
extended to include the end sill.
Set the widths of the baffle piers equal to 0.7D1 and heights equal to LOD1.
Determine the number of chute blocks and baffle piers by the following equations.
The total number of chute blocks and spaces W = (width - 2kW)/W
where
k = fractional width of block equal to side clearance, 0.375 < k < 0.50
width = total width of stilling basin
W = 0.7OD1
The TV value obtained should be rounded to the nearest odd number and then adjust
values of W and k should also be adjusted.

11. Use 0.2D1 as the top length of the baffle piers.

12. Determine end sill dimensions.
height - 0.2D2
width, W= 0.15D2
top length of end sills = 0.2 X height
The number of blocks and spaces TV = (basin width)/W
(N should be rounded to the nearest odd number and then the value of W
should be adjusted)
18.2.5 Stilling Basin with Sloping Apron
To achieve greater structural economy, a stilling basin with a sloping apron can be considered. This type of stilling basin is usually used on high dam spillways. It needs greater
tail water depth than horizontal apron. The energy dissipation is as effective as occurs in
the true hydraulic jump on a horizontal apron. The primary concern in sloping apron
design is the tail water depth which is required to move the front of the jump up the slope
to the location where the jump is expected to start. It may not be economically feasible to
design the basin to confine the entire jump, especially when sloping aprons are used in

FIGURE 18.8 Stilling basins with sloping aprons. (From Peterka, 1964)

conjunction with medium or high overfall spillways where the rock foundation is in fairly good condition. When shorter aprons are used, the riverbed downstream must act as
part of the stilling basin. On the other hand, when the quality of foundation material is
questionable, it is desirable to make the apron sufficiently long to confine the entire jump.
The total apron length may range from about 40 to 80 percent of the length of jump.
The hydraulic jump may occur in several ways on a sloping apron, as depicted in
Fig. 18.8. The jump may have its toe form on the slope and the jump itself ends over the
horizontal apron (Case B), or ends at the junction of the slope and the horizontal apron
(Case C), or the entire jumps forms on the slope (Case D). For practical purposes the action
in Cases C and D is the same. Guidelines for the design of sloping aprons are given below:
1. Determine an apron arrangement that will give the best economy for the maximum
discharge condition.
2. The first consideration should be to determine the apron slope that will require the
minimum amount of excavation, the minimum amount of concrete, or both, for the
maximum discharge and tailwater condition.
3. Position the slope so that the front of the jump will form at the upstream end of the
slope for the maximum discharge and tailwater condition (Fig. 18.9). It may be necessary to raise or lower the apron, or change the slope entirely. Data obtained from
13 existing spillways are also shown in Fig. 18.9. Each point in the figure has been
connected with an arrow to the tan(6) curve corresponding to the apron slope. The
adequacy of the tailwater depth of these spillways can then be evaluated.
4. Use Fig. 18.10 to determine the length of the jump for maximum or other
flows. Shorter basins may be used where a solid bed exists. For most installations,
an apron length of about 60 percent of the length of jump for the maximum discharge condition should be sufficient. Longer basins are needed only when the
downstream riverbed is in very poor condition.

5. Ascertain that the tail water and length of basin available for energy dissipation
are sufficient for, say 1/4, 1/2, and 3/4 capacity. If the tailwater depth is deficient,
a different slope or a new position of the sloping portion of the apron should be
considered.
6. Horizontal and sloping aprons will perform equally well for high values of the
Froude number if the proper tail water depth is provided.
7. The slope of the chute upstream from a stilling basin has no significant effect on
the hydraulic jump when the velocity distribution and depth of flow are reasonably
uniform on entering the jump.
8. A small solid triangular sill should be provided at the end of the apron to lift the
flow as it leaves the apron for scour protection. The most effective height is
between 0.05D2 and 0.1OD2 and a slope of 3:1-2:1. Several existing stilling basins
with sloping aprons are shown in Fig. 18.11. All stilling basins shown were
designed with the aid of model studies.
9. The stilling basin should be designed to operate with as nearly symmetrical flow in
the stilling basin as possible to avoid formation of large circulating eddies and
transport of riverbed material into the apron area, and the potential undermining of
the wing walls and riprap.
10. A model study is advisable where the discharge over high spillways exceeds 500
ft3/s/ft or 46.5 m3/s/m based on the apron width, where there is any form of asymmetry involved, and for the high values of the Froude number where stilling basins
become more costly and the performance becomes less acceptable.

FIGURE 18.9 Comparison of existing sloping apron designs with experimental results.
(From Peterka, 1964)

SYMBOL

TAN$
0.050 - .067
0.100
0.135
0.150-.164
0.174
0.185
0.200-.218
0.263 - .280

SOURCE
Flumes A, B and F
Flumes A, B, D and F
Flumes A,
Flumes A, B and F
Flumes E
Flumes A
Flumes A, B and F
Flumes A and F

FIGURE 18.10 Jump length in terms of conjugate depth, D2 for stilling basins with sloping aprons.(From Peterka, 1964)

NORRlS

SHASTA

CANYON FERRY

BHAKRA
( PMELl
MlNAMV )

MAOOEN
BHAKRA
( FINAL )
FIGURE 18.11(a) Existing stilling basins with sloping aprons.(From Peterka, 1964)

FOLSOM

CAPILANO

FRIANT

DICKINSON
FIGURE 18.11(b) Existing stilling basins with sloping aprons.(From Peterka, 1964)

OLYMPUS

RIHAND

KESWICK

18.2.6 Other Types of Stilling Basins

Other types of stilling basins that may be considered include: (1) positive step basin, (2)
negative step basin, (3) baffle-sill basin, (4) baffle-block basin, (5) expanding stilling
basin, and (6) bucket stilling basin. Detailed discussions of these basins have been provided by Vischer and Hager (1995). These basins are briefly discussed as follows:
1. Positive-step basin. An upward step of a given height is provided in a prismatic channel. No end sills are included. The required basin length is significantly longer than
that of a classical jump basin.
2 Negative-step basin. A downward step is provided. No end sills are included. It
requires a slightly longer basin length than the positive step basin. Basins with steps
have not been popular because it is easier to use sills or blocks in a horizontal apron
than to change the apron elevation at the step section.
3 Baffle-sill basin. A weir-type sill is provided to form a basin. The flow over the sill
may be submerged or free. The sill is capable of stabilizing the jump in a shorter
basin and with lower tailwater than is the classical jump basin. Sills can be economical and effective devices for energy dissipation even without additional appurtenance included.
4. Baffle-block basin. Baffle blocks are normally arranged in one or several rows that
are oriented perpendicular to the direction of approach flow. Standard baffle blocks
such as the USBR blocks should be used. Baffle blocks are prone to cavitation damage and should not be used for approach velocities above 20 m/s. For velocities
between 20 and 30 m/s, a chamfer on the block edges should be provided to reduce the
cavitation potential.
5. Expanding stilling basin. There are two types of expanding basins, namely gradually
expanding basin and abruptly expanding basin. The gradually expanding basin
requires less tailwater depth and can be used for highly variable tailwater. This type of
basin is suitable for approach flow with Froude numbers less than 4. Very few basins
of this type have been built. An abruptly expanding basin has been studied and reported by Vischer and Hager (Novak et al., 1990). No practical applications have been
reported.
18.2.7 Fluctuating Pressures on Stilling Basin Floors
When designing a stilling basin to achieve highest possible hydraulic efficiency in terms of
energy dissipation, one should also consider the structural aspects of the stilling basin. The
effect of transient pressures caused by turbulence in the jump can be significant and should
be considered in the design of the structure. Extensive discussions of this subject have been
provided by International Commission on Large Dams (ICOLD, 1987), Toso and Bowers
(1988), and Visher and Hager (1995), and so on. The hydromechanic characteristics and
the turbulence level of the jump in a stilling basin depends not only on the relative tailwater level but also on the geometry and the concrete finish conditions of the basin floor and
training walls. The pressure fluctuations resulting from intense macro-scale turbulence in
the jump must be carefully considered during the design of the structure. The pressure fluctuations vary widely in amplitude at all locations within the jump. The maximum halfamplitude of the fluctuation has been determined to be approximately 40 percent of the
mean approach velocity head with a frequency of about 1 Hz. The dominant pulsating
components have frequencies between O and 10 Hz. When the pressure becomes negative

at a point on the apron surface, a dangerous local instability may develop with respect to
the uplift pressure at the bottom of the concrete slab. Some projects have experienced high
uplift pressures under large areas of the basin floor and resulted in complete floor concrete
slabs being torn up (Bowers and Toso, 1988; ICOLD, 1987; Toso and Bowers, 1988). In
addition, cavitation, abrasion, and vibration due to intense turbulence and pressure fluctuations can also contribute significantly to the damage of a stilling basin.
Based on the model studies of USBR Type II and Type III stilling basins, Toso and
Bowers (1988) obtained the following useful conclusions:
1. The pressure fluctuations in the jump tend to approach a definite limit, on the order of
80 to 100 percent of the approach velocity head. This is on the order of 10-20 times
the root-mean square (rms) of the pressure fluctuation.
2. Addition of chute blocks, intermediate blocks, and end sills did not result in significantly higher maximum negative and positive deviations than those for basins without
blocks and sills. The energy dissipation was quicker.
3. Side-wall pressure fluctuations are very significant, and peak at one to two inflow
depths above the floor.
4. The longitudinal extent of extreme pressure pulsation in the zone of maximum turbulence
is on the order of eight times the inlet flow depth. The lateral extent of a characteristic
pulse is approximately 1.6 times the longitudinal extent or 13 times the inlet flow-depth.
ICOLD (1987) recommended, as a minimum precaution, that the following two conditions be considered when designing the stilling basin apron:
1. Full downstream uplift pressure applied over the entire area of the floor with basin empty.
2. Full uplift pressure equals 12 percent of the approach velocity head applied under the
whole basin, with the basin full.
If necessary, the basin floor can be strengthened by providing anchors or using thicker slabs which may be held in place by the side walls.
ICOLD (1987) also recommended following structural arrangements to minimize
potential uplift damages due to undesirable turbulent flow induced pressure fluctuations.
1. All contraction joints should be fitted with properly located and embedded seals.
2. There should be no drain openings in the training wall inside the basin. However, drain
outlets in a dentated sill at the beginning of a stilling basin have performed satisfactorily.
3. Keep the areas of the floor slabs as large as possible.
4. Connect slabs by means of dowels, shear keys, and reinforcement across the joints.
5. Keep horizontal construction joints to a minimum, with dowels across them.
6. If drainage is necessary, keep it well away (1-1.5 m at least) from the wetted surfaces
so that abrasion or cavitation erosion will not make it accessible to the turbulent flow.

18.3

DROP-TYPE ENERGY DISSIPATORS

For small drops in canals with values of the Froude number between 2.5 and 4.5, a
drop-type energy dissipator which is in the form of a grating is particularly applicable to
reduce wave actions and dissipating energy. The device causes the over falling water jet
to separate into a number of long, thin sheets of water which falls nearly vertical into the
canal below. It has excellent capability in dissipating energy and eliminating wave problems. Guidelines for developing a drop-type energy dissipator are given as follows
(Peterka, 1964):
Next Page

CHAPTER 19
FLOODPLAIN
HYDRAULICS

Roy D. Dodson
Dodson & Associates, Inc.
Houston, Texas.

This chapter deals with the practical considerations involved in identifying the floodplain for
a stream channel, and analyzing the flow characteristics of the channel system, including its
floodplain. In general, any land area that is susceptible to inundation by rising or flowing
flood waters from any source could be considered to be a floodplain area. This would include
areas affected by coastal or lacustrine (lake) flooding. However, the emphasis of this chapter
will be on riverine floodplains, which are affected by flood waters from a stream or river.
The determination of flood elevations and floodplain boundaries along stream channels has become increasingly important as more development has occurred in floodplain
areas, or in areas that have been reclaimed from the floodplain.
A floodplain analysis requires a large amount of data in order to be accurate and
complete. Cross sections of the stream channel and detailed geometric descriptions of
bridges and other structures are required. In addition, experience is usually required to
accurately assess the roughness characteristics of a channel, to properly lay out cross
sections, and to adequately address many other aspects of a good flood plain analysis.
Sophisticated computer programs are available to analyze the data and produce detailed
computations of water surface elevations, floodplain boundaries, and other results.
However, these programs can produce misleading results without a properly planned
and executed analysis.
Because of the subjective judgments required for some of the input data and the uncertainty associated with certain values such as flow rates, some people have the attitude that
"there is no right answer" to the question of determining floodplain elevations and boundaries. However, it is important not to allow this to degenerate into the attitude that "one
answer is as good as another." This is clearly not true. A floodplain study performed well
produces superior results.
This chapter describes an approach toward floodplain studies that emphasizes quality
control and thoroughness in all aspects of the study effort. The effort is focused on those
aspects of the study that will produce the best overall results within given constraints of
time and budget. This chapter provides guidance in the following areas:

Sources of information for the floodplain analysis;

Planning data collection operations to obtain required information;
Selecting the analytical approach for a particular floodplain situation;
Performing the analysis and assessing the results.

19.1 LOCATING EXISTING DATA SOURCES FOR

FLOODPLAIN STUDIES
The first step in any floodplain analysis is to collect all available information that
might be useful in the analysis. Available mapping and/or cross section data should be
pursued before requesting new surveys. One or two weeks of contacting key government agencies, as well as local engineers, surveyors, aerial photographers, and mapping companies could save significant project funds. The principal sources of information which may be available include previous studies, topographic data, aerial photography, highway or street maps, construction drawings, stream gage data, and personal observations from local residents.
Within the United States, federal agencies should be contacted to determine if they have
data pertinent to the project being initiated. Table 19.1 lists the chief types of floodplain

Floodplain

Storm Surge Data

Flood

Historic Floods

Streamflow Data

Meteorologic Data

Hydrologic Data

TABLE 19.1 Federal Government Data Sources

Bureau of Land Management

(BLM)*
Bureau of Reclamation
Federal Emergency
Management Agency (FEMA)*
National Weather Service
Natural Resources
Conservation Service
(NRCS)11
Tennessee Valley Authority
(TVA)
U.S. Army Corps of
Engineers (USACE)
U.S. Geological Survey
(USGS)
The BLM operates in the states of Alaska, Arizona, California, Colorado, Idaho, Michigan,
Nevada,
Oregon, Vermont, and Wyoming.
1
TIiC Bureau of Reclamation operates only in the states of Arizona, California, Colorado, Idaho,
Kansas, Montana, Nebraska, North Dakota, New Mexico, Nevada, Oregon, Oklahoma, South Dakota,
Texas, Utah, Washington, and Wyoming.
*FEMA Flood Insurance Studies are an especially important source of information, as described
below.
"The NRCS was formerly called the Soil Conservation Service (SCS).
The TVA operates only in the states of Alabama, Georgia, Kentucky, Mississippi, North Carolina,
Tennessee, and Virginia.

study data typically available from each of the major federal government agencies.
In addition, local and state agencies often have study data available.
The types of information which may be derived from previous studies include

Surveyed cross section data

Data on channel structures such as bridges and culverts
Survey benchmarks used in previous field survey work
Computed or measured flow rates
Accounts of previous flooding, including high-water marks

An assessment of the usability and technical accuracy should be made of all available
information, including historical hydrologic data, high-water marks, flooding problems
within the community, flood control measures, hydraulic structures that affect flooding,
available community maps showing and naming all roads in the floodplains, topographic
maps, digital data files, and elevation control data (including consideration of land subsidence* where applicable). Photographs of past major floods, if available, should be obtained.
Within the United States, federal Flood Insurance Studies are often valuable sources
of information on floodplain hydraulics. It is always advisable to obtain the data from previous Flood Insurance Studies whenever it is available. In fact, if a study is being prepared
for submittal to the Federal Emergency Management Agency (FEMA) to replace or revise
an existing Flood Insurance Study, the data from the existing Flood Insurance Study must
be used as much as possible.
Flood Insurance Study information may include hydrologic and hydraulic models, engineering and construction plans, floodplain maps, and flood profiles. In addition, any information should be obtained that may provide data for evaluating changes to the effective
hydrologic or hydraulic models. Rood Insurance Study data may be requested from FEMA.
If the original hydraulic models and work maps used in the Flood Insurance Study are
not available, FEMA requires that the Flood Insurance Study results be recreated from the
information in the Flood Insurance Study report book and the Flood Insurance Rate Maps.
The data available from the published Flood Insurance Study report should be supplemented as needed and entered into the same hydrologic and/or hydraulic model used to
create the original Flood Insurance Study. This model should then be calibrated to the
Flood Insurance Study flood profiles to obtain 100-year water-surf ace elevations within
an 0.1-ft tolerance, if possible. FEMA requires this to ensure a logical transition between
revised and unrevised data.
19.1.1 Sources of Topographic Data
Topographic maps for the area should be obtained whenever available. Although local and
state governments sometimes compile detailed topographic maps, the most widely available topographic maps within the United States are the U.S. Geological Survey's (USGS)
7.5-minute or 15-minute quadrangle maps. USGS quadrangle maps may provide all the
following information:

*Land subsidence is the sinking or settling of land to a lower level in response to various factors, both natural and
of human origin, such as earth movements, lowering of fluid pressure (or lowering of ground water level), removal
of underlying supporting materials by mining or solution of solids, either artificially or from natural causes, compaction caused by wetting, oxidation of organic matter in soils, or added load on the land surface.

1. Survey control points. The locations of permanent vertical and horizontal control
points (monuments or benchmarks) are indicated on the maps.
2. Topography. Topography is prepared to national standards and the accuracy of each
map is keyed to the contour interval. The contour interval will vary depending on the
amount of relief at the individual locations.
3. Ground cover. The color coding indicates the general type of ground cover, which
may be useful in planning survey operations and in confirming estimates of roughness
coefficients.
4. Development features. The locations of road crossings, developments, and other features are indicated on the maps.
In the United States, the USGS is actively converting existing maps to digital form.
Planimetric information is represented using digital raster graphics (DRG), digital
orthophoto quads (DOQ), or digital line graphs (DLG).
A digital raster graphic is a carefully scanned image of a USGS topographic map,
including the borders. The map image is georeferenced to the surface of the Earth, so that
most geographic information system (GIS) software can automatically position the DRG
image correctly with respect to other types of geographic data. Figure 19.1 illustrates a
portion of a USGS DRG file.
A digital orthophoto quad is a digital image of an aerial photograph in which displacements caused by camera orientation and terrain have been removed. Orthophotos
combine the image characteristics of a photograph with the geometric qualities of a map.
The standard digital orthophoto produced by the USGS is a black-and-white or colorinfrared 1-m ground resolution quarter quadrangle (3.75-minute) image. Figure 19.2
illustrates a portion of a USGS DOQ File.
Digital line graphs are vector files containing line data, such as roads and streams, digitized from USGS topographic maps. DLGs offer a full range of attribute codes, are highly accurate, and are topologically structured, which makes them ideal for use in GIS.
A digital elevation model (DEM) is a digital file consisting of terrain elevations for
ground positions at regularly spaced horizontal intervals. DEMs are developed from
stereo models or digital contour line files derived from USGS topographic quadrangle

FIGURE 19.1 Example portion of USGS DRG file.

FIGURE 19.2 Example of USGS DOQ file.

maps. The USGS produces five different digital elevation products. The most useful for
floodplain studies consists of 7.5-minute X 7.5-minute blocks corresponding to the standard 1:24,000 scale USGS quadrangle maps. The data have a resolution (grid interval) of
30 m (approximately 100 ft). Figure 19.3 illustrates ground elevation contours computed
using the data from a USGS DEM file.

FIGURE 19.3 Example of elevation contours derived from USGS DEM file.

Canada and most Western European nations are undertaking similar programs for the
development of digital elevation data from existing topographic data. Digital data are also
being developed for other regions of the earth.
Digital terrain data have two major applications to floodplain hydraulics. First, if the
data are sufficiently detailed, they may be used as a basis for the channel and floodplain
cross sections for hydraulic studies. This is emerging as a common practice when digital
terrain models (DTMS) are available. However, the grid-based data available from a digital elevation model may not be suitable for use as channel cross section data, because the
spacing between adjacent points (at least about 30 m for most DEM data) is not dense
enough to provide a sufficiently detailed channel cross section. The use of DEM data for
extending field-surveyed cross sections into floodplain areas is more promising, because
less detail is generally needed in these areas.
Digital topographic data may also be used for floodplain mapping. The results of a
hydraulic model analysis only provides the floodplain boundaries at the locations where
cross sections are available. Between these cross sections, the floodplain must be mapped
using available topographic data. The availability of digital topographic data is dramatically improving the efficiency and accuracy of floodplain mapping, through links between
GIS software and hydraulic analysis software. Figure 19.4 illustrates a stream channel for
which two floodplains have been automatically calculated and displayed using software
that links GIS software with the U.S. Army Corps of Engineers Hydrologic Engineering
Center River Analysis System (HEC-RAS) water surface profiles computer program.

FIGURE 19.4 Example of floodplain computed and mapped using data from digital terrain
model (DTM).

FIGURE 19.5 Example of USGS National Aerial Photography Project photograph.

19.1.2 Aerial Photography
Aerial photography and topographic mapping may also be available from commercial aerial survey and mapping companies. Some local and state government agencies, such as
transportation departments or taxing authorities, may obtain a complete set of aerial photographs on a regular basis. Aerial photography may also be available through the
National Aerial Photography Project (NAPP), distributed through the USGS (Fig. 19.5).
Up-to-date aerial photographs may indicate recent channel improvements, road crossings and other special structures, channel condition, land uses in the floodplain, and other
information. It is important to note that areas and distances measured from aerial photographs may not be accurate unless the photograph has been orthorectified to correct for
the effects of the camera angle, the curvature of the earth, and the apparent displacement
of points resulting from differences in elevation at different locations in the photograph.
19.1.3

Highway or Street Maps

Highway maps from state or local highway agencies or street maps from private map companies may contain information which may be missing from the USGS quadrangle maps,
including recently constructed road crossings or other structures. Like topographic data and
aerial photographs, highway and street data are also becoming available in digital form.
19.1.4 Construction Drawings
Construction drawings may be available for channel modifications, bridges, or other projects affecting the stream channel or floodplain. Local or state transportation agencies and

FIGURE 19.6 Example of digital street map (Census Bureau TIGER File).

railroad companies generally retain the construction drawings for all bridges and many
other types of projects.
As-built construction drawings are those that have been revised to reflect the actual
dimensions of a completed construction project. As-built drawings for existing bridges are
normally available from local or state transportation agencies. Local or federal agencies
normally have as-built construction drawings for channel modification projects.
The use of information from as-built construction drawings can greatly reduce the
amount of survey information required for existing bridges or channel modifications.
However, the survey benchmark used for each set of construction drawings should be
determined, to be sure that all information is based on the same horizontal and vertical datum.
19.1.5 Stream Gage Data
Stream gage data from stream gages operated by the USGS or other agencies can provide
good estimates of the floodplain elevation at a particular point. Many floodplain analysis
computer programs can accept a channel rating curve constructed from the stream gage
data for use in computing the water surface elevation corresponding to a given flow rate.
Figure 19.7 illustrates a channel rating curve.
If the purpose of the analysis is to model a very extreme event, it is important to note
that the rating curve available for the stream gage location may not have any historical
basis for the stages at extremely high flow rates. These stages may simply be extrapolated from the historical record of lower flow rates.

Water Surface Elevation

Total Flow Rate

FIGURE 19.7 Channel rating curve.
19.1.6 Personal Observations
Personal observations by local residents, such as high-water marks, can be valuable in
estimating the floodplain elevation for calibration of the computed water surface profile.
Often the best observations are available from employees of city or county engineering
departments, or other agencies with responsibilities for public facilities such as drainage
channels, roads, or utility systems.

19.2
OBTAININGFIELDSURVEYDATAFOR
FLOODPLAIN STUDIES
Even though considerable information may be available from previous studies and other
sources, it is often necessary to perform field surveys to provide additional information.
Survey operations can be difficult to manage, but they often determine the success of a
floodplain study.
The following steps should be completed before field survey operations are initiated:
1. Channel stationing. The stream length should be accurately determined from the mouth
(or other beginning point). Normally, this is done using a recent aerial photograph or
other map which indicates the current channel morphology. Channel stationing should
begin with 0 + 00 and increase in the upstream direction. Stationing should be measured along the thalweg.*

*The thalweg is the line following the lowest part of a valley, whether under water or not. Usually this is the line
following the deepest part of the bed or channel of a river.

2. Locate structures. The stream stations of all pertinent stream crossings, tributaries, and
all other significant features should be summarized in table form, including those features found on only one map. The stream stations should be transferred to a topographic "work map" for the analysis.
3. Preliminary stream profile. If a stream profile showing the channel flow-line is not
already available from a previous analysis, it may be useful to prepare a preliminary
flowline profile from the contours on the topographic map.
4. Preliminary floodplain map. If a floodplain map is not already available from a previous analysis, the floodplain width at various points along the channel should be estimated. The estimate may be based on the readings of stream flow gages, preliminary
computations using data from the topographic work map, or observations by local residents. For major studies, it may even be advisable to set up a preliminary hydraulic
analysis using data from the topographic work map only.
5. Survey control points. The location and elevation of all known survey benchmarks in
the area of the study should be indicated on the topographic work map.
6. Site reconnaissance. The engineer should then visit the study area, using the work map
as a reference. Each road crossing and other structure should be inspected, photographed, and measured with a tape, if possible. Channel and floodplain conditions
should be noted. Special notes and photograph locations should be recorded on the
topographic work map.
7. Preliminary cross section locations. The proposed location and alignment of all surveyed cross sections should be indicated on the topographic work map.
Finally, the engineer should meet with the survey coordinator to discuss the data
requirements for the project, using the topographic work map, photographs, and the engineer's knowledge of the project area and computer program data requirements. Even after
the survey work begins, the engineer should maintain frequent contact with those performing the survey work to be aware of their progress and any special problems.
19.2.1 Vertical and Horizontal Control for Field Surveys
The Global Positioning System (GPS), when used in the differential mode, is currently
the best method for extending any survey control network unless satellite visibility is
obscured (e.g., because dense forest) or severe radiofrequency disturbances are present.
Differential GPS or third-order leveling* can be used to tie temporary benchmarks to an
established datum; to determine the elevation of high-water marks; and, where needed,
to establish horizontal and vertical control for aerial survey work. Whenever possible,
available benchmarks should be used instead of field surveys, to reduce the survey cost.
In particular, benchmarks shown on official floodplain maps should be used if possible.
As a general rule, there should be approximately two temporary benchmarks per mile
of stream length or four per square mile of floodplain, as appropriate (FEMA, 1995).
Most benchmarks established in North America are referenced to the National
Geodetic Vertical Datum of 1929 (NGVD29), which was formerly referred to as Mean
Sea Level of 1929. This datum was originally established as a vertical reference representing the average sea level as measured at a series of tide gages throughout North
America. As a result of additional work since 1929, a revised vertical reference has been
"Closures within 0.05 ft. X square root of distance in miles.

established which better represents the actual gravitation force of the earth as it exists in
various locations throughout the North American continent. This reference is called the
North American Vertical Datum of 1988 (NAVD88). FEMA and other federal government
agencies are adopting NAVD88 as the standard vertical datum for new studies. The significance of the conversion from NGVD29 to NAVD88 depends on the particular location. The National Geodetic Survey has produced a computer program called VERTCON
which will compute the conversion from NGVD29 to NAVD88 at any location in North
America, given the coordinates of latitude and longitude.
Field surveys should normally be accomplished by differential leveling or differential
GPS methods, with vertical error tolerances of 0.5 ft across the 100-year floodplain.
Horizontal control may be unnecessary since cross sections can be located simply by
visual reference to identifiable points on a map or aerial photograph. However, if crosssectional data will be combined with other data sets, it is necessary to establish and maintain adequate horizontal control.
19.2.2 Cross Section Locations
The survey information required for a floodplain analysis generally consists of cross
sections of the channel and floodplain along the stream. These cross sections may be
field-surveyed or taken from digital terrain models. Some water surface profile computer programs treat each cross-section as representing a reach of the river and use only
one section at the midpoint of the reach to calculate losses through the entire reach.*
However, the most commonly used computer programs use cross sections to define
breakpoints in the geometry, and properties of adjacent sections are averaged to calculate losses through the reach.
The objective of either computation scheme is to describe flow boundaries accurately
enough to predict energy losses due to friction and changes in flow velocity. If only a few
cross sections are available and they are located far apart, a greater amount of engineering judgment is required to satisfactorily analyze the problem. Any deviation from a
smooth profile must be explained, and in some cases it can be traced back to inadequate
cross-sectional data.
A water surface profile in nature is really a curvilinear surface that follows the general slope of the channel. When the water surface profile is computed, the shape of this
curvilinear surface is approximated by a series of straight-line segments. The endpoints of
these line segments are the channel cross section locations. If many channel cross sections
are included, the line segments will be short, and the true curvilinear shape of the natural
water surface profile can be better represented. If only a few channel cross sections are
used, the line segments will be long, and the true curvilinear shape of the natural water
surface profile will not be represented well. Figure 19.8 illustrates a water surface profile
that is well represented (using many channel cross sections) and the same profile that is
poorly represented (using only a few channel cross sections). As indicated, if an adequate
number of cross sections are not provided for the hydraulic analysis, the computed water
surface elevation can vary considerably from the actual value, especially in areas in which
there is a sudden change in the channel slope.
One study indicated that reach lengths should be limited to a maximum of 800 m
(about 0.5 mi) for wide floodplains and for slopes less than about 0.04 percent, 550 m
(about 1800 ft) for slopes equal to or less than 0.06 percent, and 350 m (about 1200 ft) for
*A reach is a length of a channel that is fairly uniform with respect to discharge, depth, area, and slope; more generally, any length of a river or drainage course.

Water Surface Underestimated Because

of Excessive Cross-Section Spacing

Water Surface Overestimated Because

of Excessive Cross-Section Spacing

Main Channel Distance

FIGURE 19.8 Example of profile plot.
slopes greater than 0.06 percent (Beasley, 1973). For flood insurance studies, maximum
spacing is generally 15Om (about 500 ft) for unimproved channels and 600 m (about 2000
ft) for improved, regular channels. However, the actual spacing requirements vary according to energy considerations. A preliminary computer run using cross sections from topographic maps can be very valuable in determining the required spacing of surveyed cross
sections. These cross section spacings are only guidelines, and should not be used as fixed
spacing requirements. Instead, cross sections should be placed at all of the following locations along the stream channel:
Changes in slope. Cross-sections are needed at distinct changes in bed slope.
Changes inflow area. Cross-sections are needed at points of contraction or expansion
of the channel.
Changes in flow rate. Cross-sections are needed in the main channel immediately
above and below a confluence and in the main stream. A tributary cross section may
also be obtained just above the confluence if the tributary is to be studied.
Changes in roughness. Additional cross-sections are needed at the upstream and
downstream ends of each channel segment that has a significantly different
Manning's roughness coefficient, such as changes in plant cover, channel improvements, and so on.
Control sections. Cross sections are usually required immediately above and below
control sections (such as bridges, drop structures, weirs, and so on) to adequately
model the changing conditions in these flow transitions. A weir is a low-overflow dam
or sill for measuring, diverting, or checking flow.
Encroachments. For encroachments into the floodplain (such as landfills, bridges, road
embankments, dams, or levees), cross sections are needed at the upstream and downstream ends of the encroachment, and at regular intervals within the reach of the
encroachment.

 Bends. Additional cross sections may be required in and around channel bends to properly represent the reach lengths in the chanwnel overbank areas.
If an elevated road or dam is found in the field that was not indicated on the cross section layout map, the survey field crew should insert additional cross sections to represent
these additional structures. However, this does not apply to roads that are not elevated and
will not block large amounts of flow.
After cross sections are placed at all of the locations identified above, then additional
cross sections may be placed between these as needed to reduce the spacing between adjacent cross sections to the values recommended above. Laying out cross sections in this
way, rather than at a fixed spacing, generally provides a superior analysis.
19.2.3 Cross Section Alignment and Orientation
Cross sections should tie into high ground, so that the maximum elevation of each end of
a cross section should be at least 1 ft higher than the anticipated maximum water surface
elevation, if possible. In addition, the cross section locations should be accessible and
practical. For example, pipeline and power line crossings make good locations for surveyed cross sections in wooded areas, because the right-of-way is usually cleared.
However, cross sections must represent the average ground profile. Roadside ditches,
washouts, gravel pits, raised areas, and other nonrepresentative conditions should be
avoided. The actual orientation and extent of the surveyed cross section should be indicated on work maps.
Most computer programs used for floodplain analysis require that cross sections
should be surveyed from left to right, with these directions determined while looking

FIGURE 19.9 Example of work map showing cross section locations for field survey.

downstream. The ground elevation as well as the distance from the channel flowline
should be determined and recorded for each point on the cross section.
Negative cross section stationing may be unacceptable to the computer programs
used for floodplain analysis. Therefore, it is advisable to begin cross section stationing
at 10 + OO or 100 + OO so that the cross section can be extended later without having
to restation the entire cross section.
Cross sections should be placed perpendicular to the direction of flow. Cross sections
can and should have angles or "doglegs" in them as needed so that the channel portion of
the cross section can be perpendicular to the flow in the channel while the floodplain portions of the cross section can be perpendicular to the flow in the floodplain. Figure 19.10
provides an example.
In the case of wide floodplains in very flat areas, where flows exceed the channel
capacity and spill into the floodplains, the effective cross section is not easily defined. One
way to solve this problem is to shape the cross section alignment so that the floodplain
portions curve upstream. This curvature will contain the water in a reasonable width since
the cross section elevations on the edge of the flow will be higher than in the case of the
straight cross section alignment.
A minimum of five points is usually required for the channel portion of a surveyed crosssection. This includes one point at the top of each channel bank, one point at the toe of each
side slope, and one point at the channel flowline, as illustrated in Fig. 19.11. Additional
points may be required when discontinuities in channel cross sections are encountered.
The number of cross section points required for floodplain areas is dependent on the
width of the cross section and on the character of the terrain in the floodplain. As a general rule, enough points should be shot to give a true representation of the floodplain terrain and to define any breaks in topography.
Valley cross sections that are intended to show the typical floodplain may be relocated in the field by the survey party to avoid heavy brush or to take advantage of power line
or pipeline easements, fence lines, or pasture lands. Any cross sections that are relocated
must be clearly marked on the work maps.
The channel station at which the cross-section intersects the channel center-line should
be determined for each cross section. The stream stations determined using aerial pho-

Reach Length (Left Overbank)

Reach Length (Channel)

Reach Length (Right Overbank)

FIGURE 19.10 Cross-sectional layout and reach lengths for meandering channel.

Left Top of Bank

Right Top of Bank

Channel Invert
Left Toe of Slope

Right Toe of Slope

FIGURE 19.11 Minimum survey points for channel only.

tographs or other maps for major structures such as bridges can be used as a reference in
determining the stream stations of individual cross sections between the structures.
Distance measurement to the nearest foot or meter is usually acceptable.
The measured distances between cross sections are defined with reach lengths. Three
reach lengths are required: left floodplain, right floodplain, and channel, which are measured from the current cross section to the previous cross section looking downstream.
Channel reach lengths are measured along the channel invert. The floodplain reach lengths
are measured along the anticipated flow path of the center of mass of flow in the floodplain
area. This is generally assumed to be 1/3 the distance from the channel bank to the edge of
the floodplain.
The channel and floodplain reach lengths will often equal each other if the reach contains a straight channel with parallel flood plains. Reach lengths will not equal each other
in channel bends. One floodplain reach length will be greater than the channel reach
length while the other floodplain reach length will be less than the channel reach length
(Fig. 19.12).
Reach lengths for the channel and flood plains may be unequal where the flood plains
are parallel and the channel meanders (Fig. 19.10). The reach lengths may vary depending
on the severity of the flood event. Therefore, a preliminary hydraulic analysis may be
required. If multiple profiles (different flow rates or storm frequencies) are analyzed using
the same reach lengths, the reach lengths for all the profiles should be determined by the
most important profile, which is usually the 100-year profile.
If the reach lengths of floodplains and channel differ, the most commonly used stream
analysis computer programs calculate one effective reach length that is used in the
Standard Step Method. This is a discharge-weighted reach length based on the discharges
in the main channel and the two floodplains over the whole reach.

Reach Length (Left Overbank)

Reach Length (Channel)
Reach Length (Right Overbank)

FIGURE 19.12 Cross-sectional layout and reach lengths for channel

bends.

_
(QLOB) (XLOB) + (QCH) (XLCH) + (QROB) (XLOBR)
L=
Q

(19.1)

where L = the weighted average reach length QLOB = the total flow in the left floodplain XLOB = the specified reach length for the left floodplain QCH = the total flow in
the channel XLCH = the specified reach length for the channel QROB = the total flow in
the right floodplain XLOBR = the specified reach length for the right floodplain Q = the
total flow in the entire cross section.

19.2.4

Use of Aerial Topography and Contour Map Data

Field survey costs for a flood plain study may be reduced in several ways. For example,
the spacing of cross sections along the stream channel could be increased so as to reduce
the number of cross sections required. As an alternative, the width of each cross section
could be reduced.
Studies have indicated that one of the best ways to improve the accuracy .of a computed water surface profile is to provide more cross sections (HEC, 1986). This implies
that one of the worst ways to save money on a floodplain study is to reduce the number
of cross sections. In general, it is better to reduce the number of survey points within each
cross section rather than to eliminate some cross sections entirely and thus extend the
reach lengths between adjacent cross sections.
Several methods may be used to maintain the required density of cross sections along
the stream channel. For example, surveyed cross sections could be used more than once
in the analysis. These repeated cross sections may be acceptable if field observations indicate that the channel and floodplain conditions are highly uniform through the stream
reach represented by the repeated cross sections. Intermediate cross sections may also be
synthesized (interpolated) from field-survey cross sections. Interpolated cross sections are
described later in this chapter. During the data collection phase of the project, however,
the most common method of reducing the cost of obtaining cross section data is to obtain
all or part of the data points for the cross-section from aerial topography or contour maps.
In some circumstances, aerial topography or contour maps can be used for the entire
cross section, including the channel portion. Usually, however, the accuracy and level of
detail available from aerial topography or contour maps are not sufficient for channel data.
It is preferable to obtain at least the channel portion of the cross section by field survey
methods, if at all possible. Sometimes, with little or no increase in field survey costs, some
additional ground shots can also be made in the floodplain, using the same setup of the
survey instrument already used for the channel survey. FEMA generally recommends the
use of aerial topography or contour maps for obtaining cross sections for the portion of
the cross section between the 100- and 500-year floodplain boundaries (FEMA, 1995).
Aerial topography is very cost-effective if the size of the project justifies the fixed
costs of mobilizing the aircraft and other related activities. Usually, the mapping data
obtained by aerial photography are useful for many purposes other than the floodplain
analysis, and therefore the full cost may be shared with other projects, departments, or
even other organizational entities. Newer technologies such as LIDAR (Light Imaging
raDAR) may also be very cost-effective,
If a contour map meets commonly accepted accuracy standards, then the contour accuracy is slightly better than half of the contour interval while the accuracy of spot elevations is approximately 20 to 30 percent of the contour interval of the map being prepared.
Therefore, cross sections developed using contours from topographic maps will not be as

accurate as cross sections prepared from spot elevations. Whenever possible, spot elevations along the cross section should be determined directly using the source data set, rather
than indirectly using contours.
19.2.5

Road Crossing Data

Necessary dimensions and elevations of all hydraulic structures and underwater sections
along the streams should be obtained from available sources or by field survey where necessary. Dimensions and elevations of hydraulic structures should not be established by
aerial survey methods.
Most computer programs used for floodplain analysis require four user-defined cross
sections in the computations of energy losses due to the structure. These cross sections
may be identified by number (1 through 4), beginning downstream of the bridge. A plan
view of the basic cross section layout is shown in Fig. 19.13.
Cross section 1 is located sufficiently downstream from the structure so that the
flow is not affected by the structure (i.e., the flow has fully expanded from the narrow
bridge opening). This distance should generally be determined by field investigation
during high flows. If field investigation is not possible, there are two sets of criteria for
locating the downstream section. The USGS has established a criterion for locating
cross section 1 a distance downstream from the bridge equal to one times the bridge
opening width (the distance between points B and C on Fig. 19.13). Traditionally, the
Corps of Engineers criterion has been to locate the downstream cross section about
four times the average length of the side constriction caused by the structure abutments

FIGURE 19.13 Cross section locations at a bridge or culvert.

(the average of the distance from A to B and C to D on Fig. 19.13). The expansion distance will vary depending on the degree of constriction, the shape of the constriction,
the magnitude of the flow, and the velocity of the flow. (A constriction is a local
obstruction narrowing a waterway.) Both criteria should be used as rough guidance for
placing cross section 1. Cross sections 1 and 2 should be close enough to one another
so that friction losses can be adequately modeled. If the expansion reach requires a
long distance, intermediate cross sections should be placed within the expansion reach
to adequately model friction losses.
Cross section 2 is located immediately downstream from the bridge (within a few feet).
This cross section should represent the effective flow area of the natural channel just
below the roadway crossing. It should not include the road fills, road ditches, mounds, or
depressions that are not typical of the floodplain.
Cross section 3 should be located just upstream from the bridge. The distance
between cross section 3 and the bridge should be relatively short. This distance should
only reflect the length required for the abrupt acceleration and contraction of the flow
that occurs in the immediate area of the opening. Cross section 3 should represent the
effective flow area of the natural channel just above the roadway crossing. Like cross
section 2, it should not include the road fills, road ditches, mounds, or depressions that
are not typical of the floodplain.
Cross section 4 is an upstream cross section where the flow lines are approximately parallel and the cross section is fully effective. The USGS recommends that the distance between cross section 3 and 4 should be roughly the same as the average width
of the bridge opening. However, this criteria for locating the upstream cross section
may result in too short a reach length for situations where the width of the bridge opening is very small in comparison to the floodplain. The Corps of Engineers generally
recommends that cross-section 4 be located a distance upstream equal to the average
contraction width (the average of the distance from A to B and C to D on Fig. 19.13).
According to the Corps of Engineers, the distance between cross sections 3 and 4
should be less than the distance between cross sections 1 and 2 for constricted flow
conditions, because flow contractions can occur over a shorter distance than flow
expansions. Both of these recommendations should be considered; the Corps of
Engineers recommendation is probably better when the bridge opening is narrow compared with the width of the floodplain; the USGS recommendation is probably acceptable for all other cases.
19.2.6

Using Repeated Cross Sections for Roadway Crossings

The engineer directing the floodplain analysis should determine, during a preliminary
visit to the study area, if the channel is relatively uniform in the area of each road crossing structure. If the channel is highly uniform in the area of a particular road crossing, it
may be possible to survey a single channel and floodplain cross section at the upstream or
downstream face of the bridge, and repeat this cross section for the other three or more
cross sections in the bridge model.
If the channel is relatively uniform through the bridge itself, but changes upstream or
downstream of the structure, a single field surveyed cross section may be used to represent the channel at both faces of the bridge, but different field-surveyed cross sections
may be needed at other locations through the bridge model.
If the channel is not uniform through the bridge, separate cross sections are required
at the downstream face of the bridge and at the upstream face of the bridge. It may even
be necessary to survey a cross section of the channel at the centerline of the bridge

(underneath the bridge superstructure), where special conditions cause the channel
underneath the bridge to differ significantly from the cross section at the upstream or
downstream face.
19.2.7 Obtaining Bridge Survey Data
Plan and profile sketches of the bridge on which all pertinent data are illustrated and identified should always be included in the survey notes. Figure 19.14 is a representative
sketch of the type of data that will be required by the hydraulic engineer.
Several pieces of information about the actual bridge itself are required. Two of the
most important of these are top of road and low chord elevations.
The top of road is the highest part of the roadway that forms a significant obstruction
to flow over the roadway. The top of road can be the crest of the bridge roadway, the top
of a sidewalk, or the crest of a solid bridge railing. When taking a cross-section on a multilane highway, the cross section should be obtained along the higher lane. If the median or
shoulder is higher than the centerline, the high points should be located by side shots on
the median or shoulder. Cross sections on railroads should be obtained on top of the rail.
If the track is superelevated, the highest rail should be shot.
The low chord (or low steel) of a bridge refers to the lowest part of the bridge that
forms a significant obstruction to flow under the bridge. The low chord of most bridges is
formed by the bottom of the bridge girders.
Whenever any doubt exists as to the identification of the top of road and low chord of
a bridge, elevations defining the shapes of all structures which form obstructions to flow
should be determined and recorded. For example, if a water main or other pipe is suspended from the bridge, this condition should be recorded.

Downstream Face of Bridge

= Channel Station 101+81
Bridge Width = 30 feet
Bridge Length = 120 feet

(Bridge is Perpendicular
to Channel)
NoteThat Roadway
Embankments Are On
Fill Near the Bridge.

(No Curbs or Guard Rails.)

Depth of Bridge Deck
18-inch
Diam.
Circular
Piers
DISTANCE (feet)
FIGURE 19.14 Example of sketch of bridge cross section.

Natural ground elevations should be used instead of top of road elevations when the roadway is in a cut; that is, when natural ground is higher than the top of the roadway. In such
cases, it is necessary to visualize the total obstruction to flood flow at the road. This will be
a combination of the roadway and natural ground. Figure 19.15 illustrates this situation.
Bridge railings or curbs should sometimes be considered when defining the top of
roadway. If a railing or curb forms a substantial obstruction to flow over the bridge, the
top of the rail or curb should be considered as the effective top-of-road. Figures 19.15 and
19.16 illustrate bridge decks with solid and open rails.
Other important bridge data include the following:
Channel station: the channel station at which the bridge center-line (or one of the
bridge faces) intersects the channel flowline.

Effective Top of Road

Natural Ground

Actual Top of Roadway

Actual Top of Roadway

Low
^hord

FIGURE 19.15 Defining the bridge deck for roadways in open cuts.

Effective Top of Road

Solid Railing

,-Actual Top of Roadway

Low Chord

FIGURE 19.16 Defining the bridge deck for bridges with solid rails.

Open Railing

Effective Top of Road

Top of Curb
Actual Top of Roadway
Low Chord

FIGURE 19.17 Defining the bridge deck for bridges with open rails.

 Bridge dimensions: the length and width of the bridge.

Skew angle: the angle between the bridge roadway centerline and the flowline of the
channel.
Piers: the size, shape, number, and location of piers supporting the bridge.
19.2.8 Culvert Data
Data requirements for culvert crossings are very similar to those for bridges. Channel
cross section locations for culvert crossings are the same as those for single bridges. The
top of road of a culvert crossing is determined in the same way. For each culvert, the culvert size and shape, length, the downstream and upstream flowline elevations and centerline stations, and the type of headwall and wing walls, and their angle with the main axis
of the culvert should be recorded. If a culvert contains an accumulation of silt, the depth
of silt at the upstream and downstream ends, and the consistency of the accumulated silt
(compacted, loose, and so forth) should also be noted.
19.2.9 Channel Structures
Channel structures such as weirs, drop structures, and sections of slope paving can have
very significant effects on channel hydraulics. The locations, sizes, and configurations of
these structures should therefore be carefully measured and recorded.
If a channel section is fully or partially lined with concrete or some other type of
slope paving, the survey information should record the type of paving (concrete lining,
riprap, and so on), the beginning and ending channel station for the slope paving, and
the vertical extent of the slope paving (that is, how far up the sides of the channel the
paving extends).

19.3 SELECTINGTHEBESTAPPROACHFORA
FLOODPLAINSTUDY
There are three alternative approaches that are generally available for the computation of
water surface elevations in a stream channel system:
One-dimensional steady, gradually varied flow conditions;
Two-dimensional, steady, gradually varied flow conditions;
One-dimensional unsteady flow conditions.
It is also possible to model two-dimensional unsteady-flow conditions, although this
approach is not common. The following sections describe these various approaches.
19.3.1 One-Dimensional and Two-Dimensional Flows
In a constructed channel with a regular cross section, most of the lines of flow will be
roughly parallel to the channel's longitudinal axis (the axis that follows the alignment of
the channel from upstream to downstream). This is one-dimensional flow. However, in
natural channels with irregular cross-sections, and especially when flow overtops the

channel and enters the floodplain, there may be significant components of flow in other
directions at certain locations. For example, flow may travel in a vertical direction for a
short distance to overtop a roadway surface and spill into a lower area on the downstream
side. Flow may also travel in a lateral direction from the floodplain into the channel to
pass through a narrow culvert opening.
Energy is required to set water in motion in any direction and overcome friction losses. The computer programs used for floodplain studies account for friction losses and
other changes in energy potential. In fact, the computation of water surface elevation is
really a by-product of the computation of the total energy level (often called the "energy
grade line") at each location in the stream channel and floodplain. For those situations
where energy losses are not the primary consideration, such as at a weir or a hydraulic
jump, the programs use alternate solutions. (A hydraulic jump is a sudden transition from
supercritical flow to the complementary subcritical flow, conserving momentum and dissipating energy.)
One-dimensional computer programs used for floodplain studies account for energy
losses in the downstream direction only (longitudinal flows). Two-dimensional computer
programs consider lateral flows as well as longitudinal flows. Therefore, they analyze
flow in both dimensions of the horizontal plane.
A one-dimensional analysis is adequate for most purposes. Two-dimensional computer models are useful at complex roadway crossings, in shallow flooding areas, and whenever there is significant flow in more than one direction.
19.3.2 Changes in Flow Depth With Respect to Time and Distance
Steady flow occurs whenever the depth of flow at a cross section, for a given discharge, is
constant with respect to time. Unsteady flow occurs whenever the depth of flow at a crosssection varies with respect to time, for a given discharge.
During a flood event, the flow depth in a stream channel increases from base flow conditions, reaches a peak value, and declines to the base flow again, generally within the
span of a few hours or less. However, at any one moment, an observer standing on the
edge of the floodplain would see what appears to be a steady flow condition.
Steady flow can further be classified as uniform or nonuniform. Uniform flow occurs
when the depth of flow and quantity of water are constant at every section of the channel
under consideration. Under these conditions, the water surface and flowlines will be parallel to the stream bed and a hydrostatic pressure condition will exist (the pressure at a
given section will vary linearly with depth). Uniform flow conditions are rarely attained
in natural channels.
Gradually varied flow occurs whenever the depth of water changes slowly over the
length of the channel. Under this condition, the streamlines of flow are practically parallel. Therefore, uniform flow principles can be used to analyze the flow conditions, even
though the flow is nonuniform. With rapidly varied flow conditions, there is a pronounced
curvature of the flow streamlines. The assumption of hydrostatic pressure distribution is
not valid. Therefore, uniform-flow principles are not adequate for the analysis of rapidly
varied flow.
Most computer programs used for floodplain analysis are based on the analysis of
steady, gradually varied flow conditions. Some programs have special capabilities to analyze rapidly varied flow conditions, such as hydraulic jumps.
A floodplain analysis that assumes steady flow is a "snapshot" of the channel system, usually at peak flow conditions. However, some channel flow situations are dominated by rapidly changing flow rates or other dynamic conditions. For example, a

floodplain area may have substantial floodplain storage capacity that affects the flow
rate in the stream channel; streams may experience a reversal of flow under certain
conditions; the stream may be subject to sudden changes in flow rate due to the opening or closing of gates or other similar structures; streams may discharge into marine
bodies that are affected by tidal movements; or the stream may be a part of a complex
system of pipes, channels, ponds, and reservoirs. All of these situations may produce
sudden changes in flow rate which are not adequately represented by the assumption
of steady flow. The capability to analyze unsteady flow conditions is sometimes
required.
19.3.3

Critical Flow and Critical Depth

Critical depth is an important hydraulic parameter because it is always a hydraulic control. (Critical depth occurs when the specific energy is a minimum for a given discharge.)
Hydraulic controls are points along the channel where the water level or depth of flow is
limited to a predetermined level or can be computed directly from the quantity of flow.
When the depth of flow is greater than critical depth, the velocity of flow is less than critical velocity for a given discharge and hence, the flow regime is subcritical. Conversely,
when the depth of flow is less than critical depth, the flow regime is supercritical. Flow
must pass through critical depth in going from a subcritical flow regime to a supercritical
flow regime. Typical locations of critical depth are at
(1). abrupt changes in channel slope when a flat (subcritical) slope is sharply increase
to a steep (supercritical) slope, (2). a channel constriction such as a culvert entrance
under some conditions, (3). the unsubmerged outlet of a culvert on subcritical slope,
discharging into a widechannel or with a free fall at the outlet, and (4). the crest of an
overflow dam or weir.
Critical depth for a given channel is dependent on the channel geometry and discharge
only, and is independent of channel slope and roughness.
19.3.4 Types of Stream Systems
Stream systems may be divided into three main categories according to the complexity of
the interconnections among the various flow paths within the stream system:
Simple channels. These are single-stream reaches with no tributary streams. Therefore,
there are no confluences or junctions. The flow rates may vary from one cross section
to the next along the length of the stream channel, because of local inflows or stream
channel losses. The direction of flow in the channel is known. However, the flow
regime within the channel is not known, and in fact, may vary along the length of the
channel (changing from subcritical to supercritical or from supercritical to subcritical).(Fig. 19.18).
Dendritic channel systems. These include tributary streams that combine at confluences or junctions to form major streams. This process may be repeated through several steps. Streams within a dendritic channel system are sometimes identified by their
"order." First-order streams are the smallest tributaries, while higher-order streams are
larger streams and rivers. In a dendritic system, each of the individual streams is analyzed as a simple channel (as described in the category above) (Fig. 19.19).
Network channel systems. These include all of the features of dendritic channel system. In addition, full network systems (sometimes called fully-looped systems) can

FIGURE 19.18 Example of simple channel system.

have flow splits, where flow in a single channel (or a group of channels) can be divided among two or more flow paths which may either recombine downstream or discharge at completely different destinations. In some network channel systems, the
direction of flow may be known in advance for each stream; for others, the direction
of flow may be determined during the analysis and may even change during the analysis. A full network model can also include storage areas that can either provide water
to, or divert water from, a channel. A full network model is required to properly deal
with storage areas, because they represent flow splits (since water can be diverted into
the storage area). See Fig. 19.20.

Spruce Creek
Upper Reach

Pottsville

Lower Reach

FIGURE 19.19. Example of dendritic channel system.

19.3.5 Computer Programs Widely Used in Floodplain Analysis

Flood elevations for riverine areas are normally determined by step-backwater* computer
models, including HEC-RAS, HEC-2, WSPRO, and WSP-2.
HEC-RAS River Analysis System [developed by the U.S. Army Corps of Engineers
(USAGE) Hydrologic Engineering Center] computes water surface profiles for onedimensional steady, gradually-varied flow in rivers of any cross section (HEC, 1998a-c).
HEC-RAS can simulate flow through a single channel, a dendritic system of channels, or
a full network of open channels (sometimes called a fully looped system). However, HECRAS cannot currently determine how much flow follows each flow path at a flow split. In
addition, the direction of flow in each stream must be provided to HEC-RAS as part of
the input data.
HEC-RAS can easily model sub- or supercritical flow, or a mixture of each within the
same analysis. A graphical user interface provides input data entry, data modifications,
and plots of stream cross sections, profiles, and other data. Program options include floodway computations, inserting trapezoidal excavations on cross sections, and analyzing the
potential for bridge scour. The water surface profile through structures such as bridges,
culverts, weirs, and gates can be computed. The program includes sophisticated routines

Upper Spruce
Tusseyville

Bear Run

Middle Spruce

Coburrr
Lower Spruce
FIGURE 19.20 Example of network channel system.

*Backwater is an unnaturally high stage in stream caused by obstruction or confirnement of flow, as by a dam, a
bridge, or a levee. Its measure is the excess of unnatural over natural stage, not the difference in stage upstream and
downstream from its cause.

for the analysis of roadways with multiple bridges and culverts. Variable channel roughness and variable reach length between adjacent cross sections can be accommodated.
HEC-2 (developed by the USAGE Hydrologic Engineering Center) was the predecessor
to HEC-RAS, and has some of the same capabilities (HEC, 199Oa). However, HEC-2 does
not accommodate multiple bridge or culvert openings in most situations, and is restricted to
either subcritical or supercritical flow computations in a single analysis. Because HEC-2
was widely used for almost 30 years, many existing floodplain studies were performed using
HEC-2. HEC-RAS will read HEC-2 input data files and produce very similar results under
most situations, but HEC-2 may still continue to be used for some time because of the need
to match existing study results. HEC-2 is generally restricted to simple channel systems, but
may be used for dendritic channel systems under limited circumstances.
WSPRO Water Surface Profile [developed by the U.S. Geological Survey for the
Federal Highway Administration (FHWA)] is a step backwater program for natural channels with an orientation to bridge constrictions (USGS, 1990). The water surface profile
computation model has been designed to provide a water surface profile for six major
types of open-channel flow situations: unconstricted flow, single opening bridge, bridge
opening(s) with spur dikes, single opening embankment overflow, multiple alternatives
for a single job, and multiple openings. WSPRO was developed for use primarily around
roadway crossings. The bridge analysis routines originally used in WSPRO have now
been incorporated into the HEC-RAS computer program.
WSP2 Water Surface Profile-2 (developed by the USDA Natural Resources
Conservation Service) is a step backwater program for natural channels [Natural Resources
Conservation Service (NRCS, 1993)]. WSP2 estimates head loss at restrictive sections,
including roadways with either bridge openings or culverts.
19.3.6 Two-Dimensional Water-Surface Computer Models
The most commonly used two-dimensional models include TABS-MD (developed by
ACE Waterways Experiment Station) and FESWMS-2DH Two-Dimensional Flow in a
Horizontal Plane computer program (developed by USGS).The TABS-MD
(Multidimensional) Numerical Modeling System is a collection of generalized computer
programs and utility codes, designed for studying multidimensional hydrodynamics in
rivers, reservoirs, bays, and estuaries. The primary computational program in the TABSMD system is the RMA-2 model [Waterways Experiment Station (WES) Hydraulic
Laboratory, 1997]. The TABS-MD system can be used to study project impacts on flows,
sedimentation, constituent transport, and salinity. FESWMS-2DH Finite Element Surface
Water Modeling System-Two-Dimensional Flow in a Horizontal Plane simulates steady
and unsteady flow and is useful for simulating two-dimensional flow at width constrictions and highway crossings of rivers and floodplains (FHWA, 1989). FESWMS-2DH is
based on the momentum balance approach, with consideration of shear stresses due to
friction between the moving water and the fixed boundaries.
19.3.7 One-Dimensional Unsteady-Flow Models
There are several computer programs available for simulating one-dimensional unsteadyflow conditions in stream channels. These include UNET, Storm Water Management
Model (SWMM), DAMBRK, and NETWORK.
UNET, One-Dimensional Unsteady Flow Through a Full Network of Open
Channels (supported and maintained by USAGE HEC is a one-dimensional unsteady
flow model that can simulate flow through a single channel, a dendritic system of channels, or a full network of open channels (HEC, 199Ob). In UNET, storage areas can be the

upstream or downstream boundaries for a river reach. In addition, the river can overflow
laterally into the storage areas over a gated spillway, weir, levee, through a culvert, or a
pumped diversion.
In addition to solving the one-dimensional unsteady flow equations in a network system, UNET provides the user with the ability to apply several external and internal
boundary conditions, including flow and stage hydrographs, gated and uncontrolled spillways, bridges, culverts, and levee systems. UNET can read channel cross sections that are
input in a modified HEC-2 format.
Future versions of the HEC-RAS computer program are expected to incorporate most
of the functions of the UNET computer program, and will likely provide access to these
functions through a user interface that is more convenient to use than the original UNET
user interface, which relies heavily on the HEC-DSS (Data Storage System) for data entry
and output.
Storm Water Management Model (SWMM) (developed by the U.S. Environmental
Protection Agency) is a comprehensive water quantity and quality simulation model
developed primarily for urban areas. SWMM is a complete hydrology and hydraulics
model that is most commonly applied to urban areas with closed storm sewer systems
(Huber and Dickinson, 1988). The model uses different modules (or "blocks") for different types of computations. SWMM is not widely used for floodplain studies on natural
stream channels. However, the Extran and Transport blocks of the SWMM program can
use natural channel cross section data in the same form as required by the HEC-2 computer program. The Transport block provides simple hydraulic computations, but the more
sophisticated Extran block can perform a dynamic backwater analysis (Roesner et al.,
1988). This can be useful when a sophisticated hydraulic analysis is required, particularly when the open channel is a part of an urban drainage network.
Dam Break Flood Forecasting Model (DAMBRK) (developed by the National
Weather Service) is an unsteady flow dynamic routing model which develops an outflow
discharge hydrograph due to spillway and/or dam failure flows (Fread, 1982). This hydrograph is routed through the downstream river valley. This can be a useful method of floodplain analysis for such events.
NETWORK: Enhanced Dynamic Wave Model (developed by the National Weather
Service) is an unsteady flow dynamic routing model for a single channel or network (dendritic and/or bifurcated) of channels for free surface or pressurized flow (Fread & Lewis,
1986, Fread & Lewis, 1988 and Fread, 1993).
19.3.8 Selecting a Computer Program for a
Floodplain Analysis
A computer program that is suitable for one-dimensional steady, gradually varied flow
conditions is adequate for most floodplain studies. The most widely used programs in this
category are the Corps of Engineers HEC-RAS and the older HEC-2 computer programs.
If conditions dictate, a two-dimensional analysis or a dynamic analysis (or, in some cases,
a two-dimensional dynamic analysis) may be required.
If a floodplain study is being done to revise a FEMA flood insurance study, FEMA normally requires that revised study be completed using the same computer program used for
the original study. However, a different computer program may be accepted if documentation can be submitted justifying why a change would be an improvement. It is important to discuss such issues with FEMA in advance (FEMA, 1995).

Next Page

CYLINDRICALQUADRANT

STRAIGHT LINE

WARPED
ABRUPT

WEDGE
FIGURE 20.1 Transition types. (From Corry et al., 1975)
The design procedure for abrupt expansions is as follows (Corry et al., 1975):
1. Determine V0 and ^0 at the culvert outlet using Fig. 20.4 for box culverts or Fig. 20.5
for circular culverts.
o
2. Compute Froude number, Fr = ^-
o/o
3. Determine optimum flare angle (9) using tan 0 = Fr /3 (Blaisdell and Donnelly, 1949).
If the selected wing wall flare (0W) is greater than 0, consider reducing 0W to 6.
4. Use Fig. 20.2 (for box culverts) or Fig. 20.3 (for pipe culverts) to determine the downstream average depth (yA) knowing Fr and the desired distance L (expressed in multiples of D, diameter).

CHAPTER 20
FLOW TRANSITIONS AND
ENERGY DISSIPATORS
FOR CULVERTS

AND

CHANNELS

Larry W. Mays
Department of Civil and Environmental Engineering
Arizona State University
Tempe, Arizona

20.1

FLOWTRANSITIONSFORCULVERTS

Flow transitions are changes in the cross section of an open channel over short distances.
They are designed to have a minimum amount of flow disturbance. Figure 20.1 illustrates
the various types of transitions; the two most common ones are the abrupt (headwall) and
the straight line (wingwall).
Highway culverts typically are designed to operate with an upstream headwater pool
that dissipates the of the channel approach velocity. This type of situation does not require
an approach flow transition. Outlet transitions (expansions) should be considered in the
design of all culverts, channel protection, and energy dissipators. Transition design can be
categorized as
culverts with outlet control (subcritical flow) and
culverts with inlet control (supercritical flow).
20.1.1 Culverts with Outlet Control
Culverts with outlet control can have abrupt expansions or gradual transitions. In an
abrupt expansion, the water surface plunges or drops rapidly and the flow spreads out.
The potential energy stored as depth is converted to kinetic energy with a higher velocity of flow. The transition (apron) end velocity can be determined using the experimental
results of Watts (1968) (Figs. 20.2 and 20.3). Figure 20.2 relates the average depth brink
depth ratio (yA/y0) for a rectangular outlet to the Froude number. Figure 20.3 is a similar
relationship for pipe culverts. These curves in Figs. 20.2 and 20.3 were developed for
Fr's ranging from 1.0 to 2.5, the applicable range for most abrupt outlet transitions.
Usually, a low tailwater is encountered at culvert outlets and the flow is supercritical on
the outlet apron.

CYLINDRICALQUADRANT

STRAIGHT LINE

WARPED
ABRUPT

WEDGE
FIGURE 20.1 Transition types. (From Corry et al., 1975)
The design procedure for abrupt expansions is as follows (Corry et al., 1975):
1. Determine V0 and ^0 at the culvert outlet using Fig. 20.4 for box culverts or Fig. 20.5
for circular culverts.
o
2. Compute Froude number, Fr = ^-
o/o
3. Determine optimum flare angle (9) using tan 0 = Fr /3 (Blaisdell and Donnelly, 1949).
If the selected wing wall flare (0W) is greater than 0, consider reducing 0W to 6.
4. Use Fig. 20.2 (for box culverts) or Fig. 20.3 (for pipe culverts) to determine the downstream average depth (yA) knowing Fr and the desired distance L (expressed in multiples of D, diameter).

B = WIDTH OF CULVERT
= (D) FOR PIPE
L = DISTANCE TO DESIRED DEPT.H
MEASURED IN MULTIPLES OF (B)
Y0 = BRINKDEPTH
Y A = AVERAGE DEPTH OF FLOW

FIGURE 20.2 Average depth for abrupt exapnsion below rectangular culvert outlet. (From Cony et al., 1975)

D = DIAMETER OF CULVERT
L = DISTANCE TO DESIRED DEPTH
MEASURED IN MULTIPLES OF (D)
Y0 = BRINKDEPTH
Y A = AVERAGE DEPTH OF FLOW

FIGURE 20.3 Average depth for abrupt expansion below circular culvert outlet. (From Corry et al., 1975).

Y0 = brink depth
D = height of box culvert
B = width of barrel
Q = discharge
TW = tail water depth

TW/D

FIGURE 20.4 Dimensionless rating curves for the outlets of rectangulars culverts on horizontal and mild slopes (From Simons et al; 1970)

Y0 = brink depth
D = dia. of culvert
TW = tail water depth

TW/D

FIGURE 20.5 Dimensionless rating curve for the outlets of circular culverts on horizontal and
mild slopes (From Simons, et al., 1970)

5. Determine the average velocity (VA) for box culverts using

-^- = 1.65 - 0.3Fr
M)
and for pipe culverts, (for L ^ 3D) using
^
=1.65-0.45 f-j
v
(\W)
6. Compute downstream width W2 = W0 + 2L tan 9, where tanG = Fr/3. If 0W > ft then
use 6W to compute W2.
7. If 0 is used to compute W2, then compute the downstream depth y2 using W2 and VA.
Because the flow prism is laterally confined, y2 will be larger than VA.
If Bw is used, y2 = yA, and the average flow width is WA = Q/VAyA. If WA < W2, use W2
to compute y2 = Q/VAW2.
Example 1 (adapted from Cony et al., 1975). Determine the width of an abrupt expansion and the hydraulic condition (y and V) at the end of an abrupt expansion for a 5 ft by
5 ft, 200= ft-long reinforced concrete culvert on a 0.2 percent slope (S0 = 0.002 ft/ft). The
discharge is Q = 270 cfs and a wingwall flare (0W = 45) with a 10 = ft apron is
considered.
Given

C ftAand
A T^W ^O
d,c = 5

Solution.
-"

&-&-*<.-<@ti-4ffi-">'
TW
and *** O. From Fig. 20.4, V 0 / D = 0.68; therefore, y0 and V0 can be deter-

mined to be y0 = 0.68(5) = 3 .40 ft and V0 = 270 /[3.4 (5)] = 15.88 ft/s.

Step 2. The outlet Fr = V0 /V^0" = 15.88/V32.2 (3.4) = 1.52
Step 3. The optimum flare angle is determined using tan 0 = 1^ Fr = i (1.52)
= 0.507, so 0 = 26.9.
Step 4. With the apron length/width = 10/5 = 2 and L = 2.OB, use Fig. 20.2 to compute
the average depth where ^/V0 = 0.26 and yA = 0.26 (3.4) = 0.88 ft.
Step 5. Compute the average velocity, VA/V0 = 1.65 - 0.3 Fr = 1.65 - .3 (1.52)
= 1.19, so = V4 = 1.19 V0 = 1.19 (15.88) = 18.90 ft/s
Step 6. Determine the downstream width W2 = W0 + 2L tan 0W using 0 = 0W = 45
because 0W > 0 and W0 = 5 ft, then W2 = 5 + 2 (10) (tan 45) = 25 ft.
Step 7. 0W is used above, so y2 = yA = 0.88 ft and W2 = QI(VA yA) = 16.1 < 25 ft.
When subcritical flow is maintained throughout a culvert, gradual transitions can be
used. Referring to Fig. 20.6, upstream of section 1 where some backwater exists because
of the culvert, flow is transitional from a channel into the culvert and out. According to the
Federal Highway Administration (FHWA, 1978), a flare angle of 17.5 (4.5 to 1) or flatter
provides a gradually varied transition that can be analyzed using the energy equation.

PLAN

HYDRAULICGRADELINE

WATERSURFACE

ELEVATION
ATUM
FIGURE 20.6 Definition sketch. (From Cony et al., 1975)

Referring again to Fig. 20.6, the headless in the contraction HLC is

if
HLC

(V2 V 2^
= Cc
C \ 2- L
(2g
2gJ

(20 1 ^
(ml)

(V2
V2}
= CC \^- $- L
'(2g
2g)

(20 2)
2s)
(2

and the headloss in the expansion is

H
HLE

where transition loss coefficients are listed in Table (20.1) for the transition types
(USAGE, 1970).
The design procedure for gradual transitions is as follows (Corry et al 1975):
1. Use Manning's equation to compute y4 = yn and V4, knowing Q, S0, n and the outlet
geometry.
2. Compute the critical depth using yc = ^1 (QIW)m. For box culvert, kv = 0.315 where
Q is in cfs and W is the box culvert width in ft. In SI units ^1 = 0.467 with Q in mVs
and W in m. Compare yn and yc to ensure subcritical flow.
3. For the chosen transition type, use Table 20.1 to obtain the transition loss coefficients
C0 and Ce.
4. Choose V3 = 1.1 yc to have a culvert with a flow depth conservatively above. yc
5. Compute the culvert width using the following energy equation, ignoring the headloss
caused by friction:
V42
PV32 V42 ^l
V3 2
+ C
73 +-f Vv3 ++Z74 + vA;4 ++ +
C ^ - -J-Z
V3 = (W3V3 where V3 = 1.1 yc = IAk1 (QfW3)2 For box culverts and V4 = QtWj4.
Neglecting Z3 - Z4 for a short reach and a mild bottom slope,

^RK^'tei]''--'
><+fei 4*" '" (ir ^ [(^. iut,w-ii ij" -c->
TABLE 20.1 Transition Loss Coefficients
Transition Type

Contraction
Cr

Expansion
Ce

Warped
Cylindrical quadrant
Wedge
Straight line
Square end

1.10
0.15
0.30
0.30
0.30

0.20
0.25
0.50
0.50
0.75

Source: Corry, et al (1975).

A^-*-^+>$-

n A 1

n2/3

/^

"\ ^ 1 \2/3

^) ^'^-('-"^^s0-*)
x2/3

and

^J

W3 = ([BV[A])

(20.3)

where
=*+][wh]-c>
and
[B] = L 1

2/3+

/i2/3 i
Tfefi ( 1 -^

6. Compute J1 with J2 = y3 using the energy equation

y2

/y2

y2\

y2

>+y>+t+c^-^rz>+y>+w

(20

-4)

7. If the amount of backwater (yr yn) exceeds a preferred or required freeboard, then
select a larger culvert width and calculate y3 using Eq. (20.3) given in Step 5 (return
to Step 5).
8. If the culvert width is acceptable, use the flow conditions to compute the transition
length using a 4.5:1 flare (USAGE, 1970):
LT = 4.5 (^

(20.5)

LT =4.5^^

(20.6)

or

9. Compute the water surface profile through the structure using a standard step backwater analysis. This will include an evaluation of friction losses.
Example 2. Determine the dimensions for a culvert and gradual transition needed for a 3
m-wide rectangular flood control channel at a slope of 0.001 m/m. The culvert length is
30.5 m, and is to convey 8.5 m3/s. Use n = 0.02.
Solution:
Step 1. Assuming normal depth at Section 4 (refer to Fig. 20.6), then use Manning's
equation to compute V4 = yn = 1.99 m and Vn = 1.42 m/s.

D ^2/3
( S2^3
-^ = 0.467 -^- = 0.935 m.

Step 3. Use a straight line transition with Ce = 0.5 and C0 = 0.3.

Step 4. y3 = 1.1 vc = 1.1 (0.935) = 1.03 m.
Step 5. Compute W3 using Eq. (20.3).

Step 6. Compute V1 with y2 = y3 using Eq. (20.4):

Step 7. Because ^1 is 1.29 m above the normal depth and because only 0.6 m free board
is available, we go back to Step 5 and compute ^3. Assuming that the culvert
width is 1.45 m: Repeat step 5,

Then, returning to Step 6, V1 using Eq. (20.4):

V1 is 0.42 m above the normal depth. Because a free board of 0.6 m is

available; the design is OK.
Step 8. Determine transition length using Eq. (20.5) for a 4.5:1 flare.

LT

= 4.5 (W' " ^ = 4.5^ -2L45> = 3.49 m, or 3.5 m.

Step 9. Compute the water surface profile.

20.1.2 Culverts with Inlet Control
Culverts with inlet control require the transitioning of supercritical flow, (see Fig. 20.7)
Because supercritical flow is difficult to control without creating a hydraulic jump or
other surface irregularities, the full flow area should be maintained. Because changing the
flow smoothly requires a long structure, model studies should be performed to determine
the transition geometry if a hydraulic jump is not desired. When a hydraulic jump is
acceptable, Figs. 20.8 and 20.9 can be used (USACE, 1970). Such a design requires a rectangular channel and a long transition.
The design procedure for supercritical flow contractions is as follows (Corry et al, 1975):
1. Flow conditions for the approach channel should be computed assuming normal flow
Cv' K ^r) usmg Manning's equation or other design aids.
2. Approach sections that are not rectangular should have a transition to the rectangular
section with a bottom width of W1. This bottom width should be approximately equal
to the average of the water surface top width (T) and the trapezoidal section base width
(Bw): i.e., W = (T + Bw) I 2. Compute the normal flow condition for this rectangular
section using Manning's equation.
3. Assume a trial culvert width W2. Refer to Figure 20.7 for a definition of parameters.
4. Compute the contraction length needed to reduce W1 to W2. This is accomplished by
varying the contraction using wall angle (9W) until L = (W1 W2) I 2 tan 0W is equal
to L1 + L2 where L1 = W11 2 tan P1 and L2 = W21 2 tan (P2 - 0W). To minimize surface disturbances, L should be equal to L1 + L2 (Corry et al, 1975).
Choose 0W

CENTERLINE
ALONGWALL

FLOW
a. SCHEMATICPROFILE

b. PLAN
FIGURE 20.7 Supercritical inlet transition for rectangular channel.
(From USAGE, 1970)
a. Compute L = (W1- W2)/2 tan 9W
b. Find P1, y21 V 1 , and Fr1 from
* *
tan B1 V(I + 8Fr12 sin2 B1 - 3)
1
^= 2 ^ + Vl+8Fr1CV-I

(2

'7)

and

^ = IJVl + 8Fr^ sin* P1 - 11

(20.8)

^
I](^+ lfl
2 = f4^ ? -UJL * I2^2JUi
JU
JJ

(20.9)

and

Alternatively, Figs. 20.8 or 20.9 can be used to approximately solve Eqs.

(20.7), (20.8), and (20.9).
c. Calculate L1 = W1/2 tan p.
d. Compute P2, j3/y2, and Fr2 using the same procedure as in Step 4b (that is, using P2
for P1, Fr2 for Fr1, y3/y2 for yjyl and the same 6W).
e. Calculate L2 = W212 tan (P2 - 6W)

FIGURE 20.8 Supercritical inlet transition design curves for rectangular channels.
(USAGE, 1970)

FIGURE 20.9 Supercritical inlet transition design curves for

rectangular channels (From Ippen, 1951)
f. Compare L with L1 + L2
if L > L1 + L2, then increase 6W and repeat Steps 4a-f until L = L1 + L2.
g. Compute y3 from V3 = V1 (^ M
UV UV
5. Compare depth V3 and the width W2. Culvert should be of a standard dimension. If
not return to Step 3 using another W2 and repeat the design process until a better
combination of V3 and W2 are found.
Example 3 (adapted from Corry et al., 1975). Design the transition contraction and select
the culvert size for a discharge of 300 cfs in a trapezoidal channel (6 ft bottom width, 2:1
side slope, S0 = 0.02 ft/ft and n = 0.012)
Solution:
Step 1. The normal depth, velocity, and Fr are yn = 1.67 ft, Vn = 19.2 ft/s,
and

Fr = -L= =- 19'2
= 3.05m
VgAfT
V 32.2(15.6)712.7
Step 2. Because (T + Bw )/2 = (12.7 + 6) /2 = 9.4 ft, use W1 = 10 ft, a rectangular chan-

nel. Use Manning's equation to compute yn = 1.54 ft and Vn = 19.5 ft/s, then
compute the Fr:
Fr = -^= = . 19'5
=2.77.
Vg^ V32.2(1.54)
Step 3. Assume a trial culvert width W2 = 5 ft.
Step 4. Try 8W = 14 for Fr = 2.8:
L = ( 1 0 - 5 ) / 2 tan 14 = 10ft
P1 = 35, V2^1 = 1.8, Fr2 =1.8

L1 = 10/2tan35 = 7.1 ft.

P2 = 55, y3/y2 = 1.6,Fr 3 = 1.1
L2 = 5 / 2 tan (55 - 14) = 2.9 ft.
L1 + L2 = 10 ft = L, O.K.
y3 = 1.54 (1.6) (1.8) = 4.4 ft.
Use Ow = 14, y3 = 4.4 ft., V3 = 13.6 ft/s, Fr3 = 1.1

Step 5. Because y3 = 4.4 ft. and W3 = 5 ft, a 5 X 5 box culvert will be satisfactory.

20.2

ENERGYDISSIPATIONFORCULVERTSANDCHANNELS

20.2.1 Hydraulic Jump Basins

For supercritical flow expansions, the procedure outlined in Sec. 20.1 is applicable if the
exit Fr is less than 3, if the location where the flow conditions are desired within three culvert diameters of the outlet and if S0 is less than 10 percent (Cony, et al 1975).For expansions outside these limits, a hydraulic jump basin (Fig. 20.10) should be used. This type
of basin allows the flow to expand, drop or both, resulting in depth decreases, velocity
increases, and an Fr increase in. Higher Fr's result in more efficient jumps and shorter
basins.
The design procedure for supercritical flow expansions with hydraulic jump basin is as
follows (Corry, et al; 1975):
1. Compute the culvert brink depth y0 using Figs. 20.4 or 20.5.
2. Compute the tailwater depth Tw in the downstream channel assuming normal flow
(using Manning's equation) or perform backwater analysis.
3. To determine the basin elevation, first select Z1 and then use the following steps.
a. Select basin width W8 (W1 in Fig. 20.10) and basin slopes S5 and S7 (Fig. 20.10).
Slope of S5 or ST = 0.5(2:1) or 0.33 (3:1) are satisfactory (Corry et al, 1975).
b. Check WB using

DATUM
FIGURE 20.10 Definition sketch basin transition. (From Corry et al., 1975)

W <W +

(20 10)
'
'
^T 3Fr0+1
'
where LT = (Z0 Z1)AS7, and the right-hand side is the limit that flares naturally in the
slope distance L.
c. Compute Y1 using the following equation derived from the energy equation from the
culvert outlet brink to the basin (Sec. 1 in Fig. 20.10). Use V1 = Q/ y^W^ to determine^ and then V1:

r
i1'2
Q = J^B [Zg (Z* - ZI + % - *) + ^02J

( 2 O-U)

d. Compute the Fr
Fr 1 =A
vw
e. Compute y2 using Eq. (20.12) for the hydraulic jump basin:
J2 = y[vi + 8 F r J - I J

(20.12)

f. Compute Z3 from geometry using

(20.13)
where

and from Fig. 20.11, determine L, which is defined as L8

L8=Xy19Fr1).
g. Check value of Z1 by computing y2 + Z2 and Z3 + Tw. If y2 + Z2 > Z3 4- Tw, select
another Z1 and repeat steps 4a to 4g until a balance is reached.
4. Compute L5 and L:
_ (Z3 - Z2)
* ~^r~
L = LT + Lfi + L5 =

^o

Example 3. A supercritical flow expansion is to be designed for a reinforced concrete

box culvert measuring 3 by 2 m. Determine the dimensions for the hydraulic jump basin
using a design discharge of Q = 11.8 mVs. The slope is 6.5 percent, the invert outlet elevation is 30.5 m, the downstream channel has a bottom width of 3 m and side slopes of
2:1, and Manning's n = 0.03. The brink depth is supercritical, V0 = 0.457 m and
V0 = 8.47 m/s. S3 = 0.5 and ST = 0.5. (Refer to Figure 20.12)

FIGURE 20.11 Length of jump in terms Of^ 1 , rectangular channel. (From Corry et al.,
1975)

FIGURE 20.12 Example problem 18.2.1 hydraulic jump basin. (Not to scale).

Solution:
Step 1. The brink depth and velocity are given: Fr0 = 4.0.
Step 2. The tailwater depth is computed assuming normal depth. Manning's equation
is solved to obtain yn = 0.57 m tailwater depth and Vn = 4.85 m/s. (Refer to
Fig. 20.12)
Step 3. Assuming a basin elevation OfZ 1 = 25.9 m:
a. Select W8 = 3 m and S5 = S7 = 0.5
b. Check W8 using Eq. 20.10, where W0 = 3 m,
W8 = 3 < 3 + [2L7 V5JTTJ(3Fr0I OK.
c. Compute ^1 using Eq. (20.11)
1/2
11.8 = V1(S) 2(9.81)(30.5 - 25.9 + 0.457 - V1) + (8.47)2

Solving ^1 = 0.306 m and V1 = QIA = 11.8/[3(0.306)] = 12.85 m/s.

d. The Froude number is Fr = 12.85/V9.81(0.306) = 7.42
e. Compute y2 using Eq. (20.12). y2 = 0.306 Vl + 8(7.42)2 - 1 /2 = 3.06 m.
f. Compute Z3 using Eq. (20.15). First compute LT = (Z0 - ZJ/2
= (30.5 - 25.9)70.5 = 9.2 m. From Fig. 20.11, LJy1 = 63, so L8 = 63yl
= 63(0.306) = 19.28 m, then
_ [30.5 - (9.2 + 19.28 - 25.9/0.5) 0.065] _
= 28 34 m
(0.065/0.5 + 1)
'

Z =

g. Check value OfZ1: y2 + Z2 = 3.06 + 25.9 + 28.96 and Z3 + Tw = 28.34 + 0.58

= 28.92, y2 + Z2 > Z3 + Tw\ therefore, Z1 = 25.9 is OK.
4. Compute L8 and then L. L5 = (Z3 - Z2)JS5 = (28.33 - 25.9)70.5 = 4.86 m. Then
L = LT + L8 + L5 = 9.2 + 19.28 + 4.86 - 33.34 m. Refer to Fig. 20.12
20.2.2

Forced Hydraulic Jump Basins

20.2.2.1 Saint Anthony Falls stilling basin. The Saint Anthony Falls, (SAF) stilling basin
is a generalized design based upon model studies conducted by the U.S. Soil Conservation
Service at the St. Anthony Falls Hydraulic Laboratory, University of Minnesota. Figure
20.13 illustrates the SAF stilling basin design which is recommended for small structures
such as spillway outlet works, and for canals where the Fr ranges from 1.7 to 17 (at the dissipator entrance). Through the use of chute blocks, baffle or floor blocks, and an end sill,
the basin length is about 80 percent of the free hydraulic jump length.
The design procedure for SAF basins is as follows (Corry et al 1975):
1. Choose basin configuration and flare dimension, Z. (Refer to Fig. 20.13)

2. Use the design procedure presented in Sec. 20.2.1 for supercritical expansions into
hydraulic jump basins to determine basin width (W^), elevation (Z1), length (Le), total
length (L), incoming depth Cy1), incoming Fr (Fr1), and jump height (y2). Steps 3e and
3f in Sec. 20.2.1 are modified; for Step 3e, determine y2 using the sequent depth y.\
(20.14)
(20.15)
(20.16)
(20.17)

RECTANGULAR BASIN
HALF-PLAN
EQUATION NUMBER
FLARED BASIN
HALF-PLAN
0-90
45 PREFERRED

SIDEWALL

CHUTEBLOCK
FLOORORBAFFLEBLOCKS ENDSILL

VARIES

(1) W6 = BASINWIDTH UPSTREAM

(2) n BLOCKS AT 3/4 Y 1
(3) 0.4OW82 < AGGREGATEBLOCKWIDTH < 0.55Wg2
(4) ri BLOCKS AT 3/4 Y, -5_
(5) WB2 = WB + 2LB/3z
(6) WB3 = WB + 2LB/z
FIGURE 20.13 St. Anthony Falls stilling basin. (From Blaisdell, 1959)

For Step 3/, compute L5:

45v
LB = -^pr -16
i
3. Detennine the dimensions of the chute block:

Height:

H1 = yl

Width:

W2 = 0.1Sy1 and W1 = spacing

W
Nc ~ -^- (rounded)
1
W
W1 = W2 = j- (Nc includes the 1/2 block at each wall)

Number:
Adjusted:

(20.18)

4. Determine the dimensions of the baffle block:

Height: h3 = y t
Width: W3 = spacing and W4 = 0.75V1
Basin width at baffle blocks: WB2 = WB + 2L/3Z
Number of blocks: N8 = WB2/2W3 rounded
Adjusted W3 = W4 = WB2/2NB
Check total block width to insure that at least 40 to 55 percent of W82 is occupied
by blocks.
Distance from chute blocks to baffle blocks = LB/3
5. End sill height: h4 = 0.07^.
6 Side wall height: y2 + y. /3.
Example 4. Determine the dimensions of an SAF basin for the supercritical flow expansion described in Example 3.
Solution:
Step 1 Select a rectangular basin with no flare.
Step 2 Steps 1 through 3(a-f) (in Example 3) for a supercritical flow expansion into a
hydraulic jump basin.
Given V0 = 8.47 m/s, y0 = 0.457 m, and Fr0 = 4.0.
The tailwater depth Tw = yn = 0.57 m and Vn = 4.85 m/s.
Assume that Z1 = Z0 = 30.5 m. (Refer to Fig. 20.14)
a. Compute y.} using Eq. 20.14 with yl = y0 = 0.457 m and Fr1 = Fr0 = 4.0. This
assumes that Z1 = Z0: i.e., the basin floor is the same elevation as the culvert
outlet.

FIGURE 20.14 Example 4 St. Anthony Falls stilling basin. (Not to scale)

b. Next, use Eq. (20.15) for Fr1 = 1.7 to 5.5 to compute J2:

Because y2 > Tw = 0.57, we can lower the elevation of the basin. Use Z1 =
27.9 m with W8 = 3 m and ST = S5 = 0.5. WB is OK and no flare is used.
c. Compute J1 using the energy Eq. (20.11). The basin has been lowered so now
J1 is not J0, the brink depth.
Q = * WB\2g(ZQ - Z1 + J0 - ^1) + V02]
r
nl/2
11.8 = ^(312(9.81X30.5 - 27.9 - 0.457 - J1) + (8.47)2

Solving J1 = 0.348 m. V1 = 11.8/(3 3 0.348) = 11.3 m/s

d. The Fr1 = V1JVg^1 = 11.3/V9.81(0.348) = 6.1
e. Compute jj = y|Vl + 8Fr^ - 1 j = ^^-|Vl + 8(6.1)2 - I J = 2.83 m.
^^^h-SHi1-1-^]2-83=2-24111f. Compute L8 using Eq. (20.18):
LB = 4.5yjFrw* = 4.5 (2.83)/6.1076 = 3.22m.
Compute L7, using LT = (Z0 - Z1)JS7 = (30.5 - 27.9)/0.5 = 5.2m.
Compute Z3 using Eq. (20.13)

g. Check the assumption OfZ 1 = 27.9 m: y2 + Z2 = 2.24 + 27.9 = 30.14 and

Tw + Z3 = 0.57 + 29.7 = 30.2 m: y2 + Z2 Tw + Z3. Because they are so
close, then Z1 will be OK.
h. LT = 5.2 m, LB = 3.22 m, L8 = (Z3 - Z2)JS8 = (29.7 - 27.9)/0.5 - 3.6 m,
and
L = Lr + LB + L8 = 5.2 + 3.22 + 3.6 = 12 m.

Step 3. Chute blocks: H 1 ^ y 1 = 0.35, W1 0.75^1 = 0.26 m = W2. Nc = W8^W1 = 3/

(2X0.26) = 5.77 so use 6 blocks. Adjusted W1 = WB/(2NC) = 3/(2 X 6) = 0.25
m. This provides 5 blocks, 6 spaces, and a half block at each wall.
Step 4. Baffle blocks: H3 J1 = 0.35 m, W3 0.75V1 = 0.26 m = W4. Basin width,
^B2 = WB + 2LB/(3Z) = 3 + O = 3 m (no flare), NB = W52/(2W3) = 3/(2 X
0.26) = 5.77 so use 6 blocks. Adjusted W3 = W4 = 3/(2 X 6) = 0.25 m.
Total block width = 6(0.25) = 1.5 m. Check percentage: 0.4 < 1.5/3.0 < 0.55,
OK. This provides six blocks, five spaces, and a half-space at each wall. Distance
from chute block = LJ3 = 3.22/3 = 1.07 m.
Step 5. End sills: h4 = Q.Oly. = 0.07(2.83) = 0.2 m.
Step 6. Side walls: height = y2 + Vj/3 = 2.24 + 2.83/3 = 3.2 m. Refer to Fig. 20.14 with
the dimension for the chute block shown.
20.2.2.2 Type II, III, and IV basins. The U.S. Bureau of Reclamation's Type II, III,
and IV basins are illustrated in Chapter 18, in Figs. 18.3, 18.4 and 18.7 respectively.
Type II basin design. Use the design procedure presented in Sec. 20.2.1 for supercritical expansion into a hydraulic jump basin to determine W8, Z1, L8, L, V1, Fr1 and V2.
For Step 3e in that section, use C = 1.1 to find y2 = C1V1 Vl + 8Fr12- 1 /2. For Step
3f,use Fig. 20.15 to determine L8.
\-I'
Determine the dimensions for the chute blocks and dentated sill height using the relations in Fig. 18.3.
Type HI basin design. Use the design procedure in Sec. 20.2.1 to determine basin
dimension. For Step 3e use C= 1.0 to determiney2 = Qy1 Vl + 8Fr12- 1/2. For Step
3f, use Fig. 20.15 to determine L8. Use dimensions in Fig. 20.16 to determine chute block
dimensions and spacing.
Use Fig. 18.4 to determine dimensions for baffle blocks and the end sill height.
Type IV basin design. Use the same design procedure presented in Sec. 20.2.1 forsupercritical expansions into a hydraulic jump basin to determine the basin dimensions.
For Step 3e in the section, use Q = 1.0 to determine y2 = QyJ Vl + 8.Fr12- 1 /2. For
Step 3f, use Fig. 20.15 to determine L8.

FREEJUMP

TYPE IV
TYPE Il BASIN

TYPE III BASIN

FIGURE 20.15 U.S. Bureau of Reclamation Type II basin. (Corry et al, 1975)

ENDSILL
BAFFLE PIERS

BAFFLE PIERS

END

FIGURE 20.16 Height of baffle piers and end sill (Type III basin). (From U.S. Bureau of Reclamation, 1987)

Use dimensions in Fig. 18.7 to determine chute block dimensions and spacing.
Use Fig. 18,7 to determine the end sill height.
Example 5. Determine the dimensions of U.S. Bureau of Reclamation's Type II basin for
the supercritical flow expansion described in Example 3.
Solution:
Step 1. Use the design procedure for a supercritical flow expansion into a hydraulic jump
basin. (Refer to Examples 3 and 4).
Given V0 = 8.47 m/s, V0 = 0.457 m, and Fr0 = 4.0. (Figure 20.17)
The tailwater depth is Tw = yn = 0.57 m, and Vn = 4.85 m/s.
Assume Z1 = Z0 = 30.5 m, compute y2
y2 = C1J1[Vl + 8Fr^- l]/2
= 1.1(0.457/Vl + 8(4)2 - IJ/2
= 2.6m

y2 > Tw (2.6 > 0.57). Therefore we need to lower elevation Z1 of the basin floor.
a. Use a basin floor elevation of Z1 = Z2 = 25.76 m, with W8 = 3 m, ST =
S5 = 0.5.
b. W8 is OK, no flare.
c. Compute V1 using energy Eq. (20.11).
r
T'2
Q = ViW8^g(Z0 - Z1 + y0 - V1) + V02J
r
i172
11.8 = Vj(3) 2(9.81)(30.5 - 25.76 + 0.457 - V1) + 8.472

Solving V1 = 0.3 m and V1 = 13.1 m/s.

d. The Fr1 = 13.1/V9.81(0.3) = 7.6
e. Compute y2 = 1.1(0.3) Vl + 8(7.6)2 - 1 /2 = 3.39 m.
f. Using Fig. 20.15, L8Jy2 = 4.3, so L8 = 4.3(3.39) = 14.6 m:
LT = (Z0 - Z1)IS7 = (30.5 - 25.76)/0.5 = 9.5 m.
Using Eq. (20.13), compute Z3

Z3

_ [30.5 - (9.5 + 14.6 - 25.76/0.5)0.065] _

~
(0.065/0.5 + 1)
~ 28'57 m>

g. Check Z1, y2 + Z2 = 3.39 + 25.76 = 29.15 and Z3 + Tw = 28.57

+ 0.57 = 29.14, OK
h. Compute L5 = (Z3 - Z2)JS5 = (28.57 - 25.76)/0.5 = 5.6 m and

L = LT- LB + Ls = 9.5 + 14.6 + 5.6 = 29.6 m.

Step 2. Chute blocks: H1 = W1 = W2 = J1 = 0.3 m, Wc = | = 5, OK.
2(0>3
Side-wall spacing = y/2 = 0.15 m.
'
Ste/? 3. Dentated sill: H2 = 0.2v2 = 0.2(3.39) 0.7 m.
W3 = W4 = 0.15y2 = 0.15(3.39) = 0.51 m.
Ns = WB/W3 = 3/0.51 m = 5.88.
Use 5, which provides three blocks and two spaces, each of which is 0.6 m
wide. Refer to Fig. 20.17 for the dimensions of the basin.
20.2.3

Impact-Type Energy Dissipation (USBR Type Vl Basin)

Figure 20.18 illustrates the U.S, Bureau of Reclamation's Type VI impact-type energy dissipator, which can be used with culverts. This basin is contained in boxlike structures
requiring no tail water for operation. The structures can be used for open channels as well
as culverts, and the basin can be used at sites where the entrance velocity to the basin does
not exceed 50 ft/s, and the discharge is less than 400 cfs. This dissipator should not be
used if the buildup of debris or ice can cause substantial clogging.
The design procedure is as follows (Corry et al, 1975).
1. Compute the flow area at the end of the culvert using the maximum design discharge
and velocity. Compute the equivalent depth of flow entering the dissipator from the
culvert as
Je = (f)"2

(20.19)

where A is the cross-sectional area of flow in the culvert. This converts the cross-sectional area of flow of a pipe into an equivalent rectangular cross section with a width
twice the depth of flow. The culvert preceding the dissipator can be open, closed, or
have any cross section. This approach ignores the size and shape of the culvert entirely except for the determination of flow entering the dissipator.
2. Compute Fr and the energy at the end of the culvert H0:
V2
" = ^ + 2 F -

Then use Figure 20.19 to determine the basin width. Enter Figure with Fr to determine
H0IW then W = H0I (H0IW).
3. Use Table 20.2 to determine the dimensions of the dissipator structure.

DATUM

FIGURE 20.17 Example 5 U.S. Bureau of Reclamation type II basin

PLAN

SECTION
STILLING BASIN DESIGN

PLAN

BEDDING

SECTION
ALTERNATE ENO SILL

FIGURE 20.18 Baffle-wall energy dissipator of a U.S. Bureau of Reclamation type IV basin. (From Corry et
al., 1975)

FIGURE 20.19 Design curve for a baffle-wall dissipator. (Corry et al., 1975)

20.2.4 Drop Structures

20.2.4.1 Straight-drop spillway. The straight-drop spillway shown in Figs 20.20 and .
20.21 is generally used for subcritical flow in the upstream as well as the downstream
channel. To describe the flow geometry, the following drop number is used:
N

D =^

(20.20)

where q is the discharge per unit width of the crest overfall, g is the acceleration caused
by gravity, and h0 is the height of the drop. The dimensions L1, V1, y2, and ^3 in Fig. 20.22
are determined using the following:
^- = 4.3(W*27
AZ0

(20.21)

- = LQN*22
AZ0

(20.22)

TABLE 20.2 Baffle Wall Dissipator: Dimensions of The Basin in Feet and Inches

W
4-0
5-0
6-0
7_0
8-0
9-0
10-0
11-0
12-0
13-0
14-0
15-0
16-0
17_0
18-0
19-0
20-0

H1

3-1 5-5
3-10 6-8
4-7 8-0
5-5 9-5
6-2 10-8
6-11 12-0
7-8 13-5
8-5 14-7
9-2 16-0
10-0 17-4
10-9 18-8
11-6 20-0
12-3 21-4
13-0 22-6
13-9 23-11
14-7 25^
15-4 26-7

H2

1-6
1-11
2-3
2-7
3-0
3-5
3-9
4-2
4-6
4-11
5-3
5-7
6-0
6-4
6-8
7-1
7-6

Dimensions of Basin in Feet and Inches

H3 L1
L2
H4
W1 W2
t3
0-8 2-4
0-10 2-11
1-0 3-5
1-2 4-0
1-4 4-7
1-6 5-2
1-8 5-9
1-10 6-4
2-0 6-10
2-2 7-5
2-4 8-0
2-6 8-6
2-8 9-1
2-10 9-8
3-0 10-3
3-210-10
3-4 11-5

3-11-8
3-102-1
4-72-6
5-52-11
6-23^
6-113-9
7-84-2
8-54-7
9-25-0
10-05-5
10-95-10
11-66-3
12-36-8
13-07-1
13-97-6
14-77-11
15-48-4

I2

I1

I4

t5

0-41-1 0 - 6 0 - 6 0 - 6 0-6 0-3

0-5 1-5 0 - 6 0 - 6 0 - 6 0-60-3
0-6 1-8 0-6 0-6 0-6 0-6 0-3
0-6 1-11 0-6 0-6 0-6 0-6 0-3
0-7 2-2 0-7 0-7 0-6 0-6 0-3
0-8 2-6 0-8 0-7 0-7 0-7 0-3
0-9 2-9 0-9 0-8 0-8 0-8 0-3
0-10 3-0 0-9 0-9 0-8 0-8 0-4
0-11 3-0 0-10 0-10 0-8 0-9 0-4
1-0 3-0 0-10 0-11 0-8 (MO 0-4
1-1 3-0 0-11 1-0 0-8 0-11 0-5
1-2 3-0 1-0 1-0 0-8 1-0 0-5
1-3 3-0 1-0 1-0 0-9 1-0 0-6
\-A 3-0 1-0 1-1 0-9 1-0 0-6
1-4 3-0 1-1 1-1 0-9 1-1 0-7
1-5 3-0 1-1 1-2 0-10 1-1 0-7
1-6 3-0 1-2 1-2 0-10 1-2 0-8

Source: Cony, et al; (1975)

^- = 0.54A425
H0

(20.23)

^- = 1.66A#27
(20.24)
/I0
L2 is the length of the jump, L1 can be determined using Fig. 20.22. The sequent depth
and the tailwater depth Tw must be compared to determine whether Tw < y3, or Tw = y3,
or Tw > y3. If Tw < y3, the hydraulic jump moves downstream. In this case, it is necessary

AERATED

FIGURE 20.20 Flow geometry of a straight-drop spillway. (From Corry et al., 1975)

TOP SLOPE I TO I
UPPER NAPPE
SIDE WALL
HEIGHT

SECTION AT CENTER LINE

FLOOR BLOCKS
LONGITUDINALSILL
(OPTIONAL)

ENDSILL

PLAN
FIGURE 20.21 Straight-drop spillway stilling basin. (From Rand, 1955)

to construct the apron at the bed level and an end sill or baffles or to construct the apron
below the downstream bed level and an end sill. If Tw > y3, the hydraulic jump may
become submerged. If Tw = V3, the hydraulic jump begins at depth V2; there is no supercritical flow on the apron, and L1 is a minimum.
20.2.4.2 Grated energy dissipators. Energy dissipators with grates (Fig. 20.23) also can
be used in conjunction with drop structures. The U.S.Bureau of Reclamation (1987)
developed the following design recommendations for grates for incoming subcritical
flow:
1. Select slot width with a full slot width at each wall.
2. Compute beam length L0 using

LOWER NAPPE
(NOTAILWA1TER)

TAILWATER
LEVEL
TANGENTATPOINT
OFSUBMERGENCE
MEAN
SUBMERGED
TRAJECTORY
NAPPE
TRAJECTORY
FREE NAPPE
TRAJECTORY

FIGURE 20.22 Design chart for determination of L1. (From Corry et al., 1975)

LGG =

,
C(W)(N) V2^~

(20.25)

where C is a coefficient equal to 0.245, W is the width of the slots in feet, and N is the
number of slots or spaces between beams.Then compute the beam width = 1.5 W. The
quantity (W)(N) can be adjusted until an acceptable beam length, L0 is determined.
3. For self-cleaning, the grate can be tilted appoximately 3 in the downstream direction.
20.2.4.3 Straight drop structures. The straight drop structure shown in Fig. 20.21 consists of a horizontal apron with blocks and sills to dissipate energy. This structure is for
drops of less than 15 ft (4.57 m) and for sufficient tailwater. The design parameters
include the length of the basin, the position and size of the floor blocks, the position and
height of the end sill, the position of the wingwalls, and the geometry of the approach
channel. This structure was developed by the Agricultural Research Service at the Saint
Anthony Falls Hydraulic Laboratory.
The design procedure for a straight-drop structure is as follows (Corry et al 1975):

Figure 20.23. Energy dissipator with grate ( Cony et al, 1975)

1. Compute the minimum length of stilling basin L8

L8 = L1+ L2 + L3 = L1 + 2.55yc

(20.26)

where distances are illustrated in Fig. 20.21. The distance from the headwall to the
point where the surface of the upper nappe strikes the still basin floor L1 is
L 1 =^

(20.27)

Lf = -0.406 + J 3.195 - 4.368 [^] yc

V^c ) _

(20.28)

where

0.691+0.228 f M - f ^ l L
UJ
UJj
LS
~
I"
1
L

(20.29)

0.185 + 0.456^1
UJJ
and

Lt = -0.406 + J3.195 - 4.368p| yc

L
^c' _

(20.30)

where h2 = h0 y3 (Fig. 20.21). Alternatively, L1 can be determined using Fig. 20.22.

The distance from the point where the surface of the upper nappe strikes the

2.

3.
4.
5.
6.
7.
8.

9.

stilling basin floor to the upstream face of the floor blocks L2 is determined using
L2 = 0.8 yc. The distance between the upstream face of the floor blocks and the end of
the stilling basin, L3, is determined using L3 ^ 1.75yc.
Floor blocks are proportioned as follows:
a) height = 0.8yc;
b) width and spacing should be 0.4yc with a variation of 0.15yc permitted, and
c) blocks should be square in plan, and
d) blocts should occupy between 50 to 60 percent of the stilling basin width.
Compute the end sill height as 0.4yc.
If longitudinal sills (for structural purposes only) are used, they should be constructed
through the floor block, not between the floor blocks.
Compute sidewall height above the tailwater level as 0.85yc.
Wingwalls are constructed at an angle of 45 with the outlet center line with a top slope
of 1:1.
Compute the minimum height of the tailwater surface above the floor of the stilling
basin y3 using y3 = 2.15yc.
Modification, to the approach channel are as follows: The crest of the spillway should
be at the same elevation as the approach channel, the bottom width should be equal to
the spillway notch length W0 at the headwall, and protection with riprap or paving
should be provided for a distance upstream of the headwall of 3yc.
Using the recommendations in Step 8, no special provision for aeration is needed.

Example 6 Determine the dimensions of a straight-drop spillway stilling basin for a discharge of 7.08 m3/s. The downstream trapezoidal channel has a 3:1 side slope with a 3.05
m bottom width S0 = 0.002 m/m, n = 0.03, and a normal depth of 1.024 m. The drop h0 =
1.83 m.
Solution:
Step 1. Determine the minimum basin length L8. The critical depth yc is determined as
yc = 0.655 m. Then hjyc = 1.83/0.655 = 2.79; H2 = h0 -2.15yc = 1.83 2.15(0.655) = 0.422 m; hjyc = 0.422/0.655 = 0.644.
Using Fig. 20.22, L1Jyc = 8.2 for hjyc = 2.79 and hjyc = 0.644.
Then L1 = 8.2(0.655) = 5.37 m; L2 = 0.8yc = 0.8(0.655) = 0.52 m; and L3 = 1.75yc
= 1.75(0.655) = 1.15 m; thus, LB = L1 + L2 + L3 = 5.37 + 0.52 + 1.15 = 7.04 m.
Step 2. Proportions the floor blocks are
height = 0.8vc - 0.8(0.655) = 0.524 m,
width = 0.4yc = 0.262 m, and
spacing = 0.4yc = 0.262 m.
Step 3. Calculate the end sill height = 0.4yc = 0.262 m.
Step 4. Use the longitudinal sill passing through the floor blocks.
Step 5. Calculate the side-wall height above the tailwater = 0.85yc = 0.85(0.655)
= 0.557 m.

Step 6. Local wing walls are at a 45 angle with the outlet center line.
Step 7. Calculate the minimum height of tailwater above the floor of the stilling basin: y3
= 2.15yc - 2.15(0.655) = 1.41 m. The basin must be placed 1.43 - 1.024 =
0.406 m below the downstream bed level.
20.2.4.4 Box-inlet drop structure. The box-inlet drop structure shown in Figure 20.24
consists of two different sections that are effective in controlling flow: the crest of the box
inlet and the opening in the headwall. This structure is based on experiments by the U.S.
Soil Conservation Service at the Saint Anthony Falls Hydraulic Laboratory (Blaisdell and
Donnelly, 1956). The design procedure of box-inlet drop structures is as follows:
(FHWA, 1978):
1. Select/*0
2. Select L1, W2, and Lc where Lc is the length of the box-inlet crest, Lc = W2 + 2L1,

FIGURE 20.24 Box-inlet drop structure. (From Corry et al., 1975)

where W2 is the width of the box inlet, and L1 is the length of the box inlet.
3. Compute the head y0 for the crest using the discharge equation for a rectangular weir:
(20 31)
^(dkf
4. Compute -^
and determine the coefficient of discharge C2 from Fig. 20.25.
(W2j

(L,]
5. Compute M- and determine the relative head correction, CH, from Fig. 20.26.
6. Compute y0 for the headwall opening using
y0=

^7=
IC2W2V^]

2/3

-Cn

(20.32)

which is based upon the rectangular weir equation, Q2 = C2W2\^2g(y0 + CH)3/2.

7. Compare y0 from Step 3 for the crest and y0 from Step 6 for the headwall opening. The
larger value of y0 controls. If the crest controls, adjust V0 from Step 3 using the following procedure:
a. Compute y0/W2 and determine the correction for the head, C1, using Fig. 20.27.
b. Compute L1JW2 and determine the correction for the box inlet shape, C5, using Fig.
20.28
c. Compute W{/LC and determine the correction for the approach channel width CA,
using Fig. 20.29
d. Compute W4JW2 and determine the correction for dike effect (proximity of dike to
box inlet crest) CE using Table 20.3

COEFFICIENT OF DISCHARGE, C2

HYDRAULIC DESIGN OF THE BOX - INLET DROP SPILLWAY

FIGURE 20.25 Coefficient of discharge, with control at headwall opening. (From Corry et al.,
1975)

FIGURE 20.26 Relative head correction for HJw2 > 1V4 with control at headwall opening.
(From Corry et al., 1975)
e. Determine the adjusted y0:
? -(sW/wJ20

(2033)

8. Compute the critical depth in the straight section yc using

t/3

'-ml
9. Compute the critical depth at the exit of the stilling basin yc3 using

-tej

10. Compute the minimum length of the straight section L2 using

f
0.2
L2 y
~ < (LA
IkJ

(20.36)

for values OfL 1 XW 2 ^ 0.25.

11. Compute the minimum length of the stilling basin using

DISCHARGE COEFFICIENT, C1

CORRECTION FOR HEAD

FIGURE 20.27 Discharge coefficient and correction for head, with control
at box-inlet crest. (From Corry et al., 1975)

CORRECTION FOR BOX - INLET SHAPE, Cg

CORRECTION FOR APPROACH - CHANNEL WIDTH, CA

W
FIGURE 20.28 Correction for box-inlet shape, with control at the box-inlet crest (-c > 3).
From Corry et al., 1975)

FIGURE 20.29 Correction for approach-channel width, with control at box-inlet

crest. (From Cony et al., 1975)

TABLE 20.3 Correction for Dike Efect CE: Control at the Box-Inlet Crest
(Control at Box-Inlet Crest)
W4AV2
L1IW2
OO
0.5
0.90
1.0
.80
1.5
.76
2.0
.76
Source: Corry et al (1975)

Ol
0.96
.88
.83
.83

0.2
1.00
.93
.88
.88

LB

0.3
1.02
.96
.92
.92

0.4
1.04
.98
.94
.94

_(i)
2L
i
(W2)

0.5
1.05
1.00
.96
.96

0.6
1.05
1.01
.97
.97

(20.37)

and

L3 = ^f^

(20.38)

and choose the larger value of L3.

12. Compute the minimum tailwater depth over the basin floor using
V3 = 1.6vc3

if

W3 < 11.5vc3

(20.39)

or

?3 = yC3 + 0.052PF3

if

W3 > 11.5yc3

(20.40)

13.Compute the height of the end sill /I4 using h4 = y3/6.

14. Determine the number of longitudinal sills:
If W3 < 2.5W2, use two sills;
if W3 > 2.5W2, use four sills.
When two sills are used they should be located at a distance W5 on each side of the
centerline. When four sills are used, the two additional sills should be located parallel
to the outlet centerline and midway between the center sills and the sidewalls at the
stilling basin exit.
15.Compute the minimum height of the sidewalls above the water surface at the exit of
the stilling basin Ji3 using /I3 = v3/3. Sidewalls should extend above the tailwater surface under all conditions.
16. Wingwalls should be triangular in elevation and have a top slope of 45 with the horizontal. The top slopes can be as flat as 30. Wingwalls should flare in plan at an angle
of 60 with the outlet centerline. The flare-wall angle can be as small as 45.
Wingwalls parallel to the outlet centerline should not be used.

20.2.5

RIPRAPBASINS

The riprap basin recommended in Corry et al (1975) for culverts is shown in Fig. 20.30,
which is a preshaped basin lined with riprap. The surface of the riprapped floor of the
energy dissipating pool is at an elevation hs below the culvert invert, where hs is the
approximate depth of scour that would occur in a thick pad of riprap if subjected to the
design discharge (Corry et al 1975). The length of the pool is the larger of I0hs or W0, and
the overall length of the basin is the larger of I5hs or 4W0. The ratio hs/d5Q should be less
than 4 [(HJd50) < 4] .Figure 20.31 provides design detaills for riprapped culvert energy
basins.

NOTEB

IFT
BOARD

NOTEA - SUFFI
IF EXI f VtLOClIY OF BASIONNAL
IS SPECI
EXTENDN A-ABASISUCH
N AS REQUIRED TO OBTAIN
AREAFFIATEIED.DSECTI
SECTIOCNIENTAREACROSS-SECTI
AT SEC. A-A) - SPECI
EXITOVELOCI
TV. THAT O^ /(CROSS
NOTE B - FLOOR
WARP OFBASIBASIN TON SHOULD
CONFORMBE ATTOTHENATURAL
STREAM
CHANNEL.
TOP OF RIPRAP IN
CHANNEL BOTTOM AT SEC. A-A. SAME ELEVATION OR LOWER THAN NATURAL
DlSSlPATOR POOL
Sh1APRON
ORW0MIN.
100,OR3W
0MIN.
TOPOFBERM
TOP OF RIPRAP
CHANNEL
ZONTAL

TO SUPPORT RIPRAP
THI
COPTIONAL
KENEDORSLOPI- CONSTRUCT
NG
TOE
IF DOWNSTREAM
CHANNEL

SECTION

NOTE B
EXCAVATETOTHISLINE,
NOTE:
W0 - PIDIAMETER
FOR
PE CULVERT
W0 - BARREL
WIDTH
FOR
BOX
CULVERT
W = SPAN
OF PIPE-ARCH
CULVERT
SYMM ABOUT
APRON

CULVERT

HORIZONTAL

BERM
AS REQUIPRRAPED
TOSUPPORTRI

FIGURE 20.30 Details of riprapped culvert energy basin. (Corry et al., 1975)

DESIGNDISCHARGE-Q
WETTED AREA AT BRINK OF CULVERT
d50 = THE MEDIAN SIZE OF ROCK
BYWEIGHT.
OR
ANGULARROUNDEDROCK
ROCK.
EQUIVALENT BRINK DEPTH
BRINK DEPTH FOR BOX CULVERT
FOR
NON-RECTANGULAR
SECTIONS

CULVERTBRINK

RELATIVE DEPTH OF SCOUR HOLE -^

Ve

?. SECTION

RIPRAP MAY BE REQUIRED

ON BANKS AND CHANNEL
BOTTOM DOWNSTREAM
FROM BASIN - SEE DESIGN
EXAMPLE IN TEXT.

FROUDE NUMBER
FIGURE 20.31 Relative depth of scour hole versus Froude number at brink
of culvert with relative size of riprap as a third variable.

REFERENCES
Blaisdell, F. W., Flow Through Diverging Open Channel Transitions at Supercritical Velocities,
SCS Report No. SCS-TR-76, U.S. Department of Agriculture, April 1949.
Blaisdell, F. W., and C. A. Donnelly, "The Box Inlet Drop Spillway and Its Outlet", Transactions,
of the American Society of Civil Engineers, 121:955-986, 1956.
Blaisdell, F. W., The SAF Stilling Basin, U. S. Goverment Printing Office, 1959
Corry, M. L., P. L. Thompson, E. J. Watts, J. S., Jones, and D. L. Richards, Hydraulic Design of
Energy Dissipators for Culverts and Channels, Enqineering Circular 145, Federal Highway
Administration, U.S. Department of Transportation, Washington DC, 1975.
Fedeard Highinay Administratin , Hydraulics of Bridge Waterways, Hydraulic Design Series No. 1,
Federal Highway Administration, U.S. Department of Transportation, Washington, DC, 1978.
Ippen, A. T, "Mechanics of Supercritical Flow", Transactions of the American Society of Civil
Engineers,, lib; 268-295, 1951
Rand, W, "Flow Geometry at Straight Drop Spillways," Paper No. 791, Proceedings of the
American Society of Civil Engineers, VoI, 81, pp. 1-13, September 1955.

Simons, D. B., M. A. Stevens, and F. J. Watts, Flood Protection at Culvert Outlets, CER No.
69-70-DBS-MAS-FJWA, Colorado State University, Fort Collins, CO. 1970.
U.S. Army Corps of Engineers, Hydraulic Design of Flood Control Channels, Engineering and
Design Manual EMl 110-2-1601, pp.2026, July 1970.
U.S. Bureau of Reclamation, (USBR), Design of Small Dams, U.S. Government Printing Office,
Denver, CO, 1987.
Watts, F. J., Hydraulics of Rigid Boundary Basins, doctoral dissertation, Colorado State University,
Fort Collins, CO, 1968.

CHAPTER 21
HYDRAULIC DESIGN OF
FLOW MEASURING
STRUCTURES

John A. Replogle and Albert J. Clemmens

USDA-ARS Water Conservation Laboratory
Phoenix, Arizona
Clifford A. Pugh
US Bureau of Reclamation,
Denver, Colorado

21.1

INTRODUCTION

Experienced water providers and users can use this chapter as a quick review of hydraulic
principles related to water measurement and its relation to hydraulic design for environmental considerations.
The hydraulic design of flow measuring structures usually confronts the engineer with
two opportunities. One is the design of measurement structures in a retrofit situation and
the other is in original project design. The retrofit mode is usually difficult and requires
much innovation just to obtain passable function within the space and sizing limitations
and other constraints usually imposed. Because of the increasing emphasis on quantifying
flow rates and volumes in most aspects of water resource planning and management, the
retrofit applications currently dominate the design problems.
Most textbooks deal with recommending ideal installation situations and retrofit projects appear to be unable to comply without great economic impact. This too frequently
can lead to arbitrary compromises that produce poor measurement performance. Even
new installations may be limited by space requirements. This may force design decisions
into the final construction that compromise accuracy. This chapter will strive to show the
design concepts available, particularly those useful for designing both new and retrofit
installations, and will point out measurement behaviors to be expected from various compromises. This chapter suggests those deviations that cause least impact and guides the
designer to choices that may be hydraulically acceptable and still meet structural goals.
Of the numerous flowmetering methods available to the hydraulic engineer, most are
based on well-established hydraulic principles and are amenable to design manipulations
of size, shape, and response. While this aspect of flow measurement is documented in several handbooks and texts, the design and retrofit of sites to accommodate and facilitate
measurement is not as well described or is described in a scattered assortment of books
and articles.
Pipeline flows of water are usually less complicated to measure than open-channel
flows, most obviously because the flow area does not change significantly with flow rate.

Consequently, many applications of pipeline flows are held to stricter accuracy standards
than channel flows can reasonably achieve. Thus, channel flows and their measurement
are usually limited to large delivery volumes and to accuracies acceptable to the related
activities, such as sewer flows and irrigation deliveries.
The purpose of this chapter is to consolidate design information for evaluating a flow
measurement site, selecting a flow measuring system, and adapting the measuring site to
optimize measuring and other functions that may be desired from the site. Emphasis will
be on open-channel flow measurements because that is a likely need of the hydraulic engineer. Pipe flowmeters in water supply will also be discussed in lesser detail because the
major application of the many types of pipe flowmeters is well covered in the chemical
and petroleum industry literature.
Experienced readers may wish to further investigate and seek more advanced references in hydraulics and fluid mechanics. Extensive information on fluid meter theory and
detailed material for determining coefficients for tube-type meters is given in American
Society, of Mechanical Engineers (ASME) (1959, 1971) and revisited with modern
updates in books by Spitzer (1990) and Miller (1996). Brater and King (1982) have a thorough discussion of general critical depth relations and detailed relationships for most
common hydraulic flow section shapes in open channels. Bos (1989) covers a broad segment of open-channel water measurement devices.

21.2 HYDRAULIC CONCEPTS RELATED TO WATER

MEASUREMENT
21.2.1 Basic Concepts for Pipe and Channel Flows
Flow can be classified into closed conduit flow and open-channel flow. Open-channel
flow conditions occur whenever the flowing stream has a free or unconstrained surface
that is open to the atmosphere. Flows in canals or in vented pipelines that are not flowing
full are typical examples.
In hydraulics, a pipe is any closed conduit that carries water under pressure. The filled
conduit may be square, rectangular, or any other shape, but is usually round. If flow is
occurring in a conduit but does not completely fill it, the flow is not considered pipe or
closed conduit flow, but is classified as open-channel flow.
Flow rate in a pipeline responds mainly to the pressure gradient or head difference that
exists between two points along the pipeline, modified by the frictional resistance to flow
caused by pipe length, pipe roughness, bends, restrictions, changes in conduit shape and
size, the nature of the fluid flowing, and the cross-sectional area of the pipe.
In open-channel flows, the pressure gradient, or energy grade line, is controlled
mainly by the force due to gravity, which is influenced by the channel slope, resistance
from the channel wall roughness, the channel shape, and the flow area. The fluid is usually water.
Basic flow metering in both pipe flow and open channels depends on determining an
average flow velocity by some means and combining it with the flow cross-sectional area.
For open channels, a common means involves current meter measurements where metered
point velocities are applied to their applicable subareas and summed over a flow cross section. Exceptions include tracer-dilution techniques that do not require flow area or velocity. The uses of tracer techniques are applicable to special pump calibrations and some difficult channel flows (mountain streams). They are avoided for most city water distribution
systems, sewer flows and irrigation applications because of the general expense with han-

dling the equipment and doing the analysis. The most used techniques applicable to openchannel systems, including sewer flows and irrigation canal flow measurements, depend
on exploiting the special velocity properties of critical flow, as discussed in a section
21.2.3.
Continuity equation. The first basic equation for water flowing in either pipes or channels is the continuity equation, which simply states that discharge rate (volumetric flow
per unit time), Q, is equal to flow cross-sectional area, A, times flow mean velocity, V,
through the flow cross section, or
Q = AV

(21.1)

Bernoulli energy equation. Another basic equation involves energy relations and is
also applicable to both pipe and channel flows. The most familiar form is for closed pipe
flow, wherein the basic energy principles are described by the Bernoulli energy equation.
For two locations along a pipe at stations 1 and 2,(Fig. 21.1), the Bernoulli equation can
be expressed as
V2
V2
ZI + h\ + -^- = ^2 + h2 + -^- = constant

(21.2)

where the terms are expressed in length dimensions as z = the height from an arbitrary
reference plane (datum) h = the pressure head V = average velocity through the pipe
cross-section at the designated location V2/2g = the velocity head g = the gravitational
constant i ,2 = subscripts denoting the respective locations along the pipeline.
This equation is based on uniform velocity across the conduit area and no energy losses. However, in real fluid flows, nonuniform velocities exist and friction causes energy
conversion to heat. Typically, these velocities are zero at the walls and reach a maximum
profile velocity near the center of the flow. If the flow is viscous flow in a round pipe, the
flow profile is parabolic, that is, "bullet-shaped." If the velocity is fully turbulent, the bullet-shape is much flattened, with steep velocity gradients near the wall and nearly uniform
profile across the remainder of the pipe. These idealized profiles can be skewed drastically
by regulating valves, structures, conduit bends and other flow obstructions. Therefore,
application of these equations depends on knowing, or controlling, the velocity profile so
that the average velocity in the conduit cross section can be inferred.

Datum
FIGURE 21.1 Energy balance in pipe flow.

Equation (2.2) requires some adjustments to convert it to the energy equation, which
is useful in analyzing flows in pipes or open channels with a small slope (Chow, 1959).
First we introduce correction factors, GL1 and OC2, called the velocity distribution coefficients, to account for the computational expediency of using the average velocities, V1 and
V2, to compute the kinetic energy term, W2g, at the respective locations 1 and 2 along a
channel. These values for the usual range of turbulent flows in water usually range from
about 1.01 to 1.05, although for thick petroleum products in pipe flows and low velocity
flows, the value can approach a value of 2. Second, a term, hp for the loss of energy
between the two points is included. The result is
V2
v2
Z1 + H1 + a, -i = Z2 + /I2 + a2 -^ + hf
(21.3)
5
o
21.2.2 Pipe Hydraulics
Reynolds number. The behavior of flow in pipes is governed primarily by the viscosity of
the fluid. In pipeline flows, the ratio between the dynamic forces and the viscus forces is
important for defining the limits between laminar and turbulent flows and other functions
of pipe flow. This ratio is called the Reynolds number, Rn, and is defined as
Rn = ^-

(21.4)

where V = the velocity of the flow, Lc = characteristic length, typically the pipe diameter, D and v = the kinematic viscosity.
Headloss characteristics in pipes. The Reynolds number, Rn, defined above, represents the effect of viscosity relative to inertia and is used to define appropriate flow ranges
for headloss equations in pipe flow. For example, headloss is proportional to the square of
the velocity, when the velocities and pipe size combinations defined by a pipe-diameterbased Reynolds number, Rn, greater than about 1000. Most of the flows of interest in general hydraulic engineering have Reynolds numbers greater than 1000. Some exceptions
are found in drip or trickle irrigation systems common in agricultural and urban landscape
settings.
The headloss, hf for Rn greater than the minimum value of about 1000 is traditionally
expressed in terms of a friction factor,/, the pipe diameter, D, pipe length, L, and the velocity head, V2/2g, where g is the gravitational constant, and V is the average velocity, as
hf = / I

(21.5)

The value for/is usually obtained from a Moody diagram which is a graphical representation of the/value in terms of the Reynolds number, the roughness height of the pipe
wall material, e, and the pipe diameter, D. The Moody diagram is a graphical solution of
the Colebrook function
1 = -2log^
+ ^=]
Vf
I3-7
RnVf)

(21.6)

The e values range from 0.0000015 m for smooth plastic pipe to 0.00026 m for cast
iron pipe. Concrete pipe ranges from about 0.0003 m to 0.003 m (Daugherty and
Ingersoll, 1954). The equation can be readily solved by iteration techniques using a computer spreadsheet.

21.2.3 Channel Hydraulics

Hydraulic mean depth. The hydraulic mean depth, Dm [U.S. Bureau of Reclamation
(USBR), 1997] is the flow cross-sectional area, A, divided by the flow surface width, J1, or
Vm = J

(21-7>

For conduits such as pipes flowing nearly full, the surface flow width may be narrow,
and Dm may be a larger value than the physical water depth. For the usual natural channels and most canals, Dn is interchangeable with average depth. Sometimes it is simply
called the hydraulic depth (Chow, 1959).
Froude number. Open-channel flow behavior is governed primarily by gravity
forces. The ratio of the inertial forces to the gravity forces is called the Froude number,
Fr, and is defined by
Fr =

Y
(21.8)
v^:
where V the velocity of the flow, g = the gravitational constant, Dn = the hydraulic
mean depth.
The Froude number applies to most open channel flows and is used for defining model
scale ratios and estimating stable flow characteristics in open channels.
Specific energy. It is useful to define the energy equation in terms of the local channel bottom instead of an arbitrary datum. This is called the specific energy, ", and is given
by:
E = y + ^(21.9)
2g
That is, the specific energy is equal to the sum of the depth of flow y and the velocity
head (Fig. 21.2).
Critical flow and critical depth. In open channels a flow phenomenon occurs that
does not happen in closed pipe flows. The process is called critical flow. Critical flow is
defined for open-channel flows as the maximum discharge for the minimum specific energy, that is, critical flow represents the minimum combination of potential energy (depth
of flow, y) and kinetic energy (velocity head, V2/2g) for the given discharge (Chow, 1959).
The depth of flow then is the critical depth. By virtue of the continuity equation, for a constant discharge at critical flow, an increase in depth must necessarily be accompanied by
a decrease in velocity, which is called subcritical velocity. Conversely, a decrease in depth
for the same flow rate necessarily requires an increase in velocity, which is called supercritical velocity.
When critical flow occurs in an open channel it can be shown (Chow, 1959) that
V2 D
oc^ = -f

(21.10)

where Vc = mean flow velocity, g = gravitational constant, Dm = hydraulic mean depth,

and a = velocity distribution coefficient (Chow, 1959).
This can further be combined with the continuity equation, Eq. (21.1), to express the
critical flow discharge rate, Qc, as

EnergyGrade Line
Hydraulic Grade Line

FIGURE 21.2 Specific energy balance.

& = AJ%%

<2U1)

In practice, the water surface slope in a contraction is relatively steep and the precise plane of the critical flow section is not easily or reliably located. Thus, the data
for accurately evaluating the hydraulic depth, Dm, is not readily obtained. For critical
flow flumes, the flow depth is therefore not measured at this critical section, but
instead a depth is measured in the upstream channel, where the velocity head is computable or is minimal. The critical depth is then mathematically derived based on energy principles described by Bos et al. (1991). These flumes, sometimes referred to as
the computable flumes that rely on critical flow theory, will be discussed in more detail
in Section 21.7.
For maximum discharge for minimum energy, the condition described above for critical flow in open channels, it can be shown that
af = A = ^

<2U2)

where: A = the channel flow area, T = the top width of the channel flow, Dm = the
hydraulic mean depth, and Vc = the critical velocity.
Thus, the velocity head at critical flow is equal to half the hydraulic mean depth, sometimes called hydraulic depth, Dm = AIT (Chow, 1959).
From the above,
,Vc
= 1 = Fr
(21.13)
VgD m la
where Fr is the Froude number defined above. Thus, at critical flow the Froude number is
unity. Also note that the Froude number can be defined by Fr = VIV0 for velocities other
than critical.
Normal depth. Yet another depth is associated with open channel flows, the normal depth. When the flow in an open channel does not change from station to station,
the flow is said to be uniform and the bottom slope, the hydraulic grade line, and the
energy grade line are all parallel to each other. Figure 21.2 shows the condition when
the flow is not uniform.
Modular limit. If the downstream depth in a channel is too deep, the backwater will
prevent critical depth from occurring. The flow is considered to be submerged whenever
the downstream water surface exceeds the crest elevation of a channel control, such as a

weir or flume. For flumes, particularly, this submergence has little effect on critical depth,
and free flow exists until a certain limiting submergence for that particular flow module
called the modular limit is reached. At some point of submergence, the upstream flow
depth is affected, and the modular limit is exceeded, and free flow does not occur. The
modular limit is defined as that limiting submergence ratio, and is based on the ratio of
the downstream depth to upstream depth. The modular limit occurs when the downstream
backwater causes more than 1 percent change in the calculated discharge in a particular
flow module, or device (Bos, 1989). When the modular limit is exceeded, the flow is
called nonmodular.
21.2.4 Energy Balance Relationships in Channels
Hydraulic problems concerning fluid flow are commonly described in terms of conserving kinetic and potential energy, and are conveniently expressed using the classical
Bernoulli equation in combination with the Continuity equation. The applications of these
equations are generally well documented, particularly for pipe flows, in texts and handbooks and are not repeated here (Brater and King, 1982; Miller, 1996). The case for open
channels is less complete, but is given considerable treatment in Brater and King (1982),
Chow (1959), and Herschy (1985). The computational uncertainties evolve from the
effects of friction and viscosity that distort the classic assumptions of a uniform velocity
profile across the fluid stream. When accountings for friction and flow profile are successfully applied, the results for discharge computations are usually good to excellent for
both pipes and open channels (Bos et al., 1991).
Headloss characteristics in channels. In terms of frictional headlosses, the wetted
perimeter, Pw, of the flow is important. Hydraulic radius, Rh, is defined as the area of the
flow section, A, divided by the wetted perimeter, Pw, or
Rh =
p

(21.14)

Conversely, the wetted perimeter times the hydraulic radius is equal to the area of an
irregular flow section. The hydraulic radius of a channel can be compared to the radius of
a pipe, r, with a cross-sectional area A = nr2 and a circumference or wetted perimeter Pw
= 2 nr. Under these conditions, the hydraulic radius compares to the pipe radius, and to
the pipe diameter, D, as
r D
R
=
(2L15)
=2
^
h
The Manning's formula. Canal and stream discharge rates are usually estimated with
use of the Manning's formula. Many open-channel flow equations have been proposed,
but the most used is the Manning's formula. This expression is partly rational and uses an
empirical coefficient, rc, that is used in both the SI and American unit systems. In general
form it becomes
v=^LR2i3Sm

(2L16)

where V = average velocity, n = the Manning's roughness coefficient, Rh = the hydraulic

radius, S6 = the energy-line slope, and Cm = conversion of units: 1.0 for metric units and
1.486 for American units.
The factor Se is the slope of the energy line. Note that the bed slope of the channel, S09
and the slope of the water surface, Sw, are not to be used. These parameters are, however,

equal to Se when uniform flow, with the resulting normal depth occurs. As defined above,
normal depth occurs when a channel flow approaches uniformity from station to station
along the channel (Chow, 1959).
For design purposes, the n value for concrete lined canals is usually about 0.014. A
good finish can lower it to 0.012, while concrete in poor condition and channels constructed with shot crete or gunite, usually have n values from 0.016 to 0.018. In some
instances, concrete lined canals, with significant algae growth, have experienced n values
as high as 0.032. This latter value approaches the values usually experienced with unlined
channels, 0.03-0.04. Thus, for reliable application, the use of Manning's formula requires
field experience and on-site inspection of the channel being computed.
21.2.5 Modeling Characteristics for Open Channels
For flowing water in open channels, fluid friction is a factor as well as gravity and inertia. This would seem to present a problem for hydraulic scale modeling, because both
dynamic and kinematic similarity are difficult to achieve simultaneously. Fortunately for
most open-channel flows, there is usually fully developed turbulence. Thus, the fluid friction losses are nearly proportional to V2, and are nearly independent of Reynolds number,
Rn, with rare exceptions.
This means that in open-channel flows, inertia and gravity forces dominate over viscous forces (associated with pipe flows) and are a function of the Froude number, Fn,
alone. Geometric similarity between a model and a prototype then provides kinematic
similarity. For kinematic similarity the ratios of the respective velocities are everywhere
the same. The velocity ratio, Vr, is the velocity in the prototype, Vp, divided by the velocity in the model, Vn, or
Vr = ^(21.17)
m
For Froude modeling, and from the definition for Fn, we note that V is proportional
to the square-root of a length, L (for open channels we used the hydraulic depth, D1n)
with the gravitational constant, g, assumed to be constant. Thus, the above equation can
be written as
"^LT = Y
where Lr = the length ratio between prototype and model dimensions, Lp: Lm

(2L18)

Because the velocity varies as VZ^ and the cross-sectional area as L2 it follows
that
Qp'Qm = LT 1

(21-19)

O
(J Y/2
-SL = U
XZm \.LmJ

(21.20)

or

This equation is valid when all the physical structure dimensions and the heads are of
the same ratio. For example, it can be used to convert a flume rating for one size to that
of a similar flume of another size. Scale modeling works best for determining calibrations
in a range of Lp:Lm less than about 10:1, although ranges exceeding 50:1 have sometimes
been used for studying special situations.

21.3

BASIC PRINCIPLES OF WATER MEASUREMENT

Flow is usually measured by determining an average flow velocity and using the flow area
to compute the volume discharge. Flow meters then have the function of detecting this
velocity and combining it with the physical information of the conduit to produce a useable readout. This is easily demonstrated for closed conduits. Propeller meters, ultrasonic
meters, laser-Doppler velocimeters, electromagnetic meters, Venturi meters, and orifice
meters all are based on inferring a basic velocity measurement applied to a flow area for
a discharge rate.
For open channels, many flumes depend on determining the velocity based on energy
principles of critical flow. Weirs are usually described in terms of orifice flow integrated
over the weir width and the crest depth. Again these are basically velocity expressions for
flow through a defined area.
Dilution techniques applicable to both closed pipe and open channel flows depend on
detecting the amount of fluid added to a known starting amount of tracer material. The
dilution ratio determines the discharge ratio, in the case of constant injection of a tracer.
The tracer may be a chemical or even injected heat or heated fluid.
Electromagnetic meters depend on generating voltages by flowing a conductive fluid,
usually water, through a magnetic field to produce a velocity indication.
21.3.1 Water Meter Classification
Flow measuring devices are commonly classified into those that are rate meters and measure discharge rate as the primary reported indication and those that are quantity meters
and measure volume as the primary indication. The latter include weighing tanks and
batch volume tanks and are used mostly in laboratory settings as flow rate standards.
Devices in either of these broad classes can again be divided according to the physical
principle that is used to detect that primary indication (ASME, 1959). The meter part that
interacts with the flow to produce the primary indication is referred to as the primary
device. This interaction exploits one or more of a few physical principles, such as pressure force, energy conversion, weight, electrical properties, mixing properties, sonic properties, and so on, to generate a signal. Primary devices are thus limited in number and variety. Secondary devices convert the primary interaction into useable readout. These secondary devices are numerous and relatively unlimited in configuration and variety. The
function of one class can be converted into the response of the other with suitable secondary devices.
Some water measuring devices particularly suitable to municipal water supply, wastewater treatment, agricultural irrigation, and drainage applications are the historical rate
meters that are treated in most hydraulic text. These include (1) weirs, (2) flumes, (3) orifice meters, and (4) Venturi meters.

Head, h, or upstream depth, commonly is used for the open channel devices such as
flumes and weirs. Either pressure, /?, head, /i, or differential head, A/z, or differential pressure, Ap, is used with tube-type devices, such as Venturi meters and orifice meters.
Venturi meters in pipelines and long throated flumes in open-channel flows are examples where the energy principles and the flow accountings mentioned above give good to
excellent computational results with minor dependency on empirical coefficients (Bos,
1989; Bos et al., 1991).
21.3.2 Installation Requirements
Special difficulties arise in applying velocity profile and friction accountings when insufficient pipe or channel exists upstream from a flow measuring device. This is needed to
ensure that predictable and acceptable velocity profiles are presented to the meter.
Frequently pipe or channel lengths can be significantly shortened by special structural
flow conditioners. These structural measures then become a design option. Some of these
are discussed below and in section 21.3.3
Designs for pipe discharges are well described in textbooks and in standard handbooks. The design difficulties center around selecting appropriate metering candidates for
accomplishing the measuring function and in providing an appropriate environment for
economical, accurate, and serviceable operation.
In the case of pipe flows, recommended straight pipe lengths, in terms of pipe diameter, are to be provided upstream of the meter to assure reasonable operating accuracy. These
lengths depend on the flow pattern presented to the meter primarily caused by valves and
pipe elbows upstream from the meter. The number and orientation of elbows greatly influence the circulation patterns and flow profile distortions presented to the meter.
Open-channel flow water measurement generally requires that the Froude number of
the approach flow be less than 0.5 to prevent wave action that would hinder or possibly
prevent an accurate head determination.
Energy concepts are used to describe Venturi meters in pipe flows based on the
Bernoulli equation in which part of the pipe forms a contracted throat that necessarily
changes the flow velocity and hence converts some of the static pressure to velocity head.
The decrease in static pressure is the basis for flow detection. A similar concept can be
applied to open channels. A historical version is the so-called Venturi flume (Brater and
King, 1982) that detects the change in water surface elevation between an upstream station and in a contracted section. However, this small change is difficult to accurately
detect, so the direct concept is not used. Rather, contractions are designed to be severe
enough to force critical flow velocities in the contracted section. Thus, only an upstream
head is needed to define the flow energy and flow area which can be converted to discharge rate. These are generally called critical-flow flumes. The flow condition where
only one head measurement is needed is called freeflow.
The critical-flow flumes themselves consist of those called long-throated flumes that
force parallel flow in the contracted, or control, section, called the throat, and those that
have curvilinear flow in the throat and are called short-throated flumes. The limiting
throat control section is the sharp-crested weir consisting of a thin plate. Thus, for
flumes and weirs one unique head value exists for each discharge, simplifying the calibration procedure.
However, if the downstream flow level submerges critical depth enough to affect the
upstream reading, the modular limit is exceeded, and free flow does not occur. When
exceeded, separate calibrations at many levels of submergence are then required, and two
head measurements are needed to measure flow. This condition generally is to be avoided in meter site design because it reduces the accuracy of the measurement and increases

the difficulty of flow determination. The modular limit for sharp-crested weirs, in practice, is less than zero, requiring full clearance of the overfall nappe of at least 3 cm, while
short-throated flumes can usually tolerate 65 percent to 70 percent submergence. Longthroated flumes can tolerate from 70 percent to 90 percent depending on flow conditions
and flume size.
Designing flumes for submerged flow beyond the modular limit decreases the accuracy of the flow measurement. Sometimes flumes and weirs can be overly submerged unintentionally by poor design, construction errors, structural settling, attempts to supply
increased delivery needs with increasing downstream heads, accumulated sediment
deposits, or weed growths. Sometimes use of the submerged range beyond the modular
limit is an economic compromise.
Approach flow conditions for pipes. Water measurement devices are generally calibrated with certain approach flow conditions. The same approach conditions must be
attained in field applications of measuring devices. Poor flow conditions in the area just
upstream of the measuring device can cause large discharge indication errors. For open
channels, the approaching flow should generally be subcritical. The flow should be fully
developed, mild in slope, and free of curves, projections, and waves.
Pipeline meters commonly require 10 or more diameters of straight pipe approach.
Fittings and combinations of fittings, such as valves and bends, located upstream from a
flow meter can increase the number of required approach diameters. Several references
(ASME, 1971; ISO, 1991) give requirements for many pipeline configurations and
meters. These are discussed in detail by Miller (1996).
Flow conditioning options. Many installations, especially in retrofit situations, do
not provide for sufficient lengths of straight pipe to remove velocity profile distortions
and swirl to an acceptable level. Therefore, the designer may need to use flow conditioners in combination with straight pipe lengths. Swirl sensitivity varies widely. Some meters
are particularly sensitive to swirl, such as the propeller and turbine meters. Magnetic flow
meters are somewhat less sensitive to radial velocities than single-path ultrasonic flow
meters. Venturi meters are less sensitive than orifice meters. For a swirl angle of 20 , the
discharge coefficient changes by about 1 percent for a Venturi meter with p = 0.32 ((3 is
the ratio of meter throat diameter to the pipe diameter) and about +10 percent for a similar orifice. Thus a swirl can increase the discharge through an orifice for the same differential head reading (Miller, 1996).
In pipeline flows, contractions can produce a central jet and also increase an incoming
swirl, while expansions tend to slow swirls and produce enough secondary flow to restore
flow profiles to some semblance of acceptability. These characteristics can modify the
straight pipe lengths needed or the type of flow conditioner to recommend (Miller, 1996).
Rough pipes also tend to reduce a swirl.
For flows, such as that encountered in sewage discharges and irrigation pipeline deliveries that originate from open channels, many of the tube-bundle types of flow conditioners can gather trash and cause maintenance problems. Many meter providers in these situations use fins or vanes that protrude from the wall and have sloped upstream edges that
shed trash. The vanes protrude about one-fourth of the pipe diameter into the flow, leaving the center core of the flow open. While these vanes can vary in number and length,
the logic being that the fewer the vanes the longer they should be in the direction of flow,
common configurations are four vanes that are about two or three pipe diameters long.
Vanes in themselves do not condition wall jets well. Field experience, has shown that troublesome flow profiles can be conditioned significantly by inserting an orifice into the

pipe. The orifice diameter is about 90 percent of the pipe diameter and is used to control
wall jets and force them to mix with the general flow. The orifice in itself tends to crossmix the jets and would appear to reduce spin. However, if the jets are symmetrical and an
initial swirl exists, orifices tend to increase the swirl. Inserting an orifice appears to be
supported by recent recommendations of Miller (1996) where it is stated: "To achieve a
fully developed profile, it is important that the flow be blocked or restricted close to the
wall, with the central core having the larger flow area."
The addition of vanes when space permits is recommended. Because orifices, in general, tend to force the flow to the pipe center while increasing spin, it appears best to place
the vanes upstream from the orifice. If they are placed downstream, the spin not only may
be increased, but the spinning central flow may not be touched by the vanes.
21.3.3

Examples of Flow Conditioning in Field Situations

Flow conditioning in an irrigation delivery pipeline. As mentioned previously, measuring devices frequently must be installed in flow situations that are less than optimal. A field
example occurred in Arizona where a large pipe was used as an outlet to a secondary canal
and a single-path ultrasonic meter placed in it was subjected to flow profile distortions. The
pipe was about 0.75 m in diameter and delivered approximately 400 L/s. The flow rate
readout was unstable, with fluctuations varying by about 15 percent. The problem appeared
to be caused by slowly spiraling flow induced by the bottom jet from a partly open pipe
inlet gate and a 45 elbow. This is similar to two closely spaced pipe elbows that are not in
the same plane, which can cause a spiral flow pattern (ASME, 1971).
A successful attempt to modify the jet and cause it to cross mix so that the jet effects
and the strength of the spiral flow were reduced, was accomplished by inserting a large Pratio orifice in the pipe (Fig. 21.3). This consisted of an annular metal ring with the outside radius approximately that of the pipe and an inside diameter about 10 percent less, or
an orifice with (3 = 90 percent. The orifice was installed about three diameters downstream from the elbow. The slight increase in headless was compensated by increasing the
upstream gate opening. The orifice can be constructed by cutting notches from an appropriately sized piece of angle iron or aluminum and bending it to a polygon that approxi-

Control O gate
Ultrasonic flow
meter

Orifice plate
with large
opening

Spiral flow

Improved profile
FIGURE 21.3 An orifice plate with a large opening is used to condition a flow profile.

mates the circle diameter of the pipe interior. Some leakage around the ring is acceptable.
For propeller meters, additional vanes projecting from the walls may be needed to further
reduce spiral flow. These vanes would be placed upstream from the orifice. In this installation, the fluctuation was reduced to within about 3 percent.
Flow conditioning in channels. By analogy and using a minimum of 10 pipe diameters of a straight approach channel, open channel flow would require 40 hydraulic radii of
straight, unobstructed, unaltered approach, based on the calculation of hydraulic radius for
circular pipes being equal to one-fourth the pipe diameter, (Eq. 21.15). This would translate for very wide channels into approximately 40 times the flow depth. For narrow channels that are as deep as they are wide, this would compute to be about 13 channel depths
or top widths.
Other recommendations on approach channel criteria are presented by Bos (1989)
and USBR (1997). Major features of that criteria follow:
If the control width is greater than 50 percent of the approach channel width, 10 average approach flow widths of straight, unobstructed approach are required.
If the control width is less than 50 percent of the approach width, 20 control widths of
straight, unobstructed approach are required.
If upstream flow is below critical depth, a jump should be forced to occur. In this case,
30 measuring heads of straight, unobstructed approach after the jump should be provided.
If baffles are used to correct and smooth out approach flow, then 10 measuring heads
(10 /I1) should be placed between the baffles and the measuring station.
Approach flow conditions should be continually checked for deviation from these conditions as described in Bos (1989) and USBR (1997).
The baffles described above can become unacceptable maintenance problems in open
channels. Some field expediences are therefore described that have been found to work in
specific instances, but have not been studied for assured design generalizations. Nevertheless,
these constructions are but small extensions to currently accepted practices in pipe flows.
Applications for open-channel flow conditioners include abrupt channel turns, sluice
gate outflows, and channels downstream from a hydraulic jump. The abrupt turns may
benefit from floor and wall mounted vanes or fins. Based on pipe flow experience, and
assuming the channel is half of a closed conduit, these fins or vanes would probably be
about 10 percent to 15 percent of the channel depth.
As in pipe flow, wall jets that can develop downstream from sluice gates appear to
need treatment. This can be in the form of a structural angle bolted on the channel floor
and up the walls. Suggested size, based on the pipe flow analogy, is for the angle to be
about 5 percent into the channel flow depth. Whether the sidewalls need larger angles
when the channels are wide has not been tested.
21.3.4 Wave Suppression
Of special concern in open channels is wave suppression downstream from a sluice gate,
hydraulic jump, or an abrupt turn. Thus, the flow conditioners in channels have the additional task not present in pipe flows of surface wave suppression. Excessive waves in irrigation canals make reading sidewall gages difficult. These waves are usually caused by a
jet entry from a sluice gate or by a waterfall situation. The unstable surface can be 10-20
cm high and extend for tens of meters downstream.

Solidly attached
timber or concrete
FLOW
FIGURE 21.4 Wave suppresor design (From USBR, 1997).

Wave suppression in canals. A surface wave suppressor was tested by Schuster as

reported in USBR (1997). It basically was a constructed roof over the canal for a distance
equal to about four times the flow depth. The roof structure is inserted into the flow about
one-third the flow depth. All flow is forced to pass under the structure. Wave suppression is
between 60 percent and 93 percent (Fig. 21.4). For canals that usually flow at one level, this
wave-suppression method is appropriate. The wave suppressor shown in Figure 21.4 has
been successfully used in both large and small channels (USBR, 1997). An important aspect
is that the structure is fixed and not allowed to float. Floating suppressors are not effective.
Successful field applications of wave suppressors include some installations in trapezoidal irrigation channels, with 1:1 side slopes and 60-cm bottom width. They were flowing about 400 L/S at about 45 cm deep. While the velocity was not high, about 0.8 m/s,
the agitation from a flow entry gate was producing waves about 15 cm high. The suppressor "roof was only about 60 cm in the direction of flow, and penetrated the flow by
about 15 percent.
Another version that has worked in small channels is illustrated in Figure 21.5.
This can work with a single cross-member if the flow is usually at a fixed discharge
rate and becomes similar to the suppressor described above. In severe jet cases an
additional floor sill, about 10 percent of the flow depth in height, has been used successfully.
The length of the roof in the flow direction has not been well studied, but field observations seem to support a length greater than two lengths of the surface wave, if that can
be estimated, otherwise, use two to four times the maximum flow depth as described above.
To suppress waves in canals that do not always flow at the same depth, a staggered set
of baffles may help (Replogle, 1997). Because these will be submerged part of the time,
they must have a thickness that overlaps slightly to accommodate the vertical depth of
interest. To avoid obstructing the channel severely, these baffles probably should not
obstruct more than about 20 percent of the channel at any particular location. Staggering

FIGURE 21.5 Wave suppressor for variable-depth flows in a canal. (From

Replogle, 1997)

them as shown in Fig. 21.5 would accomplish this without excessive obstruction.
Rounding the upstream edges will help shed trash, but may be less effective in suppressing waves. Observe in the sequence of drawings in Fig. 21.5 that the staggering is upward
in the downstream direction. Note that the next baffle slightly overlaps the horizontal flow
lines so that flow passing over the top of one baffle is not allowed to free-fall and start
another wave. Fig. 21.5 a-c illustrate the general behavior as the flow becomes less deep.

21.4

MEASUREMENTACCURACY

Accurate application of water measuring devices generally depends upon standard designs
or careful selection of devices, careful fabrication and installation, good calibration data
and adequate analysis. Also needed is proper user operation with appropriate inspection
and maintenance procedures. During operation, accuracy requires continual verification
that all measuring systems, including the operators, are functioning properly. Thus, good
training and supervision are required to attain measurements within prescribed accuracy
bounds. Accuracy is the degree of conformance of a measurement to a standard or true
value. The standards are selected by users, providers, governments, or compacts between
these entities. All parts of a measuring system, including the user, need to be considered
in accessing the system's total accuracy.
As mentioned above, a measurement system usually consists of a primary element,
which is that part of the system that creates what is sensed, and is measured by a secondary element. For example, weirs and flumes are primary elements. A staff gage is a
secondary element.
Designers, purchasers, and users of water measurement devices generally rely on standard designs and manufacturers to provide calibrations and assurances of accuracy. A few
water users and providers have the facilities to check the condition and accuracy of flow
measuring devices. These facilities have comparison flow meters and/or volumetric tanks
for checking their flow meters. These test systems are used to check devices for compliance with specification and to determine maintenance needs. However, maintaining facilities such as these is not generally practical.
Various disciplines and organizations do not fully agree on some of the definitions
related to measuring device specifications, calibration, and error analysis. Therefore, it is
important to verify that a clear and mutual understanding of the specifications, calibration
terminology, and the error analysis processes is established when discussing these topics
with others.
21.4.1 Definitions of Terms Related to Accuracy
Error. Error is the deviation of a measurement, observation, or calculation from the
truth. The deviation can be small and inherent in the structure and functioning of the system and be within the bounds or limits specified. Lack of care and mistakes during fabrication, installation, and use can often cause large errors well outside expected performance bounds. Because the true value is seldom known, some investigators prefer to use
the term uncertainty. Uncertainty describes the possible error or range of error which
may exist. Investigators often classify errors and uncertainties into spurious, systematic,
and random types.
Precision. Precision is the ability to produce the same measurement value within
given accuracy bounds when successive readings of a specific quantity are measured.

Precision represents the maximum departure of all readings from the mean value of the
readings. Thus, a single observation of a measurement cannot be more accurate than
the inherent precision of the combined primary and secondary precision. It is possible to
have good precision of an inaccurate reading. Thus, precision and accuracy differ.
Spurious errors. Spurious errors are commonly caused by accident, resulting in false
data. Misreading and intermittent mechanical malfunctions can cause discharge readings
well outside of expected random statistical distribution about the mean. Spurious errors
can be minimized by good supervision, maintenance, inspection, and training.
Experienced, well-trained operators are more likely to recognize readings that are significantly out of the expected range of deviation. Unexpected spiral flow and blockages of
flow in the approach or in the device itself can cause spurious errors. Repeating measurements does not provide information on spurious error unless repetitions occur before and
after the introduction of the error. On a statistical basis, spurious errors confound evaluation of accuracy performance.
Systematic errors. Systematic errors are errors that persist and cannot be considered
random. Systematic errors are caused by deviations from standard device dimensions,
anomalies to the particular installation, and possible bias in the calibration. Systematic
errors cannot be removed or detected by repeated measurements. They usually cause persistent error on one side of the true value. The value of a particular systematic error for a
particular device may sometimes be considered as a random error. For example, an installation error in the zero setting for a flume might be + 1 mm for one flume and 2 mm for
another. For each flume the error is systematic, but for a number of flumes it would be a
random error.
Random errors. Random errors are caused by such things as the estimating required
between the smallest division on a head measurement device and water surface waves at
a head measuring device. Loose linkages between parts of flowmeters provide room for
random movement of parts relative to each other, causing subsequent random output
errors. Repeated readings decrease the average expected error resulting from random
errors by a factor of the square root of the number of readings.
Total error. Total error of a measurement is the result of systematic and random errors
caused by component parts and factors related to the entire system. Sometimes, error limits of all component factors are well known. In this case, total limits of simpler systems
can be determined by computation (Bos et al., 1991). In more complicated cases, it may
be difficult to confidently combine the limits. In this case, a thorough calibration of the
entire system as a unit can resolve the difference. In any case, it is better to do error analysis with data where entire system parts are operating simultaneously and compare discharge measurement against an adequate discharge comparison standard.
Expression of errors. Instrument errors are usually expressed by manufacturers as
either a percent of reading or a percent of full scale. The secondary devices based on electronic outputs are more frequently expressed in terms of percent full scale. The designer
must be aware that a probable error value of say 1 percent full-scale can exceed 10
percent for small value readings on the output device. When used with weirs, for example, the head reading of hl 5 in the weir equation can increase this 10 percent head measurement error to a 15 percent flow measurement error.

21.4.2 Terms Related to Measurement Capability

Linearity. Linearity usually means the maximum deviation in tracking a linearly varying quantity, such as measuring head, and is generally expressed as percent of full scale.
Discrimination. Discrimination is the number of decimals to which the measuring
system can be read. Precision is no better than the discrimination.
Repeatability. Repeatability is the ability to reproduce the same reading for the same
quantities. Thus, it is related to precision.
Sensitivity. Sensitivity is the ratio of the change of a secondary measurement, such
as head, to the corresponding change of discharge.
Range and Rangeability. Range is fully defined by the lowest and highest value that
the device can measure without damage and comply within a specified accuracy. The upper
and lower range bounds may be the result of mechanical limitations, such as friction at the
lower end of the range and possible overdriving damage at the higher end of the range.
Range can be designated in other ways: (1) as a simple difference between maximum discharge (Qn^x) and minimum discharge (Qmit), (2) as the ratio (QnJQm1n), called rangeability, and (3) as a ratio expressed as \'>(QminIQmw)' Neither the difference nor the ratios fully
define range without knowledge of either the minimum or maximum discharge.
Additional terms (hysteresis, response, lag, rise time). Additional terms related more
to dynamic variability might be important when continuous records are needed or if the
measurements are being sensed for automatic control of canals and irrigation. Hysteresis is
the maximum difference between measurement readings of a quantity established by the
same mechanical set point when set from a value above and reset from a value below.
Hysteresis can continually get worse as wear of parts increases friction or as linkage freedom increases. Response has several definitions in the instrumentation and measurement
field. For water measurement, one definition for response is the smallest change that can
be sensed and displayed as a significant measurement. Lag is the time difference of an output reading when tracking a continuously changing quantity. Rise time is often expressed
in the form of the time constant, defined as the time for an output of the secondary element
to achieve 63 percent of a step change of the input quantity from the primary element.
21.4.3 Comparison Standards
Water providers may want, or may be required, to have well-developed measurement programs that are highly managed and standardized. If so, water delivery managers may wish
to consult American Society for Testing Materials Standards (ASTM, 1988), Bos (1989),
International Organization for Standardization (ISO, 1983: ISO, 1991), and the National
Handbook of Recommended Methods for Water Data Acquisition (USGS, 1980).
Research laboratories, organizations, and manufacturers that certify measurement
devices may need to trace accuracy of measurement through a hierarchy of increasingly
rigid standards.
The lowest standards in the entire hierarchy of physical comparison standards are
called working standards, which are shop or field standards used to control quality of production and measurement. These standards might be gage blocks or rules used to ensure
proper dimensions of flumes during manufacturing or devices carried by water providers
and users to check the condition of water measurement devices and the quality of their

output. Other possible working standards are weights, volume containers, and stopwatches. More complicated devices are used, such as surveyors' levels, to check weir staff gage
zeros. Dead weight testers and electronic standards are needed to check and maintain
more sophisticated and complicated measuring devices, such as acoustic flow meters and
devices that use pressure cells to measure head.
For further measurement assurance and periodic checking, water users and organizations may keep secondary standards. Secondary standards are used to maintain integrity
and performance of working standards. These secondary standards can be sent to government laboratories, one of which is the National Bureau of Standards in Washington, D.C.,
to be periodically certified after calibration or comparison with accurate replicas of primary standards. Primary standards are defined by international agreement and maintained
at the International Bureau of Weights and Measurements in Paris, France.
Depending on accuracy needs, each organization should trace their measurement performance up to and through the appropriate level of standards. For example, turbine
acceptance testing, such as in the petroleum industry, might justify tracing to the primary
standards level.

27.5 SELECTION OF PRIMARY ELEMENTS OF WATER

MEASURING DEVICES
21.5.1 General Requirements
Design considerations involve the selection of the proper water measurement device for
a particular site or situation. Site-specific factors and variables must be considered in
extended detail. Each system has unique operational requirements and installation concerns. Knowledge of the immediate measurement needs and reliable estimates on future
demands of the proposed system is advantageous. Possible selection constraints may be
imposed by laws and compact agreements and should be consulted before selecting a
measurement device. Contractual agreements for the purchase of pumps, turbines, and
water measuring devices for water supply, sewage and drainage districts often dictate the
measurement system required for compliance prior to payment. These constraints may
be in terms of accuracy, specific comparison devices, and procedures. Bos (1989) provides an extensive and practical discussion on the selection of open channel water measurement devices. Miller (1996) provides a recent compilation of selection criteria for
pipe flow-meters suited to liquids and steam and other gas flows. Bos (1989) provides a
selection flow chart and a table of water measurement device properties to guide the
selection process for the open channel devices. Miller (1996) describes each meter in
detail for the pipe systems, but is more general in leaving the selection to the designer.
Because the design engineers for civil engineering projects are most likely to be dealing with irrigation water supply, waste water, or drainage and flood flows, the emphasis
is placed on the measuring systems deemed most appropriate to these processes. Large
closed-pipe systems for water supplies are frequently encountered, so installation situations appropriate to these will also be included. Gas flows, including steam, are more likely to be encountered by mechanical and chemical engineers and those readers are referred
to Miller (1996) and ASME (1959, 1971).

21.5.2 Types of Measuring Devices

System operators for water supply, drainage, and waste water commonly use many types
of standard water measuring devices, usually in open channels with limited applications
in closed conduits. Particularly prominent uses of open channel devices are found in irrigation delivery systems and farm distribution systems, although these measuring devices
are frequently used for sewer flows and even flood flows. However, the latter two areas
of application are frequently more difficult because of the likelihood of heavy bed loads
and floating debris.
In pipe flowmeters, the most commonly installed devices in industry are the orifice
meters, accounting for up to 80 percent of all industrial meters (Miller, 1996). Venturi
meters and flow tubes provide much of the remainder. In absolute numbers, the household meters, based on various technologies from nutating disks to paddle-wheel turbines, dominate.
For open-channel flows, weirs, flumes, submerged and free orifices, and current
meters dominate the flow measuring methods. Pipe flow meters, propeller and turbine,
acoustic, magnetic, and vortex-shedding meters are used on large water supply wells such
as those used in irrigation and municipal water supply. Differential head meters, such as
orifice meters, Venturi meters, and flow tubes, are also used in these applications. The
meters considered herein are
1. Open-channel flow devices
a. Current metering (cup, propeller, and electromagnetic probes)
b. Weirs
c. Flumes
d. Acoustic (transonic and Doppler)
e. Tracers
f. Miscellaneous
2. Pipe flow devices
a. Differential head meters
b. Acoustic (transonic and Doppler)
c. Tracers
d. Turbine/propeller/other insert mechanical
e. Vortex-shedding
f. Miscellaneous
The main factors which influence the selection of a measuring device include
(USBR, 1997):
a. Accuracy requirements
b. Cost
c. Legal constraints
d. Range of flow rates
e. Head loss
f. Adaptability to site conditions
g. Adaptability to variable operating conditions
h. Type of measurements and records needed

I. Operating requirements
j. Ability to pass sediment and debris
k. Longevity of device for given environment
1. Maintenance requirements
m. Construction and installation requirements
n. Device standardization and calibration
o. Field verification, troubleshooting, and repair
p. User acceptance of new methods
q. Vandalism potential
r. Impact on environment
Accuracy requirements. The desired accuracy of the measurement system is an important consideration in the selection of a measurement method. Most water measurement
installations, including the primary and secondary devices, can produce accuracies of 5
percent. Some systems are capable of 1 percent under laboratory settings. However, in
the field, maintaining such accuracies usually requires considerable expense or special
effort in terms of construction, secondary equipment, calibration in-place, and stringent
maintenance. Selecting a device that is not appropriate for the site conditions can result in
a nonstandard installation of reduced accuracy, sometimes exceeding 10 percent.
Accuracies are frequently reported that relate only to the primary measurement method
or device. However, many methods require secondary measurement equipment that produces the actual readout. This readout equipment typically increases the overall error of
the measurement.
Cost. The cost of the measurement method includes the cost of the device itself, the
installation, secondary devices, operation, and maintenance. Measurement methods vary
widely in their cost and in their serviceable life span. Measurement methods are often
selected based on the initial cost of the primary device with insufficient regard for the
additional costs associated with providing the desired records of flows over an extended
period of time.
Legal constraints. Governmental or administrative water board requirements may
dictate the water measurement devices or methods. Water measurement devices that
become a standard in one geographic area may not necessarily be accepted as a standard
elsewhere. In this sense, the term "standard" does not necessarily signify accuracy or
broad legal acceptance. Many water agencies require certain water measurement devices
used within their jurisdiction to conform to their standard for the purpose of simplifying
operation, employee training, and maintenance.
Flow range. Many measurement methods have a limited range of flow conditions for
which they are applicable. This range is usually related to the need for certain prescribed
flow conditions which are assumed in the development of calibrations. Large errors in
measurement can occur when the flow is not within this range. For example, using a bucket and stopwatch for large flows that engulf the bucket is not very accurate. Similarly,
sharp-edged devices, such as sharp-crested weirs, typically do not yield good results with
large channel flows. These are measured better with large flumes or broad-crested weirs,
which in turn are not appropriate for trickle flows.
Certain applications have typical flow ranges. Irrigation supply monitoring seldom
demands a low-flow-to-high-flow range above about 30, while this range on natural
stream flows may exceed 1000.

In some cases, secondary devices can limit the practical range of flow rates. For example, with devices requiring a head measurement, the accuracy of the head measurement
from a visually read wall gage may limit the measurement of low flow rates. For some
devices, accuracy is based on percent of the full-scale value. While the resulting error may
be well within acceptable limits for full flow, at low flows, the resulting error may become
excessive, limiting the usefulness of such measurements. Generally, the device should be
selected to cover the desired range. Choosing a device that can handle an unnecessary
large flow rate may result in compromising measurement capability at low flow rates, and
vice versa. This choice depends on the objective of the measurement. For example, in irrigation practice, usually choose a device that can measure the most common flow range at
the expense of poorly measuring extremes, such as flood flows. For urban drainage, the
flood peak may be important.
For practical reasons, different accuracy requirements for high and low flows may be
chosen. This is reasonable when an annual total is the primary goal and the low flows contribute a small percentage to that total. Also, if the inaccurate low flow readings are truly
a random error then this error approaches zero with large accumulations of readings. Thus
the designer needs to know if management decisions are made from individual readings
or from long term averages.
Headloss. Most water measurement devices require a drop in head. On retrofit installations, for example, to an existing irrigation project, such additional head may not be
available, especially in areas that have relatively flat topography. On new projects, incorporating additional headloss into the design can usually be accomplished at reasonable
cost. However, a tradeoff usually exists between the cost of the device and the amount of
headloss. For example, acoustic flow meters are expensive but require little headloss.
Sharp-crested weirs are inexpensive but require a relatively large headloss. The head loss
required for a particular measuring device usually varies over the range of discharges. In
some cases, head needed by a flow measuring device can reduce the capacity of the channel at that point.
Adaptability to site conditions. The selection of a flow measuring device must
address the site of the proposed measurement. Several potential sites may be available for
obtaining a flow measurement. The particular site chosen may influence the selection of
a measuring device. For example, discharge in a canal system can be measured within a
reach of the channel or at a structure such as a culvert or check structure. A different
device would typically be selected for each site. The device selected ideally should not
alter site hydraulics so as to interfere with normal operation and maintenance. Also, the
shape of the cross-sectional flow area may favor particular devices.
Adaptability to variable operating conditions. Flow demands for most water delivery systems usually vary over a range of flows and flow conditions. The selected device
must accommodate the flow range and changes in operating conditions, such as variations
in upstream and downstream head. Weirs or flumes should be avoided if downstream
water levels can, under some conditions, cause excessive submergence. Also, the information provided by the measuring device should be conveniently useful for the operators
performing their duties. Devices that are difficult and time consuming to operate are less
likely to be used and are more likely to be used incorrectly.
In some cases, water measurement and water level or flow control are desired at the
same site. A few devices are available for accomplishing both (e.g., constant-head orifice,
vertically movable weirs, and Neyrpic flow module; Bos, 1989). However, separate measurement and control devices are typically linked for this purpose and usually can exceed
the performance of combined devices in terms of accuracy and level control, if care is

exercised to assure that the separate devices are compatible and achieve both functions
when used as a system.
Type of measurements and records needed. An accurate measure of instantaneous
flow rate is useful for system operators in setting and verifying flow rate. However,
because flow rates change over time, a single (instantaneous) reading may not accurately
reflect the total volume of water delivered. Where accounting for water volume is desired,
a method of accumulated individual flow measurements is needed. Where flows are
steady, daily measurements may be sufficient to infer total volume. Most deliveries, however, require more frequent measurements. Meters that accumulate total delivered volume
are desirable where water users take water on demand. Totalizing and automatic recording devices are available for many measuring devices. For large structures, the cost for
water-level sensing and recording hardware is small relative to the structure cost. For
small structures, these hardware costs remain about the same and thus become a major
part of the measurement cost, and may often exceed the cost of the primary structure
itself.
Many water measuring methods are suitable for making temporary measurements
(flow surveys) or performing occasional verification checks of other devices. The method
chosen for such a measurement might be quite different from that chosen for continuous
monitoring. Although many of these flow survey methods are suited for temporary operation, the focus here is on methods for permanent installations.
Operating Requirements. Some measurement methods require manual labor to
obtain a measurement. Current metering requires a trained staff with specialized equipment. Pen-and-ink style water-stage recorders need operators to change paper, add ink,
and verify proper functioning. Manual recording of flows may require printed forms to be
manually completed and data to be accumulated for accounting purposes. Devices with
manometers require special care and attention to assure correct differential-head readings.
Automated devices, such as ultrasonic flowmeters and other systems that use transducers
and electronics, require operator training to set up, adjust, and troubleshoot. Setting gatecontrolled flow rates by simple canal level references or by current metering commonly
requires several hours of waiting between gate changes for the downstream canal to fill
and stabilize. However, if a flume or weir is installed near the control gate, that portion of
the canal can be brought to the stable, desired flow level and measured flow rate in a few
minutes, and the canal downstream of the flume or weir can then fill to the correct level
over a longer time without further gate adjustments. Thus, the requirements of the operating personnel in using the devices and techniques for their desired purposes must be
considered in meter selection.
Some measuring devices may inherently serve an additional function applicable to the
operation of a water supply system. For example, weirs and flumes serve to hydraulicalIy isolate upstream parts of a canal system from the influence of downstream parts. This
occurs for free overfall weirs and flumes flowing below their modular limits. Acoustic,
propeller, magnetic, and vortex-shedding flowmeters do not provide this function without
additional structural measures such as a downstream overfall. If these meters are used, and
the isolation function is desired, then the designer should be made aware of the requirement and provide a free overfall. Isolating the influence of upstream changes from affecting downstream channels, is less easily accomplished. However, it can be partly implemented with orifices that have a differential head that is large compared to the upstream
fluctuations.
The designer should be aware that a sharp-crested weir overfall requires a relatively
high head drop and may need to be excessively wide to provide the isolation function
with low absolute head drop. While a long board can be used downstream from a propeller meter to provide the necessary width of flow that will pass a required quantity of

water at small head, that small head, and the crude board would not be well suited for
measuring flow rate.
The designer may wish to take advantage of broad-crested weir behavior and provide
a thick crest that can withstand in excess of about 80 percent submergence, which usually translates into low absolute head loss. When used with a propeller meter, for example, the broad-crested weir need not be well defined and can be economically installed
(Replogle, 1997).
Ability to pass sediment and debris. Canal systems often carry a significant amount
of sediment in the water. Removal of all suspended solids from the water is usually prohibitively expensive. Thus, some sediment will likely be deposited anywhere the velocities are reduced, which typically occurs near flow measuring structures. Whether this sediment causes a problem depends on the specific structure and the volume of sediment in
the water. In some cases, this problem simply requires routine maintenance to remove
accumulated sediment; in others, the accumulation can make the flow measurement inaccurate or the device inoperative. Sediment deposits can affect approach conditions and
increase approach velocity in front of weirs, flumes, and orifices. Floating and suspended
debris such as aquatic plants, washed-out bank plants, and fallen tree leaves and twigs can
plug some flow measurement devices and cause significant flow measurement problems.
Many of the measurement devices which are successfully used in closed conduits (e.g.,
orifices, propeller meters, and so on) are not usable in culverts or inverted siphons because
of debris in the water. Attempting to remove this debris at the entrance to culverts is an
additional maintenance problem.
Flumes, especially long-throated flumes, can be designed to resist sedimentation. The
design options available are to select a structure shape that will maintain velocities that
assure erosion of sediments, or at least continued movement of incoming sediments
through the flume, at important flow rates. In large broad-crested weirs (a class of longthroated flumes) for capacities greater than 1 m3/s per m of flume width, velocities greater
than 1 m/s can be achieved for the upper 75 percent of the flow range, and is usually erosive enough to maintain flume function even for high-sediment bed loads. At the lower
flow ranges and for heavy sediment bed loads, deposition is likely and frequent maintenance may be required.
Trapezoidal sections tend to retain low velocities into the upper ranges of flow and
are less sediment worthy. Long-throated flumes with flat bottoms throughout and side
contractions maintain a high velocity for 0.5 m3/s per m width, and higher, but must
have throat lengths that are 2 to 3 times the throat width in order to be accurately computable. The sediment worthiness of a flume design depends more on these absolute
velocities than on whether the flume floor is flat throughout or raised as in a broadcrested weir. This prompts the designer to select shapes that can provide these velocities. One suggestion for broad-crested weirs in a fixed sized channel is to construct a
false floor in the head gage area to increase the velocity there and prevent changes in
area of flow there. Also, sediments can accumulate in the upstream channel to a depth
of the false floor without affecting the function of the flume. This can extend the time
between mandatory channel cleaning.
Device environment. Any measurement device with moving parts or sensors is subject to failure if it is not compatible with the site environment. Achieving proper operation
and longevity of devices is an important selection factor. Very cold weather can shrink
moving and fixed parts differentially and solidify oil and grease in bearings. Water can
freeze around parts and plug pressure ports and passageways. Acidity and alkalinity in

water can corrode metal parts. Water contaminants such as waste solvents can damage
lubricants, protective coatings, and plastic parts. Mineral encrustation and biological
growths can impair moving parts and plug pressure transmitting ports. Sediment can
abrade parts or consolidate tightly in bearing and runner spaces in devices such as propeller meters.
Measurement of wastewater and high sediment transport flow may preclude the use of
devices that require pressure taps, intrusive sensors, or depend upon clear transmission of
sound through the flow. Water measurement devices that depend on electronic devices and
transducers must have appropriate protective housings for harsh environments. Improper
protection against the site environment can cause equipment failure or loss of accuracy.
Maintenance requirements. The type and amount of maintenance varies widely with
different measurement methods. For example, current metering requires periodic maintenance of the current meter itself and maintenance of the meter site to assure that is has a
known cross section and velocity distribution. When the flow carries sediment or debris,
most weirs, flumes, and orifices require periodic cleaning of the approach channel. As
mentioned above, design and meter selection can mitigate the maintenance problems with
sediments, but are not likely to eliminate them. Electronic sensors need occasional maintenance to ensure that they are performing properly. Regular maintenance programs are
recommended to ensure prolonged measurement quality for all types of devices.
Construction and installation requirements. In addition to installation costs, the difficulty of installation and the need to retrofit parts of the existing conveyance system can
complicate the selection of water measurement devices. Clearly, devices that can be easily retrofitted into the existing canal system are much preferred because they generally
require less down time, and usually present fewer unforeseen problems.
Device standardization and calibration. A standard water measurement device infers
a documented history of performance based on theory, controlled calibration, and use. A
truly standard device has been fully described, accurately calibrated, correctly constructed, properly installed, and sufficiently maintained to fulfill the original installation
requirements and flow condition limitations. Discharge equations and tables for standard
devices should provide accurate calibration. Maintaining a standard device usually only
involves a visual check and measurement of a few specified items or dimensions to ensure
that the measuring device has not departed from the standard. Many standard devices have
a long history of use and calibration, and thus are potentially more reliable. Commercial
availability of a device does not necessarily guarantee that it satisfies the requirements of
a standard device.
When measuring devices are fabricated onsite or are poorly installed, small deviations from the specified dimensions can occur. These deviations may or may not affect
the calibration. The difficulty is that unless an as-built calibration is performed, the
degree to which these errors affect the accuracy of the measurements is largely
unknown. All too frequently, design deviations are made under the misconception that
current metering can be used to provide an accurate field calibration. In practice, calibration by current metering to within 2 percent is difficult to attain. An adequate calibration for free-flow conditions requires many current meter measurements at several
discharges. Changing and maintaining a constant discharge for calibration purposes is
often difficult under field conditions.
Field verification, troubleshooting, and repair. After construction or installation of a
device, some verification of the calibration is generally recommended. Usually, the meth-

ods used to verify a permanent device (e.g., current metering) are less accurate than the
device itself. However, this verification simply serves as a check against gross errors in
construction or calibration. For some devices, errors occur as components wear and the
calibration slowly drifts away from the original. Other devices have components that simply fail, that is, you get the correct reading or no reading at all. The latter is clearly preferred. However, for many devices, occasional checking is required to ensure that they are
still performing as intended. Selection of devices may depend on how they fail and how
easy it is to verify that they are performing properly.
User acceptance of new methods. Selection of a water measurement method must
also consider the past history of the practice at the site. When improved water measurement methods are needed, proposing changes that build on established practice are generally easier to institute than radical changes. It can be beneficial to select a new method
that allows conversion to take place in stages to provide educational examples and demonstrations of the new devices and procedures.
Vandalism potential Instrumentation located near public access is a prime target for
vandalism. Where vandalism is a problem, measurement devices with less instrumentation, or instrumentation that can be easily protected, are preferred. When needed, instrumentation can be placed in a buried vault to minimize visibility.
Impact on environment. During the selection of a water measurement device, consideration must be given to potential environmental impacts. Water measurement devices
vary greatly in the amount of disruption to existing conditions that is needed for installation, operation, and maintenance. For example, installing a weir or flume constricts the
channel, slows upstream flow, and accelerates flow within the structure. These changes in
the flow conditions can alter local channel erosion, local flooding, public safety, local
aquatic habitat, and movement of fish up and down the channel. These factors may alter
the cost and selection of a measurement device.
21.5.3 Selection Guidelines
Selection of a water measurement method can be a difficult, time-consuming design
process if one were to formally evaluate all the factors discussed above for each measuring device. This difficulty is one reason that standardization of measurement devices within water agency jurisdictions is often encouraged by internal administrators. However,
useful devices are sometimes overlooked when devices similar to previous purchases are
automatically selected. Therefore, some preliminary guidance on selection is offered to
the designer so that the number of choices can be narrowed down before a more thorough
design analysis of the tradeoffs between alternatives is performed.
Short list of devices based on application. The list of practical choices for a water
measuring device is quickly narrowed by site conditions because most devices are applicable to a limited range of channel or conduit conditions. Economics also limits applicable
devices. For example, few irrigation deliveries to farms can justify expensive acoustic
meters. Likewise, using current meters for manual flow measurement in a channel is
appropriate for intermittent information but is usually too labor intensive for use on a continuous basis. Table 21.1 provides a list of commonly used measurement methods that are
considered appropriate for each of several applications. Table 21.2 provides an abbreviated table of selection criteria and general compliance for categories of water measurement
devices. The symbols (+), (O), and () are used to indicate relative compliance for each
selection criterion. The (+) symbol indicates positive features that might make the device

TABLE 21.1 Application-Based Selection of Water Measurement Devices

1. Openchannel conveyance system
a. Natural channels (see Herschy, 1985)
(1) Rivers
Periodic current metering of a control section to establish stage=discharge relation
Broadcrested weirs
Long-throated flumes
Short-crested weirs
Acoustic velocity maters (AVMtransient time)
Acoustic Doppler velocity profiles
Float-velocity/area method
Slope-area methods
(2) Intermediate-sized and small streams
Current metering/control section
Broad-crested weirs
Long-throated flumes
Short-crested weirs
Short-throated flumes
Acoustic velocity meters (AVMtransient time)
Floatvelocity/area method
b. Regulated channels (see USBR, 1997)
(1.) Spillways
(a) Gated
Sluice gates
Radial gates
(b) Ungated
Broad-crested weirs (including special crest shapes, Ogee crest, etc.)
Short-crested weirs
(2) Large canals
(a) Control structures
Check gates
Sluice gates
Radial gates
Overshot gates
(b) Other
Long-throated flumes
Broad-crested weirs
Short-throated flumes
Acoustic velocity meters

TABLE 21.1 (Continue)

(3.) Small canals (including open channel fluid conduit flow)
Longthroated flumes
Broad-crested weirs
Shortthroated flumes
Sharpcrested weirs
Rated flow control structures (check gates, radial gates, sluice gates, overshot gates)
Acoustic velocity meters
(c) Other
Float-velocity/area methods
(4) Irrigation delivery to farm turnout
(a) Pipe turnouts (short inverted siphons, submerged culverts, etc.)
Metergates
Current meter
Weirs
Short-throated flumes
Long-throated flumes
(ft) Other
Constant head orifice
Rated sluice gates
Movable weirs
2. Closed conduit conveyance systems (see Brater and King, 1982; Miller, 1996)
(a.) Large pipes
Venturi meters, venturi tubes, nozzles
Rated control gates (orifice)
Acoustic velocity meters (transit time)
(ft.) Small and intermediate-sized pipelines
Venturi meters, Venturi tubes, nozzles
Orifices (in-line, end-cap, shunt meters, etc.)
Propeller and turbine meters
Magnetic meters
Acoustic meters (transit-time and Doppler)
Pitot tubes
Elbow meters
Vortex - shedding
Trajectory methods (e. g., full-pipe trajectory; California pipe method for part full pipe)
Other commercially available meters (household types)
SOURCE: From USBR (1997).

TABLE 21.2 Selection Guide for Water Measuring Devices.

Selection

Sharp
Crested

Broad
Crested

LongThroated

Short Submerg
Throated Orifice

Criteria

Weirs

Weirs

Flumes

Flumes (Channel) (Channel) (Channel)

Devise

Current
Metering

Acoustic Radial Propeller Differ.

Mechan. Magnetic
VeI Meter and Sluice Meters Head meters* Head meters* meters
Gates

Accuracy
O
O
O
O
O
+
Cost
O
O
+
O
O
+
+
+
Flows > 5 m-Vs
O
O
+
+
+
+
Flows < 0.25 mVs
O
O
+
Flow span
+
O
O
O
+
Headloss
+
O
O
Site Condition
+
Lined canal
+
+
O
O
O
Unliner canal O
+
O
O
O
O
O
Short, full pipe NA
NA
+
NA
NA
NA
NA
NA
Closed conduit NA
NA
NA
NA
NA
NA
NA
NA
Measurement Type
+
+
+
Rate
+
+
+
+
+
Volume
O
Sediment
+
+
Sediment pass
O
O
O
O
2
+
+
+
+
Debris pass
+
Longivity
+
+
+
Moving parts +
+
+
O
O
+
+
+
+
Electr. requir. +
O
O
+
+
+
+
Maintenance
O
O
+
+
+
+
+
Construction
O
O
+
+
+
Field verify
O
O
+
+
+
Standarization
O
O
O
Source: Adapted from USER (1997).
Symbols 1, O, 2 are used as relative indicators comparing application of the listed water measuring device to the listed criteria
Symbol V denotes that situability varies widely.
Symbol NA denotes "not applicable" to criteria
"Venturi,
orifice, pilot tube, etc.
f
Propeller meters, turbine meters, paddle wheel meters, etc.

(Pipe Exit) (Pipes)

+
O
O
+
O
O
+
+
O
+
O
O
+
O
O

(Pipes)

(Pipes)

4-

O
+

O
O

O
O
+
NA
NA
O
+
V
V

O
O
+
NA
NA

V
NA
NA
O
+
+
V
+
O
O
O
+
+

Acoustic
Doppler

Acoustic Acoustic
Transonic Transon.
Pipe
(Pipes) (Pipe, 1 Path) (Multipath)
+
O
O
+
O
O
O
+
O
+
+
+
NA
NA
NA
NA
NA
NA

+
+
O
O
O
O

+
+
O
O
O
O

+
O
O
O
O
O

+
+
+
O
O
O

O
2
O
O
O

attractive from the standpoint of the associated selection criteria. A (-) symbol indicates
negative aspects that might limit the usefulness of this method based on that criterion. A
(O) indicates no strong positive or negative aspects in general. A (V) means that the suitability varies widely for this class of devices. The letters NA mean that the device is not
applicable for the stated conditions. A single negative value for a device does not mean
that the device is not useful or appropriate, but other devices would be preferred for those
selection criteria.
Vortex-shedding flow meters are not specifically rated in this grouping. They are
expected to compete with orifice meter applications. They generally offer less head loss
that orifice meters and can cover a wider discharge range for a particular installation.
Although they have been around for many years, they have only recently been offered in
a configuration that makes them competitive with orifice meters, which they are generally expected to replace because they can produce less pipe head loss. Open-channel applications for vortex-shedding meters are not considered practical.

21.6 SELECTION OF SECONDARY DEVICES FOR DISCHARGE

READOUTAND CONTROL
While the emphasis of this chapter is on hydraulic design, it is important that the output
of the design be translated into useful information for the user.
21.6.1 Intended Uses
The secondary device that is used with a flowmeter depends strongly on the use of the
information. Immediate management decisions, such as adjusting a valve or canal gate,
require nearly instant feedback to the operator at an accuracy and precision that fully utilizes the available accuracy of the primary device. As discussed above, random errors over
many measurements tend to cancel, and long-term totals can often absorb large random
detection errors that would not be tolerable for decisions depending on a single reading.
The designer should be cognizant of the user needs and be prepared to provide an appropriate output.
21.6.2 Quality Assurance
Secondary devices provide a variety of functions, primary of which are data recording and
data quality assurance. The secondary devices necessary for these may not be the same. It
is good practice to provide manual, instantaneous flow rate output at the meter site so that
the servicing personnel can quickly know that the main secondary instrument is functioning. For weirs and flumes wall gages that show flow rates directly are recommended as a
quick visual check of secondary instrumentation. However, a wall gage that shows a head
reading that is converted to a discharge rate by the technician using a table or equation
will usually suffice.
Sometimes it is practical to provide field check capability to a secondary device by
special treatment of the installation. For example, if a pressure transducer is used to detect
head on a flume, the transducer can be mounted in a stilling well attached to the flume. If
it is further mounted on a movable rack, it can allow servicing personnel to raise and

Next Page

B = WIDTH OF CULVERT
= (D) FOR PIPE
L = DISTANCE TO DESIRED DEPT.H
MEASURED IN MULTIPLES OF (B)
Y0 = BRINKDEPTH
Y A = AVERAGE DEPTH OF FLOW

FIGURE 20.2 Average depth for abrupt exapnsion below rectangular culvert outlet. (From Cony et al., 1975)

D = DIAMETER OF CULVERT
L = DISTANCE TO DESIRED DEPTH
MEASURED IN MULTIPLES OF (D)
Y0 = BRINKDEPTH
Y A = AVERAGE DEPTH OF FLOW

FIGURE 20.3 Average depth for abrupt expansion below circular culvert outlet. (From Corry et al., 1975).

Y0 = brink depth
D = height of box culvert
B = width of barrel
Q = discharge
TW = tail water depth

TW/D

FIGURE 20.4 Dimensionless rating curves for the outlets of rectangulars culverts on horizontal and mild slopes (From Simons et al; 1970)

Y0 = brink depth
D = dia. of culvert
TW = tail water depth

TW/D

FIGURE 20.5 Dimensionless rating curve for the outlets of circular culverts on horizontal and
mild slopes (From Simons, et al., 1970)

5. Determine the average velocity (VA) for box culverts using

-^- = 1.65 - 0.3Fr
M)
and for pipe culverts, (for L ^ 3D) using
^
=1.65-0.45 f-j
v
(\W)
6. Compute downstream width W2 = W0 + 2L tan 9, where tanG = Fr/3. If 0W > ft then
use 6W to compute W2.
7. If 0 is used to compute W2, then compute the downstream depth y2 using W2 and VA.
Because the flow prism is laterally confined, y2 will be larger than VA.
If Bw is used, y2 = yA, and the average flow width is WA = Q/VAyA. If WA < W2, use W2
to compute y2 = Q/VAW2.
Example 1 (adapted from Cony et al., 1975). Determine the width of an abrupt expansion and the hydraulic condition (y and V) at the end of an abrupt expansion for a 5 ft by
5 ft, 200= ft-long reinforced concrete culvert on a 0.2 percent slope (S0 = 0.002 ft/ft). The
discharge is Q = 270 cfs and a wingwall flare (0W = 45) with a 10 = ft apron is
considered.
Given

C ftAand
A T^W ^O
d,c = 5

Solution.
-"

&-&-*<.-<@ti-4ffi-">'
TW
and *** O. From Fig. 20.4, V 0 / D = 0.68; therefore, y0 and V0 can be deter-

mined to be y0 = 0.68(5) = 3 .40 ft and V0 = 270 /[3.4 (5)] = 15.88 ft/s.

Step 2. The outlet Fr = V0 /V^0" = 15.88/V32.2 (3.4) = 1.52
Step 3. The optimum flare angle is determined using tan 0 = 1^ Fr = i (1.52)
= 0.507, so 0 = 26.9.
Step 4. With the apron length/width = 10/5 = 2 and L = 2.OB, use Fig. 20.2 to compute
the average depth where ^/V0 = 0.26 and yA = 0.26 (3.4) = 0.88 ft.
Step 5. Compute the average velocity, VA/V0 = 1.65 - 0.3 Fr = 1.65 - .3 (1.52)
= 1.19, so = V4 = 1.19 V0 = 1.19 (15.88) = 18.90 ft/s
Step 6. Determine the downstream width W2 = W0 + 2L tan 0W using 0 = 0W = 45
because 0W > 0 and W0 = 5 ft, then W2 = 5 + 2 (10) (tan 45) = 25 ft.
Step 7. 0W is used above, so y2 = yA = 0.88 ft and W2 = QI(VA yA) = 16.1 < 25 ft.
When subcritical flow is maintained throughout a culvert, gradual transitions can be
used. Referring to Fig. 20.6, upstream of section 1 where some backwater exists because
of the culvert, flow is transitional from a channel into the culvert and out. According to the
Federal Highway Administration (FHWA, 1978), a flare angle of 17.5 (4.5 to 1) or flatter
provides a gradually varied transition that can be analyzed using the energy equation.

PLAN

HYDRAULICGRADELINE

WATERSURFACE

ELEVATION
ATUM
FIGURE 20.6 Definition sketch. (From Cony et al., 1975)

Referring again to Fig. 20.6, the headless in the contraction HLC is

if
HLC

(V2 V 2^
= Cc
C \ 2- L
(2g
2gJ

(20 1 ^
(ml)

(V2
V2}
= CC \^- $- L
'(2g
2g)

(20 2)
2s)
(2

and the headloss in the expansion is

H
HLE

where transition loss coefficients are listed in Table (20.1) for the transition types
(USAGE, 1970).
The design procedure for gradual transitions is as follows (Corry et al 1975):
1. Use Manning's equation to compute y4 = yn and V4, knowing Q, S0, n and the outlet
geometry.
2. Compute the critical depth using yc = ^1 (QIW)m. For box culvert, kv = 0.315 where
Q is in cfs and W is the box culvert width in ft. In SI units ^1 = 0.467 with Q in mVs
and W in m. Compare yn and yc to ensure subcritical flow.
3. For the chosen transition type, use Table 20.1 to obtain the transition loss coefficients
C0 and Ce.
4. Choose V3 = 1.1 yc to have a culvert with a flow depth conservatively above. yc
5. Compute the culvert width using the following energy equation, ignoring the headloss
caused by friction:
V42
PV32 V42 ^l
V3 2
+ C
73 +-f Vv3 ++Z74 + vA;4 ++ +
C ^ - -J-Z
V3 = (W3V3 where V3 = 1.1 yc = IAk1 (QfW3)2 For box culverts and V4 = QtWj4.
Neglecting Z3 - Z4 for a short reach and a mild bottom slope,

^RK^'tei]''--'
><+fei 4*" '" (ir ^ [(^. iut,w-ii ij" -c->
TABLE 20.1 Transition Loss Coefficients
Transition Type

Contraction
Cr

Expansion
Ce

Warped
Cylindrical quadrant
Wedge
Straight line
Square end

1.10
0.15
0.30
0.30
0.30

0.20
0.25
0.50
0.50
0.75

Source: Corry, et al (1975).

A^-*-^+>$-

n A 1

n2/3

/^

"\ ^ 1 \2/3

^) ^'^-('-"^^s0-*)
x2/3

and

^J

W3 = ([BV[A])

(20.3)

where
=*+][wh]-c>
and
[B] = L 1

2/3+

/i2/3 i
Tfefi ( 1 -^

6. Compute J1 with J2 = y3 using the energy equation

y2

/y2

y2\

y2

>+y>+t+c^-^rz>+y>+w

(20

-4)

7. If the amount of backwater (yr yn) exceeds a preferred or required freeboard, then
select a larger culvert width and calculate y3 using Eq. (20.3) given in Step 5 (return
to Step 5).
8. If the culvert width is acceptable, use the flow conditions to compute the transition
length using a 4.5:1 flare (USAGE, 1970):
LT = 4.5 (^

(20.5)

LT =4.5^^

(20.6)

or

9. Compute the water surface profile through the structure using a standard step backwater analysis. This will include an evaluation of friction losses.
Example 2. Determine the dimensions for a culvert and gradual transition needed for a 3
m-wide rectangular flood control channel at a slope of 0.001 m/m. The culvert length is
30.5 m, and is to convey 8.5 m3/s. Use n = 0.02.
Solution:
Step 1. Assuming normal depth at Section 4 (refer to Fig. 20.6), then use Manning's
equation to compute V4 = yn = 1.99 m and Vn = 1.42 m/s.

D ^2/3
( S2^3
-^ = 0.467 -^- = 0.935 m.

Step 3. Use a straight line transition with Ce = 0.5 and C0 = 0.3.

Step 4. y3 = 1.1 vc = 1.1 (0.935) = 1.03 m.
Step 5. Compute W3 using Eq. (20.3).

Step 6. Compute V1 with y2 = y3 using Eq. (20.4):

Step 7. Because ^1 is 1.29 m above the normal depth and because only 0.6 m free board
is available, we go back to Step 5 and compute ^3. Assuming that the culvert
width is 1.45 m: Repeat step 5,

Then, returning to Step 6, V1 using Eq. (20.4):

V1 is 0.42 m above the normal depth. Because a free board of 0.6 m is

available; the design is OK.
Step 8. Determine transition length using Eq. (20.5) for a 4.5:1 flare.

LT

= 4.5 (W' " ^ = 4.5^ -2L45> = 3.49 m, or 3.5 m.

Step 9. Compute the water surface profile.

20.1.2 Culverts with Inlet Control
Culverts with inlet control require the transitioning of supercritical flow, (see Fig. 20.7)
Because supercritical flow is difficult to control without creating a hydraulic jump or
other surface irregularities, the full flow area should be maintained. Because changing the
flow smoothly requires a long structure, model studies should be performed to determine
the transition geometry if a hydraulic jump is not desired. When a hydraulic jump is
acceptable, Figs. 20.8 and 20.9 can be used (USACE, 1970). Such a design requires a rectangular channel and a long transition.
The design procedure for supercritical flow contractions is as follows (Corry et al, 1975):
1. Flow conditions for the approach channel should be computed assuming normal flow
Cv' K ^r) usmg Manning's equation or other design aids.
2. Approach sections that are not rectangular should have a transition to the rectangular
section with a bottom width of W1. This bottom width should be approximately equal
to the average of the water surface top width (T) and the trapezoidal section base width
(Bw): i.e., W = (T + Bw) I 2. Compute the normal flow condition for this rectangular
section using Manning's equation.
3. Assume a trial culvert width W2. Refer to Figure 20.7 for a definition of parameters.
4. Compute the contraction length needed to reduce W1 to W2. This is accomplished by
varying the contraction using wall angle (9W) until L = (W1 W2) I 2 tan 0W is equal
to L1 + L2 where L1 = W11 2 tan P1 and L2 = W21 2 tan (P2 - 0W). To minimize surface disturbances, L should be equal to L1 + L2 (Corry et al, 1975).
Choose 0W

CENTERLINE
ALONGWALL

FLOW
a. SCHEMATICPROFILE

b. PLAN
FIGURE 20.7 Supercritical inlet transition for rectangular channel.
(From USAGE, 1970)
a. Compute L = (W1- W2)/2 tan 9W
b. Find P1, y21 V 1 , and Fr1 from
* *
tan B1 V(I + 8Fr12 sin2 B1 - 3)
1
^= 2 ^ + Vl+8Fr1CV-I

(2

'7)

and

^ = IJVl + 8Fr^ sin* P1 - 11

(20.8)

^
I](^+ lfl
2 = f4^ ? -UJL * I2^2JUi
JU
JJ

(20.9)

and

Alternatively, Figs. 20.8 or 20.9 can be used to approximately solve Eqs.

(20.7), (20.8), and (20.9).
c. Calculate L1 = W1/2 tan p.
d. Compute P2, j3/y2, and Fr2 using the same procedure as in Step 4b (that is, using P2
for P1, Fr2 for Fr1, y3/y2 for yjyl and the same 6W).
e. Calculate L2 = W212 tan (P2 - 6W)

FIGURE 20.8 Supercritical inlet transition design curves for rectangular channels.
(USAGE, 1970)

FIGURE 20.9 Supercritical inlet transition design curves for

rectangular channels (From Ippen, 1951)
f. Compare L with L1 + L2
if L > L1 + L2, then increase 6W and repeat Steps 4a-f until L = L1 + L2.
g. Compute y3 from V3 = V1 (^ M
UV UV
5. Compare depth V3 and the width W2. Culvert should be of a standard dimension. If
not return to Step 3 using another W2 and repeat the design process until a better
combination of V3 and W2 are found.
Example 3 (adapted from Corry et al., 1975). Design the transition contraction and select
the culvert size for a discharge of 300 cfs in a trapezoidal channel (6 ft bottom width, 2:1
side slope, S0 = 0.02 ft/ft and n = 0.012)
Solution:
Step 1. The normal depth, velocity, and Fr are yn = 1.67 ft, Vn = 19.2 ft/s,
and

Fr = -L= =- 19'2
= 3.05m
VgAfT
V 32.2(15.6)712.7
Step 2. Because (T + Bw )/2 = (12.7 + 6) /2 = 9.4 ft, use W1 = 10 ft, a rectangular chan-

nel. Use Manning's equation to compute yn = 1.54 ft and Vn = 19.5 ft/s, then
compute the Fr:
Fr = -^= = . 19'5
=2.77.
Vg^ V32.2(1.54)
Step 3. Assume a trial culvert width W2 = 5 ft.
Step 4. Try 8W = 14 for Fr = 2.8:
L = ( 1 0 - 5 ) / 2 tan 14 = 10ft
P1 = 35, V2^1 = 1.8, Fr2 =1.8

L1 = 10/2tan35 = 7.1 ft.

P2 = 55, y3/y2 = 1.6,Fr 3 = 1.1
L2 = 5 / 2 tan (55 - 14) = 2.9 ft.
L1 + L2 = 10 ft = L, O.K.
y3 = 1.54 (1.6) (1.8) = 4.4 ft.
Use Ow = 14, y3 = 4.4 ft., V3 = 13.6 ft/s, Fr3 = 1.1

Step 5. Because y3 = 4.4 ft. and W3 = 5 ft, a 5 X 5 box culvert will be satisfactory.

20.2

ENERGYDISSIPATIONFORCULVERTSANDCHANNELS

20.2.1 Hydraulic Jump Basins

For supercritical flow expansions, the procedure outlined in Sec. 20.1 is applicable if the
exit Fr is less than 3, if the location where the flow conditions are desired within three culvert diameters of the outlet and if S0 is less than 10 percent (Cony, et al 1975).For expansions outside these limits, a hydraulic jump basin (Fig. 20.10) should be used. This type
of basin allows the flow to expand, drop or both, resulting in depth decreases, velocity
increases, and an Fr increase in. Higher Fr's result in more efficient jumps and shorter
basins.
The design procedure for supercritical flow expansions with hydraulic jump basin is as
follows (Corry, et al; 1975):
1. Compute the culvert brink depth y0 using Figs. 20.4 or 20.5.
2. Compute the tailwater depth Tw in the downstream channel assuming normal flow
(using Manning's equation) or perform backwater analysis.
3. To determine the basin elevation, first select Z1 and then use the following steps.
a. Select basin width W8 (W1 in Fig. 20.10) and basin slopes S5 and S7 (Fig. 20.10).
Slope of S5 or ST = 0.5(2:1) or 0.33 (3:1) are satisfactory (Corry et al, 1975).
b. Check WB using

DATUM
FIGURE 20.10 Definition sketch basin transition. (From Corry et al., 1975)

W <W +

(20 10)
'
'
^T 3Fr0+1
'
where LT = (Z0 Z1)AS7, and the right-hand side is the limit that flares naturally in the
slope distance L.
c. Compute Y1 using the following equation derived from the energy equation from the
culvert outlet brink to the basin (Sec. 1 in Fig. 20.10). Use V1 = Q/ y^W^ to determine^ and then V1:

r
i1'2
Q = J^B [Zg (Z* - ZI + % - *) + ^02J

( 2 O-U)

d. Compute the Fr
Fr 1 =A
vw
e. Compute y2 using Eq. (20.12) for the hydraulic jump basin:
J2 = y[vi + 8 F r J - I J

(20.12)

f. Compute Z3 from geometry using

(20.13)
where

and from Fig. 20.11, determine L, which is defined as L8

L8=Xy19Fr1).
g. Check value of Z1 by computing y2 + Z2 and Z3 + Tw. If y2 + Z2 > Z3 4- Tw, select
another Z1 and repeat steps 4a to 4g until a balance is reached.
4. Compute L5 and L:
_ (Z3 - Z2)
* ~^r~
L = LT + Lfi + L5 =

^o

Example 3. A supercritical flow expansion is to be designed for a reinforced concrete

box culvert measuring 3 by 2 m. Determine the dimensions for the hydraulic jump basin
using a design discharge of Q = 11.8 mVs. The slope is 6.5 percent, the invert outlet elevation is 30.5 m, the downstream channel has a bottom width of 3 m and side slopes of
2:1, and Manning's n = 0.03. The brink depth is supercritical, V0 = 0.457 m and
V0 = 8.47 m/s. S3 = 0.5 and ST = 0.5. (Refer to Figure 20.12)

FIGURE 20.11 Length of jump in terms Of^ 1 , rectangular channel. (From Corry et al.,
1975)

FIGURE 20.12 Example problem 18.2.1 hydraulic jump basin. (Not to scale).

Solution:
Step 1. The brink depth and velocity are given: Fr0 = 4.0.
Step 2. The tailwater depth is computed assuming normal depth. Manning's equation
is solved to obtain yn = 0.57 m tailwater depth and Vn = 4.85 m/s. (Refer to
Fig. 20.12)
Step 3. Assuming a basin elevation OfZ 1 = 25.9 m:
a. Select W8 = 3 m and S5 = S7 = 0.5
b. Check W8 using Eq. 20.10, where W0 = 3 m,
W8 = 3 < 3 + [2L7 V5JTTJ(3Fr0I OK.
c. Compute ^1 using Eq. (20.11)
1/2
11.8 = V1(S) 2(9.81)(30.5 - 25.9 + 0.457 - V1) + (8.47)2

Solving ^1 = 0.306 m and V1 = QIA = 11.8/[3(0.306)] = 12.85 m/s.

d. The Froude number is Fr = 12.85/V9.81(0.306) = 7.42
e. Compute y2 using Eq. (20.12). y2 = 0.306 Vl + 8(7.42)2 - 1 /2 = 3.06 m.
f. Compute Z3 using Eq. (20.15). First compute LT = (Z0 - ZJ/2
= (30.5 - 25.9)70.5 = 9.2 m. From Fig. 20.11, LJy1 = 63, so L8 = 63yl
= 63(0.306) = 19.28 m, then
_ [30.5 - (9.2 + 19.28 - 25.9/0.5) 0.065] _
= 28 34 m
(0.065/0.5 + 1)
'

Z =

g. Check value OfZ1: y2 + Z2 = 3.06 + 25.9 + 28.96 and Z3 + Tw = 28.34 + 0.58

= 28.92, y2 + Z2 > Z3 + Tw\ therefore, Z1 = 25.9 is OK.
4. Compute L8 and then L. L5 = (Z3 - Z2)JS5 = (28.33 - 25.9)70.5 = 4.86 m. Then
L = LT + L8 + L5 = 9.2 + 19.28 + 4.86 - 33.34 m. Refer to Fig. 20.12
20.2.2

Forced Hydraulic Jump Basins

20.2.2.1 Saint Anthony Falls stilling basin. The Saint Anthony Falls, (SAF) stilling basin
is a generalized design based upon model studies conducted by the U.S. Soil Conservation
Service at the St. Anthony Falls Hydraulic Laboratory, University of Minnesota. Figure
20.13 illustrates the SAF stilling basin design which is recommended for small structures
such as spillway outlet works, and for canals where the Fr ranges from 1.7 to 17 (at the dissipator entrance). Through the use of chute blocks, baffle or floor blocks, and an end sill,
the basin length is about 80 percent of the free hydraulic jump length.
The design procedure for SAF basins is as follows (Corry et al 1975):
1. Choose basin configuration and flare dimension, Z. (Refer to Fig. 20.13)

2. Use the design procedure presented in Sec. 20.2.1 for supercritical expansions into
hydraulic jump basins to determine basin width (W^), elevation (Z1), length (Le), total
length (L), incoming depth Cy1), incoming Fr (Fr1), and jump height (y2). Steps 3e and
3f in Sec. 20.2.1 are modified; for Step 3e, determine y2 using the sequent depth y.\
(20.14)
(20.15)
(20.16)
(20.17)

RECTANGULAR BASIN
HALF-PLAN
EQUATION NUMBER
FLARED BASIN
HALF-PLAN
0-90
45 PREFERRED

SIDEWALL

CHUTEBLOCK
FLOORORBAFFLEBLOCKS ENDSILL

VARIES

(1) W6 = BASINWIDTH UPSTREAM

(2) n BLOCKS AT 3/4 Y 1
(3) 0.4OW82 < AGGREGATEBLOCKWIDTH < 0.55Wg2
(4) ri BLOCKS AT 3/4 Y, -5_
(5) WB2 = WB + 2LB/3z
(6) WB3 = WB + 2LB/z
FIGURE 20.13 St. Anthony Falls stilling basin. (From Blaisdell, 1959)

For Step 3/, compute L5:

45v
LB = -^pr -16
i
3. Detennine the dimensions of the chute block:

Height:

H1 = yl

Width:

W2 = 0.1Sy1 and W1 = spacing

W
Nc ~ -^- (rounded)
1
W
W1 = W2 = j- (Nc includes the 1/2 block at each wall)

Number:
Adjusted:

(20.18)

4. Determine the dimensions of the baffle block:

Height: h3 = y t
Width: W3 = spacing and W4 = 0.75V1
Basin width at baffle blocks: WB2 = WB + 2L/3Z
Number of blocks: N8 = WB2/2W3 rounded
Adjusted W3 = W4 = WB2/2NB
Check total block width to insure that at least 40 to 55 percent of W82 is occupied
by blocks.
Distance from chute blocks to baffle blocks = LB/3
5. End sill height: h4 = 0.07^.
6 Side wall height: y2 + y. /3.
Example 4. Determine the dimensions of an SAF basin for the supercritical flow expansion described in Example 3.
Solution:
Step 1 Select a rectangular basin with no flare.
Step 2 Steps 1 through 3(a-f) (in Example 3) for a supercritical flow expansion into a
hydraulic jump basin.
Given V0 = 8.47 m/s, y0 = 0.457 m, and Fr0 = 4.0.
The tailwater depth Tw = yn = 0.57 m and Vn = 4.85 m/s.
Assume that Z1 = Z0 = 30.5 m. (Refer to Fig. 20.14)
a. Compute y.} using Eq. 20.14 with yl = y0 = 0.457 m and Fr1 = Fr0 = 4.0. This
assumes that Z1 = Z0: i.e., the basin floor is the same elevation as the culvert
outlet.

FIGURE 20.14 Example 4 St. Anthony Falls stilling basin. (Not to scale)

b. Next, use Eq. (20.15) for Fr1 = 1.7 to 5.5 to compute J2:

Because y2 > Tw = 0.57, we can lower the elevation of the basin. Use Z1 =
27.9 m with W8 = 3 m and ST = S5 = 0.5. WB is OK and no flare is used.
c. Compute J1 using the energy Eq. (20.11). The basin has been lowered so now
J1 is not J0, the brink depth.
Q = * WB\2g(ZQ - Z1 + J0 - ^1) + V02]
r
nl/2
11.8 = ^(312(9.81X30.5 - 27.9 - 0.457 - J1) + (8.47)2

Solving J1 = 0.348 m. V1 = 11.8/(3 3 0.348) = 11.3 m/s

d. The Fr1 = V1JVg^1 = 11.3/V9.81(0.348) = 6.1
e. Compute jj = y|Vl + 8Fr^ - 1 j = ^^-|Vl + 8(6.1)2 - I J = 2.83 m.
^^^h-SHi1-1-^]2-83=2-24111f. Compute L8 using Eq. (20.18):
LB = 4.5yjFrw* = 4.5 (2.83)/6.1076 = 3.22m.
Compute L7, using LT = (Z0 - Z1)JS7 = (30.5 - 27.9)/0.5 = 5.2m.
Compute Z3 using Eq. (20.13)

g. Check the assumption OfZ 1 = 27.9 m: y2 + Z2 = 2.24 + 27.9 = 30.14 and

Tw + Z3 = 0.57 + 29.7 = 30.2 m: y2 + Z2 Tw + Z3. Because they are so
close, then Z1 will be OK.
h. LT = 5.2 m, LB = 3.22 m, L8 = (Z3 - Z2)JS8 = (29.7 - 27.9)/0.5 - 3.6 m,
and
L = Lr + LB + L8 = 5.2 + 3.22 + 3.6 = 12 m.

Step 3. Chute blocks: H 1 ^ y 1 = 0.35, W1 0.75^1 = 0.26 m = W2. Nc = W8^W1 = 3/

(2X0.26) = 5.77 so use 6 blocks. Adjusted W1 = WB/(2NC) = 3/(2 X 6) = 0.25
m. This provides 5 blocks, 6 spaces, and a half block at each wall.
Step 4. Baffle blocks: H3 J1 = 0.35 m, W3 0.75V1 = 0.26 m = W4. Basin width,
^B2 = WB + 2LB/(3Z) = 3 + O = 3 m (no flare), NB = W52/(2W3) = 3/(2 X
0.26) = 5.77 so use 6 blocks. Adjusted W3 = W4 = 3/(2 X 6) = 0.25 m.
Total block width = 6(0.25) = 1.5 m. Check percentage: 0.4 < 1.5/3.0 < 0.55,
OK. This provides six blocks, five spaces, and a half-space at each wall. Distance
from chute block = LJ3 = 3.22/3 = 1.07 m.
Step 5. End sills: h4 = Q.Oly. = 0.07(2.83) = 0.2 m.
Step 6. Side walls: height = y2 + Vj/3 = 2.24 + 2.83/3 = 3.2 m. Refer to Fig. 20.14 with
the dimension for the chute block shown.
20.2.2.2 Type II, III, and IV basins. The U.S. Bureau of Reclamation's Type II, III,
and IV basins are illustrated in Chapter 18, in Figs. 18.3, 18.4 and 18.7 respectively.
Type II basin design. Use the design procedure presented in Sec. 20.2.1 for supercritical expansion into a hydraulic jump basin to determine W8, Z1, L8, L, V1, Fr1 and V2.
For Step 3e in that section, use C = 1.1 to find y2 = C1V1 Vl + 8Fr12- 1 /2. For Step
3f,use Fig. 20.15 to determine L8.
\-I'
Determine the dimensions for the chute blocks and dentated sill height using the relations in Fig. 18.3.
Type HI basin design. Use the design procedure in Sec. 20.2.1 to determine basin
dimension. For Step 3e use C= 1.0 to determiney2 = Qy1 Vl + 8Fr12- 1/2. For Step
3f, use Fig. 20.15 to determine L8. Use dimensions in Fig. 20.16 to determine chute block
dimensions and spacing.
Use Fig. 18.4 to determine dimensions for baffle blocks and the end sill height.
Type IV basin design. Use the same design procedure presented in Sec. 20.2.1 forsupercritical expansions into a hydraulic jump basin to determine the basin dimensions.
For Step 3e in the section, use Q = 1.0 to determine y2 = QyJ Vl + 8.Fr12- 1 /2. For
Step 3f, use Fig. 20.15 to determine L8.

FREEJUMP

TYPE IV
TYPE Il BASIN

TYPE III BASIN

FIGURE 20.15 U.S. Bureau of Reclamation Type II basin. (Corry et al, 1975)

ENDSILL
BAFFLE PIERS

BAFFLE PIERS

END

FIGURE 20.16 Height of baffle piers and end sill (Type III basin). (From U.S. Bureau of Reclamation, 1987)

Use dimensions in Fig. 18.7 to determine chute block dimensions and spacing.
Use Fig. 18,7 to determine the end sill height.
Example 5. Determine the dimensions of U.S. Bureau of Reclamation's Type II basin for
the supercritical flow expansion described in Example 3.
Solution:
Step 1. Use the design procedure for a supercritical flow expansion into a hydraulic jump
basin. (Refer to Examples 3 and 4).
Given V0 = 8.47 m/s, V0 = 0.457 m, and Fr0 = 4.0. (Figure 20.17)
The tailwater depth is Tw = yn = 0.57 m, and Vn = 4.85 m/s.
Assume Z1 = Z0 = 30.5 m, compute y2
y2 = C1J1[Vl + 8Fr^- l]/2
= 1.1(0.457/Vl + 8(4)2 - IJ/2
= 2.6m

y2 > Tw (2.6 > 0.57). Therefore we need to lower elevation Z1 of the basin floor.
a. Use a basin floor elevation of Z1 = Z2 = 25.76 m, with W8 = 3 m, ST =
S5 = 0.5.
b. W8 is OK, no flare.
c. Compute V1 using energy Eq. (20.11).
r
T'2
Q = ViW8^g(Z0 - Z1 + y0 - V1) + V02J
r
i172
11.8 = Vj(3) 2(9.81)(30.5 - 25.76 + 0.457 - V1) + 8.472

Solving V1 = 0.3 m and V1 = 13.1 m/s.

d. The Fr1 = 13.1/V9.81(0.3) = 7.6
e. Compute y2 = 1.1(0.3) Vl + 8(7.6)2 - 1 /2 = 3.39 m.
f. Using Fig. 20.15, L8Jy2 = 4.3, so L8 = 4.3(3.39) = 14.6 m:
LT = (Z0 - Z1)IS7 = (30.5 - 25.76)/0.5 = 9.5 m.
Using Eq. (20.13), compute Z3

Z3

_ [30.5 - (9.5 + 14.6 - 25.76/0.5)0.065] _

~
(0.065/0.5 + 1)
~ 28'57 m>

g. Check Z1, y2 + Z2 = 3.39 + 25.76 = 29.15 and Z3 + Tw = 28.57

+ 0.57 = 29.14, OK
h. Compute L5 = (Z3 - Z2)JS5 = (28.57 - 25.76)/0.5 = 5.6 m and

L = LT- LB + Ls = 9.5 + 14.6 + 5.6 = 29.6 m.

Step 2. Chute blocks: H1 = W1 = W2 = J1 = 0.3 m, Wc = | = 5, OK.
2(0>3
Side-wall spacing = y/2 = 0.15 m.
'
Ste/? 3. Dentated sill: H2 = 0.2v2 = 0.2(3.39) 0.7 m.
W3 = W4 = 0.15y2 = 0.15(3.39) = 0.51 m.
Ns = WB/W3 = 3/0.51 m = 5.88.
Use 5, which provides three blocks and two spaces, each of which is 0.6 m
wide. Refer to Fig. 20.17 for the dimensions of the basin.
20.2.3

Impact-Type Energy Dissipation (USBR Type Vl Basin)

Figure 20.18 illustrates the U.S, Bureau of Reclamation's Type VI impact-type energy dissipator, which can be used with culverts. This basin is contained in boxlike structures
requiring no tail water for operation. The structures can be used for open channels as well
as culverts, and the basin can be used at sites where the entrance velocity to the basin does
not exceed 50 ft/s, and the discharge is less than 400 cfs. This dissipator should not be
used if the buildup of debris or ice can cause substantial clogging.
The design procedure is as follows (Corry et al, 1975).
1. Compute the flow area at the end of the culvert using the maximum design discharge
and velocity. Compute the equivalent depth of flow entering the dissipator from the
culvert as
Je = (f)"2

(20.19)

where A is the cross-sectional area of flow in the culvert. This converts the cross-sectional area of flow of a pipe into an equivalent rectangular cross section with a width
twice the depth of flow. The culvert preceding the dissipator can be open, closed, or
have any cross section. This approach ignores the size and shape of the culvert entirely except for the determination of flow entering the dissipator.
2. Compute Fr and the energy at the end of the culvert H0:
V2
" = ^ + 2 F -

Then use Figure 20.19 to determine the basin width. Enter Figure with Fr to determine
H0IW then W = H0I (H0IW).
3. Use Table 20.2 to determine the dimensions of the dissipator structure.

DATUM

FIGURE 20.17 Example 5 U.S. Bureau of Reclamation type II basin

PLAN

SECTION
STILLING BASIN DESIGN

PLAN

BEDDING

SECTION
ALTERNATE ENO SILL

FIGURE 20.18 Baffle-wall energy dissipator of a U.S. Bureau of Reclamation type IV basin. (From Corry et
al., 1975)

FIGURE 20.19 Design curve for a baffle-wall dissipator. (Corry et al., 1975)

20.2.4 Drop Structures

20.2.4.1 Straight-drop spillway. The straight-drop spillway shown in Figs 20.20 and .
20.21 is generally used for subcritical flow in the upstream as well as the downstream
channel. To describe the flow geometry, the following drop number is used:
N

D =^

(20.20)

where q is the discharge per unit width of the crest overfall, g is the acceleration caused
by gravity, and h0 is the height of the drop. The dimensions L1, V1, y2, and ^3 in Fig. 20.22
are determined using the following:
^- = 4.3(W*27
AZ0

(20.21)

- = LQN*22
AZ0

(20.22)

TABLE 20.2 Baffle Wall Dissipator: Dimensions of The Basin in Feet and Inches

W
4-0
5-0
6-0
7_0
8-0
9-0
10-0
11-0
12-0
13-0
14-0
15-0
16-0
17_0
18-0
19-0
20-0

H1

3-1 5-5
3-10 6-8
4-7 8-0
5-5 9-5
6-2 10-8
6-11 12-0
7-8 13-5
8-5 14-7
9-2 16-0
10-0 17-4
10-9 18-8
11-6 20-0
12-3 21-4
13-0 22-6
13-9 23-11
14-7 25^
15-4 26-7

H2

1-6
1-11
2-3
2-7
3-0
3-5
3-9
4-2
4-6
4-11
5-3
5-7
6-0
6-4
6-8
7-1
7-6

Dimensions of Basin in Feet and Inches

H3 L1
L2
H4
W1 W2
t3
0-8 2-4
0-10 2-11
1-0 3-5
1-2 4-0
1-4 4-7
1-6 5-2
1-8 5-9
1-10 6-4
2-0 6-10
2-2 7-5
2-4 8-0
2-6 8-6
2-8 9-1
2-10 9-8
3-0 10-3
3-210-10
3-4 11-5

3-11-8
3-102-1
4-72-6
5-52-11
6-23^
6-113-9
7-84-2
8-54-7
9-25-0
10-05-5
10-95-10
11-66-3
12-36-8
13-07-1
13-97-6
14-77-11
15-48-4

I2

I1

I4

t5

0-41-1 0 - 6 0 - 6 0 - 6 0-6 0-3

0-5 1-5 0 - 6 0 - 6 0 - 6 0-60-3
0-6 1-8 0-6 0-6 0-6 0-6 0-3
0-6 1-11 0-6 0-6 0-6 0-6 0-3
0-7 2-2 0-7 0-7 0-6 0-6 0-3
0-8 2-6 0-8 0-7 0-7 0-7 0-3
0-9 2-9 0-9 0-8 0-8 0-8 0-3
0-10 3-0 0-9 0-9 0-8 0-8 0-4
0-11 3-0 0-10 0-10 0-8 0-9 0-4
1-0 3-0 0-10 0-11 0-8 (MO 0-4
1-1 3-0 0-11 1-0 0-8 0-11 0-5
1-2 3-0 1-0 1-0 0-8 1-0 0-5
1-3 3-0 1-0 1-0 0-9 1-0 0-6
\-A 3-0 1-0 1-1 0-9 1-0 0-6
1-4 3-0 1-1 1-1 0-9 1-1 0-7
1-5 3-0 1-1 1-2 0-10 1-1 0-7
1-6 3-0 1-2 1-2 0-10 1-2 0-8

Source: Cony, et al; (1975)

^- = 0.54A425
H0

(20.23)

^- = 1.66A#27
(20.24)
/I0
L2 is the length of the jump, L1 can be determined using Fig. 20.22. The sequent depth
and the tailwater depth Tw must be compared to determine whether Tw < y3, or Tw = y3,
or Tw > y3. If Tw < y3, the hydraulic jump moves downstream. In this case, it is necessary

AERATED

FIGURE 20.20 Flow geometry of a straight-drop spillway. (From Corry et al., 1975)

TOP SLOPE I TO I
UPPER NAPPE
SIDE WALL
HEIGHT

SECTION AT CENTER LINE

FLOOR BLOCKS
LONGITUDINALSILL
(OPTIONAL)

ENDSILL

PLAN
FIGURE 20.21 Straight-drop spillway stilling basin. (From Rand, 1955)

to construct the apron at the bed level and an end sill or baffles or to construct the apron
below the downstream bed level and an end sill. If Tw > y3, the hydraulic jump may
become submerged. If Tw = V3, the hydraulic jump begins at depth V2; there is no supercritical flow on the apron, and L1 is a minimum.
20.2.4.2 Grated energy dissipators. Energy dissipators with grates (Fig. 20.23) also can
be used in conjunction with drop structures. The U.S.Bureau of Reclamation (1987)
developed the following design recommendations for grates for incoming subcritical
flow:
1. Select slot width with a full slot width at each wall.
2. Compute beam length L0 using

LOWER NAPPE
(NOTAILWA1TER)

TAILWATER
LEVEL
TANGENTATPOINT
OFSUBMERGENCE
MEAN
SUBMERGED
TRAJECTORY
NAPPE
TRAJECTORY
FREE NAPPE
TRAJECTORY

FIGURE 20.22 Design chart for determination of L1. (From Corry et al., 1975)

LGG =

,
C(W)(N) V2^~

(20.25)

where C is a coefficient equal to 0.245, W is the width of the slots in feet, and N is the
number of slots or spaces between beams.Then compute the beam width = 1.5 W. The
quantity (W)(N) can be adjusted until an acceptable beam length, L0 is determined.
3. For self-cleaning, the grate can be tilted appoximately 3 in the downstream direction.
20.2.4.3 Straight drop structures. The straight drop structure shown in Fig. 20.21 consists of a horizontal apron with blocks and sills to dissipate energy. This structure is for
drops of less than 15 ft (4.57 m) and for sufficient tailwater. The design parameters
include the length of the basin, the position and size of the floor blocks, the position and
height of the end sill, the position of the wingwalls, and the geometry of the approach
channel. This structure was developed by the Agricultural Research Service at the Saint
Anthony Falls Hydraulic Laboratory.
The design procedure for a straight-drop structure is as follows (Corry et al 1975):

Figure 20.23. Energy dissipator with grate ( Cony et al, 1975)

1. Compute the minimum length of stilling basin L8

L8 = L1+ L2 + L3 = L1 + 2.55yc

(20.26)

where distances are illustrated in Fig. 20.21. The distance from the headwall to the
point where the surface of the upper nappe strikes the still basin floor L1 is
L 1 =^

(20.27)

Lf = -0.406 + J 3.195 - 4.368 [^] yc

V^c ) _

(20.28)

where

0.691+0.228 f M - f ^ l L
UJ
UJj
LS
~
I"
1
L

(20.29)

0.185 + 0.456^1
UJJ
and

Lt = -0.406 + J3.195 - 4.368p| yc

L
^c' _

(20.30)

where h2 = h0 y3 (Fig. 20.21). Alternatively, L1 can be determined using Fig. 20.22.

The distance from the point where the surface of the upper nappe strikes the

2.

3.
4.
5.
6.
7.
8.

9.

stilling basin floor to the upstream face of the floor blocks L2 is determined using
L2 = 0.8 yc. The distance between the upstream face of the floor blocks and the end of
the stilling basin, L3, is determined using L3 ^ 1.75yc.
Floor blocks are proportioned as follows:
a) height = 0.8yc;
b) width and spacing should be 0.4yc with a variation of 0.15yc permitted, and
c) blocks should be square in plan, and
d) blocts should occupy between 50 to 60 percent of the stilling basin width.
Compute the end sill height as 0.4yc.
If longitudinal sills (for structural purposes only) are used, they should be constructed
through the floor block, not between the floor blocks.
Compute sidewall height above the tailwater level as 0.85yc.
Wingwalls are constructed at an angle of 45 with the outlet center line with a top slope
of 1:1.
Compute the minimum height of the tailwater surface above the floor of the stilling
basin y3 using y3 = 2.15yc.
Modification, to the approach channel are as follows: The crest of the spillway should
be at the same elevation as the approach channel, the bottom width should be equal to
the spillway notch length W0 at the headwall, and protection with riprap or paving
should be provided for a distance upstream of the headwall of 3yc.
Using the recommendations in Step 8, no special provision for aeration is needed.

Example 6 Determine the dimensions of a straight-drop spillway stilling basin for a discharge of 7.08 m3/s. The downstream trapezoidal channel has a 3:1 side slope with a 3.05
m bottom width S0 = 0.002 m/m, n = 0.03, and a normal depth of 1.024 m. The drop h0 =
1.83 m.
Solution:
Step 1. Determine the minimum basin length L8. The critical depth yc is determined as
yc = 0.655 m. Then hjyc = 1.83/0.655 = 2.79; H2 = h0 -2.15yc = 1.83 2.15(0.655) = 0.422 m; hjyc = 0.422/0.655 = 0.644.
Using Fig. 20.22, L1Jyc = 8.2 for hjyc = 2.79 and hjyc = 0.644.
Then L1 = 8.2(0.655) = 5.37 m; L2 = 0.8yc = 0.8(0.655) = 0.52 m; and L3 = 1.75yc
= 1.75(0.655) = 1.15 m; thus, LB = L1 + L2 + L3 = 5.37 + 0.52 + 1.15 = 7.04 m.
Step 2. Proportions the floor blocks are
height = 0.8vc - 0.8(0.655) = 0.524 m,
width = 0.4yc = 0.262 m, and
spacing = 0.4yc = 0.262 m.
Step 3. Calculate the end sill height = 0.4yc = 0.262 m.
Step 4. Use the longitudinal sill passing through the floor blocks.
Step 5. Calculate the side-wall height above the tailwater = 0.85yc = 0.85(0.655)
= 0.557 m.

Step 6. Local wing walls are at a 45 angle with the outlet center line.
Step 7. Calculate the minimum height of tailwater above the floor of the stilling basin: y3
= 2.15yc - 2.15(0.655) = 1.41 m. The basin must be placed 1.43 - 1.024 =
0.406 m below the downstream bed level.
20.2.4.4 Box-inlet drop structure. The box-inlet drop structure shown in Figure 20.24
consists of two different sections that are effective in controlling flow: the crest of the box
inlet and the opening in the headwall. This structure is based on experiments by the U.S.
Soil Conservation Service at the Saint Anthony Falls Hydraulic Laboratory (Blaisdell and
Donnelly, 1956). The design procedure of box-inlet drop structures is as follows:
(FHWA, 1978):
1. Select/*0
2. Select L1, W2, and Lc where Lc is the length of the box-inlet crest, Lc = W2 + 2L1,

FIGURE 20.24 Box-inlet drop structure. (From Corry et al., 1975)

where W2 is the width of the box inlet, and L1 is the length of the box inlet.
3. Compute the head y0 for the crest using the discharge equation for a rectangular weir:
(20 31)
^(dkf
4. Compute -^
and determine the coefficient of discharge C2 from Fig. 20.25.
(W2j

(L,]
5. Compute M- and determine the relative head correction, CH, from Fig. 20.26.
6. Compute y0 for the headwall opening using
y0=

^7=
IC2W2V^]

2/3

-Cn

(20.32)

which is based upon the rectangular weir equation, Q2 = C2W2\^2g(y0 + CH)3/2.

7. Compare y0 from Step 3 for the crest and y0 from Step 6 for the headwall opening. The
larger value of y0 controls. If the crest controls, adjust V0 from Step 3 using the following procedure:
a. Compute y0/W2 and determine the correction for the head, C1, using Fig. 20.27.
b. Compute L1JW2 and determine the correction for the box inlet shape, C5, using Fig.
20.28
c. Compute W{/LC and determine the correction for the approach channel width CA,
using Fig. 20.29
d. Compute W4JW2 and determine the correction for dike effect (proximity of dike to
box inlet crest) CE using Table 20.3

COEFFICIENT OF DISCHARGE, C2

HYDRAULIC DESIGN OF THE BOX - INLET DROP SPILLWAY

FIGURE 20.25 Coefficient of discharge, with control at headwall opening. (From Corry et al.,
1975)

FIGURE 20.26 Relative head correction for HJw2 > 1V4 with control at headwall opening.
(From Corry et al., 1975)
e. Determine the adjusted y0:
? -(sW/wJ20

(2033)

8. Compute the critical depth in the straight section yc using

t/3

'-ml
9. Compute the critical depth at the exit of the stilling basin yc3 using

-tej

10. Compute the minimum length of the straight section L2 using

f
0.2
L2 y
~ < (LA
IkJ

(20.36)

for values OfL 1 XW 2 ^ 0.25.

11. Compute the minimum length of the stilling basin using

DISCHARGE COEFFICIENT, C1

CORRECTION FOR HEAD

FIGURE 20.27 Discharge coefficient and correction for head, with control
at box-inlet crest. (From Corry et al., 1975)

CORRECTION FOR BOX - INLET SHAPE, Cg

CORRECTION FOR APPROACH - CHANNEL WIDTH, CA

W
FIGURE 20.28 Correction for box-inlet shape, with control at the box-inlet crest (-c > 3).
From Corry et al., 1975)

FIGURE 20.29 Correction for approach-channel width, with control at box-inlet

crest. (From Cony et al., 1975)

TABLE 20.3 Correction for Dike Efect CE: Control at the Box-Inlet Crest
(Control at Box-Inlet Crest)
W4AV2
L1IW2
OO
0.5
0.90
1.0
.80
1.5
.76
2.0
.76
Source: Corry et al (1975)

Ol
0.96
.88
.83
.83

0.2
1.00
.93
.88
.88

LB

0.3
1.02
.96
.92
.92

0.4
1.04
.98
.94
.94

_(i)
2L
i
(W2)

0.5
1.05
1.00
.96
.96

0.6
1.05
1.01
.97
.97

(20.37)

and

L3 = ^f^

(20.38)

and choose the larger value of L3.

12. Compute the minimum tailwater depth over the basin floor using
V3 = 1.6vc3

if

W3 < 11.5vc3

(20.39)

or

?3 = yC3 + 0.052PF3

if

W3 > 11.5yc3

(20.40)

13.Compute the height of the end sill /I4 using h4 = y3/6.

14. Determine the number of longitudinal sills:
If W3 < 2.5W2, use two sills;
if W3 > 2.5W2, use four sills.
When two sills are used they should be located at a distance W5 on each side of the
centerline. When four sills are used, the two additional sills should be located parallel
to the outlet centerline and midway between the center sills and the sidewalls at the
stilling basin exit.
15.Compute the minimum height of the sidewalls above the water surface at the exit of
the stilling basin Ji3 using /I3 = v3/3. Sidewalls should extend above the tailwater surface under all conditions.
16. Wingwalls should be triangular in elevation and have a top slope of 45 with the horizontal. The top slopes can be as flat as 30. Wingwalls should flare in plan at an angle
of 60 with the outlet centerline. The flare-wall angle can be as small as 45.
Wingwalls parallel to the outlet centerline should not be used.

20.2.5

RIPRAPBASINS

The riprap basin recommended in Corry et al (1975) for culverts is shown in Fig. 20.30,
which is a preshaped basin lined with riprap. The surface of the riprapped floor of the
energy dissipating pool is at an elevation hs below the culvert invert, where hs is the
approximate depth of scour that would occur in a thick pad of riprap if subjected to the
design discharge (Corry et al 1975). The length of the pool is the larger of I0hs or W0, and
the overall length of the basin is the larger of I5hs or 4W0. The ratio hs/d5Q should be less
than 4 [(HJd50) < 4] .Figure 20.31 provides design detaills for riprapped culvert energy
basins.

NOTEB

IFT
BOARD

NOTEA - SUFFI
IF EXI f VtLOClIY OF BASIONNAL
IS SPECI
EXTENDN A-ABASISUCH
N AS REQUIRED TO OBTAIN
AREAFFIATEIED.DSECTI
SECTIOCNIENTAREACROSS-SECTI
AT SEC. A-A) - SPECI
EXITOVELOCI
TV. THAT O^ /(CROSS
NOTE B - FLOOR
WARP OFBASIBASIN TON SHOULD
CONFORMBE ATTOTHENATURAL
STREAM
CHANNEL.
TOP OF RIPRAP IN
CHANNEL BOTTOM AT SEC. A-A. SAME ELEVATION OR LOWER THAN NATURAL
DlSSlPATOR POOL
Sh1APRON
ORW0MIN.
100,OR3W
0MIN.
TOPOFBERM
TOP OF RIPRAP
CHANNEL
ZONTAL

TO SUPPORT RIPRAP
THI
COPTIONAL
KENEDORSLOPI- CONSTRUCT
NG
TOE
IF DOWNSTREAM
CHANNEL

SECTION

NOTE B
EXCAVATETOTHISLINE,
NOTE:
W0 - PIDIAMETER
FOR
PE CULVERT
W0 - BARREL
WIDTH
FOR
BOX
CULVERT
W = SPAN
OF PIPE-ARCH
CULVERT
SYMM ABOUT
APRON

CULVERT

HORIZONTAL

BERM
AS REQUIPRRAPED
TOSUPPORTRI

FIGURE 20.30 Details of riprapped culvert energy basin. (Corry et al., 1975)

DESIGNDISCHARGE-Q
WETTED AREA AT BRINK OF CULVERT
d50 = THE MEDIAN SIZE OF ROCK
BYWEIGHT.
OR
ANGULARROUNDEDROCK
ROCK.
EQUIVALENT BRINK DEPTH
BRINK DEPTH FOR BOX CULVERT
FOR
NON-RECTANGULAR
SECTIONS

CULVERTBRINK

RELATIVE DEPTH OF SCOUR HOLE -^

Ve

?. SECTION

RIPRAP MAY BE REQUIRED

ON BANKS AND CHANNEL
BOTTOM DOWNSTREAM
FROM BASIN - SEE DESIGN
EXAMPLE IN TEXT.

FROUDE NUMBER
FIGURE 20.31 Relative depth of scour hole versus Froude number at brink
of culvert with relative size of riprap as a third variable.

REFERENCES
Blaisdell, F. W., Flow Through Diverging Open Channel Transitions at Supercritical Velocities,
SCS Report No. SCS-TR-76, U.S. Department of Agriculture, April 1949.
Blaisdell, F. W., and C. A. Donnelly, "The Box Inlet Drop Spillway and Its Outlet", Transactions,
of the American Society of Civil Engineers, 121:955-986, 1956.
Blaisdell, F. W., The SAF Stilling Basin, U. S. Goverment Printing Office, 1959
Corry, M. L., P. L. Thompson, E. J. Watts, J. S., Jones, and D. L. Richards, Hydraulic Design of
Energy Dissipators for Culverts and Channels, Enqineering Circular 145, Federal Highway
Administration, U.S. Department of Transportation, Washington DC, 1975.
Fedeard Highinay Administratin , Hydraulics of Bridge Waterways, Hydraulic Design Series No. 1,
Federal Highway Administration, U.S. Department of Transportation, Washington, DC, 1978.
Ippen, A. T, "Mechanics of Supercritical Flow", Transactions of the American Society of Civil
Engineers,, lib; 268-295, 1951
Rand, W, "Flow Geometry at Straight Drop Spillways," Paper No. 791, Proceedings of the
American Society of Civil Engineers, VoI, 81, pp. 1-13, September 1955.

CHAPTER 20
FLOW TRANSITIONS AND
ENERGY DISSIPATORS
FOR CULVERTS

AND

CHANNELS

Larry W. Mays
Department of Civil and Environmental Engineering
Arizona State University
Tempe, Arizona

20.1

FLOWTRANSITIONSFORCULVERTS

Flow transitions are changes in the cross section of an open channel over short distances.
They are designed to have a minimum amount of flow disturbance. Figure 20.1 illustrates
the various types of transitions; the two most common ones are the abrupt (headwall) and
the straight line (wingwall).
Highway culverts typically are designed to operate with an upstream headwater pool
that dissipates the of the channel approach velocity. This type of situation does not require
an approach flow transition. Outlet transitions (expansions) should be considered in the
design of all culverts, channel protection, and energy dissipators. Transition design can be
categorized as
culverts with outlet control (subcritical flow) and
culverts with inlet control (supercritical flow).
20.1.1 Culverts with Outlet Control
Culverts with outlet control can have abrupt expansions or gradual transitions. In an
abrupt expansion, the water surface plunges or drops rapidly and the flow spreads out.
The potential energy stored as depth is converted to kinetic energy with a higher velocity of flow. The transition (apron) end velocity can be determined using the experimental
results of Watts (1968) (Figs. 20.2 and 20.3). Figure 20.2 relates the average depth brink
depth ratio (yA/y0) for a rectangular outlet to the Froude number. Figure 20.3 is a similar
relationship for pipe culverts. These curves in Figs. 20.2 and 20.3 were developed for
Fr's ranging from 1.0 to 2.5, the applicable range for most abrupt outlet transitions.
Usually, a low tailwater is encountered at culvert outlets and the flow is supercritical on
the outlet apron.

Simons, D. B., M. A. Stevens, and F. J. Watts, Flood Protection at Culvert Outlets, CER No.
69-70-DBS-MAS-FJWA, Colorado State University, Fort Collins, CO. 1970.
U.S. Army Corps of Engineers, Hydraulic Design of Flood Control Channels, Engineering and
Design Manual EMl 110-2-1601, pp.2026, July 1970.
U.S. Bureau of Reclamation, (USBR), Design of Small Dams, U.S. Government Printing Office,
Denver, CO, 1987.
Watts, F. J., Hydraulics of Rigid Boundary Basins, doctoral dissertation, Colorado State University,
Fort Collins, CO, 1968.

CHAPTER 21
HYDRAULIC DESIGN OF
FLOW MEASURING
STRUCTURES

John A. Replogle and Albert J. Clemmens

USDA-ARS Water Conservation Laboratory
Phoenix, Arizona
Clifford A. Pugh
US Bureau of Reclamation,
Denver, Colorado

21.1

INTRODUCTION

Experienced water providers and users can use this chapter as a quick review of hydraulic
principles related to water measurement and its relation to hydraulic design for environmental considerations.
The hydraulic design of flow measuring structures usually confronts the engineer with
two opportunities. One is the design of measurement structures in a retrofit situation and
the other is in original project design. The retrofit mode is usually difficult and requires
much innovation just to obtain passable function within the space and sizing limitations
and other constraints usually imposed. Because of the increasing emphasis on quantifying
flow rates and volumes in most aspects of water resource planning and management, the
retrofit applications currently dominate the design problems.
Most textbooks deal with recommending ideal installation situations and retrofit projects appear to be unable to comply without great economic impact. This too frequently
can lead to arbitrary compromises that produce poor measurement performance. Even
new installations may be limited by space requirements. This may force design decisions
into the final construction that compromise accuracy. This chapter will strive to show the
design concepts available, particularly those useful for designing both new and retrofit
installations, and will point out measurement behaviors to be expected from various compromises. This chapter suggests those deviations that cause least impact and guides the
designer to choices that may be hydraulically acceptable and still meet structural goals.
Of the numerous flowmetering methods available to the hydraulic engineer, most are
based on well-established hydraulic principles and are amenable to design manipulations
of size, shape, and response. While this aspect of flow measurement is documented in several handbooks and texts, the design and retrofit of sites to accommodate and facilitate
measurement is not as well described or is described in a scattered assortment of books
and articles.
Pipeline flows of water are usually less complicated to measure than open-channel
flows, most obviously because the flow area does not change significantly with flow rate.

Consequently, many applications of pipeline flows are held to stricter accuracy standards
than channel flows can reasonably achieve. Thus, channel flows and their measurement
are usually limited to large delivery volumes and to accuracies acceptable to the related
activities, such as sewer flows and irrigation deliveries.
The purpose of this chapter is to consolidate design information for evaluating a flow
measurement site, selecting a flow measuring system, and adapting the measuring site to
optimize measuring and other functions that may be desired from the site. Emphasis will
be on open-channel flow measurements because that is a likely need of the hydraulic engineer. Pipe flowmeters in water supply will also be discussed in lesser detail because the
major application of the many types of pipe flowmeters is well covered in the chemical
and petroleum industry literature.
Experienced readers may wish to further investigate and seek more advanced references in hydraulics and fluid mechanics. Extensive information on fluid meter theory and
detailed material for determining coefficients for tube-type meters is given in American
Society, of Mechanical Engineers (ASME) (1959, 1971) and revisited with modern
updates in books by Spitzer (1990) and Miller (1996). Brater and King (1982) have a thorough discussion of general critical depth relations and detailed relationships for most
common hydraulic flow section shapes in open channels. Bos (1989) covers a broad segment of open-channel water measurement devices.

21.2 HYDRAULIC CONCEPTS RELATED TO WATER

MEASUREMENT
21.2.1 Basic Concepts for Pipe and Channel Flows
Flow can be classified into closed conduit flow and open-channel flow. Open-channel
flow conditions occur whenever the flowing stream has a free or unconstrained surface
that is open to the atmosphere. Flows in canals or in vented pipelines that are not flowing
full are typical examples.
In hydraulics, a pipe is any closed conduit that carries water under pressure. The filled
conduit may be square, rectangular, or any other shape, but is usually round. If flow is
occurring in a conduit but does not completely fill it, the flow is not considered pipe or
closed conduit flow, but is classified as open-channel flow.
Flow rate in a pipeline responds mainly to the pressure gradient or head difference that
exists between two points along the pipeline, modified by the frictional resistance to flow
caused by pipe length, pipe roughness, bends, restrictions, changes in conduit shape and
size, the nature of the fluid flowing, and the cross-sectional area of the pipe.
In open-channel flows, the pressure gradient, or energy grade line, is controlled
mainly by the force due to gravity, which is influenced by the channel slope, resistance
from the channel wall roughness, the channel shape, and the flow area. The fluid is usually water.
Basic flow metering in both pipe flow and open channels depends on determining an
average flow velocity by some means and combining it with the flow cross-sectional area.
For open channels, a common means involves current meter measurements where metered
point velocities are applied to their applicable subareas and summed over a flow cross section. Exceptions include tracer-dilution techniques that do not require flow area or velocity. The uses of tracer techniques are applicable to special pump calibrations and some difficult channel flows (mountain streams). They are avoided for most city water distribution
systems, sewer flows and irrigation applications because of the general expense with han-

dling the equipment and doing the analysis. The most used techniques applicable to openchannel systems, including sewer flows and irrigation canal flow measurements, depend
on exploiting the special velocity properties of critical flow, as discussed in a section
21.2.3.
Continuity equation. The first basic equation for water flowing in either pipes or channels is the continuity equation, which simply states that discharge rate (volumetric flow
per unit time), Q, is equal to flow cross-sectional area, A, times flow mean velocity, V,
through the flow cross section, or
Q = AV

(21.1)

Bernoulli energy equation. Another basic equation involves energy relations and is
also applicable to both pipe and channel flows. The most familiar form is for closed pipe
flow, wherein the basic energy principles are described by the Bernoulli energy equation.
For two locations along a pipe at stations 1 and 2,(Fig. 21.1), the Bernoulli equation can
be expressed as
V2
V2
ZI + h\ + -^- = ^2 + h2 + -^- = constant

(21.2)

where the terms are expressed in length dimensions as z = the height from an arbitrary
reference plane (datum) h = the pressure head V = average velocity through the pipe
cross-section at the designated location V2/2g = the velocity head g = the gravitational
constant i ,2 = subscripts denoting the respective locations along the pipeline.
This equation is based on uniform velocity across the conduit area and no energy losses. However, in real fluid flows, nonuniform velocities exist and friction causes energy
conversion to heat. Typically, these velocities are zero at the walls and reach a maximum
profile velocity near the center of the flow. If the flow is viscous flow in a round pipe, the
flow profile is parabolic, that is, "bullet-shaped." If the velocity is fully turbulent, the bullet-shape is much flattened, with steep velocity gradients near the wall and nearly uniform
profile across the remainder of the pipe. These idealized profiles can be skewed drastically
by regulating valves, structures, conduit bends and other flow obstructions. Therefore,
application of these equations depends on knowing, or controlling, the velocity profile so
that the average velocity in the conduit cross section can be inferred.

Datum
FIGURE 21.1 Energy balance in pipe flow.

Equation (2.2) requires some adjustments to convert it to the energy equation, which
is useful in analyzing flows in pipes or open channels with a small slope (Chow, 1959).
First we introduce correction factors, GL1 and OC2, called the velocity distribution coefficients, to account for the computational expediency of using the average velocities, V1 and
V2, to compute the kinetic energy term, W2g, at the respective locations 1 and 2 along a
channel. These values for the usual range of turbulent flows in water usually range from
about 1.01 to 1.05, although for thick petroleum products in pipe flows and low velocity
flows, the value can approach a value of 2. Second, a term, hp for the loss of energy
between the two points is included. The result is
V2
v2
Z1 + H1 + a, -i = Z2 + /I2 + a2 -^ + hf
(21.3)
5
o
21.2.2 Pipe Hydraulics
Reynolds number. The behavior of flow in pipes is governed primarily by the viscosity of
the fluid. In pipeline flows, the ratio between the dynamic forces and the viscus forces is
important for defining the limits between laminar and turbulent flows and other functions
of pipe flow. This ratio is called the Reynolds number, Rn, and is defined as
Rn = ^-

(21.4)

where V = the velocity of the flow, Lc = characteristic length, typically the pipe diameter, D and v = the kinematic viscosity.
Headloss characteristics in pipes. The Reynolds number, Rn, defined above, represents the effect of viscosity relative to inertia and is used to define appropriate flow ranges
for headloss equations in pipe flow. For example, headloss is proportional to the square of
the velocity, when the velocities and pipe size combinations defined by a pipe-diameterbased Reynolds number, Rn, greater than about 1000. Most of the flows of interest in general hydraulic engineering have Reynolds numbers greater than 1000. Some exceptions
are found in drip or trickle irrigation systems common in agricultural and urban landscape
settings.
The headloss, hf for Rn greater than the minimum value of about 1000 is traditionally
expressed in terms of a friction factor,/, the pipe diameter, D, pipe length, L, and the velocity head, V2/2g, where g is the gravitational constant, and V is the average velocity, as
hf = / I

(21.5)

The value for/is usually obtained from a Moody diagram which is a graphical representation of the/value in terms of the Reynolds number, the roughness height of the pipe
wall material, e, and the pipe diameter, D. The Moody diagram is a graphical solution of
the Colebrook function
1 = -2log^
+ ^=]
Vf
I3-7
RnVf)

(21.6)

The e values range from 0.0000015 m for smooth plastic pipe to 0.00026 m for cast
iron pipe. Concrete pipe ranges from about 0.0003 m to 0.003 m (Daugherty and
Ingersoll, 1954). The equation can be readily solved by iteration techniques using a computer spreadsheet.

21.2.3 Channel Hydraulics

Hydraulic mean depth. The hydraulic mean depth, Dm [U.S. Bureau of Reclamation
(USBR), 1997] is the flow cross-sectional area, A, divided by the flow surface width, J1, or
Vm = J

(21-7>

For conduits such as pipes flowing nearly full, the surface flow width may be narrow,
and Dm may be a larger value than the physical water depth. For the usual natural channels and most canals, Dn is interchangeable with average depth. Sometimes it is simply
called the hydraulic depth (Chow, 1959).
Froude number. Open-channel flow behavior is governed primarily by gravity
forces. The ratio of the inertial forces to the gravity forces is called the Froude number,
Fr, and is defined by
Fr =

Y
(21.8)
v^:
where V the velocity of the flow, g = the gravitational constant, Dn = the hydraulic
mean depth.
The Froude number applies to most open channel flows and is used for defining model
scale ratios and estimating stable flow characteristics in open channels.
Specific energy. It is useful to define the energy equation in terms of the local channel bottom instead of an arbitrary datum. This is called the specific energy, ", and is given
by:
E = y + ^(21.9)
2g
That is, the specific energy is equal to the sum of the depth of flow y and the velocity
head (Fig. 21.2).
Critical flow and critical depth. In open channels a flow phenomenon occurs that
does not happen in closed pipe flows. The process is called critical flow. Critical flow is
defined for open-channel flows as the maximum discharge for the minimum specific energy, that is, critical flow represents the minimum combination of potential energy (depth
of flow, y) and kinetic energy (velocity head, V2/2g) for the given discharge (Chow, 1959).
The depth of flow then is the critical depth. By virtue of the continuity equation, for a constant discharge at critical flow, an increase in depth must necessarily be accompanied by
a decrease in velocity, which is called subcritical velocity. Conversely, a decrease in depth
for the same flow rate necessarily requires an increase in velocity, which is called supercritical velocity.
When critical flow occurs in an open channel it can be shown (Chow, 1959) that
V2 D
oc^ = -f

(21.10)

where Vc = mean flow velocity, g = gravitational constant, Dm = hydraulic mean depth,

and a = velocity distribution coefficient (Chow, 1959).
This can further be combined with the continuity equation, Eq. (21.1), to express the
critical flow discharge rate, Qc, as

EnergyGrade Line
Hydraulic Grade Line

FIGURE 21.2 Specific energy balance.

& = AJ%%

<2U1)

In practice, the water surface slope in a contraction is relatively steep and the precise plane of the critical flow section is not easily or reliably located. Thus, the data
for accurately evaluating the hydraulic depth, Dm, is not readily obtained. For critical
flow flumes, the flow depth is therefore not measured at this critical section, but
instead a depth is measured in the upstream channel, where the velocity head is computable or is minimal. The critical depth is then mathematically derived based on energy principles described by Bos et al. (1991). These flumes, sometimes referred to as
the computable flumes that rely on critical flow theory, will be discussed in more detail
in Section 21.7.
For maximum discharge for minimum energy, the condition described above for critical flow in open channels, it can be shown that
af = A = ^

<2U2)

where: A = the channel flow area, T = the top width of the channel flow, Dm = the
hydraulic mean depth, and Vc = the critical velocity.
Thus, the velocity head at critical flow is equal to half the hydraulic mean depth, sometimes called hydraulic depth, Dm = AIT (Chow, 1959).
From the above,
,Vc
= 1 = Fr
(21.13)
VgD m la
where Fr is the Froude number defined above. Thus, at critical flow the Froude number is
unity. Also note that the Froude number can be defined by Fr = VIV0 for velocities other
than critical.
Normal depth. Yet another depth is associated with open channel flows, the normal depth. When the flow in an open channel does not change from station to station,
the flow is said to be uniform and the bottom slope, the hydraulic grade line, and the
energy grade line are all parallel to each other. Figure 21.2 shows the condition when
the flow is not uniform.
Modular limit. If the downstream depth in a channel is too deep, the backwater will
prevent critical depth from occurring. The flow is considered to be submerged whenever
the downstream water surface exceeds the crest elevation of a channel control, such as a

weir or flume. For flumes, particularly, this submergence has little effect on critical depth,
and free flow exists until a certain limiting submergence for that particular flow module
called the modular limit is reached. At some point of submergence, the upstream flow
depth is affected, and the modular limit is exceeded, and free flow does not occur. The
modular limit is defined as that limiting submergence ratio, and is based on the ratio of
the downstream depth to upstream depth. The modular limit occurs when the downstream
backwater causes more than 1 percent change in the calculated discharge in a particular
flow module, or device (Bos, 1989). When the modular limit is exceeded, the flow is
called nonmodular.
21.2.4 Energy Balance Relationships in Channels
Hydraulic problems concerning fluid flow are commonly described in terms of conserving kinetic and potential energy, and are conveniently expressed using the classical
Bernoulli equation in combination with the Continuity equation. The applications of these
equations are generally well documented, particularly for pipe flows, in texts and handbooks and are not repeated here (Brater and King, 1982; Miller, 1996). The case for open
channels is less complete, but is given considerable treatment in Brater and King (1982),
Chow (1959), and Herschy (1985). The computational uncertainties evolve from the
effects of friction and viscosity that distort the classic assumptions of a uniform velocity
profile across the fluid stream. When accountings for friction and flow profile are successfully applied, the results for discharge computations are usually good to excellent for
both pipes and open channels (Bos et al., 1991).
Headloss characteristics in channels. In terms of frictional headlosses, the wetted
perimeter, Pw, of the flow is important. Hydraulic radius, Rh, is defined as the area of the
flow section, A, divided by the wetted perimeter, Pw, or
Rh =
p

(21.14)

Conversely, the wetted perimeter times the hydraulic radius is equal to the area of an
irregular flow section. The hydraulic radius of a channel can be compared to the radius of
a pipe, r, with a cross-sectional area A = nr2 and a circumference or wetted perimeter Pw
= 2 nr. Under these conditions, the hydraulic radius compares to the pipe radius, and to
the pipe diameter, D, as
r D
R
=
(2L15)
=2
^
h
The Manning's formula. Canal and stream discharge rates are usually estimated with
use of the Manning's formula. Many open-channel flow equations have been proposed,
but the most used is the Manning's formula. This expression is partly rational and uses an
empirical coefficient, rc, that is used in both the SI and American unit systems. In general
form it becomes
v=^LR2i3Sm

(2L16)

where V = average velocity, n = the Manning's roughness coefficient, Rh = the hydraulic

radius, S6 = the energy-line slope, and Cm = conversion of units: 1.0 for metric units and
1.486 for American units.
The factor Se is the slope of the energy line. Note that the bed slope of the channel, S09
and the slope of the water surface, Sw, are not to be used. These parameters are, however,

equal to Se when uniform flow, with the resulting normal depth occurs. As defined above,
normal depth occurs when a channel flow approaches uniformity from station to station
along the channel (Chow, 1959).
For design purposes, the n value for concrete lined canals is usually about 0.014. A
good finish can lower it to 0.012, while concrete in poor condition and channels constructed with shot crete or gunite, usually have n values from 0.016 to 0.018. In some
instances, concrete lined canals, with significant algae growth, have experienced n values
as high as 0.032. This latter value approaches the values usually experienced with unlined
channels, 0.03-0.04. Thus, for reliable application, the use of Manning's formula requires
field experience and on-site inspection of the channel being computed.
21.2.5 Modeling Characteristics for Open Channels
For flowing water in open channels, fluid friction is a factor as well as gravity and inertia. This would seem to present a problem for hydraulic scale modeling, because both
dynamic and kinematic similarity are difficult to achieve simultaneously. Fortunately for
most open-channel flows, there is usually fully developed turbulence. Thus, the fluid friction losses are nearly proportional to V2, and are nearly independent of Reynolds number,
Rn, with rare exceptions.
This means that in open-channel flows, inertia and gravity forces dominate over viscous forces (associated with pipe flows) and are a function of the Froude number, Fn,
alone. Geometric similarity between a model and a prototype then provides kinematic
similarity. For kinematic similarity the ratios of the respective velocities are everywhere
the same. The velocity ratio, Vr, is the velocity in the prototype, Vp, divided by the velocity in the model, Vn, or
Vr = ^(21.17)
m
For Froude modeling, and from the definition for Fn, we note that V is proportional
to the square-root of a length, L (for open channels we used the hydraulic depth, D1n)
with the gravitational constant, g, assumed to be constant. Thus, the above equation can
be written as
"^LT = Y
where Lr = the length ratio between prototype and model dimensions, Lp: Lm

(2L18)

Because the velocity varies as VZ^ and the cross-sectional area as L2 it follows
that
Qp'Qm = LT 1

(21-19)

O
(J Y/2
-SL = U
XZm \.LmJ

(21.20)

or

This equation is valid when all the physical structure dimensions and the heads are of
the same ratio. For example, it can be used to convert a flume rating for one size to that
of a similar flume of another size. Scale modeling works best for determining calibrations
in a range of Lp:Lm less than about 10:1, although ranges exceeding 50:1 have sometimes
been used for studying special situations.

21.3

BASIC PRINCIPLES OF WATER MEASUREMENT

Flow is usually measured by determining an average flow velocity and using the flow area
to compute the volume discharge. Flow meters then have the function of detecting this
velocity and combining it with the physical information of the conduit to produce a useable readout. This is easily demonstrated for closed conduits. Propeller meters, ultrasonic
meters, laser-Doppler velocimeters, electromagnetic meters, Venturi meters, and orifice
meters all are based on inferring a basic velocity measurement applied to a flow area for
a discharge rate.
For open channels, many flumes depend on determining the velocity based on energy
principles of critical flow. Weirs are usually described in terms of orifice flow integrated
over the weir width and the crest depth. Again these are basically velocity expressions for
flow through a defined area.
Dilution techniques applicable to both closed pipe and open channel flows depend on
detecting the amount of fluid added to a known starting amount of tracer material. The
dilution ratio determines the discharge ratio, in the case of constant injection of a tracer.
The tracer may be a chemical or even injected heat or heated fluid.
Electromagnetic meters depend on generating voltages by flowing a conductive fluid,
usually water, through a magnetic field to produce a velocity indication.
21.3.1 Water Meter Classification
Flow measuring devices are commonly classified into those that are rate meters and measure discharge rate as the primary reported indication and those that are quantity meters
and measure volume as the primary indication. The latter include weighing tanks and
batch volume tanks and are used mostly in laboratory settings as flow rate standards.
Devices in either of these broad classes can again be divided according to the physical
principle that is used to detect that primary indication (ASME, 1959). The meter part that
interacts with the flow to produce the primary indication is referred to as the primary
device. This interaction exploits one or more of a few physical principles, such as pressure force, energy conversion, weight, electrical properties, mixing properties, sonic properties, and so on, to generate a signal. Primary devices are thus limited in number and variety. Secondary devices convert the primary interaction into useable readout. These secondary devices are numerous and relatively unlimited in configuration and variety. The
function of one class can be converted into the response of the other with suitable secondary devices.
Some water measuring devices particularly suitable to municipal water supply, wastewater treatment, agricultural irrigation, and drainage applications are the historical rate
meters that are treated in most hydraulic text. These include (1) weirs, (2) flumes, (3) orifice meters, and (4) Venturi meters.

Head, h, or upstream depth, commonly is used for the open channel devices such as
flumes and weirs. Either pressure, /?, head, /i, or differential head, A/z, or differential pressure, Ap, is used with tube-type devices, such as Venturi meters and orifice meters.
Venturi meters in pipelines and long throated flumes in open-channel flows are examples where the energy principles and the flow accountings mentioned above give good to
excellent computational results with minor dependency on empirical coefficients (Bos,
1989; Bos et al., 1991).
21.3.2 Installation Requirements
Special difficulties arise in applying velocity profile and friction accountings when insufficient pipe or channel exists upstream from a flow measuring device. This is needed to
ensure that predictable and acceptable velocity profiles are presented to the meter.
Frequently pipe or channel lengths can be significantly shortened by special structural
flow conditioners. These structural measures then become a design option. Some of these
are discussed below and in section 21.3.3
Designs for pipe discharges are well described in textbooks and in standard handbooks. The design difficulties center around selecting appropriate metering candidates for
accomplishing the measuring function and in providing an appropriate environment for
economical, accurate, and serviceable operation.
In the case of pipe flows, recommended straight pipe lengths, in terms of pipe diameter, are to be provided upstream of the meter to assure reasonable operating accuracy. These
lengths depend on the flow pattern presented to the meter primarily caused by valves and
pipe elbows upstream from the meter. The number and orientation of elbows greatly influence the circulation patterns and flow profile distortions presented to the meter.
Open-channel flow water measurement generally requires that the Froude number of
the approach flow be less than 0.5 to prevent wave action that would hinder or possibly
prevent an accurate head determination.
Energy concepts are used to describe Venturi meters in pipe flows based on the
Bernoulli equation in which part of the pipe forms a contracted throat that necessarily
changes the flow velocity and hence converts some of the static pressure to velocity head.
The decrease in static pressure is the basis for flow detection. A similar concept can be
applied to open channels. A historical version is the so-called Venturi flume (Brater and
King, 1982) that detects the change in water surface elevation between an upstream station and in a contracted section. However, this small change is difficult to accurately
detect, so the direct concept is not used. Rather, contractions are designed to be severe
enough to force critical flow velocities in the contracted section. Thus, only an upstream
head is needed to define the flow energy and flow area which can be converted to discharge rate. These are generally called critical-flow flumes. The flow condition where
only one head measurement is needed is called freeflow.
The critical-flow flumes themselves consist of those called long-throated flumes that
force parallel flow in the contracted, or control, section, called the throat, and those that
have curvilinear flow in the throat and are called short-throated flumes. The limiting
throat control section is the sharp-crested weir consisting of a thin plate. Thus, for
flumes and weirs one unique head value exists for each discharge, simplifying the calibration procedure.
However, if the downstream flow level submerges critical depth enough to affect the
upstream reading, the modular limit is exceeded, and free flow does not occur. When
exceeded, separate calibrations at many levels of submergence are then required, and two
head measurements are needed to measure flow. This condition generally is to be avoided in meter site design because it reduces the accuracy of the measurement and increases

the difficulty of flow determination. The modular limit for sharp-crested weirs, in practice, is less than zero, requiring full clearance of the overfall nappe of at least 3 cm, while
short-throated flumes can usually tolerate 65 percent to 70 percent submergence. Longthroated flumes can tolerate from 70 percent to 90 percent depending on flow conditions
and flume size.
Designing flumes for submerged flow beyond the modular limit decreases the accuracy of the flow measurement. Sometimes flumes and weirs can be overly submerged unintentionally by poor design, construction errors, structural settling, attempts to supply
increased delivery needs with increasing downstream heads, accumulated sediment
deposits, or weed growths. Sometimes use of the submerged range beyond the modular
limit is an economic compromise.
Approach flow conditions for pipes. Water measurement devices are generally calibrated with certain approach flow conditions. The same approach conditions must be
attained in field applications of measuring devices. Poor flow conditions in the area just
upstream of the measuring device can cause large discharge indication errors. For open
channels, the approaching flow should generally be subcritical. The flow should be fully
developed, mild in slope, and free of curves, projections, and waves.
Pipeline meters commonly require 10 or more diameters of straight pipe approach.
Fittings and combinations of fittings, such as valves and bends, located upstream from a
flow meter can increase the number of required approach diameters. Several references
(ASME, 1971; ISO, 1991) give requirements for many pipeline configurations and
meters. These are discussed in detail by Miller (1996).
Flow conditioning options. Many installations, especially in retrofit situations, do
not provide for sufficient lengths of straight pipe to remove velocity profile distortions
and swirl to an acceptable level. Therefore, the designer may need to use flow conditioners in combination with straight pipe lengths. Swirl sensitivity varies widely. Some meters
are particularly sensitive to swirl, such as the propeller and turbine meters. Magnetic flow
meters are somewhat less sensitive to radial velocities than single-path ultrasonic flow
meters. Venturi meters are less sensitive than orifice meters. For a swirl angle of 20 , the
discharge coefficient changes by about 1 percent for a Venturi meter with p = 0.32 ((3 is
the ratio of meter throat diameter to the pipe diameter) and about +10 percent for a similar orifice. Thus a swirl can increase the discharge through an orifice for the same differential head reading (Miller, 1996).
In pipeline flows, contractions can produce a central jet and also increase an incoming
swirl, while expansions tend to slow swirls and produce enough secondary flow to restore
flow profiles to some semblance of acceptability. These characteristics can modify the
straight pipe lengths needed or the type of flow conditioner to recommend (Miller, 1996).
Rough pipes also tend to reduce a swirl.
For flows, such as that encountered in sewage discharges and irrigation pipeline deliveries that originate from open channels, many of the tube-bundle types of flow conditioners can gather trash and cause maintenance problems. Many meter providers in these situations use fins or vanes that protrude from the wall and have sloped upstream edges that
shed trash. The vanes protrude about one-fourth of the pipe diameter into the flow, leaving the center core of the flow open. While these vanes can vary in number and length,
the logic being that the fewer the vanes the longer they should be in the direction of flow,
common configurations are four vanes that are about two or three pipe diameters long.
Vanes in themselves do not condition wall jets well. Field experience, has shown that troublesome flow profiles can be conditioned significantly by inserting an orifice into the

pipe. The orifice diameter is about 90 percent of the pipe diameter and is used to control
wall jets and force them to mix with the general flow. The orifice in itself tends to crossmix the jets and would appear to reduce spin. However, if the jets are symmetrical and an
initial swirl exists, orifices tend to increase the swirl. Inserting an orifice appears to be
supported by recent recommendations of Miller (1996) where it is stated: "To achieve a
fully developed profile, it is important that the flow be blocked or restricted close to the
wall, with the central core having the larger flow area."
The addition of vanes when space permits is recommended. Because orifices, in general, tend to force the flow to the pipe center while increasing spin, it appears best to place
the vanes upstream from the orifice. If they are placed downstream, the spin not only may
be increased, but the spinning central flow may not be touched by the vanes.
21.3.3

Examples of Flow Conditioning in Field Situations

Flow conditioning in an irrigation delivery pipeline. As mentioned previously, measuring devices frequently must be installed in flow situations that are less than optimal. A field
example occurred in Arizona where a large pipe was used as an outlet to a secondary canal
and a single-path ultrasonic meter placed in it was subjected to flow profile distortions. The
pipe was about 0.75 m in diameter and delivered approximately 400 L/s. The flow rate
readout was unstable, with fluctuations varying by about 15 percent. The problem appeared
to be caused by slowly spiraling flow induced by the bottom jet from a partly open pipe
inlet gate and a 45 elbow. This is similar to two closely spaced pipe elbows that are not in
the same plane, which can cause a spiral flow pattern (ASME, 1971).
A successful attempt to modify the jet and cause it to cross mix so that the jet effects
and the strength of the spiral flow were reduced, was accomplished by inserting a large Pratio orifice in the pipe (Fig. 21.3). This consisted of an annular metal ring with the outside radius approximately that of the pipe and an inside diameter about 10 percent less, or
an orifice with (3 = 90 percent. The orifice was installed about three diameters downstream from the elbow. The slight increase in headless was compensated by increasing the
upstream gate opening. The orifice can be constructed by cutting notches from an appropriately sized piece of angle iron or aluminum and bending it to a polygon that approxi-

Control O gate
Ultrasonic flow
meter

Orifice plate
with large
opening

Spiral flow

Improved profile
FIGURE 21.3 An orifice plate with a large opening is used to condition a flow profile.

mates the circle diameter of the pipe interior. Some leakage around the ring is acceptable.
For propeller meters, additional vanes projecting from the walls may be needed to further
reduce spiral flow. These vanes would be placed upstream from the orifice. In this installation, the fluctuation was reduced to within about 3 percent.
Flow conditioning in channels. By analogy and using a minimum of 10 pipe diameters of a straight approach channel, open channel flow would require 40 hydraulic radii of
straight, unobstructed, unaltered approach, based on the calculation of hydraulic radius for
circular pipes being equal to one-fourth the pipe diameter, (Eq. 21.15). This would translate for very wide channels into approximately 40 times the flow depth. For narrow channels that are as deep as they are wide, this would compute to be about 13 channel depths
or top widths.
Other recommendations on approach channel criteria are presented by Bos (1989)
and USBR (1997). Major features of that criteria follow:
If the control width is greater than 50 percent of the approach channel width, 10 average approach flow widths of straight, unobstructed approach are required.
If the control width is less than 50 percent of the approach width, 20 control widths of
straight, unobstructed approach are required.
If upstream flow is below critical depth, a jump should be forced to occur. In this case,
30 measuring heads of straight, unobstructed approach after the jump should be provided.
If baffles are used to correct and smooth out approach flow, then 10 measuring heads
(10 /I1) should be placed between the baffles and the measuring station.
Approach flow conditions should be continually checked for deviation from these conditions as described in Bos (1989) and USBR (1997).
The baffles described above can become unacceptable maintenance problems in open
channels. Some field expediences are therefore described that have been found to work in
specific instances, but have not been studied for assured design generalizations. Nevertheless,
these constructions are but small extensions to currently accepted practices in pipe flows.
Applications for open-channel flow conditioners include abrupt channel turns, sluice
gate outflows, and channels downstream from a hydraulic jump. The abrupt turns may
benefit from floor and wall mounted vanes or fins. Based on pipe flow experience, and
assuming the channel is half of a closed conduit, these fins or vanes would probably be
about 10 percent to 15 percent of the channel depth.
As in pipe flow, wall jets that can develop downstream from sluice gates appear to
need treatment. This can be in the form of a structural angle bolted on the channel floor
and up the walls. Suggested size, based on the pipe flow analogy, is for the angle to be
about 5 percent into the channel flow depth. Whether the sidewalls need larger angles
when the channels are wide has not been tested.
21.3.4 Wave Suppression
Of special concern in open channels is wave suppression downstream from a sluice gate,
hydraulic jump, or an abrupt turn. Thus, the flow conditioners in channels have the additional task not present in pipe flows of surface wave suppression. Excessive waves in irrigation canals make reading sidewall gages difficult. These waves are usually caused by a
jet entry from a sluice gate or by a waterfall situation. The unstable surface can be 10-20
cm high and extend for tens of meters downstream.

Solidly attached
timber or concrete
FLOW
FIGURE 21.4 Wave suppresor design (From USBR, 1997).

Wave suppression in canals. A surface wave suppressor was tested by Schuster as

reported in USBR (1997). It basically was a constructed roof over the canal for a distance
equal to about four times the flow depth. The roof structure is inserted into the flow about
one-third the flow depth. All flow is forced to pass under the structure. Wave suppression is
between 60 percent and 93 percent (Fig. 21.4). For canals that usually flow at one level, this
wave-suppression method is appropriate. The wave suppressor shown in Figure 21.4 has
been successfully used in both large and small channels (USBR, 1997). An important aspect
is that the structure is fixed and not allowed to float. Floating suppressors are not effective.
Successful field applications of wave suppressors include some installations in trapezoidal irrigation channels, with 1:1 side slopes and 60-cm bottom width. They were flowing about 400 L/S at about 45 cm deep. While the velocity was not high, about 0.8 m/s,
the agitation from a flow entry gate was producing waves about 15 cm high. The suppressor "roof was only about 60 cm in the direction of flow, and penetrated the flow by
about 15 percent.
Another version that has worked in small channels is illustrated in Figure 21.5.
This can work with a single cross-member if the flow is usually at a fixed discharge
rate and becomes similar to the suppressor described above. In severe jet cases an
additional floor sill, about 10 percent of the flow depth in height, has been used successfully.
The length of the roof in the flow direction has not been well studied, but field observations seem to support a length greater than two lengths of the surface wave, if that can
be estimated, otherwise, use two to four times the maximum flow depth as described above.
To suppress waves in canals that do not always flow at the same depth, a staggered set
of baffles may help (Replogle, 1997). Because these will be submerged part of the time,
they must have a thickness that overlaps slightly to accommodate the vertical depth of
interest. To avoid obstructing the channel severely, these baffles probably should not
obstruct more than about 20 percent of the channel at any particular location. Staggering

FIGURE 21.5 Wave suppressor for variable-depth flows in a canal. (From

Replogle, 1997)

them as shown in Fig. 21.5 would accomplish this without excessive obstruction.
Rounding the upstream edges will help shed trash, but may be less effective in suppressing waves. Observe in the sequence of drawings in Fig. 21.5 that the staggering is upward
in the downstream direction. Note that the next baffle slightly overlaps the horizontal flow
lines so that flow passing over the top of one baffle is not allowed to free-fall and start
another wave. Fig. 21.5 a-c illustrate the general behavior as the flow becomes less deep.

21.4

MEASUREMENTACCURACY

Accurate application of water measuring devices generally depends upon standard designs
or careful selection of devices, careful fabrication and installation, good calibration data
and adequate analysis. Also needed is proper user operation with appropriate inspection
and maintenance procedures. During operation, accuracy requires continual verification
that all measuring systems, including the operators, are functioning properly. Thus, good
training and supervision are required to attain measurements within prescribed accuracy
bounds. Accuracy is the degree of conformance of a measurement to a standard or true
value. The standards are selected by users, providers, governments, or compacts between
these entities. All parts of a measuring system, including the user, need to be considered
in accessing the system's total accuracy.
As mentioned above, a measurement system usually consists of a primary element,
which is that part of the system that creates what is sensed, and is measured by a secondary element. For example, weirs and flumes are primary elements. A staff gage is a
secondary element.
Designers, purchasers, and users of water measurement devices generally rely on standard designs and manufacturers to provide calibrations and assurances of accuracy. A few
water users and providers have the facilities to check the condition and accuracy of flow
measuring devices. These facilities have comparison flow meters and/or volumetric tanks
for checking their flow meters. These test systems are used to check devices for compliance with specification and to determine maintenance needs. However, maintaining facilities such as these is not generally practical.
Various disciplines and organizations do not fully agree on some of the definitions
related to measuring device specifications, calibration, and error analysis. Therefore, it is
important to verify that a clear and mutual understanding of the specifications, calibration
terminology, and the error analysis processes is established when discussing these topics
with others.
21.4.1 Definitions of Terms Related to Accuracy
Error. Error is the deviation of a measurement, observation, or calculation from the
truth. The deviation can be small and inherent in the structure and functioning of the system and be within the bounds or limits specified. Lack of care and mistakes during fabrication, installation, and use can often cause large errors well outside expected performance bounds. Because the true value is seldom known, some investigators prefer to use
the term uncertainty. Uncertainty describes the possible error or range of error which
may exist. Investigators often classify errors and uncertainties into spurious, systematic,
and random types.
Precision. Precision is the ability to produce the same measurement value within
given accuracy bounds when successive readings of a specific quantity are measured.

Precision represents the maximum departure of all readings from the mean value of the
readings. Thus, a single observation of a measurement cannot be more accurate than
the inherent precision of the combined primary and secondary precision. It is possible to
have good precision of an inaccurate reading. Thus, precision and accuracy differ.
Spurious errors. Spurious errors are commonly caused by accident, resulting in false
data. Misreading and intermittent mechanical malfunctions can cause discharge readings
well outside of expected random statistical distribution about the mean. Spurious errors
can be minimized by good supervision, maintenance, inspection, and training.
Experienced, well-trained operators are more likely to recognize readings that are significantly out of the expected range of deviation. Unexpected spiral flow and blockages of
flow in the approach or in the device itself can cause spurious errors. Repeating measurements does not provide information on spurious error unless repetitions occur before and
after the introduction of the error. On a statistical basis, spurious errors confound evaluation of accuracy performance.
Systematic errors. Systematic errors are errors that persist and cannot be considered
random. Systematic errors are caused by deviations from standard device dimensions,
anomalies to the particular installation, and possible bias in the calibration. Systematic
errors cannot be removed or detected by repeated measurements. They usually cause persistent error on one side of the true value. The value of a particular systematic error for a
particular device may sometimes be considered as a random error. For example, an installation error in the zero setting for a flume might be + 1 mm for one flume and 2 mm for
another. For each flume the error is systematic, but for a number of flumes it would be a
random error.
Random errors. Random errors are caused by such things as the estimating required
between the smallest division on a head measurement device and water surface waves at
a head measuring device. Loose linkages between parts of flowmeters provide room for
random movement of parts relative to each other, causing subsequent random output
errors. Repeated readings decrease the average expected error resulting from random
errors by a factor of the square root of the number of readings.
Total error. Total error of a measurement is the result of systematic and random errors
caused by component parts and factors related to the entire system. Sometimes, error limits of all component factors are well known. In this case, total limits of simpler systems
can be determined by computation (Bos et al., 1991). In more complicated cases, it may
be difficult to confidently combine the limits. In this case, a thorough calibration of the
entire system as a unit can resolve the difference. In any case, it is better to do error analysis with data where entire system parts are operating simultaneously and compare discharge measurement against an adequate discharge comparison standard.
Expression of errors. Instrument errors are usually expressed by manufacturers as
either a percent of reading or a percent of full scale. The secondary devices based on electronic outputs are more frequently expressed in terms of percent full scale. The designer
must be aware that a probable error value of say 1 percent full-scale can exceed 10
percent for small value readings on the output device. When used with weirs, for example, the head reading of hl 5 in the weir equation can increase this 10 percent head measurement error to a 15 percent flow measurement error.

21.4.2 Terms Related to Measurement Capability

Linearity. Linearity usually means the maximum deviation in tracking a linearly varying quantity, such as measuring head, and is generally expressed as percent of full scale.
Discrimination. Discrimination is the number of decimals to which the measuring
system can be read. Precision is no better than the discrimination.
Repeatability. Repeatability is the ability to reproduce the same reading for the same
quantities. Thus, it is related to precision.
Sensitivity. Sensitivity is the ratio of the change of a secondary measurement, such
as head, to the corresponding change of discharge.
Range and Rangeability. Range is fully defined by the lowest and highest value that
the device can measure without damage and comply within a specified accuracy. The upper
and lower range bounds may be the result of mechanical limitations, such as friction at the
lower end of the range and possible overdriving damage at the higher end of the range.
Range can be designated in other ways: (1) as a simple difference between maximum discharge (Qn^x) and minimum discharge (Qmit), (2) as the ratio (QnJQm1n), called rangeability, and (3) as a ratio expressed as \'>(QminIQmw)' Neither the difference nor the ratios fully
define range without knowledge of either the minimum or maximum discharge.
Additional terms (hysteresis, response, lag, rise time). Additional terms related more
to dynamic variability might be important when continuous records are needed or if the
measurements are being sensed for automatic control of canals and irrigation. Hysteresis is
the maximum difference between measurement readings of a quantity established by the
same mechanical set point when set from a value above and reset from a value below.
Hysteresis can continually get worse as wear of parts increases friction or as linkage freedom increases. Response has several definitions in the instrumentation and measurement
field. For water measurement, one definition for response is the smallest change that can
be sensed and displayed as a significant measurement. Lag is the time difference of an output reading when tracking a continuously changing quantity. Rise time is often expressed
in the form of the time constant, defined as the time for an output of the secondary element
to achieve 63 percent of a step change of the input quantity from the primary element.
21.4.3 Comparison Standards
Water providers may want, or may be required, to have well-developed measurement programs that are highly managed and standardized. If so, water delivery managers may wish
to consult American Society for Testing Materials Standards (ASTM, 1988), Bos (1989),
International Organization for Standardization (ISO, 1983: ISO, 1991), and the National
Handbook of Recommended Methods for Water Data Acquisition (USGS, 1980).
Research laboratories, organizations, and manufacturers that certify measurement
devices may need to trace accuracy of measurement through a hierarchy of increasingly
rigid standards.
The lowest standards in the entire hierarchy of physical comparison standards are
called working standards, which are shop or field standards used to control quality of production and measurement. These standards might be gage blocks or rules used to ensure
proper dimensions of flumes during manufacturing or devices carried by water providers
and users to check the condition of water measurement devices and the quality of their

output. Other possible working standards are weights, volume containers, and stopwatches. More complicated devices are used, such as surveyors' levels, to check weir staff gage
zeros. Dead weight testers and electronic standards are needed to check and maintain
more sophisticated and complicated measuring devices, such as acoustic flow meters and
devices that use pressure cells to measure head.
For further measurement assurance and periodic checking, water users and organizations may keep secondary standards. Secondary standards are used to maintain integrity
and performance of working standards. These secondary standards can be sent to government laboratories, one of which is the National Bureau of Standards in Washington, D.C.,
to be periodically certified after calibration or comparison with accurate replicas of primary standards. Primary standards are defined by international agreement and maintained
at the International Bureau of Weights and Measurements in Paris, France.
Depending on accuracy needs, each organization should trace their measurement performance up to and through the appropriate level of standards. For example, turbine
acceptance testing, such as in the petroleum industry, might justify tracing to the primary
standards level.

27.5 SELECTION OF PRIMARY ELEMENTS OF WATER

MEASURING DEVICES
21.5.1 General Requirements
Design considerations involve the selection of the proper water measurement device for
a particular site or situation. Site-specific factors and variables must be considered in
extended detail. Each system has unique operational requirements and installation concerns. Knowledge of the immediate measurement needs and reliable estimates on future
demands of the proposed system is advantageous. Possible selection constraints may be
imposed by laws and compact agreements and should be consulted before selecting a
measurement device. Contractual agreements for the purchase of pumps, turbines, and
water measuring devices for water supply, sewage and drainage districts often dictate the
measurement system required for compliance prior to payment. These constraints may
be in terms of accuracy, specific comparison devices, and procedures. Bos (1989) provides an extensive and practical discussion on the selection of open channel water measurement devices. Miller (1996) provides a recent compilation of selection criteria for
pipe flow-meters suited to liquids and steam and other gas flows. Bos (1989) provides a
selection flow chart and a table of water measurement device properties to guide the
selection process for the open channel devices. Miller (1996) describes each meter in
detail for the pipe systems, but is more general in leaving the selection to the designer.
Because the design engineers for civil engineering projects are most likely to be dealing with irrigation water supply, waste water, or drainage and flood flows, the emphasis
is placed on the measuring systems deemed most appropriate to these processes. Large
closed-pipe systems for water supplies are frequently encountered, so installation situations appropriate to these will also be included. Gas flows, including steam, are more likely to be encountered by mechanical and chemical engineers and those readers are referred
to Miller (1996) and ASME (1959, 1971).

21.5.2 Types of Measuring Devices

System operators for water supply, drainage, and waste water commonly use many types
of standard water measuring devices, usually in open channels with limited applications
in closed conduits. Particularly prominent uses of open channel devices are found in irrigation delivery systems and farm distribution systems, although these measuring devices
are frequently used for sewer flows and even flood flows. However, the latter two areas
of application are frequently more difficult because of the likelihood of heavy bed loads
and floating debris.
In pipe flowmeters, the most commonly installed devices in industry are the orifice
meters, accounting for up to 80 percent of all industrial meters (Miller, 1996). Venturi
meters and flow tubes provide much of the remainder. In absolute numbers, the household meters, based on various technologies from nutating disks to paddle-wheel turbines, dominate.
For open-channel flows, weirs, flumes, submerged and free orifices, and current
meters dominate the flow measuring methods. Pipe flow meters, propeller and turbine,
acoustic, magnetic, and vortex-shedding meters are used on large water supply wells such
as those used in irrigation and municipal water supply. Differential head meters, such as
orifice meters, Venturi meters, and flow tubes, are also used in these applications. The
meters considered herein are
1. Open-channel flow devices
a. Current metering (cup, propeller, and electromagnetic probes)
b. Weirs
c. Flumes
d. Acoustic (transonic and Doppler)
e. Tracers
f. Miscellaneous
2. Pipe flow devices
a. Differential head meters
b. Acoustic (transonic and Doppler)
c. Tracers
d. Turbine/propeller/other insert mechanical
e. Vortex-shedding
f. Miscellaneous
The main factors which influence the selection of a measuring device include
(USBR, 1997):
a. Accuracy requirements
b. Cost
c. Legal constraints
d. Range of flow rates
e. Head loss
f. Adaptability to site conditions
g. Adaptability to variable operating conditions
h. Type of measurements and records needed

I. Operating requirements
j. Ability to pass sediment and debris
k. Longevity of device for given environment
1. Maintenance requirements
m. Construction and installation requirements
n. Device standardization and calibration
o. Field verification, troubleshooting, and repair
p. User acceptance of new methods
q. Vandalism potential
r. Impact on environment
Accuracy requirements. The desired accuracy of the measurement system is an important consideration in the selection of a measurement method. Most water measurement
installations, including the primary and secondary devices, can produce accuracies of 5
percent. Some systems are capable of 1 percent under laboratory settings. However, in
the field, maintaining such accuracies usually requires considerable expense or special
effort in terms of construction, secondary equipment, calibration in-place, and stringent
maintenance. Selecting a device that is not appropriate for the site conditions can result in
a nonstandard installation of reduced accuracy, sometimes exceeding 10 percent.
Accuracies are frequently reported that relate only to the primary measurement method
or device. However, many methods require secondary measurement equipment that produces the actual readout. This readout equipment typically increases the overall error of
the measurement.
Cost. The cost of the measurement method includes the cost of the device itself, the
installation, secondary devices, operation, and maintenance. Measurement methods vary
widely in their cost and in their serviceable life span. Measurement methods are often
selected based on the initial cost of the primary device with insufficient regard for the
additional costs associated with providing the desired records of flows over an extended
period of time.
Legal constraints. Governmental or administrative water board requirements may
dictate the water measurement devices or methods. Water measurement devices that
become a standard in one geographic area may not necessarily be accepted as a standard
elsewhere. In this sense, the term "standard" does not necessarily signify accuracy or
broad legal acceptance. Many water agencies require certain water measurement devices
used within their jurisdiction to conform to their standard for the purpose of simplifying
operation, employee training, and maintenance.
Flow range. Many measurement methods have a limited range of flow conditions for
which they are applicable. This range is usually related to the need for certain prescribed
flow conditions which are assumed in the development of calibrations. Large errors in
measurement can occur when the flow is not within this range. For example, using a bucket and stopwatch for large flows that engulf the bucket is not very accurate. Similarly,
sharp-edged devices, such as sharp-crested weirs, typically do not yield good results with
large channel flows. These are measured better with large flumes or broad-crested weirs,
which in turn are not appropriate for trickle flows.
Certain applications have typical flow ranges. Irrigation supply monitoring seldom
demands a low-flow-to-high-flow range above about 30, while this range on natural
stream flows may exceed 1000.

In some cases, secondary devices can limit the practical range of flow rates. For example, with devices requiring a head measurement, the accuracy of the head measurement
from a visually read wall gage may limit the measurement of low flow rates. For some
devices, accuracy is based on percent of the full-scale value. While the resulting error may
be well within acceptable limits for full flow, at low flows, the resulting error may become
excessive, limiting the usefulness of such measurements. Generally, the device should be
selected to cover the desired range. Choosing a device that can handle an unnecessary
large flow rate may result in compromising measurement capability at low flow rates, and
vice versa. This choice depends on the objective of the measurement. For example, in irrigation practice, usually choose a device that can measure the most common flow range at
the expense of poorly measuring extremes, such as flood flows. For urban drainage, the
flood peak may be important.
For practical reasons, different accuracy requirements for high and low flows may be
chosen. This is reasonable when an annual total is the primary goal and the low flows contribute a small percentage to that total. Also, if the inaccurate low flow readings are truly
a random error then this error approaches zero with large accumulations of readings. Thus
the designer needs to know if management decisions are made from individual readings
or from long term averages.
Headloss. Most water measurement devices require a drop in head. On retrofit installations, for example, to an existing irrigation project, such additional head may not be
available, especially in areas that have relatively flat topography. On new projects, incorporating additional headloss into the design can usually be accomplished at reasonable
cost. However, a tradeoff usually exists between the cost of the device and the amount of
headloss. For example, acoustic flow meters are expensive but require little headloss.
Sharp-crested weirs are inexpensive but require a relatively large headloss. The head loss
required for a particular measuring device usually varies over the range of discharges. In
some cases, head needed by a flow measuring device can reduce the capacity of the channel at that point.
Adaptability to site conditions. The selection of a flow measuring device must
address the site of the proposed measurement. Several potential sites may be available for
obtaining a flow measurement. The particular site chosen may influence the selection of
a measuring device. For example, discharge in a canal system can be measured within a
reach of the channel or at a structure such as a culvert or check structure. A different
device would typically be selected for each site. The device selected ideally should not
alter site hydraulics so as to interfere with normal operation and maintenance. Also, the
shape of the cross-sectional flow area may favor particular devices.
Adaptability to variable operating conditions. Flow demands for most water delivery systems usually vary over a range of flows and flow conditions. The selected device
must accommodate the flow range and changes in operating conditions, such as variations
in upstream and downstream head. Weirs or flumes should be avoided if downstream
water levels can, under some conditions, cause excessive submergence. Also, the information provided by the measuring device should be conveniently useful for the operators
performing their duties. Devices that are difficult and time consuming to operate are less
likely to be used and are more likely to be used incorrectly.
In some cases, water measurement and water level or flow control are desired at the
same site. A few devices are available for accomplishing both (e.g., constant-head orifice,
vertically movable weirs, and Neyrpic flow module; Bos, 1989). However, separate measurement and control devices are typically linked for this purpose and usually can exceed
the performance of combined devices in terms of accuracy and level control, if care is

exercised to assure that the separate devices are compatible and achieve both functions
when used as a system.
Type of measurements and records needed. An accurate measure of instantaneous
flow rate is useful for system operators in setting and verifying flow rate. However,
because flow rates change over time, a single (instantaneous) reading may not accurately
reflect the total volume of water delivered. Where accounting for water volume is desired,
a method of accumulated individual flow measurements is needed. Where flows are
steady, daily measurements may be sufficient to infer total volume. Most deliveries, however, require more frequent measurements. Meters that accumulate total delivered volume
are desirable where water users take water on demand. Totalizing and automatic recording devices are available for many measuring devices. For large structures, the cost for
water-level sensing and recording hardware is small relative to the structure cost. For
small structures, these hardware costs remain about the same and thus become a major
part of the measurement cost, and may often exceed the cost of the primary structure
itself.
Many water measuring methods are suitable for making temporary measurements
(flow surveys) or performing occasional verification checks of other devices. The method
chosen for such a measurement might be quite different from that chosen for continuous
monitoring. Although many of these flow survey methods are suited for temporary operation, the focus here is on methods for permanent installations.
Operating Requirements. Some measurement methods require manual labor to
obtain a measurement. Current metering requires a trained staff with specialized equipment. Pen-and-ink style water-stage recorders need operators to change paper, add ink,
and verify proper functioning. Manual recording of flows may require printed forms to be
manually completed and data to be accumulated for accounting purposes. Devices with
manometers require special care and attention to assure correct differential-head readings.
Automated devices, such as ultrasonic flowmeters and other systems that use transducers
and electronics, require operator training to set up, adjust, and troubleshoot. Setting gatecontrolled flow rates by simple canal level references or by current metering commonly
requires several hours of waiting between gate changes for the downstream canal to fill
and stabilize. However, if a flume or weir is installed near the control gate, that portion of
the canal can be brought to the stable, desired flow level and measured flow rate in a few
minutes, and the canal downstream of the flume or weir can then fill to the correct level
over a longer time without further gate adjustments. Thus, the requirements of the operating personnel in using the devices and techniques for their desired purposes must be
considered in meter selection.
Some measuring devices may inherently serve an additional function applicable to the
operation of a water supply system. For example, weirs and flumes serve to hydraulicalIy isolate upstream parts of a canal system from the influence of downstream parts. This
occurs for free overfall weirs and flumes flowing below their modular limits. Acoustic,
propeller, magnetic, and vortex-shedding flowmeters do not provide this function without
additional structural measures such as a downstream overfall. If these meters are used, and
the isolation function is desired, then the designer should be made aware of the requirement and provide a free overfall. Isolating the influence of upstream changes from affecting downstream channels, is less easily accomplished. However, it can be partly implemented with orifices that have a differential head that is large compared to the upstream
fluctuations.
The designer should be aware that a sharp-crested weir overfall requires a relatively
high head drop and may need to be excessively wide to provide the isolation function
with low absolute head drop. While a long board can be used downstream from a propeller meter to provide the necessary width of flow that will pass a required quantity of

water at small head, that small head, and the crude board would not be well suited for
measuring flow rate.
The designer may wish to take advantage of broad-crested weir behavior and provide
a thick crest that can withstand in excess of about 80 percent submergence, which usually translates into low absolute head loss. When used with a propeller meter, for example, the broad-crested weir need not be well defined and can be economically installed
(Replogle, 1997).
Ability to pass sediment and debris. Canal systems often carry a significant amount
of sediment in the water. Removal of all suspended solids from the water is usually prohibitively expensive. Thus, some sediment will likely be deposited anywhere the velocities are reduced, which typically occurs near flow measuring structures. Whether this sediment causes a problem depends on the specific structure and the volume of sediment in
the water. In some cases, this problem simply requires routine maintenance to remove
accumulated sediment; in others, the accumulation can make the flow measurement inaccurate or the device inoperative. Sediment deposits can affect approach conditions and
increase approach velocity in front of weirs, flumes, and orifices. Floating and suspended
debris such as aquatic plants, washed-out bank plants, and fallen tree leaves and twigs can
plug some flow measurement devices and cause significant flow measurement problems.
Many of the measurement devices which are successfully used in closed conduits (e.g.,
orifices, propeller meters, and so on) are not usable in culverts or inverted siphons because
of debris in the water. Attempting to remove this debris at the entrance to culverts is an
additional maintenance problem.
Flumes, especially long-throated flumes, can be designed to resist sedimentation. The
design options available are to select a structure shape that will maintain velocities that
assure erosion of sediments, or at least continued movement of incoming sediments
through the flume, at important flow rates. In large broad-crested weirs (a class of longthroated flumes) for capacities greater than 1 m3/s per m of flume width, velocities greater
than 1 m/s can be achieved for the upper 75 percent of the flow range, and is usually erosive enough to maintain flume function even for high-sediment bed loads. At the lower
flow ranges and for heavy sediment bed loads, deposition is likely and frequent maintenance may be required.
Trapezoidal sections tend to retain low velocities into the upper ranges of flow and
are less sediment worthy. Long-throated flumes with flat bottoms throughout and side
contractions maintain a high velocity for 0.5 m3/s per m width, and higher, but must
have throat lengths that are 2 to 3 times the throat width in order to be accurately computable. The sediment worthiness of a flume design depends more on these absolute
velocities than on whether the flume floor is flat throughout or raised as in a broadcrested weir. This prompts the designer to select shapes that can provide these velocities. One suggestion for broad-crested weirs in a fixed sized channel is to construct a
false floor in the head gage area to increase the velocity there and prevent changes in
area of flow there. Also, sediments can accumulate in the upstream channel to a depth
of the false floor without affecting the function of the flume. This can extend the time
between mandatory channel cleaning.
Device environment. Any measurement device with moving parts or sensors is subject to failure if it is not compatible with the site environment. Achieving proper operation
and longevity of devices is an important selection factor. Very cold weather can shrink
moving and fixed parts differentially and solidify oil and grease in bearings. Water can
freeze around parts and plug pressure ports and passageways. Acidity and alkalinity in

water can corrode metal parts. Water contaminants such as waste solvents can damage
lubricants, protective coatings, and plastic parts. Mineral encrustation and biological
growths can impair moving parts and plug pressure transmitting ports. Sediment can
abrade parts or consolidate tightly in bearing and runner spaces in devices such as propeller meters.
Measurement of wastewater and high sediment transport flow may preclude the use of
devices that require pressure taps, intrusive sensors, or depend upon clear transmission of
sound through the flow. Water measurement devices that depend on electronic devices and
transducers must have appropriate protective housings for harsh environments. Improper
protection against the site environment can cause equipment failure or loss of accuracy.
Maintenance requirements. The type and amount of maintenance varies widely with
different measurement methods. For example, current metering requires periodic maintenance of the current meter itself and maintenance of the meter site to assure that is has a
known cross section and velocity distribution. When the flow carries sediment or debris,
most weirs, flumes, and orifices require periodic cleaning of the approach channel. As
mentioned above, design and meter selection can mitigate the maintenance problems with
sediments, but are not likely to eliminate them. Electronic sensors need occasional maintenance to ensure that they are performing properly. Regular maintenance programs are
recommended to ensure prolonged measurement quality for all types of devices.
Construction and installation requirements. In addition to installation costs, the difficulty of installation and the need to retrofit parts of the existing conveyance system can
complicate the selection of water measurement devices. Clearly, devices that can be easily retrofitted into the existing canal system are much preferred because they generally
require less down time, and usually present fewer unforeseen problems.
Device standardization and calibration. A standard water measurement device infers
a documented history of performance based on theory, controlled calibration, and use. A
truly standard device has been fully described, accurately calibrated, correctly constructed, properly installed, and sufficiently maintained to fulfill the original installation
requirements and flow condition limitations. Discharge equations and tables for standard
devices should provide accurate calibration. Maintaining a standard device usually only
involves a visual check and measurement of a few specified items or dimensions to ensure
that the measuring device has not departed from the standard. Many standard devices have
a long history of use and calibration, and thus are potentially more reliable. Commercial
availability of a device does not necessarily guarantee that it satisfies the requirements of
a standard device.
When measuring devices are fabricated onsite or are poorly installed, small deviations from the specified dimensions can occur. These deviations may or may not affect
the calibration. The difficulty is that unless an as-built calibration is performed, the
degree to which these errors affect the accuracy of the measurements is largely
unknown. All too frequently, design deviations are made under the misconception that
current metering can be used to provide an accurate field calibration. In practice, calibration by current metering to within 2 percent is difficult to attain. An adequate calibration for free-flow conditions requires many current meter measurements at several
discharges. Changing and maintaining a constant discharge for calibration purposes is
often difficult under field conditions.
Field verification, troubleshooting, and repair. After construction or installation of a
device, some verification of the calibration is generally recommended. Usually, the meth-

ods used to verify a permanent device (e.g., current metering) are less accurate than the
device itself. However, this verification simply serves as a check against gross errors in
construction or calibration. For some devices, errors occur as components wear and the
calibration slowly drifts away from the original. Other devices have components that simply fail, that is, you get the correct reading or no reading at all. The latter is clearly preferred. However, for many devices, occasional checking is required to ensure that they are
still performing as intended. Selection of devices may depend on how they fail and how
easy it is to verify that they are performing properly.
User acceptance of new methods. Selection of a water measurement method must
also consider the past history of the practice at the site. When improved water measurement methods are needed, proposing changes that build on established practice are generally easier to institute than radical changes. It can be beneficial to select a new method
that allows conversion to take place in stages to provide educational examples and demonstrations of the new devices and procedures.
Vandalism potential Instrumentation located near public access is a prime target for
vandalism. Where vandalism is a problem, measurement devices with less instrumentation, or instrumentation that can be easily protected, are preferred. When needed, instrumentation can be placed in a buried vault to minimize visibility.
Impact on environment. During the selection of a water measurement device, consideration must be given to potential environmental impacts. Water measurement devices
vary greatly in the amount of disruption to existing conditions that is needed for installation, operation, and maintenance. For example, installing a weir or flume constricts the
channel, slows upstream flow, and accelerates flow within the structure. These changes in
the flow conditions can alter local channel erosion, local flooding, public safety, local
aquatic habitat, and movement of fish up and down the channel. These factors may alter
the cost and selection of a measurement device.
21.5.3 Selection Guidelines
Selection of a water measurement method can be a difficult, time-consuming design
process if one were to formally evaluate all the factors discussed above for each measuring device. This difficulty is one reason that standardization of measurement devices within water agency jurisdictions is often encouraged by internal administrators. However,
useful devices are sometimes overlooked when devices similar to previous purchases are
automatically selected. Therefore, some preliminary guidance on selection is offered to
the designer so that the number of choices can be narrowed down before a more thorough
design analysis of the tradeoffs between alternatives is performed.
Short list of devices based on application. The list of practical choices for a water
measuring device is quickly narrowed by site conditions because most devices are applicable to a limited range of channel or conduit conditions. Economics also limits applicable
devices. For example, few irrigation deliveries to farms can justify expensive acoustic
meters. Likewise, using current meters for manual flow measurement in a channel is
appropriate for intermittent information but is usually too labor intensive for use on a continuous basis. Table 21.1 provides a list of commonly used measurement methods that are
considered appropriate for each of several applications. Table 21.2 provides an abbreviated table of selection criteria and general compliance for categories of water measurement
devices. The symbols (+), (O), and () are used to indicate relative compliance for each
selection criterion. The (+) symbol indicates positive features that might make the device

TABLE 21.1 Application-Based Selection of Water Measurement Devices

1. Openchannel conveyance system
a. Natural channels (see Herschy, 1985)
(1) Rivers
Periodic current metering of a control section to establish stage=discharge relation
Broadcrested weirs
Long-throated flumes
Short-crested weirs
Acoustic velocity maters (AVMtransient time)
Acoustic Doppler velocity profiles
Float-velocity/area method
Slope-area methods
(2) Intermediate-sized and small streams
Current metering/control section
Broad-crested weirs
Long-throated flumes
Short-crested weirs
Short-throated flumes
Acoustic velocity meters (AVMtransient time)
Floatvelocity/area method
b. Regulated channels (see USBR, 1997)
(1.) Spillways
(a) Gated
Sluice gates
Radial gates
(b) Ungated
Broad-crested weirs (including special crest shapes, Ogee crest, etc.)
Short-crested weirs
(2) Large canals
(a) Control structures
Check gates
Sluice gates
Radial gates
Overshot gates
(b) Other
Long-throated flumes
Broad-crested weirs
Short-throated flumes
Acoustic velocity meters

TABLE 21.1 (Continue)

(3.) Small canals (including open channel fluid conduit flow)
Longthroated flumes
Broad-crested weirs
Shortthroated flumes
Sharpcrested weirs
Rated flow control structures (check gates, radial gates, sluice gates, overshot gates)
Acoustic velocity meters
(c) Other
Float-velocity/area methods
(4) Irrigation delivery to farm turnout
(a) Pipe turnouts (short inverted siphons, submerged culverts, etc.)
Metergates
Current meter
Weirs
Short-throated flumes
Long-throated flumes
(ft) Other
Constant head orifice
Rated sluice gates
Movable weirs
2. Closed conduit conveyance systems (see Brater and King, 1982; Miller, 1996)
(a.) Large pipes
Venturi meters, venturi tubes, nozzles
Rated control gates (orifice)
Acoustic velocity meters (transit time)
(ft.) Small and intermediate-sized pipelines
Venturi meters, Venturi tubes, nozzles
Orifices (in-line, end-cap, shunt meters, etc.)
Propeller and turbine meters
Magnetic meters
Acoustic meters (transit-time and Doppler)
Pitot tubes
Elbow meters
Vortex - shedding
Trajectory methods (e. g., full-pipe trajectory; California pipe method for part full pipe)
Other commercially available meters (household types)
SOURCE: From USBR (1997).

TABLE 21.2 Selection Guide for Water Measuring Devices.

Selection

Sharp
Crested

Broad
Crested

LongThroated

Short Submerg
Throated Orifice

Criteria

Weirs

Weirs

Flumes

Flumes (Channel) (Channel) (Channel)

Devise

Current
Metering

Acoustic Radial Propeller Differ.

Mechan. Magnetic
VeI Meter and Sluice Meters Head meters* Head meters* meters
Gates

Accuracy
O
O
O
O
O
+
Cost
O
O
+
O
O
+
+
+
Flows > 5 m-Vs
O
O
+
+
+
+
Flows < 0.25 mVs
O
O
+
Flow span
+
O
O
O
+
Headloss
+
O
O
Site Condition
+
Lined canal
+
+
O
O
O
Unliner canal O
+
O
O
O
O
O
Short, full pipe NA
NA
+
NA
NA
NA
NA
NA
Closed conduit NA
NA
NA
NA
NA
NA
NA
NA
Measurement Type
+
+
+
Rate
+
+
+
+
+
Volume
O
Sediment
+
+
Sediment pass
O
O
O
O
2
+
+
+
+
Debris pass
+
Longivity
+
+
+
Moving parts +
+
+
O
O
+
+
+
+
Electr. requir. +
O
O
+
+
+
+
Maintenance
O
O
+
+
+
+
+
Construction
O
O
+
+
+
Field verify
O
O
+
+
+
Standarization
O
O
O
Source: Adapted from USER (1997).
Symbols 1, O, 2 are used as relative indicators comparing application of the listed water measuring device to the listed criteria
Symbol V denotes that situability varies widely.
Symbol NA denotes "not applicable" to criteria
"Venturi,
orifice, pilot tube, etc.
f
Propeller meters, turbine meters, paddle wheel meters, etc.

(Pipe Exit) (Pipes)

+
O
O
+
O
O
+
+
O
+
O
O
+
O
O

(Pipes)

(Pipes)

4-

O
+

O
O

O
O
+
NA
NA
O
+
V
V

O
O
+
NA
NA

V
NA
NA
O
+
+
V
+
O
O
O
+
+

Acoustic
Doppler

Acoustic Acoustic
Transonic Transon.
Pipe
(Pipes) (Pipe, 1 Path) (Multipath)
+
O
O
+
O
O
O
+
O
+
+
+
NA
NA
NA
NA
NA
NA

+
+
O
O
O
O

+
+
O
O
O
O

+
O
O
O
O
O

+
+
+
O
O
O

O
2
O
O
O

attractive from the standpoint of the associated selection criteria. A (-) symbol indicates
negative aspects that might limit the usefulness of this method based on that criterion. A
(O) indicates no strong positive or negative aspects in general. A (V) means that the suitability varies widely for this class of devices. The letters NA mean that the device is not
applicable for the stated conditions. A single negative value for a device does not mean
that the device is not useful or appropriate, but other devices would be preferred for those
selection criteria.
Vortex-shedding flow meters are not specifically rated in this grouping. They are
expected to compete with orifice meter applications. They generally offer less head loss
that orifice meters and can cover a wider discharge range for a particular installation.
Although they have been around for many years, they have only recently been offered in
a configuration that makes them competitive with orifice meters, which they are generally expected to replace because they can produce less pipe head loss. Open-channel applications for vortex-shedding meters are not considered practical.

21.6 SELECTION OF SECONDARY DEVICES FOR DISCHARGE

READOUTAND CONTROL
While the emphasis of this chapter is on hydraulic design, it is important that the output
of the design be translated into useful information for the user.
21.6.1 Intended Uses
The secondary device that is used with a flowmeter depends strongly on the use of the
information. Immediate management decisions, such as adjusting a valve or canal gate,
require nearly instant feedback to the operator at an accuracy and precision that fully utilizes the available accuracy of the primary device. As discussed above, random errors over
many measurements tend to cancel, and long-term totals can often absorb large random
detection errors that would not be tolerable for decisions depending on a single reading.
The designer should be cognizant of the user needs and be prepared to provide an appropriate output.
21.6.2 Quality Assurance
Secondary devices provide a variety of functions, primary of which are data recording and
data quality assurance. The secondary devices necessary for these may not be the same. It
is good practice to provide manual, instantaneous flow rate output at the meter site so that
the servicing personnel can quickly know that the main secondary instrument is functioning. For weirs and flumes wall gages that show flow rates directly are recommended as a
quick visual check of secondary instrumentation. However, a wall gage that shows a head
reading that is converted to a discharge rate by the technician using a table or equation
will usually suffice.
Sometimes it is practical to provide field check capability to a secondary device by
special treatment of the installation. For example, if a pressure transducer is used to detect
head on a flume, the transducer can be mounted in a stilling well attached to the flume. If
it is further mounted on a movable rack, it can allow servicing personnel to raise and

Next Page

CHAPTER 22
WATER AND WASTEWATER
TREATMENT PLANT
HYDRAULICS

Federico E. Maisch, P. E., Partner

Thomas J. Sullivan, P. E., Managing Partner
Roger J. Cronin, P. E., Partner
Frank J. Tantone, P. E., Partner
David V. Hobbs, P. E., Associate
William L. Judy, P. E., Associate
Sharon L Cole P. E., Senior Engineer
Greeley and Hansen
Chicago, IL
22.1

INTRODUCTION

Designers of water treatment plants and wastewater treatment plants are faced with the
need to design treatment processes which must meet the following general hydraulic
requirements:
Water treatment plants. Provide the head required to allow the water to flow through
the treatment processes and to be delivered to the transmission/distribution system in
the flow rates and at the pressures required for delivery to the users.
Wastewater treatment plants. Provide the head required to raise the flow of wastewater
from the sewer system to a level which allows the flow to proceed through the treatment processes and be delivered to the receiving body of water.
The above requires knowledge of open-channel, closed-conduit, and hydraulic
machine flow principles. It also requires an understanding of the interaction between these
elements and their impact on the overall plant (site) hydraulics. Head is either available
through the difference in elevation (gravity) or it has to be converted from mechanical
energy using hydraulic machinery. Distribution of flows using open channels or closed
conduit is critical for proper hydraulic loading and process performance.
This chapter brings together information on commonly used hydraulic elements and
specific applications to water treatment plants and wastewater treatment plants. The
development of hydraulic profiles through the entire treatment process with examples for
water treatment plants and wastewater treatment is also presented.
Many processes and flow control devices are similar in both water treatment plants and
wastewater treatment plants. Both types of plants require flow distribution devices, gates
and valves, and flowmeters. These devices are discussed in Section 22.2. The development of water treatment plant hydraulics, including examples from in-place facilities, are
presented in Section 22.3. Wastewater treatment plant hydraulics are discussed in Section
22.4, and Section. 22.5 is devoted to non-Newtonian flow principles.

22.2

GENERAL

22.2.1 Introduction
This section addresses some elements which are common to both water treatment plants
and wastewater treatment plants including:

Flow distribution-manifolds
Gates and valves
Flowmeters
Local losses

22.2.2

Flow distribution-manifolds

In the design of water and wastewater treatment plants, proper flow distribution can be as
critical as process design considerations, which typically receive much more attention.
Plant failures resulting from unequal and unmanageable flow distribution are possibly as
common and as serious as those resulting from errors in process design.
Flow distribution devices, such as distribution channels, pipe manifolds or distribution
boxes, are commonly used to distribute or equalize flow to parallel treatment units, such
as flocculation tanks, sedimentation basins, aeration tanks, or filters.
22.2.2.1 Distribution boxes. The simplest of these devices, the distribution box, typically consists of a structure arranged to provide a common water surface as the supply to two
or more outlets. The outlets are typically over weirs and the key to equal flow distribution
is to provide independent hydraulic characteristics between the downstream system and
the water level in the distribution box. In other words, provide a free discharge weir (nonsubmerged under all conditions) for each outlet to eliminate the impact of downstream
physical system differences on the flow distribution. Velocity gradients across the distribution box must be nearly zero to equalize flow conditions over each outfall weir. Weirs
clearly should be of uniform design in terms of physical arrangement length and materials of construction. They should also be adjustable to account for any minor flow differences noted in actual operation. The same principles apply if the designer wishes to distribute flows in specific proportions which are not necessarily equal. In this case the
designer could control the proportions of flow distribution by varying the relative geometry of the weirs (i.e., change the width or invert of each weir to achieve a desired flow
distibution). The specifics of weir hydraulics are covered in various texts in the literature.
Attention should always be paid to the selection of the proper coefficients to model the
specific weir geometry and the geometry of the approach flow.
22.2.2.2 Distribution channels and pipe manifolds. Distribution channels and manifolds
are also common in plant design but a bit more complex in their function and design. The
distribution of flow in these devices is impacted by the flow distribution itself. Since a
portion of the flow leaves the channel or manifold along the length of the device, the
velocity of flow and, therefore, the relationship of energy grade line, velocity head and
hydraulic grade line varies along the length of the device. This is more clearly visible in
a distribution channel of uniform cross section, using side weirs along its length for flow
distribution. At each weir, flow leaves the channel, resulting in less velocity head in the

channel and possibly a higher water surface at each ensuing weir. Chao and Trussell
(1980), Camp and Graber (1968), and Yao (1972) have presented comprehensive
approaches for the design of distribution channels and manifolds and should be reviewed
for details of design.
As in distribution boxes, the most important consideration to achieving equalized flow
distribution is to minimize the effects of unequal hydraulic conditions relative to each
point of distribution. In channels this can be accomplished by tapering the channel cross
section, varying weir elevations, making the channel large enough to cause velocity head
changes to be insignificant or a combination of these. Similar considerations may be
applied to manifolds with submerged orifice outlets. A reliable approach here is to provide
a large enough manifold, resulting in a total headloss along the length of the distribution
of less than one tenth the loss through any individual orifice. This approach essentially
results in the orifices becoming the only hydraulic control and the accuracy of the flow
distribution is then dependent on the uniformity of the orifices themselves.
22.2.3

Gates and Valves

Gates and valves generally serve to either control the rate of flow or to start/stop flow.
Gates and valves in treatment plants are typically subjected to much lower pressures
than those in water distribution systems or sewage force mains and can be of lighter
construction.
22.2.3.1 Gates. Gates are typically used in channels or in structures to start and stop flow
or to provide a hydraulic control point which is seldom adjusted. Because of the time and
effort required to operate gates, they are not suited for controlling flow when rapid
response, frequent variation, or delicate adjustments are needed. Primary design considerations when using gates are the type of gate fabrication and the installation conditions
during construction.
There are many fabrication details including materials used, bottom arrangement, and
stem arrangement. For instance, for solids bearing flows, a flush bottom, rising stem gate
can be used to avoid creating a point of solids deposition and to minimize solids contact
with the threaded stem. Gate manufacturers are a good source of information for gate fabrication details and can assist with advice regarding specific applications.
Most commonly used gates are designed to stop flow in a single direction. They may
use upstream water pressure to assist in achieving a seal (seating head), but typically also
must be designed to resist static water pressure from downstream (unseating head). Both
seating and unseating heads must be evaluated in design of a gate application. For most
manufacturers, the seating or unseating head is expressed as the pressure relative to the
center line of the gate.
22.2.3.2 Valves. Table 22.1 provides a summary of several types of valves and their
applications. Valves are used to either throttle (control) flow or start/stop flow.
Start/stop valves are intended to be fully open or fully closed and nonthrottling. They
should present minimum resistance to flow when fully open and should be intended for
infrequent operation.
Gate valves, plug valves, cone valves, ball valves, and butterfly valves are the most
common start/stop valve selections. Butterfly valves have a center stem, are most common
in clean water applications and should not be used in applications including materials that
could hang-up on the stem. Therefore, they are seldom used at wastewater plants prior to
achieving a filter effluent water quality.

TABLE 22.1 Typical Valves and Their Application*

Type

Open/Close

Sluicegate
Slide gate
Gate valve
Plug valve
Cone valve
Ball valve
Butterfly valve
Swing check
Lift check
Ball check
Spring check
Globe valve
Needle valve
Angle valve
Pinch/diaphragm

X
X
X
X
X
X
X
X
X
X
X
X

Throttling

X
X
X
X

X
X
X

Water

Wastewater

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X

*Typical applications-exceptions are possible, but consultation with valve manufacturers is recommended.

Check valves are a special case of a start/stop valve application. Check valves offer
quick, automatic reaction to flow changes and are intended to stop flow direction reversal. Typical configurations include swing check, lift check, ball check and spring loaded.
These valves are typically used on pump discharge piping and are opened by the pressure
of the flowing liquid and close automatically if pressure drops and flow attempts to
reverse direction. The rapid closure of these valves can result in unacceptable "waterhammer" pressures with the potential to damage the system. A detailed surge analysis may
be required for many check valve applications (see Chapter 12). At times, mechanically
operating check valves should be avoided in favor of electrically or pneumatically operated valves (typically plug, ball, or cone valves) to provide a mechanism to control time
of closing and reduce surge pressure peaks.
Throttling valves are used to control rate of flow and are designed for frequent or nearly continuous operation depending on whether they are manually operated or electronically controlled. Typical throttling valve types include globe valves, needle valves, and
angle valves in smaller sizes, and ball, plug, cone, butterfly, and pinch/diaphragm valves
in larger sizes. Throttling valves are typically most effective in the mid-range of loose line
open/close travel and for best flow control should not be routinely operated nearly fully
closed or nearly fully open.
22.2.4 Flow meters
The most common types of flow meters used in water and wastewater treatment plants are
summarized in Table 22.2 and fall into the following categories:

TABLE 22.2 Common Types of Flow Meters

Type

Headloss

Cost

WW

1-120 in
Any size

Low
Medium

Medium
Low

X
X

X
X

Pitottube
0.5-5% of scale
Parshall flume 5% of rate

1/2-96 in
Wide range

Low
Low

Low
Medium

X
X

Magnetic
Doppler
Propeller
Turbine

1/10-120 in
1/8-120 in
Up to 24 in
Up to 24 in

None
None
High
High

High
High
High
High

X
X
X
X

Venturi
Orifice plate

Typical Accuracy
0.75% of rate
2% of scale

0.5% of rate
1-2.5% of rate
2% of rate
0.5-2% of rate

Size Range

X
X

Pressure differential/pressure measuring meters (e.g., Venturi, orifice plate, pitot tube,
and Parshall flume meters)
Magnetic meters
Doppler (ultrasonic) meters
Mechanical meters (e.g., propeller and turbine meters)
Accurate flow measurements require uniform flow patterns. Most meters are
significantly impacted by adjacent piping configurations. Typically a specific number of
straight pipe diameters is required both upstream and downstream of a meter to obtain
reliable measurements. In some cases, 15 straight pipe diameters upstream and 5 straight
pipe diameters downstream are recommended. However, different types of meters have
varying levels of susceptibility to the uniformity of the flow pattern. Meter manufacturers
should be consulted.
22.2.4.1 Pressure differential/pressure measuring meters. Pressure differential/pressure
measuring flow meters include Venturi meters, orifice plates, averaging pitot meters, and
Parshall flume s. These meters measure the change in pressure through a known flow cross
section-or in the case of the pitot meter, measure the difference in pressure at a point in
the flow versus static pressure just downstream in a uniform section of pipe.
Venturi meters and orifice plates are commonly used in water and wastewater. Solids
in wastewater could plug the openings of a pitot tube meter-limiting their use to relatively clean liquids. The Venturi meter and orifice plate meter use pressure taps at the wall of
the device and can be arranged to minimize potential for debris from clogging the taps.
The Parshall flume can be arranged with a side stilling well and level measuring float system or an ultrasonic level sensing device to measure water level.
22.2.4.2 Magnetic meters. In a magnetic flowmeter, a magnetic field is generated around
a section of pipe. Water passing through the field induces a small electric current proportional to the velocity of flow. Because a magnetic meter imposes no obstruction to the
flow, it is well suited to measuring solids bearing liquids as well as clean liquids and produces no headless in addition to the normal pipe loss. Magnetic meters are among the least
susceptible to the uniformity of the stream lines in the approaching flow.

22.2.4.3 Ultrasonic meters. In an ultrasonic flow meter, a pair of transceivers (transmitter/receiver) are positioned diagonally across from each other on the pipe wall. The
transmitter sends out a signal which is affected by the speed of the flow. The receiver measures the difference between the speed of the signal when directed counter to the flow
(slowed by the flow) and when directed with the flow (speeded up by the flow). The time
difference is a function of fluid velocity, which is used to compute the flow. As with magnetic meters, no flow obstruction is imposed resulting in no headloss in addition to the
normal pipe loss. Ultrasonic meters are also available for open-channel applications.
22.2.4.4 Mechanical meters. Mechanical meters include propeller and turbine-type
equipment. The two meters are similar in function in that in each a device is inserted into
the flowpath. The device is rotated by the flow and the speed of rotation is used to compute rate of flow. These devices impose an obstruction to flow, are recommended for clean
water only, and generally result in significant headloss.
22.2.5

Local Losses

In any piping system as flow travels along the pipe, pressure drops as a result of headloss
due to friction along the pipe and local losses at bends, fittings, and valves. The local
losses at bends, fittings, and valves are least significant in long, straight piping systems
and most significant at treatment plants where the length of straight pipe is relatively short
and therefore, the frictional pipe losses comprise a smaller fraction of the total losses
when compared to the summation of all local losses. A term often used to refer to local
losses is "minor losses," however, because of the later consideration the term "minor losses" can be misleading.
Traditionally, local losses have been computed in terms of "equivalent length" of
straight pipe or in terms of multiples of velocity head. The "equivalent length" or loss factor K methods attempt to estimate the local losses based on the characteristic of the specific bend, fitting or valve. The K loss factor method is discussed here. Essentially, a local
loss is computed as follows:
L =K^

(22-D

where hL = local loss, K = loss factor, V = velocity, g = gravitational acceleration.

The values for K reported by various sources vary considerably for some local losses
and are relatively consistent for others. See references. There are many literature sources
for K values. The Bureau of Reclamation (1992) is one such source of information regarding energy loss equations. Table 22.3 shows a range of K factors from additional sources
as well as a typically used value for each. Judgment must be applied in computing local
losses, taking into account any unique system conditions. Throughout this chapter K values were obtained from equipment manufacturers when available. Values from Table 22.3
were used only as an approximation when more specific data were unavailable. The reader is cautioned that there are application-specific characteristics which have significant
influence on the K factors. One of these characteristics, for example, is size. A K value of
0.6 is often encountered in literature to characterize the losses associated with flow
through the run of a tee. However, for flow past tees in large pipes this factor can be very
small and nearly zero.

Gate valve
100% open
0.39
75% open
1.1
50% open
4.8
25% open
27
Globe valve-open
10
Angle valve^open
4.3
Check valve-ball
4.5
Swing check
Butterfly valve-open
1.2
Foot valve-hinged
2.2
Foot valve-poppet
12.5
Elbows
45 regular
45 long radius
90 regular
90 long radius
180 regular
180 long radius (flanged)
Tees
Std. teee-flowthrough run 0.6
Std. teee-flow-through branch 1.8
Return bend
1.5
Mitre bend
90
1.8
60
0.75
30
0.25
Expansion
d/D = 0.75
0.18
d/D = 0.5
0.55
d/D = 0.25
0.88
Contraction
d/D = 0.75
0.18
d/D = 0.5
0.33
d/D = 0.25
0.43
Entrancee-projecting
0.78
Entrancee-sharp
0.5
Entrancee-well rounded
0.04
Exit
1.0

10
5 2.1-3.1
65-70
0.6-2.3

0.19
1.15
5.6
24

0.30-0.42
0.18-0.20
0.21-0.3
0.14-0.23
0.38
0.25

0.19

0.1-0.3

10
5

4.0-6.0
1.8-2.9

0.42

0.6
1.8
2.2

0.6
1.8 1.8
2.2
1.129-1.265
0.471-0.684
0.130-0.165

0.78
0.5
0.04
1.0

Typically Used Value

Committee on Pipeline Planning (1975)

Sanks (1989)

Simon (1986)

Daugherty (1977)
Cameron Hydraulic Data

Bulletin No. 2552, University of Wisconsin

Crane Co. (1987)

Walski (1992)

Valve and Fitting Types

Ten-State Standards (1978)

TABLE 22.3 Typical K Factors for Computing Local Losses

0.2
1.2
5.6
24
10
2.5

06-2.2
0.16-0.35
1.0-1.4
5.0-14.0

0.2
1.2
5.6
25
10
5
5
0.6-2.5 2.5
0.5
2.2
14

0.18
0.25
0.18

0.5
0.7
0.6

0.3
0.75
0.4

1.8

0.8
0.35
0.1

0.42
0.2
0.25
0.19
0.38
0.25
0.6
1.8
2.2
1.3
0.6
0.16

0.19
0.56
0.92

0.2
0.6
0.9

0.2
0.6
0.9

0.19
0.33
0.42
0.83
0.5

0.2
0.3
0.4
0.78
0.5
0.04
1.0

0.2
0.33
0.43
0.8
0.5
0.04
1.0

0.8
0.5
0.040.04

0.8
0.5
0.25
1.0

22.3

HYDRAULICS OF WATER TREATMENT PLANTS

22.3.1 Introduction
Water treatment comprises the withdrawal of water from a source of supply and the
treatment of raw water through a series of unit processes for the beneficial use of the
system customers. Raw water quality can vary widely. The ultimate uses of water by the
system customer (e.g., drinking, fire protection, irrigation, aquifer recharge, etc.) can also
vary and be subject to different treatment level requirements and regulations. Therefore,
the selected treatment processes vary widely over a multitude of treatment technologies
in use. Water treatment consists of a series of chemical, biological, and physical processes connected by channels and pipelines. Figures 22.1 and 22.2 illustrate process
flow diagrams (flowsheets) for typical surface water and groundwater treatment plants,
respectively. The designer of the water treatment process must carefully evaluate source
water characteristics and desired water quality characteristics of the treated water to
design treatment processes capable of purifying the source water to water suitable for the
system customers. The objective of this chapter is to review the hydraulic considerations
required to convey water through the treatment process.
Design of a plant's treatment process is closely linked with the hydraulic design of the
treatment plant. This chapter presumes that the designer has evaluated and selected treatment processes for the water treatment plant. Although design flows are discussed below,
we have also assumed that the designer has chosen a design flow requirement for the treatment process. For municipal treatment plants, design flows are based on the service area
FLOCCULATION/COAGULATION
RAPID MIX

CLEARWELL

TO
DISTRIBUTION

RAW
WATER

SEDIMENTATION
FILTRATION
- RAW WATER CONTROL
CHAMBER/RAPID MIX

FINISHED
WATER
PUMPING

FIGURE 22.1 Typical surface water treatment plant process flow diagram.
PRESSURE FILTRATION

TO
DISTRIBUTION

WELL
PUMP

WELL

CHLORINE
CONTACT
CHAMBER

FIGURE 22.2 Typical ground water treatment plant process flow

diagram with dual trains (#1 and #2).

population and the per capita use of water by the population served. The per capita use of
water can be obtained from literature sources as an initial approximation. However, these
initial estimations must be corroborated with actual site specific population counts and
water usage. For nonmunicipal treatment facilities, treated water needs of the service area
must be individually evaluated.
22.3.1.1 Sources of supply. Natural sources of supply include groundwater and surface
water supplies. Groundwater supplies typically are smaller in daily delivery but serve
more systems than surface water supplies. Groundwater supplies normally come from
wells, springs, or infiltration galleries.
Wells constitute the largest source of groundwater. Except in rare circumstances of
artesian wells (wells under the influence of a confined aquifer) and springs, groundwater
collection involves pumping facilities. Hydraulics of groundwater treatment plants are
frequently based on hydraulics of conduits under pressure, such as pipelines, pressure
filters, and pressure tanks. Raw water characteristics of groundwaters are uniform in
quality compared with surface supplies.
Surface water supplies are normally larger in daily delivery. Surface supplies are used
to service larger population centers and industrial centers. In areas where groundwater
supplies are limited in yield or where groundwater supplies contain undesirable chemical
characteristics, smaller surface water treatment plants may be utilized. Surface water
sources of supply include rivers, lakes, impoundments, streams, and ponds. The treatment
processes chosen in plants treating surface water favor nonpressurized systems such as
gravity sedimentation. The larger flow volumes characteristic of surface water supplies
also favor open channel hydraulic structures for conveying water through the treatment
process. Raw water characteristics of surface supplies can vary rapidly over short periods
of time and also experience seasonal variation.
22.3.1.2 Treatment requirements. Treatment requirements for municipal water treatment plants are normally defined by regulatory agencies having authority over the plant's
service area. In the United States, regulatory agencies include national government regulations promulgated through the Environmental Protection Agency and state government
regulations. Water treatment plants are designed to meet these regulations. Treatment regulations change through improved knowledge of health effects of water constituents and
through identification of possible new water-borne threats. The designer therefore should
attempt to select treatment processes which will also meet treatment requirements which
are expected to be promulgated over the next few years. To the extent possible, treatment
plant process design should provide flexibility for future plant expansions or for possible
additional treatment processes. Because hydraulic design of plants must go hand-in-hand
with the process selection, plant hydraulic design should provide for the flexibility to add
future plant facilities.
Treatment requirements for industrial water treatment plants are dictated by process
needs of the industry and less by regulatory agency requirements. Industrial water treatment plants that result in contact between or ingestion of the treated water by humans
must conform to the local regulatory requirements.
22.3.1.3 General design philosophy. Effective design of water treatment plant
hydraulics requires that the hydraulic designer have a thorough knowledge of all aspects
of the water system. The overall treatment system hydraulic design must be integrated and
coordinated including the treatment plant, the raw water intake and pumping facilities, the
treated water storage, and treated water pressure/head requirements. The design within the
water treatment plant must also be integrated between the various treatment processes.

Additionally, design considerations must address the availability of operating personnel

and hours of operation such that the process and hydraulic requirements conform to available resources.
22.3.2

Hydraulic Design Considerations in Process Selection

Water treatment plant process selections are controlled principally by characteristics of

the raw water and by the desired water quality characteristics of the finished water. Flow
through each unit process and each conduit connecting processes results in loss of
hydraulic head. Most treatment plants have limited head available.
The selection of a particular unit process will include evaluation of numerous criteria
including costs, operability, performance, energy use and similar items. One criteria
which must also be evaluated for each process is the hydraulic head requirements of the
process.
22.3.2.1 Head available. For the design flow to pass through a water treatment plant, the
total available head must exceed the head requirements of the unit processes and connecting conduits. The head available is the difference in energy grade line (EGL) in the
hydraulic profile between the head works of the plant and the end of the plant. Additional
head may be provided by pumping or by lowering the elevation of treatment units at the
end of the plant. See Figure 22.3 for a typical water treatment plant hydraulic profile.
For most surface water plants, the hydraulic profile at the head of the plant is
controlled by raw water pumps pumping from the intake facilities. The hydraulic profile
at the head of a plant in a groundwater system is typically determined by the well pumps
serving the plant.
22.3.2.2 Typical unit process head requirements. Following below is a table of typical
head requirements for water treatment plant processes. This table may be used for initial
evaluation of unit processes. More detailed hydraulic evaluations must be performed after
plant operating modes and design flows are determined. Detailed hydraulic evaluations
must also include headlosses in connecting conduits.

Unit Process

Head Requirement
at Rated Capacity, m (ft)

Intakes, including bar screens

Rapid mixing
Flocculation
Sedimentation
Filtration
- Gravity
- Pressure
Disinfection
Aeration
- Spray
- Cascade
- Compressed air
Softening
Ion exchange softening
Iron and manganese removal

0.3-0.9 (1-3)
0.15-0.30 (0.5-1)
0.06-0.15 (0.2-0.5)
0.6-2.4 (2-8)
3-4.6 (10-15)
3-7.6 (10-25)
0.15-0.6 (0.5-2)
3-4.6 (10-15)
3-4.6 (10-15)
0.15-0.6 (0.5-2)
0.15-0.6(0.5-2)
0.6-1.5 (2-5)
0.6-1.5 (2-5)

POINT

RAW WATER
CONTROL CHAMBER

MIXING
CHAMBER

FLOCCULATION/
SEDIMENTATION
BASINS

CLEARWELL
FINISHED WATER
PUMPING STATION
WATER SURFACE AT DESIGN
MAXIMUM HOUR FLOW
FIGURE 22.3 Hydraulic profile.

OVERFLOW WEIR CREST

22.3.3 Hydraulic Design Considerations in Plant Siting

Plant sites are normally selected before the hydraulic designer initiates design of the treatment system. If a plant site has not been selected, the designer should be aware of
hydraulic considerations which may influence site selection.
Site elevation has the most significant impact on plant hydraulics. A plant site located
above the service area will eliminate or reduce pumping requirements from the plant to
the service areas. Typical municipal distribution system pressures are 40-70 psi, therefore
the elevation of the treatment plant should be at least 100 ft above the service area to eliminate finished water pumping. Similarly, plant sites which permit gravity intake of the
source water may reduce or eliminate raw water pumping. Few plants are able to meet
these optimal conditions.
The typical surface water plant must pump both raw and finished water. Raw water
(low-lift) pumps are used to pump water from the water source into the treatment facilities and finished water (high-lift) pumps are used to pump from the treatment plant into
the service area distribution system.
22.3.4

Hydraulic Design Consideration in Plant Layout

After the plant site has been identified, the plant design may be arranged for optimal
hydraulic benefit. In particular, arrangement of treatment processes to allow flow to move
down gradient minimizes excavation needs for structures. Arrangements which are
designed for future expansion should consider the hydraulic needs of the expanded plant
as well as the process needs. Grouping of processes together facilitates movement of
water through the treatment process train.
The designer should also consider secondary hydraulic systems for optimal design.
Chemical feed systems and dewatering systems are examples of secondary hydraulic
systems which must be coordinated with the treatment flow system. Normally it is
desirable to minimize the length of chemical piping systems. Dewatering systems are usually based on gravity drainage of basins and conduits.
22.3.5

Bases for Design

After evaluation and selection of a source of supply and development of the treatment
plant process train, the designer is prepared to develop the plant Bases for Design. The
Bases for Design is a summary of design flow and capacity, and proposed treatment
processes, including the chemical storage and feed facilities.
22.3.5.1 Design flows. Design flows for water treatment plants serving municipalities
are typically based on the projected population within the water service area for the
design life of the treatment facilities. Population data is normally determined from
census records, land use zoning information, and studies of existing and projected
population densities. Service area per capita demands are affected by the mix of
domestic, commercial, and industrial water users which are unique to each service
area.Typically water consumption records are available for water service areas. For new
facilities, the use of generalized water consumption data may be needed. In the United
States, water consumption varies widely but generally ranges between 100-200 gallons per capita per day.

From studies of projected population and per capita demand, planned design flows for
the water treatment facilities may be developed. These demands include the following:
Annual average demand. The average daily water consumption for the water service
areas, generally computed by multiplying the average daily consumption (gallons per
capita) by the projected population of the service area.
Maximum demand. Maximum demand experienced by the water plant throughout its
service life. The maximum hour demand is generally 200 to 300 percent of the average demand but numerous factors affect the peak demand experienced by water treatment plants. These factors include seasonal demands (particularly for plants where service areas are located in extremes of hot and cold temperatures), normal daily flow
variations, the community size, industrial usage, and system storage. Normally system
storage is provided to service peak hour demands, allowing the treatment facilities to
be designed on peak day demands. Peak day demands generally range between 125
and 200 percent of the average demand.
Minimum flow. As the name suggests, the minimum flow expected to be processed
through the treatment facilities. Minimum flow depends upon system operations. In
general, minimum flows for municipal plants may be estimated as 50 percent of the
average demand, but range between 25 and 75 percent of the average demand.
22.3.5.2 Rated treatment capacity. The rated treatment capacity of a plant is that capacity for which each of the unit processes are designed. For municipal treatment plants with
adequate system storage, the rated treatment capacity is the system's maximum day
demand. Where storage is limited, the rated treatment capacity may be greater, for example, the system maximum hour demand or greater. Smaller systems may be designed to
produce the rated treatment capacity in one or two 8-h shifts rather than over the entire
24-h day.
22.3.5.3 Hydraulic treatment capacity. Treatment plants are normally designed for a
hydraulic capacity greater than the rated treatment capacity. Hydraulic treatment capacities are normally equal to 125 to 150 percent of the rated treatment capacity. The hydraulic
treatment capacity provides flexibility for future process changes or alternative flow routings through the plant. Hydraulic capacities in excess of the rated treatment capacity provide some margin of safety for operations which may not be optimal (e.g., control gates
inadvertently left partially open).
22.3.5.4 Treatment process bases for design. The development of the water treatment
plant's "Bases for Design" is a key step in establishing the criteria to which the plant will
be designed. This document must be reviewed carefully with the water treatment plant
owner representatives and understood and agreed to by all before the final design proceeds. The Bases for Design presents a summary of each treatment process including
design flows (minimum, average, rated capacity), specification of dimension of major elements (e.g., tanks, pumps), both hydraulic and process loading characteristics, required
performance, and design data for the chemical storage and feed system. Table 22.4 presents an example of the bases for design for sedimentation basins (one of the many unit
processes in a water treatment plant).
22.3.6 Plant Hydraulic Design
As noted above, a water treatment plant consists of a series of treatment processes
connected by free surface flow channels and pipelines. During development of the plant's

TABLE 22.4 Treatment Process Bases for DesignSedimentation Basins

Item

Number of basins
Basin characteristics
Plan-75 ft X 230-6 in
Nominal side water
depth-12 ft (SWD)
Surface area/basin- 17, 28 8 ft2
Volume/basin-207,456 f3
Channels/basin-2
L:Wratio-6.1:l
Displacement time (h)
Surface loading [(gal'm)/ft2]
Flowthrough velocity (ft/min)
Sludge collectors
Longitudinal collectors
Type: chain flight
Number per basin
Cross collectors
Type: chain flight
Number per basin
Settled sludge pumps
Type: progressive cavity
Number:
100 gal/min capacity
400 gal/min capacity
200 gal/min capacity
Capacity (gal/min)
Installed
Firm

Stage I

Stage II

StageIII

Annual
Average
4

Maximum
Day
4

Annual
Average
8

Maximum
Day
8

Annual
Average
12

Maximum
Day
12

3.17
0.47
1.21

1.99
0.75
1.93

3.17
0.47
1.21

1.99
0.75
1.93

3.17
0.47
1.21

1.99
0.75
1.93

4
4

4
4

4
4
8

4
4
8

4
4
16

4
4
16

2000
1600

2000
1600

3600
3200

3600
3200

5200
4800

5200
4800

Bases for Design, the designer determines the rated treatment capacity, average flow,
minimum flow and hydraulic capacity of the plant.
Following development of the Bases for Design, the designer must evaluate plant
operating modes to develop a detailed plant flow diagram and hydraulic profile
through the plant.

22.3.6.1 Plant operating modes. Operating modes describe the sequence of treatment
processes the water goes through to achieve the required level of purification. Operational
modes are normally presented in the form of simplified block diagrams which illustrate
the flow path through the plant from one process to the next. These operational mode
block diagrams are useful in visualizing stages during construction, future planned plant
expansions or simply alternative operating modes.
Figures 22.4 through 22.9 show an example of a sequence of plant operating modes
for a surface water treatment plant which illustrate three stages of a plant expansion program with alternatives for the flocculation and sedimentation basins to work in series or
in parallel. Plant processes proposed include raw water control chambers, rapid mix
chambers, flocculation/sedimentation basins, ozone contact chambers, and filters. In this
example, the raw water control chambers are used to split flow between plant process
groups and also as a rapid mix chamber for chemical addition.

STAGE I RAW WATER CONTROL

CHAMBERS/RAPID MIX
(FUTURE)

STAGE I
FLOC/SED

STAGE I
FILTER
FIGURE 22.4 Stage Ioperational mode diagram.

RAW WATER CONTROL

CHAMBERS/RAPID MIX

(FUTURE)

STAGE I
FLOC/SED

- NEW OZONE
CONTACT CHAMBER

STAGE Il
FLOC/SED

STAGE Il OZONE
CONTACT CHAMBER
STAGE
Il
RLTER

STAGE I
FILTER

RLTER BUILDINGS

FIGURE 22.5 Stage IIparallel operational mode diagram.

The Stage I facilities including raw water control chamber, flocculation/sedimentation

basins and filters are depicted in Fig. 22.4. Operational modes for a proposed plant expansion to double the plant capacity (Stage II) are shown in Figs. 22.5 through 22.7 and operating modes for a second plant expansion to triple the plant capacity (Stage III) are shown
in Figs. 22.8 and 22.9. Settled water ozone contact chambers were added to the expanded
plant, which illustrates treatment upgrades.
Operational modes for the Stage II treatment plant include parallel and series flocculation/sedimentation. When the plant is operated in the parallel mode, influent raw
water for each set of sedimentation basins flows by gravity from the raw water control
chamber serving the basin set. Raw water flow is divided between each sedimentation
basin in service at the raw water control chamber. Settled water from each set of basins
is routed to an ozone contact chamber. Ozonated settled water is then combined prior to
flowing to the filters.
STAGE I
RAW WATER CONTROL
CHAMBERS/RAPID MIX

STAGE I
FLOC/SED

STAGE Il OZONE
CONTACT CHAMBER

STAGE Il
FLOC/SED

NEW OZONE
CONTACT CHAMBER
STAGE
FILTER

STAGE I
FILTER

FILTER BUILDINGS

FIGURE 22.6 Stage IIseries flocculation/sedimentation basin

operational mode diagram.

STAGE I
RAW WATER CONTROL
CHAMBERS/RAPID MIX

(FUTURE)

STAGE I
FLOC/SED

NEW OZONE
CONTACT CHAMBER
STAGE
Il
RLTER

STAGE I
RLTER

STAGE Il
FLOC/SED

STAGE Il OZONE
CONTACT CHAMBER
FILTER BUILDINGS

FIGURE 22.7 Stage IIsplit parallel operational mode diagram.

STAGE I
RAW WATER CONTROL
CHAMBER/RAPID MIX-

STAGE I
FLOC/SED

STAGE III
FLOC/SED

NEW OZONE
CONTACT CHAMBER
RLTER BUILDINGS

- STAGE HI
RAW WATER CONTROL
CHAMBER/RAPID MIX

STAGE
III
FILTER

STAGE
I
RLTER

STAGE Il
FLOC/SED

STAGE Il OZONE
CONTACT'CHAMBER

STAGE I
RLTER

FIGURE 22.8 Stage IIIparallel operational mode diagram.

STAGE I
RAW WATER CONTROL
CHAMBER/RAPID MIX

STAGE III
RAW WATER CONTROL
CHAMBER/RAPID MIX

STAGE I
FLOC/SED

STAGE III
FLOC/SED

NEW OZONE
CONTACT CHAMBER
RLTER BUILDINGS

STAGE
III
FILTER

STAGE
I
RLTER

STAGE I
RLTER

STAGE Il
FLOC/SED

STAGE Il OZONE
CONTACT CHAMBER

FIGURE 22.9 Stage IIIsplit parallel operational mode diagram.

The Stage III split parallel operational mode is similar to the parallel operational mode
except that the ozonated settled water from each set of basins is not combined prior to
flowing to the filters. Side-by-side plant scale treatment studies are possible with the
future split parallel mode since part of the flocculation/ sedimentation/filtration processes
can be operated as a "control" while the remainder of the plant can be operated in a
controlled experimental mode.
The series flocculation/sedimentation operational mode is designed to permit operation of the sedimentation basins in two stages in lieu of the single-stage parallel mode.
Under certain raw water conditions, operation of the basins in series may enhance performance of the basins. Chemical feed for the first and second sedimentation stages may be
adjusted to respond to raw water conditions and settled water quality after the first-stage
sedimentation. Series flocculation/sedimentation increases hydraulic losses through the

plant. Under this mode, twice as much flow is routed to each basin and the flow pattern
is longer, since the settled water from the first sedimentation stage must be returned to the
influent of the second sedimentation stage.
Operational mode block diagrams are also a convenient means to illustrate the effect
of side stream flows which may impact the overall plant flow. For example, removal of
sludge from the sedimentation basins is accompanied by a decrease in flow leaving the
basins compared with flow entering the basins. In a similar manner, filter backwash water
removes a certain amount of flow. A plant designed to produce a certain rated capacity
may have to treat more than the rated capacity through certain processes. The impact of
these side stream flows must be evaluated on an individual basis. In many treatment
plants, backwash water treatment facilities are installed to recycle backwash water to the
head of the plant.
22.3.6.2 Plant flow diagrams. After establishing plant operating modes, more detailed
flow diagrams are developed by the designer. The diagrams normally start with possible
valving and gating arrangements and are then expanded with tentative valve, sluice gate,
pipeline, and conduit sizes.
Valving arrangements are designed to enable any of the major operational units (e.g.,
sedimentation basin, ozone contact chamber) to be removed from service. The arrangement may include design of temporary flow stop devices, such as stop logs (sectional barriers which were originally constructed of logs but are now commonly metal plates). The
arrangement should be designed to permit maintenance work on major valves and sluice
gates while minimizing the impact on plant process. Major channel sections should be
designed so they can be removed from service and dewatered while minimizing impacts
on the rest of the plant.
The designer should distinguish between units taken out of service frequently (such
as filters), periodically (such as sedimentation basins), or rarely (such as conduits).
Filter backwashing occurs so frequently that the rated treatment capacity can be met
with one filter out for backwashing. Sedimentation basins may be removed from service
once or twice per year for equipment maintenance. Since the basins outages occur at
widely scattered intervals, it is reasonable to design the units to be removed from service during lower flow periods. Conduits and pipelines are rarely removed from service,
but the hydraulic impacts can be significant. Depending on the conduit location,
removal of a conduit can remove a portion of the plant from service. Effective design
will provide redundant conduits so that a portion of the plant can remain in service during conduit dewatering.
The focus of this section has been on the main plant hydraulics, but the hydraulic
designer must also design for hydraulic subsystems. An important group of these subsystems include dewatering of all basins and conduits. Where plant elevations will allow,
gravity de watering is recommended. In most cases, dewatering pumps are necessary.
These pumps may be located in the unit being dewatered or may be located in a separate
structure connected to the process unit by de watering pipelines.
22.3.6.3 Hydraulic Profile. One of the most important tools in the hydraulic design of a
water treatment plant is the development of a hydraulic profile. The hydraulic profile is a
diagram showing the energy grade line (EGL) at each unit process. For open tanks with
flows at minimal velocities, which is the case in most water treatment plants, the velocity head is negligible and the hydraulic grade line (HGL) or water surface elevation
(WSEL) provide an adequate representation of the EGL. Profiles normally include critical
structural elevations of processes and conduits. The profile may also include ground surface profiles and other site information.

Hydraulic profiles are developed for each of the design flows. In the case of water
treatment plants, the design flows may include rated treatment capacity, hydraulic capacity, average flow, and minimum flow. Hydraulic profiles should also take into consideration unit processes or conduits which may be taken out of service. Hydraulic profiles are
valuable design and operational tools to assist in scheduling routine maintenance activities and for evaluating the impact to the treatment plant capacity during outages of process
units or conduits.
Computations of hydraulic profiles begin at control points where there is a definite
relationship between the plant flow and water surface depth. For gravity flow plants, the
most common forms of control points are weirs and tank water surface elevations (e.g.,
clear well water surface elevations), but other types of control points may be used. From
each control point, head losses associated with local losses, plant piping, and open channel flow are added to the control water surface. Since flows in water treatment plant's are
mostly in the subcritical regime (Froude number < 1), most hydraulic designers will work
upstream from the control point. For pressure plants, control points are typically pressure
regulating or pressure control points, frequently in the service area distribution system.
From these control points and knowledge of the flow velocity, both the EGL and HGL
may be computed back to the treatment facilities.
Hydraulic profiles are valuable design tools to identify overall losses through the plant.
Profiles are also valuable to identify units with excessive losses. Since total head available is normally limited, units with excessive losses should be considered for redesign to
reduce local loss coefficients or to reduce velocities.
Figure 22.3 is an example hydraulic profile for a gravity surface water treatment plant
with conventional treatment processes. The method of computing headlosses is presented
in Section 22.3.7.
22.3.7 Water Treatment Plant Process Hydraulics
In this section calculations required to establish the WSEL through a medium-sized water
treatment plant will be presented. A schematic of the water treatment plant is shown in
Fig. 22.10. Notice that future growth has been considered in the initial design. Three
examples are included which illustrate typical hydraulic calculations. The first example
calculates the WSEL from the sedimentation basin effluent chamber back through the
flocculation/sedimentation basins to the Raw Water Control Chamber. The second follows
the flow from the clear well back through the filters. Filter hydraulics are illustrated in the
third example. All examples are presented in a spreadsheet format which is designed to
facilitate calculating the EGL, HGL, and WSEL at various points through the treatment
process and for multiple flow rates (i.e., minimum, daily average, peak hour, future
conditions).
22.3.7.1 Coagulation. Process criteria and key hydraulic design parameters. The coagulation process, used to reduce particulates and turbidity, is carried out in three steps: mixing (often referred to as rapid or flash mixing), flocculation, and sedimentation. Each of
these steps is briefly discussed below.
Rapid mixing. The mixing process imparts energy to increase contact between
existing solids and added coagulants. Possible mixer types include turbine, propeller,
pneumatic, and hydraulic. Headless that occurs in mixing chambers depends on the chosen mixing device. Most mechanical mixers do not create significant head losses. The
headloss coefficient (K) associated with a specific mixer can be obtained from the manufacturer. Pneumatic mixing, which is not common, has associated losses similar to those

RAW WATER
CONTROL
CHAMBER

RAPID MIX
CHAMBER
FLOCCULATION/SEDIMENTATION
BASINS
BASIN BASIN BASIN BASIN
1
4
3
2
FILTER BOX
LEGEND
INITIAL FACILITIES
FUTURE FACILITIES
CLEARWELL
EFFLUENT
PUMPING STATION
FIGURE 22.10 Schematic of a water treatment plant.
for aeration (see table in Section 22.3.2.2, above). Hydraulic mixing takes place using
weirs, swirl chambers, throttled valves, Parshall flumes, or other devices to induce turbulence. Head loss coefficients for these devices can be obtained from the manufacturer.
Important considerations during the initial design of a mixing chamber include:
Velocity gradient. This is mixerspecific information and can be obtained from the
manufacturer. The system should be designed to provide a velocity gradient that is
optimal for the coagulation process taking place.
Dead spots and short circuiting. An ideal mixing system will have minimal dead spots
and short circuiting. These can be avoided with proper sizing and placement of
mixers.
Flocculation. Coagulated particles form larger particles (floes) during the gentle mixing of flocculation, where the flow travels slowly through a series of flocculator paddles,
baffles, or conduits. Inlets and weirs are designed to provide low turbulence for protection
of the floes. The energy provided to the system by the flocculators (manufacturer-specific) or baffling is decreased as the flow approaches the sedimentation basins.
Sedimentation. Gravity sedimentation removes coagulated solids prior to filtration.
There are four zones in a clarifier as shown in Fig. 22.11 and listed below:
Inlet zonewhere upstream flow conditions transition smoothly to uniform flow settling conditions
Sedimentation zonewhere sedimentation takes place
Sludge zonewhere solids collect and are removed
Outlet zonewhere settling conditions smoothly transition to downstream flow
conditions

OUTLET ZONE

INLET ZONE
SEDIMENTATION ZONE
SLUDGE ZONE

FIGURE 22.11 Hypothetical zones in a rectangular sedimentation basin.

Each of the zones is designed to minimize turbulence and avoid short circuiting. The
velocity in the sedimentation zone is limited to 0.3 m/s (1 ft/s) for average flow. Sludge
removal equipment moves slowly so that settling patterns are not disturbed. Because the
process is designed for smooth flow and minimal turbulence, very little head loss occurs
in sedimentation basins. Ports at the inlet and outlet produce the greatest head losses in
this process.
Hydraulic design example. Table 22.5 illustrates the calculation of the WSEL, using
metric units, through the coagulation process at the medium-sized water treatment plant
shown in Fig. 22.10. Figs. 22.12 through 22.14 show plan views and details of the
TABLE 22.5 Hydraulic Calculations of a Typical Coagulation Process, SI Units
Parameter
1. Plant now(mVs)
Note: For Points 1 through 8, see Fig. 22.12
2. WSEL (see Note 1) at Point 1 (Calculation done
in Table 22.6) (m)
3. Point 1 to Point 2
Average flow = 210/32 (mVs)
Flow depth = WSEL @ 1 - invert (106.60
m) (m)
Flow area = 5.13 m width X depth (m2)
Velocity = flow/area (m/s)
Hydraulic Radius r = AIP2/3(P 2= w + 2d) (m)
Conduit loss = [(V X n)/(r )] X L (m)
where Manning's n = 0.014 and L = 28.96 m
WSEL at Point 2 (m)
4. Point 2 to Point 3
Average Flow = 50/16 (mVs)
Flow depth = WSEL @ 2 - invert (106.60 m) (m)
Flow area = 5.13m width X depth (m3)
Velocity = flow/area (m/s)
r = AIP (P = w + Id) (m)
Conduit loss = [(V X n)/(r2/3)]2 X L (m)
where Manning's n = 0.014 and L = 14.63 m
WSEL at Point 3 (m)
5. Point 3 to Point 4
Average flow = 0/8 (mVs)
Flow depth = WSEL @ 3 - invert (106.60 m) (m)
Flow area = 5.13 m width X depth (m3)

Initial Operation
Design Operation
Min Day Avg Day Avg Day Max Hour
2.19
109.73

3.06
109.73

3.28
109.74

4.38
109.74

1.44
3.13
16.05
0.09
1.41

2.01
3.13
16.06
0.13
1.41

2.15
3.13
16.07
0.13
1.41

2.87
3.14
16.10
0.18
1.41

0.00
109.73

0.00
109.73

0.00
109.74

0.00
109.74

0.68
3.13
16.05
0.04
1.41

0.96
3.13
16.06
0.06
1.41

1.03
3.13
16.07
0.06
1.41

1.37
3.14
16.10
0.08
1.41

0.00
109.73

0.00
109.73

0.00
109.74

0.00
109.74

0.27
3.13
16.05

0.38
3.13
16.06

0.41
3.13
16.07

0.55
3.14
16.10

(1) The Energy Grade Line (EGL) is equal to the Hydraulic Grade Line (HGL) + the Velocity Head (W2g). At
points in the system where the velocities are very low (i.e., V => O) the velocity head approaches O (i.e., W2g =>
O). Consequently, EGL => HGL => WSEL. Where: WSEL = Water Surface Elevation.

TABLE 22.5

(Continued)
Parameter

Initial Operation
Design Operation
Min Day Avg Day Avg Day Max Hour

Velocity = flow/area (m/s)

0.02
r = AIP(P = w + 2d)(m) 2
1.41
Conduit loss = [(V X n)/(r)] X L (m)
where n = 0.014 and L = 21.95 m
0.00
WSEL at Point 4 (m)
109.73
6. Point 4 to Point 5
Flow = 0/32 (mVs)
0.07
Port area = 0.30 m deep X 0.76 m wide (m2)
0.23
Velocity = flow/area (m/s)
0.29
Submerged entrance loss = 0.8 W2g (m)
0.00
WSEL at Point 5 (in Sedimentation Tank) (m)
109.73
7. Point 5 to Point 6
Width of sedimentation basin (W) (m)
23.16
Flow ((2/4) (mVs)
0.55
Invert elevation of sedimentation baffles (m)
105.97
Flow depth (H) (WSEL at Point 5 - baffle
invert) (m)
3.76
Area downstreams of baffle (W X H) (m2)
87.21
Horizontal openings in baffle, 2.54 cm wide
spaced every 22.86 cm. Area of
openings = A = W X .0254 X H/.2286 (m2)
9.69
Velocity of downstream baffle (V downstream)
0.01
(0/A) (m/s)
Velocity of 2.54 cm opening section (Vl) (QIA) (m/s) 2 0.06
Local losses = sudden expansion (1.0 X (V downstream) /2g)
+ sudden contraction (0.36 X (VI)2/ 2g) (m)
0.00
WSEL at Point 6 (Upstream of sedimentation baffles) (m) 109.73
8. Point 6 to Point 7
Loss per stage (provided by flocculator manufacturer) (m) 0.01
Total loss (three stages) (m)
0.04
WSEL at Point 7 (m)
109.77
9. Point 7 to Point 8
Flow = Q/24 (m3/s)
0.09
Port area = 0.30 m deep X 0.46 m wide (m2)
0.14
Velocity = flow / area (m/s)
0.65
Entrance loss = 1.25 W2g (m)
0.03
WSEL at Point 8 (inlet port) (m)
109.80
Note: For Points 8 through 14, see Fig. 22.13
10. Point 8 to Point 9
Average flow = Q/24 (mVs)
0.09
Flow depth = WSEL @ 8 - invert (109.12
m)
(m)
0.68
Flow area = 0.91 m width X depth (m2)
0.62
Velocity = flow/area (m/s)
0.15
r = A/P (P = w + Id) (m)
0.27
Conduit loss [(V X n)/(r2'3)]2X L (m)
where n = 0.014 and L = 3.86 m
0.00
WSEL at Point 9 (m)
109.80
11. Point 9 to Point 10
Average flow = Q/l2 (m3/s)
0.18

0.02
1.41

0.03
1.41

0.03
1.41

0.00
109.73

0.00
109.74

0.00
109.74

0.10
0.23
0.41
0.01
109.74

0.10
0.23
0.44
0.01
109.74

0.14
0.23
0.59
0.01
109.76

23.16
0.77
105.97
3.77
87.36

23.16
0.82
105.97
3.77
87.41

23.16
1.09
105.97
3.79
87.68

9.71
0.01

9.71
0.01

9.74
0.01

0.08

0.08

0.11

0.00
109.74

0.00
109.74

0.00
109.76

0.01
0.04
109.78

0.03
0.09
109.83

0.05
0.15
109.91

0.13
0.14
0.92
0.05
109.83

0.14
0.14
0.98
0.06
109.89

0.18
0.14
1.31
0.11
110.02

0.13
0.72
0.65
0.19
0.28

0.14
0.77
0.71
0.19
0.29

0.18
0.90
0.82
0.22
0.30

0.00
109.83

0.00
109.89

0.00
110.02

0.26

0.27

0.36

TABLE 22.5

(Continued)
Parameter

Flow depth = WSEL @ 9 -invert (109.12

m) (m)
Flow area = 0.91 m width X depth (m2)
Velocity = flow/area (m/s)
r = AIP (P = \v + 2d) (m)
Conduit loss = [(V X n)/(r2'3)]2 X L (m)
where n = 0.014 and L = 3.86 m
WSEL at Point 10 (m)
12. Point 10 to Point 11
Flow = (2/8, m3/s
Flow depth = WSEL @ 10 - invert2 (109.12 m) (m)
Flow area = 0.91 width X depth (m )
Velocity = flow/area (m/s)
Loss at two 45 bends = 2 X 0.2 V2/2g (m)
WSEL at Point 11 (m)
13. Point 11 to Point 12
Flow = Q/4 (mVs)
Flow depth = WSEL @ 11 - invert (109.12
m) (m)
Flow area = 1.52 m width X depth (m2)
Velocity = flow/area (m/s)
Loss at two 45 bends = 2 X 0.2 V2/2g (m)
r = AIP(P = w + 2d)(m)
Conduit loss = [(V X n)/(r2/3)]2 X L (m)
where n = 0.014 and L = 9.75 m
WSEL at Point 12 (m)
14. Point 12 to Point3 13
Flow = (2/4, (m /s)
Flow depth = WSEL @ 12 - invert (109.12
m) (m)
Inlet area= 1.52m width X depth (m2)
Velocity = flow/area
(m/s)
Inlet loss = 1 V2/2g (m)
WSEL at Point 13 (Mixing Chamber No. 2 outlet) (m)
15. Point 13 to Point 14
Note: Mixers provide negligible head loss
Flow = Q/4 (mVs)
Chamber area = 1.83 m X 1.83 m (m2)
Velocity = flow/area 2(m/s)
Losses = Mixer (1 V /2g) + Sharp bend (1.8 V2l2g) (m)
WSEL at Point 14 (Mixing Chamber No. 2 inlet) (m)
Note: For Points 14 through 21, see Fig. 22.14
16. Point 14 to Point 15
Flow = Q/2 (mVs)
Conduit area = 2.29 m wide X 1.22 m deep (m2)
Velocity = flow/area (m/s)
R = AIP (P = 2w + Id) (m)
Conduit losses = L X [V/(0.849 X C X /?o.63)] 1/0.54 (m)
where L = 47.24 m and Hazen-Williams C = 120
Local losses2 = flow split 2(0.6 V2/2g) + contraction
(0.07 V /2g) = 0.67 V /2g (m)
WSEL at Point 15 (at Mixing Chamber No. 1) (m)

Initial Operation
Min Day Avg Day

Design Operation
Avg Day Max Hour

0.68
0.62
0.29
0.27

0.72
0.65
0.39
0.28

0.77
0.71
0.39
0.29

0.90
0.82
0.44
0.30

0.00
109.80

0.00
109.84

0.00
109.89

0.00
110.02

0.27
97.34
89.01
0.00
0.00
109.80

0.38
97.38
89.04
0.00
0.00
109.84

0.41
97.44
89.09
0.00
0.00
109.89

0.55
97.56
89.21
0.01
0.00
110.02

0.55
0.68
1.04
0.52
0.00
0.36

0.77
0.72
1.09
0.70
0.00
0.37

0.82
0.78
1.18
0.69
0.00
0.38

1.09
0.90
1.37
0.80
0.00
0.41

0.00
109.81

0.00
109.84

0.00
109.90

0.00
110.03

0.55
0.69
1.05
0.52
0.01
109.82

0.77
0.72
1.10
0.69
0.02
109.87

0.82
0.78
1.19
0.69
0.02
109.92

1.09
0.91
1.38
0.79
0.03
110.06

0.55
3.34
0.16
0.00
109.82

0.77
3.34
0.23
0.01
109.87

0.82
3.34
0.25
0.01
109.93

1.09
3.34
0.33
0.02
110.07

1.09
2.79
0.39
0.40

1.53
2.79
0.55
0.40

1.64
2.79
0.59
0.40

2.19
2.79
0.78
0.40

0.00

0.01

0.01

0.02

0.01
109.83

0.01
109.89

0.01
109.95

0.02
110.11

TABLE 22.5

(Continued)
Parameter

17. The above calculations (for Points 1 through 15) have

been for flow routed through Tank No. 4. When the
flow is routed through Tank No. !.the WSEL (m) is:
In reality, the headloss through each basin is equal.
The flow through the basin naturally adusts to
equalize headlosses, i. e. flow through Tank
No. 1 is greater than Q/4 and flow through Tank
No. 4 is less than Q/4. The actual headloss through
each basin can be estimated as the average of losses
through Tank No's. 1 and 4
and the WSEL (m) at Point 15 is:
18. Point 15 to Point 16
Flow = Q (mVs)
Conduit area = 2.29 m wide X 1.22 m deep (m2)
Velocity = flow/area (rn/s)
R = AIP (P = 2w + 2d) (m)
Conduit losses = L X [W(0.849 X C X /jo.63)]i/o.s4
(m) where L= 125.58 m and
Hazen-Williams C = 120
WSEL at Point 16 (m)
19. Point 16 to Point 17
Flow = Q (mVs)
Conduit area @ 16 = 2.29 m wide X 1.22 m deep (m22)
Conduit area @ 17 = 1.68 m wide X 1.68 m deep (m )
Average area (m2)
Velocity = flow /Area (m/s)
R @ 16 = A16/ (2 X (2.29 m + 1.22 m)) (m)
R @ 17 = A17/ (2 X (1.68 m + 1.68 m)) (m)
Average R, (m)
Conduit losses = L X [W(0.849 X C X
/jo.63)] 1/0.54 (m) where L = 9.14 m
and Hazen-Williams C = 120
WSEL at Point 17 (m)
20. Point 17 to Point 18
Fk)W = (MmVs)
Conduit area @ 17 = 1.68 m wide X 1.68 m
deep(m2)
Velocity 17 = flow/area
17 (m/s)
Pipe area @ 18 = (j)2 X TT (m) where D = 1.68 m
Velocity 18 = flow/area
18 (m/s)
Exit /osses = V182/2g - V172/2g (m/s)
WSEL at Point 18 (m)
21. Point 18 to Point 19
R = AfP(P = dX TT) (m)
Local losses = 3 elbows
(3 X 0.25V2/2g) +
entrance (0.5 X V2/2g) = 1.25 X V2/2g (m)
Conduit losses = L X [W(0.849 X C X
/{0.63)] 1/0.54 (m) where L = 138-68 m
and Hazen-Williams C = 120
WSEL at Point 19 (exit of Control Chamber) (m)

Initial Operation
Design Operation
Min Day Avg Day Avg Day Max Hour

109.82

109.88

109.94

110.08

109.83

109.89

109.95

110.10

2.19
2.79
0.78
0.40

3.06
2.79
1.10
0.40

3.28
2.79
1.18
0.40

4.38
2.79
1.57
0.40

0.04
109.87

0.08
109.97

0.10
110.04

0.16
110.26

2.19
2.79
2.81
2.80
0.78
0.40
0.42
0.41

3.06
2.79
2.81
2.80
1.09
0.40
0.42
0.41

3.28
2.79
2.81
2.80
1.17
0.40
0.42
0.41

4.38
2.79
2.81
2.80
1.56
0.40
0.42
0.41

0.00
109.88

0.01
109.98

0.01
110.05

0.01
110.27

2.19

3.06

3.28

4.38

2.81
0.78
2.21
0.99
0.02
109.90

2.81
1.09
2.21
1.39
0.04
110.01

2.81
1.17
2.21
1.49
0.04
110.09

2.81
1.56
2.21
1.98
0.8
110.35

0.42

0.42

0.42

0.42

0.06

0.12

0.14

0.25

0.07
110.03

0.13
110.27

0.15
110.39

0.26
110.86

TABLE 22.5 (Continued)

Parameter
22. Point 19 to Point 20
Weir elevation (m)
Depth of flow over weir = (WSEL @
19 - weir elevation), (m)
Length of weir, L, (m)
Flow over weir = q = 1.71 X hm X [ 1 - (d I nf2 ]0385
XL
Note: Rather than solve for /i, find an h by trial
and error that gives a q equal to the flow
for the given flow scenarios (given in Item 1)
assume h (m) =
then q (mVs) =
assume h (m)
=
then q (m3/s) =
Note: These #'s equal the flows for the given
scerios (Item 1)
h(m)
WSEL at Point 20 (h + WSEL @ Point 19) (m)
23. Point 20 to Point 21
Flow = Q (mVs)
Sluice gate area = 1.37 m X 1.37 m (m2)
Velocity = Flow/Area (m/s)
Gate Losses = 1.5 X V2/2g (m)
WSEL at Point 21 (Raw Water Control
Chamber) (m)
The overflow weir in the Raw Water Control
Chamber is 3.05 m long
and is sharp crested
Q = 1.82 X L X hm soh = (0/1.82L)2'3 (m)
The water surface must not rise above elevation 112.78 m
The overflow weir elevation may be safely set at 111.86 m

Initial Operation
Design Operation
MinDay Avg Day Avg Day Max Hour
109.73

109.73

109.73

109.73

0.30
2.74

0.54
2.74

0.66
2.74

1.13
2.74

0.60
1.84
0.66
2.18

0.90
3.14
0.89
3.07

0.95
3.12
0.97
3.27

1.35
4.21
1.37
4.42

0.66
110.39

0.89
110.62

0.97
110.70

1.37
111.10

2.19
1.88
1.16
0.10

3.06
1.88
1.63
0.20

3.28
1.88
1.74
0.23

4.38
1.88
2.33
0.41

110.49

110.82

110.93

0.54

0.67

0.70

111.51
0.85

hydraulic reaches analyzed in the example. The circled numbers indicate points at which
the WSEL is calculated. Hydraulic calculations start downstream of the sedimentation
basins (Fig. 22.12) and proceed upstream through the mixing chamber (Fig. 22.13) and
the Raw Water Control Chamber (Fig. 22.14). Mechanical mixers and mechanical flocculators are used. Conduit losses between the rapid mix chambers and the Raw Water
Control Chamber are also calculated in the example. Three different flow rates (i.e., minimum day, average day, and, maximum hour) are used in the calculations. This is a range
of design flow conditions that a design engineer would typically take into consideration.
The longest path through the flocculation and sedimentation processes, through Basin
No. 4, is followed (Points 1 through 15). Although not shown, losses along the shortest
path have also been calculated. As would be expected, the calculated head loss is smaller
for the shorter path. The actual losses are equal for each path. The flows through each path
naturally adjust to equalize losses. The flow through the longest path is slightly smaller
than the flow through the shortest path. In the example, the WSEL at Point 15 is adjusted
to reflect the average losses through the basins. The WSEL calculations upstream of Point

FLOW

INFLUENT PORT
(SEE DETAIL BELOW)
(TYP OF 24)

FLOCCULATION
ZONE BAFFLES

SEDIMENTATION
ZONE BAFFLE
TANK 4

TANK 3

TANK 2

TANK 1

EFFLUENT PORT
FLOCCULATION/SEDIMENTATION BASIN
MIXED WATER
CHANNEL
BOT EL 109.12mBAFFLE
DETAIL
INLET PORT
BETWEEN (7) AND (S)

EL = 108.20m-

SECTION A-A
INLET PORT

FIGURE 22.12 Flocculation/ sedimentation basin

15 are based on the adjusted WSEL. Alternatively the weirs or ports feeding flow into
each basin may be adjusted to create an equal distribution of flows in all basins as discussed in Sec. 22.2.1.
22.3.7.2.2 Filtration. Process criteria. Suspended solids are removed from the water as
it passes through a porous medium during filtration. Filters operate under either gravity or
pressure. Filters also differ in the type and distribution of the media used (fine, course,
uniformly graded, graded coarse to fine, etc.) and the direction of flow through the media
(upflow, downflow, and biflow). Pressure filter hydraulics information is very product
specific and should be obtained from the manufacturer. The design engineer using pressure filters should then apply this information to the project using project-specific
hydraulic considerations. This section presents information on gravity filters.
Key hydraulic design parameters. The headloss through a filter increases with use as
the voids become filled with solid particles. When the headloss reaches a certain point
(terminal headloss), the filter is backwashed to remove the solids. The rate of headloss
buildup is dependent on several factors, including how the filter is graded (the arrangement of media particle sizes). The rate of headloss buildup is reduced (and filtration is
more effective) when the flow first goes through the coarse media and then the fine media.

MIXING CHAMBER NO.2

1.83m SQUARE

TANK 4

TANK 3

FLOCCULATION BASIN
INFLUENT DETAIL
MIXED WATER CHANNEL
BOT. EL=109.12m

RAW WATER CONDUIT

BOT. EL= 103.78rr
FLOW
SECTION B-B
M I X I N G CHAMBER
FIGURE 22.13 Mixing chamber

MIXING
CHAMBER
NO. 2

MIXING
CHAMBER
NO. 1
1.68m DIAM
RAW WATER
CONDUIT
RAW WATER CONTROL CHAMBER
AND CONDUITS

FIGURE 22.14 Raw water control chamber

However, during backwash, the high rate of flow expands the filter bed and, over time,
the media are regraded so that the more coarsely graded grains are located at the bottom
and the fines are located at the top. To benefit from the coarse-to-fine grading, an upward
flow pattern can be used, but is very uncommon. More often the filter media are selected
such that the fine media have a higher specific gravity than the coarse media to maintain
the course-to-fine gradation during backwash. The most commonly used filter media are
natural silica sand and crushed anthracite coal; however garnet and ilmenite are used in
mixed media beds. Granular carbon is often used if taste and odor control is desired.
The terminal headloss is determined by a combination of factors including filter breakthrough (when the filter bed loses its adsorptive capacity), available static head, and outlet pressure required. The filter should be designed so that the headloss in any level of the
filter bed does not exceed the static pressure. A negative head can result in air binding in
the filter which will, in turn, further increase headloss.
Filter influent piping is sized to limit velocities to about 0.6 m/s (2 ft/s). Washwater and effluent piping flow velocities are kept below 1.8 m/s (6 ft/s) so that
hydraulic transients (waterhammer) and excessive headlosses are minimized and controlled to within tolerable limits.
Hydraulic design example. Table 22.6 illustrates the calculation of the WSEL from the
clear well back upstream to the Sedimentation Basin effluent at the medium-sized water
treatment plant shown in Fig. 22.10. Figures 22.15 and 22.16 show details of the hydraulic
reaches analyzed in the example. Table 22.7 illustrates the filter hydraulic calculation, the
details of which are shown in Figs. 22.17 and 22.18.
The hydraulic profile of the plant (based on hydraulic calculations done in Tables 22.5,
22.6 and 22.7) is shown in Figure 22.3.
TABLE 22.6 Hydraulic Calculations in a Medium-Sized Water Treatment Plant from the Filter
Effluent to the Effluent Clearwell
Parameter
1. Flow(m3/s)
Note: for Points 22 through 28, see Figure 22.15
2. Point 22 to Point 23
Maximum water level in Clearwell (Point 22) (m)
Invert in Clearwell (m)
Flow = 0/2 (mVs)
Stop logs @ A
Flow area (2 openings,
1.52 m wide,
3.66 m deep) (m2)
Velocity = flow/area (m/s)
Loss = 0.5 V2/2g (m)
Baffles
Flow area (3.05 m wide, 3.66 m deep) (m2)
Velocity = flow/area
(m/s)
Loss = 1.0 V2/2g (m)
Stop logs @ B and C
Same as the losses @ A, times 2 (m)
WSEL (see Note 1) at Point 23 (m)

Initial Operation
Min Day Avg Day

Design Operation
Avg Day Max Hour

2.19

3.06

3.28

4.38

105.16
101.50
1.09

105.16
101.50
1.53

105.16
101.50
1.64

105.16
101.50
2.19

11.15
0.20
0.00

11.15
0.27
0.00

11.15
0.29
0.00

11.15
0.39
0.00

11.15
0.20
0.00

11.15
0.27
0.00

11.15
0.29
0.00

11.15
0.39
0.01

0.00
105.16

0.00
105.17

0.00
105.17

0.01
105.18

(1) The Energy Grade Line (EGL) is equal to the Hydraulic Grade Line (HGL) + the Velocity Head (W2g). At
points in the system where the velocities are very low (i.e., V => O) the velocity head approaches O (i.e., W2g =>
O). Consequently, EGL => HGL => WSEL. Where: WSEL = Water Surface Elevation.

TABLE 22.6

(Continued)
Parameter

3. Point 23 to Point 24
Flow = QI2 (mVs)
1.68 (m) diameter pipe
Flow area = d2/4 X TT (m2)
Velocity = flow/area (m/s)2
Exit loss @ clearwell = V /2g (m)
Loss @ 2 - 90bends = (0.25 V2/2g) X 22 (m)
Entrance loss @ Filter Building
= 0.5 V /2g (m)
Pipe loss
= (3.022
X V1-85 X L)/
L8S
1 165
(C X D - ) where C = 120 and
L = 57.91 m (m)
WSEL at Point 24 (m)
4. Point 24 to Point 25
Flow = Q/4 (mVs)
Flow area = 1.52m X 1.52 (m2)
Velocity = Ql'A (m/s)
Loss as flows merge = 1.0 2/3
V2/2g2 (m)
Conduit loss = [(V X n)/(/? )] X L (m)
where n = 0.013, L = 16.76 m and R = AIP
(P = 6.1Om)
WSEL at Point 25 (m)
5. Point 25 to Point 26
Sluice Gate No. 1 flow area = 1.22 m X 0.91 m (m2)
Velocity = QIA
(m/s)
Loss = 0.5 V2/2g (m)
WSEL at Point 26 (m)
6. Point 26 to Point 27
Sluice Gate No. 2 Loss = 0.8 V2/2g (m)
WSEL at Point 27 (m)
7. Point 27 to Point 28
Port to Filter Clearwell: Calculate losses through port
as if were a weir when depth of flow is below top
of port. Port dimmensions = 2.74 m wide
by 0.813 m deep. Flow = Q/4 (m's)
Weir (bottom of port) elevation (m)
Depth of flow over weir =
(WSEL @ 27 - weir elevation) (m)
Flow over submergedweir = q 1.71 X hm
X [1 - (d/hf2]03*5 X L
Note: Rather than solve for h, find an h, by trial
and error, that gives a q equal to the flow for the
given flow scenario
assume h (m) =
then q (mVs) =
assume h (m) =
then?(m3/s)=
Note: These q's equal the flows for the given
scenarios
h(m)

Initial Operation
Design Operation
Min Day Avg Day Avg Day Max Hour
1.09

1.53

1.64

2.19

2.21
0.50
0.01
0.01
0.01

2.21
0.69
0.02
0.01
0.01

2.21
0.74
0.03
0.01
0.01

2.21
0.99
0.05
0.03
0.03

0.00
105.19

0.00
105.22

0.00
105.23

0.00
105.28

0.55
2.32
0.24
0.00

0.77
2.32
0.33
0.01

0.82
2.32
0.35
0.01

1.09
2.32
0.47
0.01

0.00

0.00

0.00

0.00

1.11
0.49
0.01
105.20

1.11
0.69
0.01
105.24

1.11
0.74
0.01
105.24

1.11
0.98
0.02
105.32

0.01
105.21

0.02
105.25

0.02
105.27

0.04
105.36

0.55
104.85

0.77
104.85

0.82
104.85

1.09
104.85

0.36

0.40

0.42

0.51

0.40
0.59
0.39
0.52

0.45
0.69
0.46
0.76

0.50
0.95
0.48
0.82

0.60
1.23
0.58
1.09

0.39

0.46

0.48

0.58

TABLE 22.6

(Continued)
Parameter

WSEL at Point 28 (m)

FiltersSee Filter Hydraulics in Table 22.7
Note: for Points 29 through 33, see Fig. 22.16
8. Point 29
WSEL above filters (m)
9. Point 29 to Point 30
Entrance to Filter #4
Flow, G/8 (mVs)
Channel velocity = flow/area
(area = 1.22m X 1.22m) (m/s)
Submerged entrance loss = 0.8 V2/2g (m)
1.22 m pipe velocity
= flow/area
(area = d2/4 X TT) (m/s) 2
Butterfly valve loss = 0.25 V IIg (m)
Sudden enlargement loss - 0.25 V2/2g (m)
WSEL in influent channel (Point 30) (m)
10. Point 30 to Point 31
Flow depth = WSEL @ 30 - invert (107.29
m) (m)
Flow area = 1.83 m width X depth (m2)
Velocity = flow/area (m/s)
R = AIP (P = w + 2d) (m) 2 3 2
Conduit Loss = [(V X )/(fl ' )] X L
where n = 0.014 and L = 10.77 m (m)
WSEL at Point 31 (m)
11. Point 31 to Point 32
Flow = Q/4 (mVs)
Flow depth = WSEL @ 31 -invert (107.29m) (m)
Flow area = 1.83 m width X depth (m2)
Velocity = flow/area (m/s)
R = AIP (P = w + 2d) (m) 2/3 2
Conduit loss = [(V X n)/(/? )] X L (m)
where n = 0.014 and L = 10.77 m
WSEL at Point 32 (m)
12. Point 32 to Point 33
Flow = 3<2/8 (mVs)
Flow depth = WSEL @ 32 - invert (107.29
m) (m)
Flow area = 1.83 m width X depth (m2)
Velocity = flow/area (m/s)
R = A/P(P = w + 2d) (m)2/3 2
Conduit loss = [(V X n)/(/? )] X L (m)
where n = 0.014 and L = 10.77 m
WSEL at Point 33 (m)
13. Point 33 to Point 1
Flow = Q/2 (mVs)
Flow depth = WSEL @ 33 - invert (107.29
m) (m)
Flow area = 1.83 m width X depth (m2)
Velocity = flow/area (m/s)
R = AIP (P = w + 2d) (m) 2/3 2
Conduit loss = [(V + )/(/? )] + L (m)
where n = 0.014 and L= 11.07 m
WSEL at Point 1 (m)

Initial Operation
Design Operation
Min Day Avg Day Avg Day Max Hour
105.24

105.31

105.33

105.43

109.73

109.73

109.73

109.73

0.27

0.38

0.41

0.55

0.18
0.00

0.26
0.00

0.28
0.00

0.37
0.01

0.23
0.00
0.00
109.73

0.33
0.00
0.00
109.73

0.35
0.00
0.00
109.73

0.47
0.00
0.00
109.74

2.44
4.46
0.06
0.67

2.44
4.47
0.09
0.67

2.44
4.47
0.09
0.67

2.45
4.48
0.12
0.67

0.00
109.73

0.00
109.73

0.00
109.73

0.00
109.74

0.55
2.44
4.46
0.12
0.67

0.77
2.44
4.47
0.17
0.67

0.82
2.44
4.47
0.18
0.67

1.09
2.45
4.48
0.24
0.67

0.00
109.73

0.00
109.73

0.00
109.73

0.00
109.74

0.82
2.44
4.46
0.18
0.67

1.15
2.44
4.47
0.26
0.67

1.23
2.44
4.47
0.28
0.67

1.64
2.45
4.48
0.37
0.67

0.00
109.73

0.00
109.73

0.00
109.73

0.00
109.74

1.09
2.44
4.46
0.24
0.67

1.53
2.44
4.47
0.34
0.67

1.64
2.45
4.47
0.37
0.67

2.19
2.45
4.48
0.49
0.67

0.00
109.73

0.00
109.73

0.00
109.74

0.00
109.74

FILTER
EFFLUENT
CONDUIT
-CONTROL WEIR
SLUICE GATE *2
FILTER
BUILDING
SLUICE GATE #1

STOP LOG OPENINGS

JO FINISHED WATER
"PUMPING STATION
CLEARWELL TO FILTER EFFLUENT
FIGURE 22.15 Clearwell to filter effluent

SEDIMENTATION BASIN
EFFLUENT CONDUIT
OPENING(TYP)

1.22m BUTTERFLY
VALVE(TYP)
FILTER

INFLUENT

FIGURE 22.16 Filter effluent

TABLE 22.7 Example Hydraulic Calculation of a Typical Filter

Parameter

Initial Operation
Min Day Avg Day

Design Operation
Avg Day Max Hour

Plant flow (mVs)

2.19
3.06
3.28
4.38
Filter loading, [(m3 m)/m2]
0.083
0.167
0.250
0.334
Filter area per filterseven (7) out of eight (8)
115
115
115
115
filters in operation (m2)
Flow = loading X area (mVs)
0.16
0.32
0.48
0.64
Losses through filter effluent piping (Fig. 22.17)
0.51 m piping (Q):
Pipe velocity = QIA (m/s)
0.79
1.58
2.37
3.16
Local losses = Exit (0.5) + butterfly
valves (2 X 0.25) + 90 elbows (2 X 0.4)
+ tee (1.8) = 3.6V 2 /2g(m)
0.11
0.46
1.03
1.83
R=AIP = (d2/4 X Tr)AW X TT) = d/4 (m)
0.13
0.13
0.13
0.13
Conduit losses = L X [V/(0.849 X C X /?063)]1/0-54
where L = 6.10 m and HazenWilliams C = 120 (m)
0.01
0.03
0.06
0.11
0.51 m piping (QII):
Pipe velocity = Q/A (m/s)
0.40
0.79
1.19
1.58
Local Losses = Butterfly Valve (0.25) (m)
0.00
0.01
0.02
0.03
R = AIP = (d2/4 X ir)/(d X TT) = d/4 (m)
0.13
0.13
0.13
0.13
Conduit losses = L X [V/(0.849 X C X /?o.63)]i/o.54
where L = 3.05 m and HazenWilliams C = 120 (m)
0.00
0.00
0.01
0.02
0.61 m piping (0/2):
Pipe velocity = QIA (m/s)
0.27
0.55
0.82
1.10
Local losses = entrance (1.0) + tee (1.8)
2
= 2.8V /2g(m)
0.01
0.04
0.10
0.17
Filter (clean) and underdrain losses (obtain from
manufacturer) (m)
0.09
0.15
0.23
0.34
Total losses (effluent pipe and clean filters) (m)
0.23
0.70
1.45
2.50
Assume that headless will be allowed to increase 2.44 m before the filters are backwashed. A rate controller
will be used to maintain a constant flow through the filters. Determine the ranges of available head over
which the rate controller will operate.
Static Head (Fig. 22.18)
WSEL (see Note 1) above filters (m)
109.73
109.73
109.73
109.73
WSEL in filter effluent conduit, Point 29
(see Example 22.2) break Maximum (m)
105.61
105.61
105.61
105.61
Minimum (m)
105.16
105.16
105.16 105.16
Static head = WSEL above filtersWSEL at
Point 29 (Filter effluent conduit-2)
Maximum (m)
4.57
4.57
4.57
4.57
Minimum (m)
4.11
4.11
4.11
4.11
Available head = static head -2.44 m
Maximum (m)
2.13
2.13
2.13
2.13
Minimum (m)
1.68
1.68
1.68
1.68
(1) The Energy Grade Line (EGL) is equal to the Hydraulic Grade Line (HGL) + the Velocity Head (W2g). At
points in the system where the velocities are very low (i.e., V => O) the velocity head approaches O (i.e., W2g =>
O). Consequently, EGL => HGL => WSEL. Where: WSEL = Water Surface Elevation.

NORTH HALF
FILTER
SOUTH HALF
FILTER
W.S. EL 109.72mPLAN AT 100.58m

WEIR

CLEAR
WELL

SECTION A-A

SECTION B-B

Head, meters

FIGURE 22.17 Filter effluent piping

Maximum Static Head

Minimum Static Head
Maximum Available Head
Minimum Available Head
Effluent Piping and Clean Filter Losses

Filter Loading, (m2 min)/m2

FIGURE 22.18 Available head over which filter effluent rate controller operatesmetric units.

22.3.8

Membrane Technology

Membranes are synthetic filtering media manufactured from a variety of materials including polypropylene, polyamide, polysulfone, and cellulose acetate. The membrane material can be arranged in various configurations, including the following:

Spiral wound
Hollow fiber
Tubular
Plate frame

Examples of these configurations are presented in Fig. 22.19. In water and wastewater
treatment applications, the most common configurations are spiral wound and hollow fiber.
In general, there are four classes of membranes: microfllters (MF), ultrafilters (UF),
nanofilters (NF), and hyperfilters. Treatment through hyperfilters is referred to as hyperfiltration, or reverse osmosis (RO).
The hydraulics associated with membranes are membrane-specific and can be obtained
from the manufacturer. This section presents general considerations pertinent to flow
through membranes.
As with natural particle media filters, clean membranes have a specific headless and,
over time, as the membranes become covered with a cake buildup, the effectiveness of the
membrane decreases and headloss increases. Fouling (excessive buildup) may damage the
membrane.
The need for pretreatment ahead of membranes is determined by the raw water quality and the membrane type. In general, microfilters and ultrafilters do not require pretreatment for treating surface or groundwater. Nanofilters and reverse osmosis membranes may require pretreatment depending on the type of fouling. Membrane fouling
can result from particulate blocking, chemical scaling, and biological growth within the
membranes.
An estimate of particulate blocking can be made using indices such as the Silt Density
Index (SDI) and the Modified Fouling Index (MFI). These fouling indices are determined
from simple bench membrane tests using 0.45 micron Millipore filters and monitoring
flow through the filter at a given pressure, usually 30 psig. Approximate values of suitable SDIs for nanofiltration are 0-3 units, and for reverse osmosis, 0-2 units.
Corresponding values of MFI are, for nanofiltration O to 10 s/L2, and for RO, 0-2 s/L2.
Scaling control is essential in RO and nanofilter membrane filtration, especially when
the filtration provides water softening. Controlling precipitation or scaling within the
membrane element requires identification of limiting salt, acid addition for prevention of
calcium carbonate precipitation within the membrane, and/or the addition of an
antiscalant. The amount of antiscalant or acid addition is determined by the limiting salt.
A diffusion controlled membrane process will naturally concentrate salts on the feed side
of the membrane. As water is passed through the membrane, this concentration process
will continue until a salt precipitates and scaling occurs. Scaling will reduce membrane
productivity and, consequently, recovery is limited by the allowable recovery just before
the limiting salt precipitates. The limiting salt can be determined from the solubility products of potential limiting salts and the actual feed stream water quality. Ionic strength must
also be considered in these calculations as the natural concentration of the feed stream
during the membrane process increases the ionic strength, allowable solubility and recovery. Calcium carbonate scaling is commonly controlled by sulfuric acid addition, although
sulfate salts, such as barium sulfate and strontium sulfate, are often the limiting salt.
Commercially available antiscalants can be used to control scaling by complexing the

FIGURE 22.19 Membrane configurations, (a) Spiral wound, (b) hollow fiber, (c) tubular, (d) plate
and frame.

metal ions in the feed stream and preventing precipitation. Equilibrium constants for these
antiscalants are not available which prohibits direct calculation. However, some manufacturers provide computer programs for estimating the required antiscalant dose for a
given recovery, water quality, and membrane.
Biological fouling is controlled with some membranes such, as cellulose acetate, by
maintaining a free chlorine residual of not more than 1 mg/L. Other membranes, such as
the thin-film composites, are not chlorine tolerant and must rely on upstream disinfection
by, for example, ultraviolet disinfection or chlorination-dechlorination. The extent of fouling for a specific application and its influence in the design of nanofiltration and RO
membrane systems is best determined by pilot studies.
It has been suggested that some buildup on the membrane may be beneficial to treatment by providing an additional filtering layer. At facilities operated by the Metropolitan
Water District of Southern California (MWD), removal rates of 1.7-2.9 logs were

observed for seeded virus MS2 bacteriophage through microfliters that had a pore size an
order of magnitude larger than the nominal size of MS2 (1).
The microfiltration system used by MWD utilizes an air backwash procedure whereby compressed air at 90-100 psig is introduced into the filtrate side of the hollow fiber
membranes. Accumulated particulates dislodged by the compressed air are swept away
by raw water introduced to the feed side of the membranes. The backwash sequence is
carried out automatically at preset time intervals. MWD found the best interval to be
every 18 minutes. The total volume of backwash represents approximately 5-7 percent
of influent flow.
The difference between influent and effluent pressure across the membrane is termed
the transmembrane pressure (TMP). Despite the frequent air and water backwashes, the
TMP gradually increases over time. Generally, when the TMP reaches approximately
15 psig, chemical cleaning of the membranes is carried out. If the TMP is allowed
to increase beyond 15 psig, particulates can become deeply lodged within the lattice structure of the membranes and will not be removed, even by chemical cleaning. Chemical
cleaning typically lasts 2-3 hours and involves circulating a solution of sodium hydroxide and a surfactant through the membranes, and soaking them in the solution.
The membranes at the MWD microfilter plants have a surface loading rate of 40-67
gal/day ft2. The lower flux rate of 40 ft2 has the advantage that the rate of increase of
TMP is reduced and the interval between chemical cleanings is increased. A possible
explanation for this is that particulates are not forced as deeply into the lattice structure of
the membranes, thereby allowing the air-water backwash to clean the membranes more
effectively. By reducing the flux rate from 67-40 gal/day ft2, the interval between chemical cleanings was increased from 2 to 3 weeks to almost 20 weeks. However, MWD has
instituted a maximum run time of 3 months between chemical cleanings to ensure the
long-term integrity of the membranes.
Nanofiltration is widely used for softening groundwaters in Florida. A typical nanofiltration plant would include antiscalant for scale control added to the raw water. Cartridge
filters, usually rated at 5 microns, remove particles that may foul the membrane system.
Feed water pumps boost the pretreated water pressure to about 90-130 pounds per square
inch (psi) before entering the membrane system. The membranes typically are spiral
wound nanofiltration membranes generally with molecular weight cutoff values in the
200-500 dalton range.

22.4

WASTEWATER TREATMENT

Many factors and considerations influence the hydraulic design of a wastewater treatment
plant. This section describes typical phases of wastewater treatment planning required for
design of new plants or additions to existing plants, and then presents typical unit process
hydraulic computations.
22.4.1 Wastewater Treatment Planning
Hydraulic decision making for a new wastewater treatment plant or expansion of an existing plant involves several planning phases. Typical planning phases are presented below
in their common order of consideration.
22 A.I.I Service area and flows. More than 15,000 municipal wastewater treatment plants
are in operation in the United States today. The plants are designed to treat a total of about
Next Page

CHAPTER 23
HYDRAULIC DESIGN
FOR GROUNDWATER
CONTAMINATION

Paul C. Johnson
Department of Civil and Environmental Engineering
Arizona State University
Tempe, Arizona

23.1

INTRODUCTION

This chapter focuses on in situ treatment systems designed to eliminate or minimize the
impacts of hazardous chemicals on groundwater quality. This chapter builds on the principles introduced in Chapter 4. First, typical soil and groundwater contamination scenarios of interest are introduced. Then a discussion of general remediation design principles
and strategies follows. Finally, the design, monitoring, and refinement of selected in situ
remediation technologies is presented. These include both conventional and developing
technologies, and include containment, reaction barrier, active remediation, and passive
remediation systems.
23.1.1 Unique Features of In Situ Treatment Technology Design
The design of in situ treatment systems for groundwater contamination is uniquely different from the design of above-ground treatment and hydraulic conveyance systems. The
subsurface is an environment that is difficult to describe precisely; it is naturally heterogeneous and, due to practical constraints, characterization data are generally limited.
Thus, treatment systems must be designed under conditions of great uncertainty-usually
the amount of chemical spilled is unknown, the timing of the release is unknown, and the
system into which the spill occurred is poorly characterized. For this reason, the design of
in situ treatment systems relies heavily on

conceptual models,
screening-level calculations,
empiricism, heuristics, and experience, and
monitoring and refinement of the design.

A conceptual model is a realization of how the subsurface might look and how contaminants might move, based on the available data. Often, there are several plausible conceptual models, and the treatment system designer must allow for a range of possibilities
in the design. It is also recognized that the conceptual model is continuously refined as
new data become available.

Screening level calculations are used to estimate treatment performance limits. They
are generally biased toward estimating best-case performance, and are useful for making
rough estimates of treatment times and treatment limits. They are primarily used for feasibility screening and creating the initial treatment system design. More sophisticated
models are available and can be used; however, given the typical level of site characterization data and gross approximations that must be made, their use is rarely warranted,
except at those sites where treatment costs are projected to be very high.
Empiricism and experience play a key role in the design as well. In many ways, each
site is unique; yet, the collective experience from many sites is invaluable. This design
field is still relatively young and most experiences are not well-documented. As is often
the case, successful applications are more likely to be reported than failures, although the
knowledge from both would be equally valuable. In practice, there are many "rules of
thumb" that designers use and some are well founded in theory, while others are indefensible (yet they still persist in practice). In any case, it is useful to be aware of these rulesof-thumb even if their bases are not well documented in the literature.
Finally, given the inherent large initial uncertainty, it is important to recognize that the
design phase continues well past the construction, installation, and operation of the initial
system. Following sound design practices does not guarantee success; it does, however,
provide a higher probability of success at most sites. Appropriate monitoring is essential
to verify assumptions built into the conceptual model and initial design, and system
design refinement should be anticipated in the initial design. For this reason, it is important to build robust systems that are both flexible and expandable.
23.1.2

Overview

It has been estimated that costs associated with the remediation of known hazardous
waste sites in the United States will exceed $700 billion (all costs are quoted in U.S. dollars) and that the current average rate of expenditure is about $20 billion (Bredehoeft,
1994). Despite the fact that society has been tackling this difficult problem since the
1970s, the environmental profession still struggles to find cost-effective remediation
processes. Remediation costs for service station sites now range from $100,000 to
$500,000 per site, while costs for the larger landfill and manufacturing sites often exceed
$10,000,000 per site. Costs associated with sites where groundwater is contaminated
generally exceed those where only soil is impacted by at least an order of magnitude.
Clearly, these are not trivial expenses; thus, appropriate technology selection, design, and
optimization are important activities.
23.1.3 Groundwater Contamination Scenarios-Point Versus
Area Sources
Groundwater contamination is prevalent in most industrial societies. Leaking fuel tanks,
petroleum storage and refining, solvent degreasing activities, chemical manufacturing
and storage, landfills, and mining activities have all contributed to groundwater contamination problems. Groundwater contamination is also prevalent in most modern agricultural areas, where the combination of irrigation and fertilizer, pesticide, and herbicide use
have contributed to many large-scale groundwater contamination problems. Fig. 23.1
[US Enviromental Protection Agency (USEPA), 1990] summarizes the more common
sources of groundwater contamination and the relative frequency at which they are
expected to cause groundwater impacts. Septic tanks and leaking underground fuel storage tanks are considered to be the largest single contributor to the number of groundwa-

Underground Storage Tanks

Septic Tanks
Agricultural Activity
Municipal Landfills
Surface Impoundments
Abandoned Hazardous Waste Sites
Industrial Landfills
Other Landfills
Injection Wells
Regulated Hazardous Waste Sites
Land Application
Oil and Gas Brine Pits
Other

Number of States and Territories

FIGURE 23.1 Common sources of ground water contamination and the number of states and territories considering each to be major threat to groundwater quality.
(USEPA, 1990, after Fetter, 1993).

ter contamination sources. Agricultural activities are likely the largest contributor to
groundwater contamination in terms of the actual extent of impact.
For the purposes of this chapter, we first segregate the universe of groundwater contamination scenarios into two main categories:
Point sources,
Area, or distributed sources
Underground storage tanks, above-ground storage tanks, landfills, pipeline releases,
chemical manufacturing and petroleum refining locations, wood treating facilities, and so
forth are all considered to be point sources. Agricultural activities are often considered to
be area, or distributed sources.
By virtue of their large areal extent, the environmental engineering profession has yet
to develop effective solutions for the remediation of area source contamination problems.
These often involve widespread region- or basin-scale nitrate or pesticide contamination.
Currently, the only practicable solution for these problems appears to be above-ground
water blending and direct well-head treatment. In other words, contaminated groundwater is either diluted with water of higher quality as it is pumped from the aquifer, or else
it is treated as necessary above-ground as it is removed from the aquifer. As stated, area
source problems are rarely addressed with in situ treatment systems, and therefore, these
will not be considered further in this chapter.
23.1.4 Groundwater Contamination Scenarios-Segregation
by Contaminant Type
Point source groundwater contamination problems are divided here into three main categories according to the contaminant properties. The three categories are
Light nonaqueous phase liquids (LNAPLs)
Dense nonaqueous phase liquids (DNAPLs)
Inorganics and other dissolved constituents
LNAPLs, in their pure liquid form, are less dense than water (p < 1 g/cm3). Most
LNAPL sites involve the release of petroleum products or crude oil (e.g., service stations,
refineries, pipeline spills). DNAPLs, in their pure liquid form, are more dense than water
(p > 1 g/cm3). DNAPL contamination is generally found near sites where dry cleaning and
aviation, automobile, and electronic circuit board degreasing operations took place.
Historically these used chlorinated solvents, such as trichloroethylene (TCE) and perchloroethylene (PCE). Metals and salts fall into the category of inorganics and other dissolved constituents. Mining operations, electroplating operations, leaking wastewater
treatment facilities, and landfills are examples of sources that historically have added contaminants to groundwater in dissolved form.
23.1.5 Groundwater Contamination Scenarios-Subsurface
Contaminant Distributions
Figures 23.2a, 23.2&, and 23.2c schematically illustrates release scenarios involving
LNAPLs, DNAPLs, and inorganics. While very simplistic, these figures highlight key
characteristics that impact subsurface characterization activities, remediation design,
and monitoring.

Underground
Storage Tank

, residual (immobile) hydrocarbon liquid

. hydrocarbon vapors
vadose zone

free-product (mobile)
Jiy&ocwbon ItiyM

dissolved hydrocarbon plume

dry cleaners

residual (immobile) hydrocarbon liquid

hydrocarbon vapors
vadose zone

dissolved
hydrocarbon
plume
dissolved
hydrocarbon
plume
FIGURE 23.2 Example contaminant release scenarios: (a) LNAPL release.(b)
DNAPL release.

land fill
leachate containing
dissolved inorganic
chemicals
vadose zone

dissolved
inorganic
plume

FIGURE 23.2 (Continued) (c) Example contaminant release scenarios: (c) dissolved
contaminant release.
As their density is less than that of water, one can expect to find LNAPLs in the general vicinity of the water table. Provided that a given spill or leak is of sufficient size and
that there are not any impermeable vertical flow barriers, LNAPLs travel downward and
pool on top of the water table. Then over time, water table fluctuations cause a vertical
smearing so that immiscible-phase LNAPL can be found trapped by capillary forces in the
soil pores above and below the current ground water table level. LNAPL solubilities are
generally so low that dissolution (and other natural loss processes) occurs very slowly and
pools of LNAPL will persist for decades.
Dissolution of LNAPLs into groundwater leads to the formation of dissolved contaminant plumes. Aquifer characteristics and chemical properties control the growth and
extent of the dissolved contaminant plume. Fortunately, most LNAPL sites involve the
release of petroleum hydrocarbons and some of these are aerobically degradable by
indigenous microorganisms. The result is that natural chemical degradation often limits
the growth of the dissolved LNAPL plume. For example, two survey studies have shown
that most benzene, toluene, ethylbenzene, and xylenes (BTEX) dissolved plumes at
leaking underground storage tank (LUST) sites do not extend more than 300 ft (100 m)
from the downgradient edge of the source zone (Mace et al., 1997; Rice et al., 1995). It is
important to note, however, that not all fuel-related contaminants degrade at an appreciable rate. For example, some oxygenates (e.g., MTBE), are highly soluble and resistant to
degradation. Consequently, these compounds generally migrate much further distances
than the more degradable monoaromatic chemicals (e.g., benzene, toluene, xylenes). The
behavior of these compounds is currently the focus of intense research.
DNAPL sites are generally more complex and challenging. Provided that a given
spill or leak is of sufficient size and that there are not any impermeable vertical flow
barriers, DNAPLs travel downward to the water table. Initially, they too will pool on
top of the water table; but, provided that the spill or leak is large enough, the weight

of the DNAPL pool will exceed the entry pressure of the medium and at that point the
DNAPL will penetrate through the groundwater table. It will continue to migrate
downward until a vertical flow barrier is encountered, or until all of the DNAPL liquid is bound by capillary forces. DNAPL vertical migration in aquifers appears to be
very sensitive to small changes in soil structure, and as a DNAPL migrates downward,
it may also spread horizontally along thin, more permeable, soil layers. DNAPL solubilities are also generally low so that dissolution (and other natural loss processes)
occurs very slowly and DNAPL found below the water table will persist for decades.
The assessment and remediation of DNAPL sites is generally much more challenging
than that of LNAPL sites.
Most DNAPLs of interest are chlorinated solvents (e.g., TCE, PCE, and so on) and
these degrade slowly, if at all, under natural conditions. This results in dissolved plumes
that are much longer than the dissolved LNAPL plumes that might be found in similar settings, and plumes > 1500 ft (500 m) in length are not uncommon; in fact, plumes several
miles (kilometers) in length can be found in regions of concentrated historical aviation or
electronics manufacturing.
In the case of inorganics, such as those originating from the landfill in Fig. 23.2, the
contaminants are already dissolved in solution when they reach the water table. Thus,
these chemicals follow groundwater flow once they reach the aquifer. Dissolved inorganic chemical migration is governed by complex geochemical interactions; however,
in some cases of interest (e.g., nitrate contamination), the contaminants typically are not
transformed or retarded, and may also form long dissolved plumes that persist for
extended periods of time. These solutions generally behave as neutrally buoyant liquids
(p ~ 1 g/cm3); however, if the total concentration of inorganics exceeds about 10,000
mg/L H2O, then density gradients will cause these solutions to migrate vertically downward within aquifers.

23.2

REMEDIATIONGOALS

When subsurface soil and/or groundwater contamination have been identified, decisions
must be made as to whether or not treatment is necessary, and if so, to what levels. Often
regulatory guidelines prescribe the actions that must be taken. In most cases, contaminant
concentrations in soil and groundwater are compared with levels that have been deemed
acceptable. These levels are sometimes referred to as target levels. If exceedences are
noted, then treatment or further study is required; otherwise, no action other than monitoring is typically required.
While the development and enforcement of target levels is not the focus of this chapter, it is important to recognize that target levels play a significant role in the selection and
design of in situ treatment systems. There is no universal set of target levels, and they vary
from state to state and country to country. Thus, it is worthwhile to briefly discuss the
range of situations that might be encountered in practice.
Target levels may be prescribed in terms of acceptable groundwater concentrations,
or in terms of soil concentrations that are expected to be sufficiently protective of groundwater quality. In the United States, target levels are often linked to
maximum contaminant levels (MCLs) or maximum contaminant level goals (MCLGs)
risk-based considerations, or
resource protection goals

23.2.1 Maximun Contaminant Levels (MCL)

In the United States, maximum contaminant levels (MCLs) are concentrations that must
not be exceeded at any drinking water supply point (e.g., the home faucet). Water suppliers are responsible for meeting these criteria. MCLs are promulgated by USEPA, and are
based on considerations of health effects, aesthetic impacts, available treatment technologies, analytical limitations, and cost. Table 23.1 lists MCLs and MCLGs for some chemicals of interest (USEPA 1996a). Historically, it was common for regulatory agencies to
uniformly adopt MCLs as target levels for groundwater quality.
23.2.2 Risk-Based Target Levels
In recent years some regulatory agencies have moved away from using MCLs as fixed
cleanup levels, and have begun to consider risk-based target levels [e.g., American Society
of Testing and Materials (ASTM), 1995]. Like MCLs, risk-based target levels are concentrations also deemed to be protective; however, they differ from MCLs in that they take into
account site-specific considerations, the beneficial uses of the aquifer (or soil), the distance
from the contaminant source zone and actual water supply wells (or other potential receptors), and any dilution that might occur during groundwater pumping. Risk-based target
levels may be larger, or smaller, than MCLs, depending on the site-specific conditions.
23.2.3

Resource Protection Goals

In some areas aquifers are considered to be valued resources, and therefore must be protected from any impacts. In such areas, no level of impact is acceptable. These anti degradation policies require that all contamination must be remediated and the resource must
be restored to pristine, or background, conditions. Of the three approaches, this is typically the most stringent as the cleanup goal is either zero, or the background concentration. It is also typically the most difficult goal to achieve.
23.2.4 Application of the Target LevelsRemediation,
Points of Compliance, and Containment
Once target levels are selected, one must choose how and where to apply them. Two obvious choices are
apply target levels everywhere in the soil and groundwater, or
apply the target levels to the boundary of a compliance zone.
While total cleanup of soils and groundwater is always preferred, it is not always practicable. For this reason, a compliance zone is sometimes negotiated. Contaminant concentrations must not exceed the target levels outside of this compliance zone, but may
exceed them within the compliance zone. This approach is similar to that taken when
permitting mixing, or dilution zones for surface water discharges.
The allowance of a compliance zone greatly impacts the range of technologies that can
be selected at a site. In this case complete remediation of the source zone and groundwater is not always necessary and this opens the door for consideration of contaminant con-

TABLE 23.1 Regulatory and Human Health Benchmarks Used for Soil and Groundwater Target
Cleanup Levels

Chemical

Maximum
Contaminant
Level Goal
(MCLG)
WQ

Acenapthene
Acetone
Aldrin
Anthracene
6
Antimony
Arsenic
Barium
2000
Benz(a)anthracene
Benzene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzoic acid
Benzo(a)pyrene
4
Beryllium
Bis (2-chloroethyl) ether
Bis (2-ethylhexyl) phthalate
Bromodichloromethane
Bromoform (tribromomethane)
Butanol
Butyl benzyl phthalate
Cadmium
5
Carbazole
Carbon disulfide
Carbon tetrachloride
Chlordane
p-Chloroaniline
100
Chlorobenzene
Chlorodibromomethane
60
Chloroform
2-Chlorophenol
Chromium
100
Chromium (III)
Chromium (VI)
Chrysene
Cyanide (amenable)
200
DDD

Maximum
Contaminant
Level
(MCL)
(g/L)

Water HealthBased Limit*

Mq/L

2,000
4,000
0.005
10,000

6
50
2,000
0.1

0.005
0.1
1.0
100,000
0.2
4

0.08
6
100*
100*

4,000
7,000
5
4
4,000
5
2
100
100
100*
100*
200
100

40,000
100

10
200
0.4

TABLE 23.1.

(Continued)

Chemical

DDE
DDT
Dibenze(a,h)anthracene
Di-n-butyl phthalate
1 ,2-Dichlorobenzene
1 ,4-Dichlorobenzene
3 ,3-Dichlorbenzidene
1 , 1 -Dichloroethane
1 ,2-Dichloroethane
1 , 1 -Dichloroethylene
cis-1 ,2-Dichloroethylene
trans- 1 ,2-Dichloroethylene
2,4-Dichlorophenol
1 ,2-Dichloropropane
1 ,3-Dichloropropene
Dieldrin
Diethylphthalate
2,4-Dimethylphenol
2,4-Dinitrophenol
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Di-rc-octyl phthalate
Endosulfan
Endrin
Ethylbenzene
Fluoranthene
Fluorene
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachloro- 1 ,3-butadiene
a-HCH (a-BCH)
P-HCH O-BCH)
HCH (Lindane)
Hexachlorocyclopentadiene
Hexachloroethane
Indeno( 1 ,2,3-cd)pyrene

Maximum
Contaminant
Level Goal
(MCLG)
WU)

Maximum
Contaminant
Level
(MCL)
WV

Water HealthBased Limit*

Mq/L

0.3
0.3
0.01
4,000
600
75

600
75
0.2
4,000

7
70
100

5
7
70
100
100
5
0.5
0.005
30,000
700
40
0.1
0.1
700
200

2
700

2
700

1,000
1,000
0.4
0.2
1
1

0.2
50

1
0.01
0.05
0.2
50
6
0.1

TABLE 23.1

(Continued)

Chemical

Maximum
Contaminant
Level Goal
(MCLG)
WQ

Maximum
Contaminant
Level
(MCL)
(V8/L)

Water HealthBased Limit*

90
Isophorone
2
2
Mercury
40
40
Methoxychlor
50
Methylbromide
Mthylene Chloride
5
2,000
2-Methylphenol (0-cresol)
1,000
Napthalene
100
Nickel
20
Nitrobenzene
20
N-Nitrosodiphenylamine
0.01
N-Nitrosodi-n-propylamine
1
Pentachlorophenol
20
Phenol
1,000
Pyrene
Selenium
50
50
200
Silver
100
100
Styrene
0.4
1 , 1 ,2,2-Tetrachloroethane
5
Tetrachloroethylene
2
Thallium
0.5
Toluene
1,000
1,000
3
Toxaphene
70
70
1 ,2,4-Trichlorobenzene
200
200
1,1,1-Trichloroethane
3
5
1 , 1 ,2-Trichloroethane
Trichloroethylene
O
5
4,000
2,4,5-Trichlorophenol
8
2,4,6-Trichlorophenol
300
Vandium
40,000
Vinyl acetate
2
Vinyl chloride (chloroethene)
10,000
10,000
w-, o, and p-Xylene
Zinc
10,000
proposed MCL=SO ng/L, Drinking Water Regulations and Health Advisories, USEPA (1995a)

tainment systems. In summary, prior to selecting and designing in situ treatment technologies, the user must
define the target treatment levels (both in soil and ground water),
define the points of compliance where these goals are to be applied, and
define the time frame within which compliance must be achieved.
23.3 INSITU TREATMENT TECHNOLOGIES-GENERAL
CLASSIFICATIONS
In situ treatment technologies are generally designed to perform one, or more of the following functions:
Contaminant source zone removal
Aquifer restoration
Prevent, or minimize continued contaminant migration
Rarely does a single technology accomplish all three of these goals, and so combinations and sequences of complementary technologies are often used.
23.3.1 Source Zone Treatment Technologies
Source zone treatment technologies target removal and destruction of the residual contaminants in soil that serve as the source for groundwater contamination. The goal here is
to treat the cause of the groundwater contamination, rather than the symptom. Example
source zone treatment technologies include

Free-product recovery
Excavation and disposal or above-ground treatment
In situ soil venting
Bioventing
In situ air sparging
Enhanced in situ soil venting with soil heating and/or soil fracturing
In situ vitrification
Phytoremediation
Groundwater pump and treat systems

Fig 23.3 depicts simple process schematics of the most common of these source zone
treatment technologies.
23.3.2 Aquifer Restoration Technologies
Aquifer restoration technologies target treatment of dissolved contaminant plumes. These
technologies may be employed before, during, or after source zone treatment. Examples
of aquifer restoration technologies include

blower
water
treatment

oil-water
separator

oil
collection

compressor
air

vapor
treatment

LNAPL
Free
Product

blower

blower

vapor
treatment
source zone

sotttte gone

FIGURE 23.3. Example source zone treatment technology processes schematic: (a) free-product recovery, (b) in situ air sparging, (c)
soil vapor extraction, and (d) bioventing.

Groundwater pump and treat systems

Natural attenuation
In situ air sparging
Enhanced bioremediation

Fig. 23.4 displays simple process schematics of the first three of these aquifer restoration technologies. Restoration of groundwater quality and use of these technologies is
only feasible if a complementary source zone treatment technology is also utilized.
23.3.3 Contaminant Migration Prevention
Contaminant migration prevention technologies are designed to minimize future impacts
of contaminants on groundwater. These technologies are often employed at sites where:
a) the source zone location is not known and/or b) there are no practicable source zone
and aquifer restoration options. Examples of these technologies, or strategies, include

Natural attenuation
Groundwater pump and treat systems
In situ reaction walls
In situ air sparging curtains
Infiltration barriers
In situ containment options (grout walls, sheet piling walls, and so on)

Figure 23.5 displays simple process schematics of common contaminant migration

prevention technologies.
23A GENERIC TECHNOLOGY SELECTION AND
DESIGNPROCESS
Fig. 23.6 presents a generalized sequence of actions to be followed when selecting and
designing any in situ treatment technology. Each key step is discussed briefly below in
this section, and then discussions specific to common source zone treatment, aquifer
restoration, and contaminant migration prevention technologies follow in Sees. 23.5,
23.6, and 23.7.
23.4.1 Site Assessment and Conceptual Model Development
As stated above in Sec. 23.1.1, the design of in situ treatment systems for groundwater
contamination is challenging and unique because the subsurface is an environment that is
difficult to describe precisely; it is naturally heterogeneous and, due to practical constraints, characterization data is generally limited.
Thus, the first step of any design process involves characterizing the site and, from the
limited data available, proposing a conceptual model (or models) of how the subsurface
might look and how contaminants might move. At a minimum, the following data are
essential to the design process:

blower
discharge or
reinjection

compressor

water
treatment
dissolved
plume

monitoring wells

FIGURE 23.4. Example aquifer restoration process schematics: (a) groundwater pump and treat, (b) in situ air sparging, (c) natural attenuation.

compressor
sheet piling

low permeability cap

sheet piling
source zone

source zone

FIGURE 23.5 Example contaminant migration technology processes schematics: (a) physical barriers, (b) in situ air sparging curtain, (c)
reaction barrier.

Site Assessment and Conceptual

Model Development
I Select Target Treatment Levels

No Further Action

Remediation Necessary?

Compliance Monitoring

Identify Potential Technologies

Yes
Treatment Goals Met?

Screening Level Calculations

Refine Design

Technology Appropriate?

Operate &
Monitor

Pilot Test

Select Alternate
Technology or
Change Remedial
Goals

Initial Design

Technology Appropriate
FIGURE 23.6 Generalized technology selection and design flowchart.

A historic assessment
A geologic/hydrogeologic assessment
A contaminant distribution assessment
A receptor assessment
A political assessment

The historic assessment reviews the history of activities at the site, including chemical inventory records, plot plans (locations of structures, pipelines, rail roads, storage
areas, maintenance areas, and so on); records of known spills, leaks, and other releases, and any existing site assessment data and reports for this site and any other nearby
sites.
The geologic/hydrogeologic assessment often involves collecting soil cores, installing
groundwater wells, performing aquifer characterization tests, monitoring groundwater
elevations, and using other geotechnical tools (e.g., cone penetrometer) to characterize the
subsurface.
The contaminant distribution assessment involves conducting chemical analyses of
soil, water, groundwater, and soil gas samples, and measuring free-product levels in
monitoring wells.
The receptor assessment involves visiting the site and nearby area and reviewing water
well inventory records for the immediate vicinity. The goal is to identify persons or
resources that have been, or could be impacted by the contamination.

Finally, the political assessment involves identifying the appropriate regulatory

authorities, the potentially responsible parties, and any other persons or groups that may
be involved in the decision-making process for the site.
The collective goal of these various assessment activities is to identify the key components of the conceptual model(s):

Primary sources (storage tanks, pipelines, degreasing operations, etc.)

Time of release and amount of contaminant released
Depth to groundwater
Direction of groundwater flow and groundwater velocity
Subsurface geology (soil type, distinct layers)
Chemicals present
Contaminant distribution in the subsurface
Existence and location of any mobile free-product liquid
Potential receptors and any existing adverse impacts
Persons that will be involved in the decision makingprocess

Often the time of the release and the quantity released are unknown, unless a single
catastrophic well-documented event took place (many slow leaks from underground storage tanks go undetected for years). However, if they are known, then this information can
be used to better choose the number and locations of soil and groundwater samples. For
example, for a spill size of volume Vspill (m3 or ft3), the maximum depth dspm (m or ft) penetrated by the spill can be estimated by
4- = A^fV
^spiii ^R VT
dspin =
Vspin =
A spin =
SR
=

(23 1}

'

maximum depth penetrated by the spill (m or ft)

lume of liquid contaminant spilled (m3 or ft3)
cross-sectional area (plan view) impacted by spill (m2 or ft2)
residual saturation of contaminant liquid in soil (m3 fluid/in3 pores or ft3
fluid/ft3 pores), stet
(f>T
= total soil porosity ( 0.40 m3 pores/m3 soil or ft3 pores/ft3 soil)
Where the residual saturation SR is specific to the contaminant fluid and porous medium. SR increases with increasing surface/interfacial tension and decreasing pore size (finer
soils). Table 23.2 [American Petroleum Institute (API), 1989] provides reasonable values
for SR$T for petroleum fuels as a function of soil type. Also given are equivalent soil concentrations C50il (mgcontaminajkgsoil). While these values are specific to the petroleum mixtures and soils given in the table, they also provide reasonable estimates for many other
liquid contaminants of concern.
As an example, Fig. 23.7 presents depths of spill penetration as a function of spill size
and infiltration area, for a range of soil types.
In addition to soil concentration measurements, groundwater and soil gas samples can
also be used to help define contaminant source zone locations. A general rule of thumb is
to look for areas where groundwater or soil gas concentrations exceed 1 percent of the
maximum chemical-specific concentration expected for the contaminants released.
v

TABLE 23.2 Ranges of Residual Liquid Hydrocabon Concentrations in the Unsaturated Zone.
Residual Liquid Hydrocarbon Concentrations in Unsaturated Zone Soils (= SR$T)
Gasolines
Soil

Middle Distillates

(gal/ft ) (Um ) (mg/kg)

(gal/ft )

(Um ) (mg/kg)

Fuel Oils
3

(gal/ft )

(Um3)

(mg/kg)

Coarse
0.02 2.5
950
0.04
5
2200
0.07
10
4,800
gravel
Coarse
0.06 7.5 2800
0.1
15
6500
0.22 30 15,000
sand
Fine
sand/
0.15 20.0 7500
0.3
40
17000
0.6
80 39,000
silts
Abreviations: (gal/ft3) = gallons of liquid per ft3 of soil; L/m3] b = L3 of liquid/m3 of soil, (mg/kg) = mg of liq- 3
uid/kg of soil; (mg/kg) concentrations based on psoil = 1.85 g soil/cm soil, pliquid = 0.7, 0.8, and 0.9 g,-liquid/cm
liquid for gasoline, middle distillates, and fuel oils, respectively.
Source: API, 1989.
In the case of single-component spills, the maximum dissolved concentration Cwmax
[mg/LH20] is the pure component solubility S (mg/L H2O), while the maximum vapor concentration Cvmax (mg/Lvapor) is derived from tabulated values of vapor pressure Pv (torr),
the molecular weight Mw (g/mol), and the Ideal Gas Law. In summary, for single-component spills:
cL, (mg/LH20] = S
(Pv/160 torr I aim) X (Mw X 103 mg/g]
CL [mg/Lvapor] =
0.0821 L - atm lmol IK X 293 K

coarse fine
sands' sands silts

finer soils

FIGURE 23.7 Depth of spill penetration vs. spill size and infiltration area.

(23.2)

(23.3)

Table 23.3 contains Cy10x and CJJ0x for a range of chemicals of interest. For chemicalspecific properties of other chemicals, the reader is referred to the USEPA Superfund
Chemical Data Matrix databases (USEPA, 1996b).
For the case of mixtures, the appropriate equations for estimating the maximum dissolved and vapor concentrations of each compound are
<, i (ngfrnj

=
c

max, i (m8'LvaPor)

= X1S [mg ILHJ

(23.4)

X1 (Pv/760torr I atm) X (Mw X 1Q3 mg I g)

0.0821L - atmlmol IK X 293 K

>

TABLE 23.3 Chemical Properties of Selected Soil and Groundwater Contaminants of Interest
(derived from USEPA SCDM 1996b)

Chemical
Acetone
Aldrin
Anthracene
Atrazine
Benz(a)anthracene
Benzene
Benzo(a)pyrene
Benzo(b)fluoranthene
1,3 -Butadiene
Butanol
Carbon tetrachloride
Chlordane
Chlorobenzene
Chloroform
Chloromethane
Chrysene
m-Cresol
Cyclohexane
DDD
DDE
DDT
1,3 Dichlorobenzene
1 ,2-Dichloroethane
trans 1,2-Dichloroethylene
1 ,2-Dichloropropane
Dieldrin

Molecular
Weight
(g/mole)
58
365
178
216
238
78
252
252
54
74
154
410
113
119
51
228
108
84
320
318
354
147
99
97
113
381

Pure
Component
Solubility
(mg/LH20)

Vapor
Pressure
(torr)

106
0.017
0.043
70
0.009
1800
0.002
0.002
740
106
790
0.056
470
7900
5300
0.006
23,000
55
0.09
0.12
0.025
130
8500
6300
2800
0.2

23
730
6 X 10-6
0.0001
3 X 10-6
0.00003
0.000004
3 X 10-7
1 X 10-7
0.000001
9
405
6 X 10-9
0.00000008
5 X 10-7
0.000007
6200
2100
74,000
300,000
1010
120
5 X 10-5
0.001
12
14
200
1300
4300
12,000
9.5 X 10-5
0.001
0.14
0.83
97
450
7 X 10-7
0.00001
6 X 10-6
0.0001
2 X 10-7
0.000004
2.2
18
79
430
1800
330
52
320
5 X 10-6
0.0001

Maximun
Vapor
Cone.
(m/Lvapor)

TABLE 23.3 (Continued)

Chemical

Molecular
Weight
(g/mole)

Endrin
Ethylbenzene
Ethylene glycol
Formaldehyde
Heptachlor
Hexane
Isobutanol
Lindane
Methanol
Methylethylketone
Napthalene
PCBs
Phenol
Styrene
Tetrachloroethylene
Toluene
Toxaphene
Trichlorbenzene
1 , 1 ,2-Trichlorethane
Trichloroethylene
Vinyl chloride
m-xylene
Gasoline (fresh)*
Gasoline (weathered) *

381
106
62
30
373
86
74
291
32
72
128
356
94
104
166
92
413
181
133
131
63
106
95
111

Pure
Component
Solubility
(mg/LH20)
0.2
170
106
550,000
0.18
12
85,000
7.3
106
220,000
31
0.07
83,000
310
200
530
0.74
49
4400
1500
8800
160

Vapor
Pressure
(Torr)

Maximun
Vapor
Cone.
(m8/Lvapor)

3 X 10-6
9.6
0.092
5200
0.0004
152
10
0.0004
130
95
0.085
7.7 X 10-5
0.28
6.1
19
28
9.8 X 10-7
0.43
23
73
3000
8.5

0.00006
0.056
0.31
8500
0.008
720
40
0.006
230
370
0.60
0.0015
1.4
35
170
141
0.00002
4.3
170
523
10,000
49
1300
220

Source: Johnson, et al., (199Oa, b).

where ^(moles-i/total moles in mixture) is the mole fraction of the chemical of interest
in the mixture. In many cases, when the complete composition of the mixture is unknown,
the mole fraction can be approximated by the mass fraction W1 (mass-//mass of mixture):
X1 (moles Utotal moles) W1 (mass iltotal mass)

(23.6)

For example, if a spill of benzene liquid occurred, we would expect to see dissolved
and soil gas vapor benzene concentrations approaching 1800 wg/LH20 and 400 mg/Lvapor,
respectively, near the source zone. If, on the other hand, a gasoline spill occurred, then
we would expect to see dissolved and soil gas vapor benzene concentrations approaching 18 mglLmo and 4 mglLvapor, respectively, for a gasoline containing 1 percent by
weight benzene.
For the site assessment, how much data collection is enough? Clearly, it is desirable to
be able to completely characterize a site, but this option is cost-prohibitive. There are no
universal guidelines that define minimum numbers of soil borings, samples, and so on;
however, some useful questions to consider are

 Have you adequately defined primary sources, depth to groundwater, direction of

groundwater flow, groundwater velocity, subsurface geology (soil type, distinct layers), chemicals present, contaminant distribution in the subsurface, existence and location of any mobile free-product liquid, potential receptors and any existing adverse
impacts?
Is it likely that the additional data will change the proposed site conceptual model?
Are additional data critical to the prescreening of treatment technologies?
What are the risks of designing a system based on the current data?
One other important factor to consider is that some treatment technologies are more
robust than others. Here "robustness" is a measure of how easily an initial system design
can be refined once operation begins. It is also a measure of how sensitive treatment effectiveness is to small variations in subsurface conditions and contaminant distributions. In
other words, a very robust technology is one that can be designed with little site-specific
information, and that can be made to perform as desired through minor operating condition changes rather than major system equipment changes. Table 23.4 provides a qualitative indication of the robustness of many treatment technologies in use today.
23.4.2 Select Target Treatment Levels
At this step, appropriate target treatment levels are identified for the chemicals that have
been identified at the site. The discussion in Sec. 23.2 outlines the range of target treatment goals that might be encountered.
Briefly, at this stage the user must
define the target treatment levels (both in soil and groundwater),
define the points of compliance where these goals are to be applied, and
define the time frame within which compliance must be achieved.
After this step is completed, site concentrations are compared with the target treatment
values. Any exceedences will trigger further assessment, and treatment may be required.
23.4.3

Identify Potential Technologies

If treatment is necessary, a gross pre-screening of technologies takes place. This prescreening considers the site conditions, documented performance of the available technologies at similar sites, the state of understanding of the technology, economic factors,
and the treatment goals (target levels and target time frame). Table 23.4 contains information that can be used to provide a rough screen of technologies.
At the end of this step the user is generally considering one to three treatment options.
23.4.4 Screening Level Calculations
At this stage a more refined screening of technologies occurs based on basic fundamental
screening calculations. These calculations are generally biased toward predicting optimal
system performance. Typical calculations involve estimating maximum treatment rates.
Examples of these screening level calculations are presented in Sees. 23.5 and 23.7.

TABLE 23.4 Example Technology Prescreening Matrix.

Technology

Application

Relative Time Frame

Robustness

Typical Settings

Excavation and
above-ground
treatment
Groundwater
pump and treat

Source treatment

Short

Containment

Long

Bioventing

Source treatment

Medium

Soil vapor
extraction

Source treatment

Medium

Physical barriers

Containment

Long

Natural
attenuation

Dissolved plume
management

Long

Petroleum
release sites

Free-product
recovery
systems

Source zone
treatment

Medium

Sites with
mobile and
pumpable freeproduct liquid

In situ air
sparging

Source treatment

Medium

Petroleum fuels
and solvents
trapped below
the water table
in permeable soils

Chemical
migration
/reaction
barriers

Dissolved plume
containment

Long

Chlorinated
solvent spill sites

Shallow
contamination
(<20 ft BGS)
All, except very
low permeability
settings
Petroleum
release sites and
permeable soils
Petroleum fuel
and solvent
release sites and
permeable soils
Mixed waste
sites involving
very hazardous
chemicals, low
permeability
settings

23.4.5

Decision PointIs the Technology Appropriate?

The user then compares the output of the screening level calculations with treatment
expectations. If, even under optimal conditions, the performance predicted by the
screening-level calculations does not meet treatment expectations, then the user has
two options:
eliminate that technology as an option, or
revise treatment expectations (target levels or treatment time) and retain the technology.
23.4.6 Pilot Testing
In situ pilot tests are usually conducted for the most promising treatment technologies.
Often these are designed to test critical conceptual model assumptions, such as achievable
flow rates, maximum effluent concentrations, and so on. Convention and economic constraints dictate that most pilot tests will be less than 1 week in duration.
Key guidelines for planning and conducting pilot tests include
Design the pilot test so that the equipment used is easily integrated into any full-scale
system that might be installed
As it is very unlikely that the short-term performance of a pilot test can be accurately
extrapolated to long-term full-scale system performance, pilot tests should be designed
to look for infeasibility. In other words, the pilot test should be designed to identify
behavior or fatal flaws that suggest that the technology would not work at the site
(e.g., maximum achievable flow is too low, geology prohibits access to the contaminants, and so on).
23.4.7 Decision PointIs the Technology Appropriate?
The user then compares the pilot test results with treatment expectations. If the performance does not meet treatment expectations, then the user has two options:
eliminate that technology as an option or,
revise treatment expectations (target levels or treatment time) and retain the technology
23.4.8 Initial Design
Based on all available data, the site conceptual model is revised, and an initial system
design is created and installed. The design process often relies on a combination of screening level models, heuristics, and sometimes more sophisticated computational design
tools. It is important at this stage to
Design a robust system that can succeed under a range of actual site conditions
Design a system that can easily be modified and expanded
Design the system to be easily monitored

23.4.9 Operation and Monitoring

Once the system is installed and operated, adequate monitoring should be conducted to
assess if the system is meeting the treatment goals and regulatory objectives, and if there
are modifications that can be made to improve system performance.
23.4.10

Design Refinement

As stated above, given the inherent large initial uncertainty, the design phase continues
well past the construction, installation, and operation of the initial system. It should be
recognized that following sound design practices does not guarantee success; it does,
however, provide a higher probability of success. For this reason, it is likely that refinements will be made to the original system design based on the performance monitoring
data. These refinements may include changes in operating conditions (e.g., flow rates) or
the addition of new hardware (e.g., additional wells).
23.4.11 Decision Point-Have the Treatment Goals Been Met?
To determine when it is appropriate to terminate the treatment process, the following
questions must be addressed:
Has the treatment system achieved the treatment goals?
If the system is turned off, will treatment goals continue to be met?
If the treatment goals have not yet been met, is there another technology that can more
effectively be used to progress toward the treatment goals?
If the answer to any of these questions is yes, then operation of the current treatment
system should be terminated. If treatment goals have yet to be met, it is time to transition
to the next technology in the treatment train. For example, at petroleum fuel release sites,
a typical sequence would involve free-product liquid recovery soil vapor
extraction/bioventing > natural attenuation.
23.5

SOURCEZONETREATMENT

In this section, general design principles are presented for common source zone treatment
technologies, including free-product recovery, pump and treat, soil vapor extraction,
bio venting and in situ air sparging.
23.5.1 Free-Product Recovery
Most regulatory programs require removal of any mobile and pumpable immiscible freeproduct liquid that can be collected in monitoring wells. In the case of LNAPLs, free
product is found floating on top of groundwater in a monitoring well; in the case of
DNAPLs, free=product liquid is found at the bottom of the well. This of course assumes
that the well is screened across the interval containing the free=product liquid. The following discussion is specific to LNAPL sites, as our understanding of these sites greatly
exceeds our knowledge of free-product recovery at DNAPL sites.
Next Page

CHAPTER 24
ARTIFICIAL RECHARGE
OF GROUNDWATER:
SYSTEMS, DESIGN,
AND MANAGEMENT

Herman Bouwer
USDA-ARS Water Conservation Laboratory
Phoenix, Arizona.

24.1

ABSTRACT

The type of system to be selected for artificial recharge of groundwater and how it should
be designed and managed for optimum performance depend entirely on local conditions
of soil, hydrogeology, topography, water availability (quality, continuous, or interrupted
supply), and climate. Unlike water or wastewater treatment plants, guidelines, standards,
and "blueprints" for artificial recharge systems cannot be given. Rather, the responsible
planner should have extensive knowledge of the various types of recharge systems that
can be used, how they work, the processes, cause and effect relations, what experiences
have been obtained elsewhere, and how they should be adapted to local conditions. Once
the type system has been selected, a protocol of site investigations should be followed for
preliminary design. Key factors in the design and management of successful artificial
recharge systems are site and system selection, maintenance of adequate infiltration rates,
hydraulic continuity between the recharge system and the aquifer (no clay layers in the
vadose zone), and groundwater control for effective water recovery and prevention of
undue groundwater rises in the recharge area. Another important factor in the selection of
the type of recharge system is the pretreatment of the water required before recharge to
minimize physical, chemical, or biological clogging of the infiltrating surface (bottom in
basins and walls in trenches, shafts, and wells). Since clogging layers are easier to control and remove in surface infiltration systems, proper pretreatment is especially important for trenches, shafts, and wells. Since trenches and shafts are relatively inexpensive,
they can be replaced when their useful life has come to an end. On the other hand,
recharge wells, while much more expensive than trenches and shafts, are more amenable
to clogging prevention and control by frequent pumping and periodic redevelopment. All
these aspects have to be factored in, however, when selecting the type system that gives
recharge at minimum cost per unit volume of water put into the aquifer. While presedimentation in settling ponds may be adequate pretreatment for surface infiltration systems,
recharge with trenches, shafts, and wells may require coagulation and sand filtration to
remove suspended solids to acceptable levels. Where sewage effluent is used for groundwater recharge, conventional primary and secondary treatment may be sufficient for surface infiltration systems. However, recharge with effluent through wells may require

advanced treatment processes including nitrogen and phosphorus removal, sand filtration
for removal of suspended solids and parasitic organisms like helminths, Giardia, and
Cryptosporidium, disinfection for removal of viruses and bacteria, and reverse osmosis
(RO) for removal of organic compounds and other chemicals.
If the site investigations indicate no fatal flaws, a small pilot or test project should be
installed to gain a better understanding of how the system behaves and how the full-scale
system should be designed and managed for best performance. Even then, a phased construction should be used for the full-scale system, so that operating experience is obtained
and refinements can be made as the rest of the system is designed and constructed. This
is especially true for "new" areas where artificial recharge has not yet been practiced.
Where local recharge experience already is available, additional systems then are essentially expansions of existing systems and the normal protocol of preinvestigations and
pilot testing can be reduced. The objective of this chapter is to provide information on the
types of recharge systems, how they work, and how they should be designed and managed, so that planners, designers, and managers can make best available decisions.

24.2

INTRODUCTION

Artificial recharge of groundwater is accomplished by applying water to the soil for infiltration and downward movement through the unsaturated or "vadose" zone to the groundwater. Such "surface systems" (Fig. 24.1A) require soils that are sufficiently permeable,
vadose zones that have no clay or other restricting layers, and aquifers that are unconfined. Where permeable surface soils are not available but sufficiently permeable material is found at relatively small depth (about 1 m, for example), artificial recharge can be
achieved with excavated basins that are sufficiently deep to reach the permeable material
(Fig. 24.1B). If the permeable material is too deep for removal of overlying material, but
is within trenchable depth (less than about 7 m, for example) seepage trenches can be used
(Fig. 24.1C). Such trenches are also suitable where soils are highly stratified with alternating layers of fine and coarse materials. Where permeable material is too deep for
trenches, large-diameter wells, pits, or shafts in the vadose zone can be used (Fig. 24.1D).
Such shafts can be drilled with bucket augers to a depth of about 50 m with a diameter of
about 1 m. Both trenches and shafts are backfilled with coarse sand or fine gravel to prevent caving. Where permeable surface soils are not available, vadose zones are not sufficiently permeable to transmit water to the underlying aquifer, or aquifers are confined,
artificial recharge can be achieved by applying water to recharge wells penetrating the
aquifer (Fig. 24.1E).

24.3

SYSTEMS

The first step in planning and designing an artificial recharge system is selecting the type
of system to be used, based on soil and hydrogeologic information. Often, the choice is
obvious or determined by other factors such as high cost of land in urban areas which preclude the use of surface systems or excavated basins. Another important factor in the
selection of the type of recharge system is the pretreatment of the water required before
recharge to minimize physical, chemical, or biological clogging of the infiltrating soil surface in basins or of the walls in trenches, shafts, and wells. Since clogging layers are easier to control and remove in surface systems, proper pretreatment is especially important
for trenches, shafts, and wells. Since trenches and shafts are relatively inexpensive, they

FIGURE 24.1 Recharge systems for increasingly deep permeable materials: surface basin (A), excavated basin (B),trench (C), shaft or vadose zone well (D),
and aquifer well (E).

can be replaced when their useful life has come to an end. On the other hand, recharge
wells, while much more expensive than trenches and shafts, are more amenable to clogging prevention and control by frequent pumping and periodic redevelopment. All these
aspects have to be factored in, however, when selecting the type of recharge system.
While presedimentation in settling ponds may be adequate pretreatment for surface infiltration systems, recharge with trenches, shafts, and wells may require coagulation and
sand filtration to remove suspended solids to low enough levels. Where sewage effluent
is used for groundwater recharge, conventional primary and secondary treatment may be
sufficient for surface infiltration systems. However, effluent recharge with wells may
require advanced treatment processes (nitrogen and phosphorus removal, sand filtration,
removal of suspended solids and parasitic organisms like helminths, Giardia, and
Cryptosporidium, disinfection for removal of viruses and bacteria, and RO for removal of
organic compounds and other chemicals).
While conjunctive use of surface and groundwater often is the main objective of artificial recharge, recharge is also used for other purposes such as creating hydraulic barriers
against seawater intrusion, stopping land subsidence, stopping spread of contaminated
groundwater (plumes) in the aquifer from point sources of pollution, raising groundwater
levels for reduction of pumping costs, protecting streams, wetlands or other ecosystems,
protecting wooden building foundations that would decay when above groundwater, etc.
Dams for surface storage of water have a number of disadvantages, such as evaporation
losses (about 2 m/year in warm, dry climates), sediment accumulation, potential of structural failure, and increased malaria, schistosomiasis, and other diseases in the local population (Devine, 1995; Knoppers and van Hulst, 1995; Pearce, 1992). New dams often are
more and more difficult to build because of poor economics and public opposition. They
interfere with the river ecology and may flood sensitive areas (cultural, religious, archeological, environmental, and recreational, scenic, and so on.). People living on the reservoir
site have to relocate and good dam sites are becoming scarcer. Dams are not sustainable
because eventually they all silt up, and they are not effective for long-term storage of water
(years or decades), which may become increasingly necessary as increases of carbon dioxide and other greenhouse gases in the atmosphere may cause global climatic changes and
increase the probability of extremes in weather such as more prolonged droughts and periods of excessive rain. This increases the need for storing excess water in wet periods to
meet water demands in dry periods. Underground storage via artificial recharge then is preferred because of essentially zero evaporation, economics, and other factors. The water
sources for artificial recharge include water from perennial or intermittent streams that may
or may not be regulated with dams, storm runoff (including from urban areas), aqueducts
or other water conveyance facilities, irrigation districts, drinking water treatment plants,
and sewage treatment plants. Of all the water in the world, 97 percent is salt water in the
oceans (Bouwer, 1978). Of the remaining fresh water, two-thirds is in the form of ice in arctic and mountainous regions. Of the remaining liquid fresh water, less than 2 percent is surface water in streams and lakes, and much of that is fed by groundwater. More than 98 percent of the world's liquid fresh water thus is groundwater. Not only is groundwater the dominant water resource, but the potential for underground storage of water also vastly exceeds
that for surface storage.
24.3.1 Surface Systems
Surface infiltration systems for artificial recharge of underlying unconfined aquifers can
be constructed in streambeds (in-channel systems) or outside streambeds (off-channel systems). In-channel systems (Fig. 24.2) typically consist of weirs or dams across the

FIGURE 24.2 Schematic of in-channel infiltration systems consisting of low weirs in narrow steep channel (upper left), bigger dams in wider, milder sloping channel (upper right),
and T-levees in wide, flat channels (bottom).

streambed to back the water up and spread it over the entire width of the streambed to
increase the water depth and the wetted area and, hence, the infiltration of water. The
weirs or dams are small and closely spaced where channel slopes are steep, and larger and
more widely spaced where slopes are flat. The groundwater table for these systems must
be well below the streambeds so that the streams are "losing." The weirs or dams are made
of sheet metal (low weirs), earth, concrete, or inflatable rubber (fabridam). The dams
should have adequate spillway capacity or structural washout sections to handle large
flows. Such flows can clean the channel from fine sediment that may have accumulated
above the dam. Otherwise, the channel periodically must be mechanically cleaned to
remove accumulated sediment. Such sediment can be a "layercake" of coarse and fine
material deposited during high and low flows, respectively. This sediment greatly restricts
infiltration of water and storage of water behind the dams when flood waters recede and,
hence, must be removed. In coastal plains or other areas of mild channel slopes, dikes or
berms about 1 m high are pushed up by bulldozers to make T- or L-dikes in the streambed
that force the water into circuitous paths that cover the entire streambed (Fig. 24.2) and,
hence, increase the infiltration and groundwater recharge. Periodic high flows can destroy
the dikes but also clean the bottom from fine sediment and other clogging material. For a
system in southern California, this happens at least once a year. When the high flows have
receded, bulldozers restore the dikes in a few days.
In-channel systems are most effective if the stream is dammed higher up in the watershed; for example, in the mountains. This removes sediment from the water and also regulates the flow by storing water during high-flow events and releasing it slowly for
recharge. Storage dams themselves are, of course, ineffective for groundwater recharge
because of the sediment that accumulates on the bottom of the reservoirs.
Off-channel systems consist of basins constructed by building berms from soil that is
normally excavated from the basin areas. They can be square or rectangular, or of irregular form to adapt to existing topographies or for environmental enhancement. The latter
may include nature and wildlife benefits, scenic enhancement (landscaping, and the like),
trails for walking and bicycle riding, recreational facilities (picnic tables, playgrounds),
and so on. Each basin should have its own water supply. Deeper basins may also need a
drainage facility so that water can be drained out for drying and maintenance where dry-

ing by natural infiltration would take too long. Each basin is then hydraulically independent and can be operated according to its best schedule of flooding, drying, and cleaning.
The inflow facility should be carefully designed to dissipate the energy of the inflowing
water to avoid soil erosion in the basin. In finer textured soils such erosion will cause the
water to become "muddy" which will lead to clogging as the fine particles settle out in the
rest of the basin. Basin banks also need to be protected against erosion by rainfall and
wave action.
Occasionally, agricultural irrigation-type systems are also used for recharge, such as
borders, furrows, and terraces. On very permeable soil in irregular topographies like
kames, eskers, and other glacial deposits, low-pressure perforated pipe (hole diameter
about 5 mm) can be used to apply the water so as to avoid construction of basins and
destruction of trees or other natural vegetation.
24.3.2 Trenches, Shafts, and Wells.
Trenches (Fig. 24.3) are constructed with backhoes or other trenching equipment.
Vadose zone shafts (Fig. 24.3) are drilled with bucket augers. Trenches and shafts are
backfilled with fine gravel or coarse sand. For trenches, water is supplied with a perforated pipe on top of the backfill and the trench can be covered to avoid exposure to
sunlight and public access. For shafts, water is supplied with a vertical perforated pipe
or well screen in the center of the backfill. Recharge wells in aquifers are drilled with
conventional well drilling techniques and are constructed much like pumped wells,
including sand or gravel envelopes in unconsolidated materials. Careful grouting is
necessary where large injection pressures are used as, for example, in fractured rock
aquifers. Injection water flows into the well with one or more pipes inside the well.
These pipes should terminate some distance below the water level in the well to avoid
free falling water that can cause air entrainment and subsequent air blocking of pores
in the aquifer. For the same reason, the pipe or pipes themselves also should have full
flow. This can be achieved by using several pipes with different diameters so that the
small flows can go through the smaller pipe to maintain full-pipe flow. Alternatively,
one pipe can be used if it has an exit valve at the bottom that can be adjusted to always

FIGURE 24.3 Vadose zone recharge well with sand or gravel fill and perforated supply pipe
(left) and recharge trench with sand or gravel fill, supply pipe, and cover (right).

create full-pipe flow. Special valves are available for this purpose. Recharge wells
should be equipped with permanent, dedicated pumps to allow periodic pumping of the
well for clogging control.

24.4

INFILTRATION

24.4.1 Infiltration Rates

Infiltration rates are expressed as volume of water moving into the soil or aquifer per unit
infiltration area and per unit time. The dimension of infiltration rate thus is length/time,
for example, cm/h or in/h, or m/day or ft/day. For infiltration basins, this rate can be visualized as the rate of fall of the water surface in the basin if there is no inflow to or outflow
from the basin. Infiltration rates for surface infiltration recharge systems typically range
from 0.3 to 3 m/day (1-10 ft/day), with most systems in the 0.5-1.5 m/day (2-5 ft/day)
range. For systems that are operated year-round, long-term infiltration rates that include
the time that basins are periodically dried and cleaned are called hydraulic loading rates.
These rates may vary from 30 m/year to 500 m/year, depending on soil type, water quality and climate. Since evaporation rates from free water surfaces and wet soils vary from
0.5 m/year or less in cool, humid climates to 2 m/year or more in hot dry climates, water
losses by evaporation normally are negligible.
Hydraulic loading rates must be known to determine how much water can be put
underground per year for a given recharge area. If, on the other hand, the recharge system must be designed to accept a certain flow rate throughout the year, as, for example,
where the entire flow from sewage treatment plants must go underground for recharge
and soil-aquifer treatment (SAT) and there is no other disposal or temporary surface storage, the land area must be based on minimum infiltration rates. These usually occur during the winter when the water temperature is low and, hence, the viscosity of the water
is high which gives proportionally lower infiltration rates (Bouwer, 1978). Also, drying
is slower when the weather is cold so that basins need to stay dry longer to achieve infiltration recovery. Drying also can be delayed by more rainfall during the drying period
which also can produce crusting effects on the soil.
Predicting and managing infiltration rates are the most important aspects in planning,
designing, and managing of recharge systems, because they determine how much land
area is needed to put a certain flow of water underground or how much groundwater
recharge can be achieved with a certain amount of available land and how much water will
be lost by evaporation. For surface infiltration systems in uniform soils without surface
clogging, infiltration rates will be about equal to the vertical hydraulic conductivity of the
soil (Bouwer, 1978), which has the following approximate orders of magnitude:
Clay soils
<0.1 m/day
Loams
0.2 m/day
Sandy loams
0.3 m/day
Loamy sands
0.5 m/day
Fine sands
1 m /day
Medium sands
5 m /day
Coarse sands
>10 m/day
These values can be used for preliminary site evaluation and system selection.

24.4.2 Cylinder lnfiltrometers

After promising soils have been identified by studying soil maps and doing on-site soil
inspections, "wet" infiltration tests should be performed. These are typically done with
metal, cylinder infiltrometers about one foot in diameter. However, use of such small infiltrometers can seriously overestimate the large-area infiltration rates because of lateral
flow (divergence) around the cylinder due to capillary suction in the soil (Bouwer, 1986
and 1995; Bouwer et al. 1999b). Double-ring or "buffered" infiltrometers are not the solution because the divergence also causes overestimation of infiltration in the center portion
of the cylinder. The obvious approach then is to use larger infiltration test areas like, for
example, 10 X 10 ft bermed areas, where divergence effects are less significant. These
tests are laborious and they can also require large volumes of water because it may take
more than a day to reach or approach "final" infiltration rates. Another approach would
be to use conventional single cylinders with significant water depth to speed up the infiltration process so that tests can be completed in a relatively short time (5 h, for example).
The resulting data are then corrected for water depth, limited depth of wetting, and divergence effects to get an estimate of the long-term infiltration rate for a large inundated area.
This rate should be about equal to the hydraulic conductivity K of the wetted zone.
The infiltrometers used for this procedure are single steel cylinders 24 in in diameter
and 12 in high with beveled edge (Figure 24.4). A piece of (2 in X 4 in) lumber is placed
on top of the cylinder and the cylinder is driven straight down with a sledgehammer to a
depth of about 1 in to 2 in into the ground. The soil is packed against the inside and outside of the cylinder with a piece of 1 in X 2 in furring strip that is held at an angle on the
soil against the cylinder and tapped with a light hammer to get good soil-cylinder contact.
If the soil contains clay, the water used for the test should be of the same chemical composition as the water used in the recharge project to avoid complications resulting from
effects of water on status of clay (coagulated or dispersed; Bouwer, 1978). A plate or pie
tin is placed on the soil for erosion prevention. The cylinder is filled to the top, and clock
time is recorded. Water is allowed to drop about 2 in to 5 in, the drop is measured with a

WETTED

ZONE

FIGURE 24.4 Geometry and symbols for single-ring infiltrometer.

yardstick, clock time is recorded, and the cylinder is filled back to the top. This is repeated for about 6 hours or when the accumulated infiltration has reached about 20 in,
whichever comes first. The last drop yn is measured and clock time is recorded to obtain
the time increment Arn for yn. A shovel is used to dig outside the cylinder to determine
the distance x of lateral wetting or divergence (Fig. 24.4). The infiltration rate In inside
the cylinder during the last water level drop is calculated as yjtn. The corresponding
downward flow rate or flux iw in the wetted area below a cylinder or radius r is then calculated as
'.-^
The depth L of the wet front at the end of the test is calculated from the accumulated
drop yt of the water level in the cylinder as
-^s?
where n is the tillable porosity of the soil. The value of n is estimated from soil texture and
initial water content. For example, n may be about 0.3 for dry uniform soils, 0.2 for moderately moist soils, and 0.1 for relatively wet soils. Well-graded soils would have lower values of n than uniform soils. The value of L can also be determined by digging down with a
shovel right after the test and seeing how deep the soil has been wetted. This works best if
the soil initially is fairly dry, there is good contrast between wet and dry soil, and there are
no rocks. Applying Darcy's law (see section 4.2.1) to the downward flow in the wetted zone
then yields
( . = K^L-HJ
(243)

where z is the average depth of water in the cylinder during the last water level drop. The
term hwe is the water-entry value of the soil and it is used to estimate the suction at the wet
front as it moves downward. The water-entry value is about half the air-entry value (due
to hysteresis) and may be estimated as follows (in inches of water):
Coarse sands
Medium sands
Fine sands
Loamy sands-sandy loams
Loams
Structured clays (aggregated)
Dispersed clays

2
4
6
-10
14
14
-40

Since K is now the only unknown in Eq. (24.3), it can be solved for K as
i*L
)
(24.4)
(z + L-hwe
The calculated value of K can be used as an estimate of long-term infiltration rates in
large and shallow inundated areas, without clogging of the surface and without restricting
layers deeper down. Because of entrapped air, K of the wetted zone is less than K at saturation, for example, about 0.5 of K at saturation for sandy soils, and about 0.25 of K at
saturation for finer soils. If the lvalues calculated with Eq. (24.4) look promising for an
infiltration system, the next step is to put in some test basins of about 1 acre for long-term
K=

flooding to evaluate clogging effects and potential for infiltration reduction by restricting
layers deeper down. Good agreement has been obtained between predicted infiltration
rates (K in Eq. 24.4) and those of larger basins. Six infiltrometers were installed in a field
west of Phoenix, Arizona. They gave an average K of 40 cm/day. Two test basins of 0.75
acre each in the same field yielded final infiltration rates of 30 and 35 cm/day (Bouwer
et al, 1997), which is a good agreement. If the infiltrometer tests give infiltration rates
that are too low for surface infiltration systems, alternate systems such as excavated
basins, recharge trenches, recharge shafts or vadose zone wells, or aquifer wells can be
considered.
The above procedure is by no means exact. However, in view of spatial variability
(vertical as well as horizontal) of soil properties, exact procedures and measuring water
level drops with vernier equipped hook gages are not necessary. The main idea is to
account somehow for divergence and limited depth of wetting, rather than applying a flat
reduction percentage to go from cylinder infiltration rates to long-term large area infiltration rates. Because of spatial variability, cylinder infiltration tests should be carried out at
various locations within a given site. Finally, the resulting infiltration rates should never
be expressed in more than two significant figures!
Hypothetical Example: A cylinder 24 in in diameter and 12 in high was driven lin into
a relatively dry sandy loam soil. The following time and drop-of-water level data were
recorded:
0800
0830
0900
1000
1100

Filled with water to top

Water dropped 4 in, cylinder refilled to top
Water dropped 3 in, cylinder refilled to top
water dropped 5 in, cylinder refilled to top
water dropped 4 in, end of test

Digging with a shovel showed a lateral wetting of 12 in outside the cylinder, n was
taken as 0.2 and hwe as - 10 in. The value of in is 4/1 = 4 in/h which when substituted
into Eq. (24.1) yields iw = 4 n 122In (12 + 12)2 = 1 in/h. The value of yt is 16 in, which
when substituted into Eq. (24.2) yields L = I6nl22/Q.2n (12 + 12)2 = 20 in. Since the
average water depth in the cylinder during the last measured water level drop was 11
4/2 = 9 in, /Hs calculated with Eq. (24.4) as 1 X 20/(9 + 20 + 10) = 0.5 in/h. Thus, the
last measured cylinder infiltration rate of 4 in/h is reduced to 0.5 in/h to eliminate effects
of divergence, limited depth of wetting, and water depth in the cylinder to produce a value
that can be used to predict potential infiltration rates for extended inundation of large
areas.
24.4.3

Clogging

The main enemy of infiltration systems for artificial recharge is clogging of the infiltrating surface (bottoms of basins or other surface infiltration systems, and walls of trenches,
shafts, and wells). Clogging is caused by inorganic (clay, silt) and organic (algae, sludge)
suspended solids in the water that accumulate on the infiltrating surface, and by microorganisms that grow on the soil particles (biofilms) and produce polysaccharides and other
insoluble metabolites to form a soil-clogging biomass. Bacteria also can produce gases
(nitrogen, methane, carbon dioxide) that can block soil pores. Gas also can be formed in
aquifers when water from recharge wells contains entrained air or is cooler than the

groundwater itself. As the recharge water then warms up in the aquifer, dissolved air may
go out of solution to form pore blocking air pockets in the aquifer (called air binding).
Also, well recharge can lead to precipitation of iron and manganese oxides or hydroxides
as dissolved oxygen levels change, and to solution or precipitation of calcium carbonates
due to changes in pH and carbon dioxide levels.
Since clogging layers are much less permeable than the natural soil material, they
reduce infiltration rates and become the controlling factor or "bottleneck" in the infiltration process (Fig. 24.5). When the infiltration rate in surface systems has become
less than the hydraulic conductivity of the soil below the clogging layer, this soil
becomes unsaturated to a water content whereby the corresponding unsaturated
hydraulic conductivity becomes numerically equal to the infiltration rate (Bouwer,
1982). The resulting unsaturated downward flow then is entirely due to gravity with a
hydraulic gradient of 1. The thickness of clogging layers may range from 1 mm or less
(biofilms, thin clay, and silt layers or "blankets") to several cm or more for thick sediment deposits.
Clogging is best controlled by prevention; that is, by removing the parameters that
cause clogging. For surface water, this typically means presedimentation to settle clay,
silt, and other suspended solids. This can be accomplished by dams in the river or aqueduct system (which would also regulate the flow) or by passing the water through dedicated presedimentation basins prior to recharge. Coagulants like alum and organic polymers can be used to accelerate settling. For well recharge, filtration may also be necessary. Algae growth and other biological clogging in basins can be reduced by removing
nutrients (nitrogen and phosphorus) from the water. This is also important where trenches, shafts, or wells are used for recharge with sewage effluent or effluent contaminated
water. Organic carbon levels then must also be reduced, using activated carbon filtration
and/or possibly reverse osmosis or other membrane filtration. Disinfection with chlorine
or other disinfectants with residual effects reduces biological activity on and near the
walls of the trenches, shafts, or wells and, hence, clogging. Clogging rates increase with
increasing infiltration rates, because of the increased loading rates of suspended solids,
nutrients, and organic carbon on the surface. Because of this, increasing the injection pres-

CLOGGING LAYER

UNSATURATED

CAPILLARY FRINGE
WATER TABLE
AQUIFER
FIGURE 24.5 Schematic of infiltration basin with clogging layer, unsaturated flow to aquifer, and capillary fringe above water table.

sures in recharge wells that show signs of clogging can actually hasten the clogging
process. Regular pumping of recharge wells and periodic redevelopment of the wells can
control and delay clogging, but probably not "forever." Increasing the water depth in
recharge basins or the injection pressure in recharge wells can compress the clogging layer
which reduces its permeability and, hence, the infiltration rates (see Sec. 24.4.4).
For surface infiltration systems, clogging is controlled by periodically drying the
basins or other infiltration facility, and letting the clogging layer dry, decompose, shrink,
crack, and curl up. This may be sufficient to restore infiltration rates to satisfactory values. If clogging materials continue to accumulate, they must be periodically removed at
the end of a drying period. This can be done mechanically with scrapers, front-end loaders, graders, or manually with rakes. After removal of the clogging material, the soil
should be disked or harrowed to break up any crusting that may have developed at or near
the surface. Disking or plowing clogging layers as such into the soil gives short-term
relief, but eventually fines and other clogging materials will accumulate in the topsoil and
the entire disk or plow layer must be removed. For good quality surface water with very
low suspended solids contents, drying and cleaning may be necessary once a year or
maybe every few years. Where soils are relatively fine textured or have many stones, clogging control becomes a major challenge (see Sec. 24.6). Where the water is extremely
muddy or where inadequately treated sewage effluent is used, drying and cleaning may be
needed after each flooding period which may then be as short as 1 or 2 days.
24.4.4 Effect of Water Depth on Infiltration
If there is no clogging layer on the bottom, the groundwater is relatively deep, and the
vadose zone is uniform, the downward flow in the vadose zone is governed by gravity and
water depth in recharge basins only has a minor effect on infiltration (Bouwer and Rice,
1989). For example, if the general groundwater table away from the infiltration system is
at a depth of 10 m, increasing the water depth in the basin from 0.1 to 1.1 m (about 10fold) would increase the infiltration rate only by about 10 percent. If, on the other hand,
infiltration is controlled by a clogging layer on the bottom and the underlying material is
unsaturated, the infiltration rate theoretically increases almost linearly with water depth,
so that doubling the water depth would double the infiltration rate (Bouwer, 1982). In reality, however, this seldom occurs because clogging layers often are loose deposits that are
quite compressible. Since increasing water depths produce increasing intergranular pressures in the clogging layer, this layer becomes compressed as the water depth in the basin
increases and, hence, becomes denser and less permeable (Bouwer and Rice, 1989). Thus,
in practice, increasing the water depth in an infiltration basin with a clogging layer will
cause a less than linear increase and maybe even a decrease (if the clogging layer is very
compressible) in infiltration rate. If the infiltration rate does not increase linearly with
water depth, increasing the water depth will then reduce the rate of turnover or the
replacement rate of the water in the basin. This increases the time that water is exposed to
sunlight which, for eutrophic water with relatively high concentrations of nitrogen, phosphorus, and organic carbon, means additional algae growth and especially the single-cell
type algae like Carteria klebsii that give the water a green appearance. On infiltration,
these algae form an algae filter cake on the bottom which increases the clogging process.
Also, photosynthezing algae absorb dissolved carbon dioxide from the water, which at
high algae concentrations increases the pH of the water to 9, 10, or maybe even 11. This
in turn causes precipitation of dissolved CaCO3, which further aggravates the clogging
process and, hence, decreases infiltration rates. This explains why increasing water depths
in infiltration basins has actually produced decreases in infiltration rates, to the surprise
and dismay of operators who wanted to increase infiltration rates!

For this reason, and also to allow quick draining of basins by natural infiltration when
a drying period is needed, shallow basins (water depth of about 20 cm) are preferred. If
plant growth becomes a problem, water depths may periodically have to be increased to
kill the plants before they emerge above the water level. Deep basins (10 m or more), like
old gravel pits, are not very good for recharge because clogging layers become compressed and infiltration rates are so low that the pits need to be pumped out to start a drying period so that clogging layers can be removed. This is expensive and slow, so that drying often is delayed and infiltration rates remain disappointingly low. An advantage of
deep basins is, however, that they can store more water in times of high runoff events. For
subsequent recharge, however, this water should then be conveyed to shallow basins
where clogging is less and easier controlled by drying and cleaning.
Where groundwater levels are so high that they are close to the bottom of the basin
or even above the bottom, the flow in the aquifer is mainly in the horizontal direction
so that infiltration rates are controlled by the slope of the water table away from the
basins. Under these conditions, increasing the water depth in the basins will increase
infiltration rates in direct proportion to the vertical distance between the water surface
in the basin and the groundwater level some distance away from the basin (Bouwer,
1990; Bouwer and Rice, 1989).
If the groundwater table is above the bottom of a recharge basin, the basin can never
be dried for infiltration recovery and removal of the clogging layer. Thus, sediment continues to accumulate in those basins and infiltration occurs mainly through the sides of the
basins which are then made and kept as steeply as possible to minimize sediment accumulation on the basin banks. Under those conditions, increasing the water depth in the
basin will also increase the infiltration or recharge rate.
24.4.5 Effect of Groundwater Depth on Infiltration Rate
Usually, bottoms and banks of infiltration basins are covered with a clogging layer that
controls and reduces infiltration rates so that the underlying soil material is unsaturated
(Fig. 24.5). The water content in the unsaturated zone then is at a value whereby the corresponding unsaturated hydraulic conductivity is numerically equal to the infiltration rate,
since the downward flow is due to gravity alone with a hydraulic gradient of 1 (Bouwer,
1982). The unsaturated zone breaks the hydraulic continuity between the basin and the
aquifer, so that infiltration rates are independent of the depth to groundwater, as long as
the water table is deep enough so that the top of the capillary fringe above the water table
is below the bottom of the basin. This capillary fringe may be about 1 ft (30 cm) thick for
medium sands, more for finer sands or soils, and less for coarse sands. Thus, a conservative conclusion is that as long as the groundwater table is more than 3 ft (about 1 m) below
the bottom of a basin where infiltration is controlled by a clogging layer on the bottom,
infiltration rates are unaffected by changes in groundwater levels. If the groundwater table
rises, infiltration rates will start to decrease when the capillary fringe reaches the bottom
of the basin, and continue to decrease linearly with depth of groundwater below the water
level in the basin, until they become zero when the groundwater table has risen to the
water surface in the basin.
Where the water for recharge is exceptionally clear and free from nutrients and organic carbon, temperatures are low, and soils are relatively coarse, infiltration can go on for
considerable time without development of a clogging layer on the bottom. In that case,
there is direct hydraulic continuity between the basin and the aquifer with the groundwater table joining the water surface in the basin (Fig. 24.6; Bouwer, 1969, 1978).
Groundwater levels are then characterized by the depth Dw of the groundwater table below

the water surface in the basin at sufficient distance from the basin to be relatively unaffected by the recharge flow system (Fig. 24.6). If Dw is relatively large, the flow below the
recharge basin is mainly downward and gravity controlled (Bouwer 1969, 1978) with a
hydraulic gradient of about 1 (Fig. 24.6). In that case, infiltration rates are essentially
unaffected by depth to groundwater. However, if groundwater levels rise and Dw decreases, the flow from the recharge basin becomes more and more lateral until it is controlled
by the slope of the water table away from the basin (Fig. 24.6; Bouwer 1969, 1978).
Modeling these flow systems on an electrical resistance network analog has shown that
the change from gravity controlled flow to slope-of-the-water-table-controlled flow
occurs when Dw is about twice the width W (or diameter) of the recharge system (Bouwer,
1990). This relation is shown in Fig. 24.7 where I is the infiltration rate per unit area of
water surface in the basin and K is the hydraulic conductivity in the wetted zone or
aquifer. Thus, as long as Dw < 2K, infiltration rates decrease linearly with decreasing Dw
and reach zero when Dw = O (Fig. 24.7). These relations apply to uniform, isotropic underground formations. Anisotropic or stratified situations need to be considered on a case-bycase basis.
Infiltration rates in clean basins (no clogging layers) thus are more sensitive to depth
to groundwater than in clogged basins. Clogged basins are the rule rather than the exception and groundwater mounds can rise much higher there than below clean basins before
reductions in infiltration rates occur. Sometimes maximum mound heights are dictated by
circumstances other than their effect on infiltration rates, such as presence of sanitary
landfills, underground sewers or other pipelines, basements (especially deep basements of
commercial buildings), cemeteries, and deep = rooted vegetation like old trees that may
die when groundwater levels rise too high.
24.4.6 Induced Recharge
One method for increasing recharge of groundwater from losing streams is to lower
groundwater levels near the stream by groundwater pumping. As described in the previ-

FIGURE 24.6 Recharge basin with high groundwater table and

lateral flow in aquifer controlled by slope of water table (top)
and with deep groundwater table with downward flow below basin
controlled by gravity (bottom).

FIGURE 24.7 Dimensionless plot of seepage (expressed as I/K) versus depth to groundwater (expressed as DJW) for clean stream channel or basin (no clogging layer on bottom).

ous section, this will only increase recharge or seepage losses from the stream if the
stream bottom is not covered by a clogging layer and the groundwater table is relatively
high so that the depth to groundwater below the water level in the stream is less than two
stream widths (Fig. 24.7). If the groundwater level is already more than two stream widths
below the water surface in the stream, recharge is already near maximum and further lowering of the groundwater will have little or no effect on recharge (Fig. 24.7). If the stream
wetted perimeter is covered by a clogging layer, lowering of groundwater levels will only
increase recharge if the groundwater table is above the stream bottom or only a small distance below the stream bottom so that the capillary fringe still touches the bottom. If the
groundwater level already is deep enough for the top of the capillary fringe to be below
the stream bottom, the soil material below the stream bottom is unsaturated. There is then
no direct hydraulic connection between the stream and the aquifer, so that further lowering of the groundwater level by pumping from wells will not increase the recharge or
seepage losses from the stream.
Groundwater pumping can decrease streamflow in two ways. If the groundwater table
is above the water surface in the stream, the stream is gaining and groundwater moves into
the stream and supports the base flow in the stream. Any pumping of groundwater will
then diminish the flow of groundwater into the stream. If the groundwater table is below
the water surface in the stream, lowering the groundwater table will increase recharge or
seepage losses from the stream when the groundwater table is high, but not when it is relatively low, as explained above. These relationships can then be used in resolving situations where there are real or perceived conflicts between owners of surface water rights
and holders of groundwater pumping rights (Bouwer and Maddock, 1997).
Where groundwater levels are high enough to support growth of phreatophytes (plants
and trees that live off groundwater), a lowering of groundwater levels can reduce the
water uptake by phreatophytes (Bouwer, 1975) and, hence, increase the capture of
groundwater by the aquifer (Bouwer and Maddock, 1997). This means that groundwater
pumping can then lessen the reduction in flow of tributary groundwater to gaining
streams, and reduce water losses from losing streams.

24.5

GROUNDWATERMOUNDING

24.5.1 Perched Groundwater Mounds

When infiltrometer data indicate surface infiltration rates that are acceptable for artificial
recharge, the next step is to investigate the vadose zone to make sure that the infiltrated
water can move unimpeded to the underlying groundwater. Trenches or pits dug by backhoes allow inspection of the soil profile to a depth of about 7 m. If fine-textured or
cemented (caliche) materials are found at a certain depth, their hydraulic conductivity can
be determined with infiltrometer tests or other methods (Bouwer, 1978) on the bottom of
a trench or pit excavated to that depth. Doing these measurements in a sloping trench bottom will give a profile of hydraulic conductivity. Soil borings are necessary to evaluate
deeper layers. Large-diameter (about 1 m) holes drilled with a bucket auger enable assessment of infiltration rates and hydraulic conductivities at great depth. After removing loose
material from the bottom, a few cm of water on the bottom are maintained, and the infiltration rate is measured. This can be managed from the top. Bucket augers can drill holes
to about 50 m.
If a restricting layer is detected, the height of the perching mound can be calculated by
applying Darcy's equation to the vertically downward flow in the perching mound and the

flow through the restricting layer (Fig. 24.8). If the material below the restricting layer is
relatively coarse, the pressure head of the water at the bottom of the restricting layer can
be taken as zero. For finer materials below the restricting layer, the water-entry values listed in Sec. 24.4.2 be taken as a first estimate. The resulting equation for the perched
mound, assuming zero pressure head for the water at the bottom of the restricting layer, is
then (Bouwer et al., 1997).
F' 1
Lp = I^-r

(24.5)

K.
where Lp = height of perching mound above restricting layer, Lr = thickness of restricting layer, / = infiltration rate, Kr = hydraulic conductivity of restricting layer, and K5 =
hydraulic conductivity of soil above restricting layer.

Often, / will be much smaller than Ks because surface soils are finer textured than
deeper soils, or there is a clogging layer on the surface soil that reduces infiltration.
Also, i often will be much larger than Kr. For these conditions, Eq. 24.5 can be simplified to
^ = I-TJT
Ar

(24.6)

Lp should be small enough so that the top of the perched mound is deep enough to
avoid reductions in infiltration rates, as discussed in the Sec. 24.4.5. Perched mounds
above noncontinuous perching layers (lenses or strips) are not as high as above continuous perching layers with the same Lr and Kr because there is then also lateral flow in the
perched mound. Thus, Eqs. 24.5 or 24.6 will then overestimate the height of the perched
mound. Where perched mounds would be too high, recharge may be achieved with long
narrow basins that allow lateral flow in the mound and reduce its height as the perched
mound spreads laterally on the restricting layer until all the recharge water will pass
through the restricting layer (Bouwer, 1978). Another possibility is to use infiltration
basins that are excavated through the restricting layer (Fig. 24.1B). If the restricting layer

FIGURE 24.8 Geometry and symbols for perched

mound above restricting layer in vadose zones.

is too deep, recharge trenches (Fig. 24.1C) or recharge shafts (large diameter holes in the
vadose zone, also called dry wells or vadose zone wells) can be used (Fig. 24.1D).
24.5.2 Groundwater Mounds
Numerous papers have been published on the rise of groundwater mounds on the aquifer
in response to infiltration from a recharge system, and some also on the fall of the mound
after infiltration has stopped (Glover, 1964: Hantush, 1967; Marino, 1975a, b; Warner et
al., 1989). The usual assumption is a uniform aquifer of infinite extent with no other
recharges or discharges. One of the difficulties in getting meaningful results from the equations is getting a representative value of aquifer transmissivity. The most reliable transmissivity data come from calibrated models. Next come Theis-type pumping tests, step-drawdown and other pumped well tests, and slug tests, in decreasing order of sampling size.
Slug tests, while simple to carry out, always have the problem of how to get representative
areal values from essentially point measurements. Averages from various tests often seriously underestimate more regional values (Bouwer, 1996, and references therein).
Piezometers at two different depths in the center of a mound enable the determination of
both vertical and horizontal hydraulic conductivity of an aquifer already being recharged
with an infiltration system (Bouwer et al. 1974). In deep or thick unconfined aquifers,
streamlines of recharge flow systems are concentrated in the upper or "active" portion of
the aquifer, with much less flow and almost stagnant water in the deeper or "passive" portion of the aquifer. Use of transmissivities of the entire aquifer between the water table and
the impermeable lower boundary for mound calculations can then seriously underestimate
the rise of the mound. Older work (Bouwer, 1962) with resistance network analog modeling showed that for rectangular recharge areas, the thickness of the active, upper portion of
the aquifer is about equal to the width of the recharge area. This thickness should then be
multiplied by K of the upper aquifer to get an "effective" transmissivity for mounding predictions. Also, if the Hantush or other equation is used to calculate long-term mound formation, as for water banking in areas with deep groundwater levels, larger transmissivity
values should be used to reflect the increase in transmissivity as the groundwater level
rises. Otherwise, the Hantush equation will seriously overestimate the mound rise.
The best way to get representative transmissivity values for artificial recharge systems
is to have a large enough infiltration area that produces a groundwater mound, and to calculate the transmissivity from the rise of the groundwater mound using, for example,
Hantush's equation (Eq. 24.7). The fillable porosity to be used in the equations for mound
rise usually is larger than the specific yield of the aquifer, because vadose zones often are
relatively dry, especially in dry climates and if they consist of coarse materials like sands
and gravels. The fillable porosity should be taken as the difference between existing and
saturated water contents of the material outside the wetted zone below the infiltration system. The Hantush equation is
hx,y,t ~H=V~^

{^ W/2 + *> <L/2 + ?)]

+ F[W/2 + jc)n, (L/2 - y)n]
+ F[W/2 - jc)w, (L/2 + y)w]
+ F[W/2 - jc)n, (L/2 - y)n]

(24.7)

where hx t = height of water table above impermeable layer at x, v, and time (Fig. 24.9),
H = original height of water table above impermeable layer, va = arrival rate at water
table of water from infiltration basin, t = time since start of recharge, / = fillable porosi-

ty (1 >/> 0)> = length of recharge basin (in y direction), W = width of recharge basin
(in x direction), n = (4t 77/)-1/2, and F(OC, p) = /1Q erf(ai - 1/2) erf($i - m)dx [which was
tabulated by Hantush (Table 24.1)]
where a = (W/2 + x) n or (W/2 - x)n and (3 = (L/2 + y)w or (1/2 - y)n.
Of most interest to operators and managers often, is the long-term effect of recharge
on groundwater; that is, where the groundwater mound will be 10, 20, or 50 years from
now, how much water can be stored or "banked" underground, will the whole area become
waterlogged, and how must the water be recovered from the aquifer to prevent waterlogging? Computer models can be used to simulate regional recharge inputs and pumped
well outputs for the aquifer. However, a quick idea about ultimate or quasi-equilibrium
mound heights can be obtained from a steady-state analysis where the mound is considered to be in equilibrium with a constant water table at some depth and at some distance
from the infiltration system. The constant "faraway " water table is obtained by groundwater pumping, discharge into surface water like rivers or lakes, or some other "control."
Usually, recharge systems consist of a number of basins or other infiltration facilities.
Steady-state equations were developed for two general geometries of the entire recharge
area: (1) the basins form a long strip with a length of at least 5 times the width so that after
long times it still performs about the same as an infinitely long strip (Glover, 1964), and
(2) the basins are in a round or square or irregular area that can be handled as an equivalent circular area (Bouwer et al., 1999b). For the long strip (Fig. 24.10) the groundwater
flow away from the strip was taken as linear horizontal flow (Dupuit-Forchheimerflow).
Below the infiltration area, the lateral flow was assumed to increase linearly with distance
from the center. The lateral flow was then assumed to be constant between the edge of the
recharge system at distance W/2 from the center and the constant control water table at
distance Ln from the edge (Fig. 24.10). This yielded the equation

FIGURE 24.9 Geometry and symbols for recharge system and groundwater mound.

"<-"- = f (T+ L)

(24.8)

for the ultimate rise of the groundwater mound below the center of the recharge strip when
there is equilibrium between recharge and pumping from the aquifer (Bouwer et al.,
1999b). The symbols in this equation are (Fig. 24.10):
H0 = height of groundwater mound in center of recharge area,
Hn = height of groundwater table at control area,
/ = average infiltration rate in recharge area (total recharge divided by total area),
W = width of recharge area,
Ln = distance between edge of recharge area and control area, and
T = transmissivity of aquifer.
For a round or square recharge area (Fig. 24.11), the groundwater flow will be radially away from the area. The equilibrium height of the mound below the center of the
recharge system above the constant groundwater table at distance Rn from the center
of the recharge system then can be calculated with radial flow theory (Bouwer et al.,
1999b) as
H0-Hn = ^ [l + 2/n^]

(24.9)

where R is the radius or equivalent radius of the recharge area, Rn is the distance from the
center of the recharge area to the water-table control area (Fig. 24.11) and the other symbols are as defined above.
Equations (24.8) and (24.9) thus can be used to predict the final mound height below
a recharge area for a given elevation of the control water table at distance Rn or Ln from
the recharge area. As indicated for the Hantush equation, the value of Tin Eq. (24.8) and
(24.9) must reflect the average transmissivity of the aquifer at the ultimate equilibrium
mound height. If the ultimate mound height is too high, Rn or Ln must be reduced by
groundwater pumping from wells closer to the recharge area or Hn must be reduced by
pumping more groundwater. Equations (24.8) and (24.9) can then indicate where groundwater should be recovered and to what depth groundwater levels should be pumped to prevent groundwater tables below the recharge areas from rising too high. Ultimate mound
heights can also be reduced by making the recharge area longer and narrower, or by reducing recharge rates as can be achieved by using less water for recharge or by spreading the
infiltration facilities out over a larger area.

24.6

CHALLENGINGSOILS

Because artificial recharge of groundwater offers many advantages, municipalities and

water districts, or other entities may still wish to do artificial recharge even though permeable soils are not available and they may have to work with finer soils (sandy loams to
loams) with marginal hydraulic conductivity. Such systems may only have infiltration
rates of 0.05-0.1 m/day and hydraulic loading rates of 10 to 20 m/yr. At these rates, evaporation losses (0.5-2.5 m/year for year-round systems, depending on climate) can become
significant. To maximize infiltration rates, operators of such systems are tempted to disk
Next Page

Index

Index terms

Links

A
Acoustic (sonic) velocity

12.2

Aguado

1.14

Ancient dams

1.13

Ancient drainage systems

1.14

Egyptian systems

1.7f

Kariz
Mesopotamian systems
Mexican systems
North American systems

1.7
1.7f
1.8

1.9f

1.10

Chaco systems

1.13f

Hohokam systems

1.11f

1.12f

Qanat

1.7

1.8f

Shaduf

1.4

1.6f

Swape

1.4

Ancient water systems

1.14

Minoan systems

1.14

1.15f

Roman aqueducts

1.20f

1.22f

Roman systems

1.16
1.19f

1.17f
1.20f

Annual expected flood damage cost

Apodyterium
Aquifer restoration process schematics

1.10f

1.18f
1.21f

7.47
1.17
23.15f

Groundwater pump and treat

23.15f

In-situ air sparging

23.15f

Natural attention

23.15f

This page has been reformatted by Knovel to provide easier navigation.

I.1

I.2

Index terms
Aquifer transmissivity

Links
4.8

Artificial recharge,
Subsurface systems, dry wells
Aquifer wells
Aquifer storage and recovery wells

24.7
24.31
24.3

Recharge shafts

24.29

Vadose zone shafts

24.6f

Vadose zone wells

24.29

Surface systems

24.4

In-channel systems

24.4

24.5f

Off-channel systems

24.4

24.7

Atmospheric heat exchange

5.3

Atmospheric radiation

5.4

B
Benefit-cost analysis

1.27

Benefit-cost ratio

1.28

Flowchart of

1.29f

Incremental benefit-cost ratios

1.28

Marginal values

1.28

Bernoulli's equation

2.13

Bioventing

23.51

Design calculations

23.55

Design principles

23.53

In-situ respirometry tests

23.53

Pilot test schematic

23.54

System schematic

23.52

Zero-order biodegradation rate

23.53

Boundary layer theory

2.14

Boundary shear stress

16.29

Average shear stress

21.3

16.29

This page has been reformatted by Knovel to provide easier navigation.

I.3

Index terms

Links

Boundary shear stress (Continued)

Bend adjustment factor

16.32

Local boundary shear adjustments

16.29

Mean shear stress

16.29

Bowen ratio
Bridge-deck drainage design

5.7
13.36

Constant-grade bridge inlet design

13.38

Efficiency curves for scuppers

13.37

Flat bridge inlet design

13.39

Off-bridge inlets

13.36

Scuppers

13.36

Bulk modulus of elasticity

13.37f

2.36

C
Caldarium
Capillary fringe
Castellum
Catchment runoff modeling

1.17
24.13
1.21
14.67

Calibration

14.71

Modeling procedure

14.69

Models

14.72t

Simulation error measures

14.75t

Verification

14.71

Cavitation

10.22

Cavitation index

17.40

Cavitation potential

17.40

Cavitation number

17.40

Cenote

1.14

Centrifugal pumps, types

10.6f

Kinetic type pumps

10.6f

Typical discharge curves

10.7f

10.8

This page has been reformatted by Knovel to provide easier navigation.

I.4

Index terms

Links

Channel bends

8.2

Channel linings

16.1

Flexible linings

16.1

Rigid linings

16.1

Channel rating curve

19.9

Channel stationing

19.9

Characteristic curves

2.27

Chezy equation

3.11

Chezy's law

6.4

Chezy resistance coefficient

3.11

Chultun

1.14

Clogging of,
Aquifer wells

24.31

Assimilable organic carbon content

24.31

Membrane filtration index

24.31

Parallel filter index

24.31

Infiltration basins
Colebrook-White equation
Contaminant migration barriers

24.10
2.17
23.61

Air sparging cut-off trenches

23.65

Hydraulic contaminant systems

23.61

Design process

23.61

Recirculation wells

23.62

Reaction-based contaminant migration barriers

10.12

23.63f

23.62

Chemical, physical/chemical, and biological

treatment barriers
Design approach
Funnel and gate reaction wall chemical barrier
schematic
In-situ reaction barriers

23.63
23.63
23.64f
23.61

This page has been reformatted by Knovel to provide easier navigation.

I.5

Index terms
Contaminant mitigation prevention technologies

Links
23.14

In-situ air sparging curtain

23.16f

Physical barriers

23.16f

Reaction barrier

23.16f

Contaminant release scenarios

23.5f

Dissolved contaminant release

23.6f

DNAPL release

23.5f

LNAPL release

23.5f

23.15f

23.6f

Continuity, see conservation laws

Conservation laws of

2.6

Chemical species

2.7

Continuity

14.32

Energy

9.3

14.32

Mass

9.3

14.32

Molecular species

2.7

Momentum

9.24

Control volume

2.7

Convective heat flux

5.7

Critical depth

3.5

Critical flow

21.5

Critical shear stress for particle motion

Culvert design
Computer models for

15.16
15.66
15.67

CDS

15.67

CULVERT2

15.67

HY8

15.66
15.21f

Design example

15.22

Inlet control

15.17

Constraints for design equations

15.21f

6.6

CAP

Critical depth

14.32

15.18t

This page has been reformatted by Knovel to provide easier navigation.

21.5

I.6

Index terms

Links

Culvert design (Continued)

Inlet control nomographs

15.17

Submerged (weir) equation

15.17

Unsubmerged (orifice) conditions

15.17

Location and alignment

Beveled edges

15.34f
15.30

End treatments

15.32

Inlet types

15.33

Side-tapered inlet

15.35f

Slope-tapered inlet

15.36f

Siphon

15.32

Top location placement

15.31

Special considerations

15.37

Debris control

15.38

Erosion

15.37

Sedimentation

15.38

Bend losses
Culvert design

15.9

15.10t

15.16
15.2

Flow characteristics

15.4

Headwater

15.2

Allowable

15.3

Maximum

15.3

Outlet velocity

15.31f

15.1

Design parameters

Inlet control

15.20f

15.30

Bottom location placement

Culverts

15.19f

15.3f

15.4

15.25

15.5

15.25

15.14f

Types of

15.6f

Junction losses

15.10

Nomographs

15.12

Outlet control

15.4

This page has been reformatted by Knovel to provide easier navigation.

I.7

Index terms

Links

Culverts (Continued)
Entrance loss coefficients for

15.8t

Hydraulics of

15.6

Outlet velocity

15.14f

Types of

15.7f

Outlet velocity

15.13

Performance curves

15.23

Roadway overtopping

15.14

Discharge coefficients for

15.15f

Submergence factors for

15.15f

Weir crest length determination for

15.16f

Tailwater
Culvert flow transitions
Abrupt expansions

15.3
20.1
20.2

Design procedure

20.2

Example design

20.7

Gradual transitions

20.9

Design procedure

20.9

Example

20.10

Inlet control

20.13

Design procedure

20.13

Example

20.16

Supercritical inlet transition

Outlet control
Transition types
Culvert performance curves
Example

20.13f
20.1
20.2f
15.25
15.25

D
DAMBRK

19.26

Darcy's law

4.1

Anisotropic media

4.3

This page has been reformatted by Knovel to provide easier navigation.

I.8

Index terms

Links

Darcy's law (Continued)

Isotropic media

4.3

Saturated flow of variable density

4.3

Unsaturated flow of variable density

4.3

Vapor flow

4.18

Wetted zone

24.9

Darcy-Weisbach equation
Sewer pipe

2.16
14.5
14.8

Darcy-Weisbach friction factor

3.12

Deflector buckets (submerged)

18.40

Desander

8.3

Desanding basin

8.3

Desilting basin

8.3

Detention storage
Computer models for design

14.76
14.91

BASINOPT

14.91

PONDOPT

14.91

Watershed modeling standalone

14.91

Detention basins

14.76

Basic elements

14.77f

Design aid equations

14.82t

Design guidelines

14.79

Design procedure

14.81

Detention outlet structures

14.78f

Multiple outlet structures

14.80f

Orifice outlet equation

14.79

Pond design aids

14.81

Sharp-crested weir equations

14.70

Stage-storage parameters
Stage-storage relationships

9.9
21.4

14.81f
14.80

This page has been reformatted by Knovel to provide easier navigation.

10.12

I.9

Index terms

Links

Detention storage (Continued)

Weir equations

14.79

Extended detention basins

14.85

Design considerations

14.88

Extended detention outlet structures

14.86

14.87f

14.88

14.89f

Retention basins
Design considerations

14.89

Design curves for solids settling

14.90

Outlet structures
Permanent pool volume

14.90f
14.88

Digital elevation model

19.5

Digital line graphs

19.4

Digital orthophoto quad

19.4

Digital raster graphic

19.4

Digital terrain models

19.6

Dispersion,
Density current mixing

5.17

Entrainment coefficient

5.18

Fraction entrainment

5.17

Interflow mixing

5.18

Bouyancy frequency

5.19

Grashof number

5.19

Intrusion length

5.20

Prandtl number

5.19

Thickness, equation for

5.19

Lakes and reservoirs

5.12

Longitudinal

5.11

Equation for
Meteorological forces

5.13f

5.11
5.26

Internal hydraulic jumps

5.27

Internal waves

5.27

This page has been reformatted by Knovel to provide easier navigation.

I.10

Index terms

Links

Dispersion (Continued)
Langmuir circulation

5.27

Plunging breaker

5.27

Shortwave motion

5.26

Spilling breaker

5.27

Swash zone

5.27

Waves

5.26

Wind mixed depth

5.26

Mixing dispersion

5.12

Plunge point

5.14

Separation point

5.15

Sources of energy for

5.12

Open channels

5.9

Outflow mixing

5.21

BETTER model

5.22

CE-QUAL-R1

5.22

CE-QUAL-W2

5.22

SELECT model

5.22

Withdrawal characteristics

5.23f

Overflow

5.17

Underflow

5.17

Draft tubes
Dupuit-Forchheimer flow

5.15

8.26
24.19

E
Energy dissipation for culverts and channels
Drop structures

20.17
20.33

Box-inlet drop structure

20.38

Grated energy dissipator

20.35

Riprap culvert energy basin

Straight-drop structure

20.44f

20.45

20.36

This page has been reformatted by Knovel to provide easier navigation.

I.11

Index terms

Links

Energy dissipation for culverts and channels (Continued)

Example design
Forced hydraulic jump basins
Saint Anthony Falls (SAP) basin

20.37
20.21
20.21

Design procedure

20.22

Example design

20.23

USER type II, III, IV basins

Example design type II basin
USER type VI basin
Hydraulic jump basins

20.26
20.26
20.29
20.17

Definition sketch

20.17f

Design procedure

20.17

Example design

20.19

Energy dissipators

18.1

Channels

20.17

Culverts

20.17

Drop type energy dissipators

18.25

Guidelines

18.25

Hydraulic jump-type dissipators

18.1

Impact-type dissipators

18.1

Engineering economics

1.25

Benefit cost analysis

1.27

Benefit - cost ratio

Flowchart of

1.28
1.29f

Incremental benefit-cost ratios

1.28

Marginal values

1.28

Discount factors

1.25

Life spans of hydraulic projects

1.28

1.29t

Energy gradeline

2.14

22.18

Entrapped air

12.2

EPANET

9.33

This page has been reformatted by Knovel to provide easier navigation.

I.12

Index terms
Euler number
Evaporation
From natural water surface

Links
12.10
5.6
5.6

Evaporative heat flux

5.7

Evaporative heat loss

5.6

Bowen ratio

5.7

Exner equation

6.27

Failure probability

7.2

FESWMS-2DH

19.26

FLDWAV

19.26

Fick's law
Flip bucket
Fix grade node
Flood control channel
Components

5.7

4.4
17.38
9.2
16.1
16.5

Channel crossing

16.6

Channel inlet

15.5

Channel outlet

16.6

Channel reach

16.5

Confluence

16.6

Side inlet

16.6

Transition

16.6

Design

16.2

Design parameters

16.8

Channel profiles

16.9

Channel section

16.9

Channel slope

16.9

Duration of flooding

16.8

16.9

Flood frequency

16.8

16.9t

Linings
Curled mat

16.10
16.17

This page has been reformatted by Knovel to provide easier navigation.

I.13

Index terms

Links

Flood control channel (Continued)

Degradable flexible linings

16.17

Flexible

16.11

Fiberglass roving

16.19f

Gravel

16.13f

Jute net

16.16f

Performance data
Rock channel

16.10
16.12f

Rock riprap

16.12

Turf-reinforcement mat

16.14

Types

16.10

Vegetative

16.12f

Wire enclosed stone

16.13f

Woven-paper net channel

16.15f

Stable channels
Boundary shear stress

16.16

16.14

16.6
16.8

Bends

16.8f

Channel sections

16.8f

Channel alignment

16.8

Dynamic-stable channels

16.6

Modes

16.6

Permissible shear stress

16.7

Static-stable channel

16.6

Tractive force

16.7

F
Flood control channel design, composite sections

16.38

Manning's roughness

16.39

Special considerations

16.39

Flood control channel design, steep gradient

Bathhurst equation

16.35
16.35

This page has been reformatted by Knovel to provide easier navigation.

I.14

Index terms

Links

Flood control channel design, steep gradient (Continued)

Boundary shear stress

16.36

Permissible shear stress

16.36

Resistance to flow

16.35

Bathhurst equation

16.35

Effect of roughness geometry

16.36

Relative roughness area

16.36

Flood control channel design, symmetric sections, mild

gradients
Boundary shear stress

16.19
16.29

Average shear stress

16.29

Bend adjustment factor

16.32

Local boundary shear adjustments

16.29

Mean shear stress

16.29

Critical shear stress

Adjustment factor

16.26
16.27

Density stiffness parameter

16.21

Effective roughness height

16.21

Flexible lining coefficients

16.22

Manning's roughness

16.21

Mild gradients

16.20t

Vegetative classes

16.22

16.29f

Permissible shear stresses

Cohesive soils

16.25f

Lining materials

16.24t

Noncohesive soils

16.24f

Retardance classes
Manning's roughness for

16.21
16.22

Tractive force design

16.24

Vegetative critical shear stress

16.21

This page has been reformatted by Knovel to provide easier navigation.

I.15

Index terms
Flood control channel design, with sediment transport

Links
16.39

Aggradtion

16.4

Bed forms

16.42

Degradation

16.41

Lane's relationship, non-cohesive soils

16.40

Resistance to flow

16.42

Revised universal soil loss equation

16.40

Sediment discharge

16.41

Sediment mass balance

16.42

Sediment supply

16.39

Sediment transport

16.40

Bed load

16.40

Suspended bed load

16.40

Total sediment load

16.40

Floodplain area

19.1

Floodplain studies

19.9

Approaches

19.21

Critical depth

19.23

Critical flow

19.23

Gradually varied flow

19.22

Hydraulic controls

19.23

One-dimensional flow

19.22

Rapidly varied flow

19.22

Steady flow

19.22

Subcritical flow

19.23

Supercritical flow

19.23

Two-dimensional flow

19.22

Uniform flow

19.22

Aerial topography

19.16

Bridge survey data

19.19

Channel structures

19.21

19.20f

This page has been reformatted by Knovel to provide easier navigation.

I.16

Index terms

Links

Floodplain studies (Continued)

Computer programs

19.25

DAMBRK

19.26

FESWMS-2DH

19.25

FLDWAV

19.26

HEC-2

19.25

HEC-RAS

19.25

NETWORK

19.26

RMA-2

19.25

SWMM

19.26

TABS-MD

19.26

UNET

19.26

WSPRO

19.25

WSP-2

19.25

Contour map data

19.16

Cross-section alignment

19.13

Cross-section interpolation

19.37

Cross-section locations

19.11

19.14f

19.17
Cross-section orientation

19.13

Culvert data

19.21

Field surveys

19.10

Floodway

19.6

Floodway determination

19.45

Performing studies

19.28

Starting conditions

19.28

Tributary stream analysis

19.30

Quality of analysis

19.38

Road crossing data

19.17

Standard step computations

19.31

Stream system types

19.23

This page has been reformatted by Knovel to provide easier navigation.

19.15f

I.17

Index terms

Links

Floodplain studies (Continued)

Dendritic channel systems

19.23

First-order streams

19.23

Higher-order streams

19.23

Network channel systems

19.23

19.25f

Simple channels

19.23

19.24f

Floodway

19.45

Floodway determination

19.45

Flow control

19.24f

2.1

Flow measurement,
Flow conditioning

21.11

Channels

21.12

Irrigation delivery pipeline

21.12

Options

21.11

Orifice plates

21.12

Flumes,
Critical flow flumes

21.10

Long-throated flumes

21.10

Short-throated flumes

21.10

Venturi flume

21.10

Installation requirements

21.10

Measurement accuracy

21.14

Accuracy terms

21.15

Comparison standards

21.17

Measurement capability terms

21.16

Water meter classification

21.9

Primary device

21.9

Quantity meters

21.9

Rate meters

21.9

Secondary devices

21.9

Wave suppression

22.5

21.13

This page has been reformatted by Knovel to provide easier navigation.

I.18

Index terms

Links

Flow measurement (Continued)

Variable-depth flows
Wave suppressor design

21.14
21.14f

Flow measuring devices,

Expedient measurement techniques
Air bubbles

21.44
21.47f

Herschel-type Venturi tube

21.48

Low pressure pipe Venturi

21.46

Portable flow measurement flumes

21.45

Rising bubbles

21.46

Surface floats

21.45

Water surface profile measurement

21.44

Long-throated flumes, applications

21.29

Broad-crested weir sizes

21.30t

Circular channels

21.41

Concrete-lined canal, broad crested weir

21.34f

Example calculations

21.36

Rating equations, rectangular weirs

21.39t

Rating equations for trapezoidal broad-crested

weirs

21.32t

Rating table for lined canals

21.30t

Rectangular structures for unlined canals

21.38

Primary elements of devices

Types of devices
Secondary devices, selection

21.18
21.18
21.25

Application based selection

21.26t

Quality assurance

21.29

Selection guide

21.28t

Flow meters

22.4

Magnetic meters

22.5

Mechanical meters

22.6

This page has been reformatted by Knovel to provide easier navigation.

I.19

Index terms

Links

Flow meters (Continued)

Pressure differential/pressure measuring meters

22.5

Orifice plates

22.5

Parashall flumes

22.5

Pitot meters

22.5

Venturi meters

22.5

Ultrasonic meters

22.6

Flow transitions, culverts

21.8

Forebay

8.2

Freeboard

8.2

Friction slope

3.22

Average conveyance friction slope

3.22

19.34

Average friction slope

3.22

19.34

Harmonic mean friction slope

3.22

19.34

Geometric mean friction slope

3.22

19.34

Frigidarium

1.17

Froude number

3.1
16.3

Densimetric Froude number

5.14

Internal Froude number

5.14

Normal densimetric Froude number

5.16

Plunge point

5.16

Separation point

5.16

3.6

5.14

5.18

5.19

G
Gates

22.3

Seating head

22.3

Unseating head

22.3

Geometric and pavement guidelines

13.2

Cross slope

13.3t

Grade

13.2

This page has been reformatted by Knovel to provide easier navigation.

I.20

Index terms

Links

Geometric and pavement guidelines (Continued)

Pavement

13.2

Hydroplaning

13.2

Pavement design factors

13.2

Sheetflow

13.2

Geometric similarity
Global positioning system
Gradually varied flow

21.8
19.10
3.16

Profile classification

3.20t

Standard step method

3.19

Water surface profiles

3.23

Gradually varied unsteady flow, models

3.25

Diffusive wave model

3.25

Dynamic model

3.25

Kinematic wave model

3.25

Grate inlet design

13.29

Efficiency

13.29

Frontal flow intercepted

13.31

Frontal flow interception efficiency

13.31f

Interception capacity

13.32

Overall efficiency

13.31

Side flow interception efficiency

13.31

Splash-over velocity

13.29

Total frontal flow

13.31

Grate inlet efficiencies

13.30t

Groundwater contamination,
Aquifer restoration process schematics

23.15f

Groundwater pump and treat

23.15f

In-situ air sparging

23.15f

Natural attention

23.15f

Bioventing

23.51

This page has been reformatted by Knovel to provide easier navigation.

I.21

Index terms

Links

Groundwater contamination (Continued)

Design calculations

23.55

Design principles

23.53

In-situ respirometry tests

23.53

Pilot test schematic

23.54

System schematic

23.52

Zero-order biodegradation rate

23.53

Chemical properties, contaminants

23.20t

Common sources

23.3f

Contaminant migration barriers

23.61

Air sparging cut-off trenches

23.65

Hydraulic contaminant systems

23.61

Design process

23.61

Recirculation wells

23.62

Reaction-based contaminant migration barriers

23.63f

23.62

Chemical, physical/chemical, and biological

treatment barriers
Design approach

23.63
23.63

Funnel and gate reaction wall chemical barrier

schematic
In-situ reaction barriers
Contaminant mitigation prevention technologies

23.64f
23.61
23.14

In-situ air sparging curtain

23.16f

Physical barriers

23.16f

Reaction barrier

23.16f

Contaminant release scenarios

23.5f

Dissolved contaminant release

23.6f

DNAPL release

23.5f

LNAPL release

23.5f

Design flowchart

23.15f

23.6f

23.14f

Dissolved plume treatment

23.59

Natural attenuation

23.60

This page has been reformatted by Knovel to provide easier navigation.

I.22

Index terms

Links

Groundwater contamination (Continued)

Free product recovery
Dual pump recovery system

23.25
23.31

Interceptor trench

23.29f

Liquid monitoring

23.26

Maximum achievable liquid recovery

23.27

Operation with groundwater drawdown

23.32f

Skimming system

23.30f

Total fluids recovery system

23.30

Trench systems

23.28

Vertical recovery well schemes

23.28

Vertical recovery well

23.29f

Groundwater pump and treat systemssource zone

treatment

23.50

In-situ air sparging (IAS)

23.55

Air injection wells

23.57

Biosparging

23.58

Blowers

23.58

Compressors

23.58

Design parameters

23.57t

Design principles

23.56

Pilot tests

23.58

Process principles

23.56

In-situ pilot testing

23.24

In-situ treatment technologies

23.12

Aquifer restoration technologies

23.12

Source zone treatment technologies

23.12

Maximum contaminant levels (MCLs)

23.7

23.9t

Maximum contaminant level goals (MCLGs)

23.7

23.9t

Maximum depth penetrated by spill

23.18

Pump and treat systems

23.50

This page has been reformatted by Knovel to provide easier navigation.

I.23

Index terms

Links

Groundwater contamination (Continued)

Remediation goals

23.7

Scenarios

23.2

Site characterization

23.16

Contaminant distribution assessment

23.17

Geologic/hydrogeologic assessment

23.17

Historic assessment

23.17

Political assessment

23.17

Receptor assessment

23.17

Soil vapor extraction

23.34

Efficiency

23.39

Equation for flow rates

23.35

Example site data for design

23.50

Example design worksheet

23.49

Extraction pump/blower

23.45

Feasibility assessment

23.35

Injection blower

23.45

Instrumentation

23.46

Macroscopic mass transfer limiting scenarios

23.39

Manifolding

23.46

Maximum achievable mass removal rates

23.36

Minimum total vapor flowrate

23.38

Pilot tests

23.40

Treatment system schematic

23.34

System design

23.42

Vapor treatment

23.45

Well construction

23.43

23.44f

23.12

23.25

Source zone treatment technologies

Free-product recovery

23.13f

In-situ air sparging

23.13f

Soil vapor extraction

23.13f

23.41f

This page has been reformatted by Knovel to provide easier navigation.

I.24

Index terms

Links

Groundwater contamination (Continued)

Technology prescreening matrix

23.23

Technology selection flowchart

23.14f

Groundwater flow:
Capture zone

4.10

Darcy's law

4.1

Interception wells

4.10

Capture zone

4.10

Dividing streamline

4.10

Stagnation point

4.10

Upgradient divide

4.10

Jacob's correction
Partially penetrating wells

4.9
4.11

Radial flow theory

24.20

Retardation factor

4.13

Saturated flow
Convolution
Mass transport equations

4.9
4.9
4.12

Steady flow, confined aquifer

4.8

Steady flow, unconfined aquifer

4.8

Superposition

4.9

Transient flow

4.9

Transmissivity
Well duplets
Groundwater mounding
Groundwater mounds

4.8
4.12
24.16
24.18

24.19f

24.24f
Hantush equation

24.18

Perched mounds

24.16

Equation for height

24.17

Restricting layer

24.16

24.17f

This page has been reformatted by Knovel to provide easier navigation.

24.23f

I.25

Index terms

Links

Groundwater transport, saturated flow:

Analytical solutions
Continuous injection, retardation and degradation

4.13
4.14

Continuous point injection, 2-D dispersive

transport
One dimensional, step change in concentrationno
degradation

4.14
4.13

One dimensional, step change in concentration

first order degradation
Point slug injection3-D transport and retardation

4.13
4.14

Mass transport equations

4.12

Retardation factor

4.13

Groundwater transport, unsaturated flow

Unsaturated zone flow

4.15
4.15

Continuity equation

4.15

Flow equations

4.15

Relative hydraulic conductivity

4.16

Specific moisture capacity

4.15

Unsaturated zone transport

Groundwater flow and transport in vapor phase

4.16
4.17

Soil vapor flow

4.17

Darcy's law

4.18

Discharge potential

4.18

Molar flux

4.18

One dimensional flow

4.18

Radial flow

4.18

Vapor radial flow

4.18

Transport in vapor phase

4.19

Adjective vapor flow

4.19

Fate and transport equation

4.19

Retardation factor

4.19

This page has been reformatted by Knovel to provide easier navigation.

I.26

Index terms

Links

Gutter flow velocity

13.10

Gutter design

13.14

Depressed flow

13.15

Manning's roughness coefficients

13.15t

Manning's equation

13.14

Gutter design

13.14

Depressed flow

13.15

Manning's roughness coefficients

13.15t

Manning's equation

13.14

Typical gutter sections

13.14f

Typical gutter sections

13.14f

H
Hantush equation
Hardy Cross method
Hazen-Williams equation

24.18
9.11
2.18
10.10

9.6
22.84

10.9

3.19

15.39

19.25

3.19
19.25

15.39

19.6

Coefficients for

10.11t

Coefficients for raw thickended Osludge

22.86t

Headrace channel

8.1

Heat flux

5.6

Convective heat flux

5.7

Evaporative heat flux

5.6

HEC-2
HEC-RAS
Highway drainage
Bridge-deck drainage design

13.36

Constant-grade bridge inlet design

13.38

Efficiency curves for scuppers

13.37

Flat bridge inlet design

13.39

This page has been reformatted by Knovel to provide easier navigation.

I.27

Index terms

Links

Highway drainage (Continued)

Off-bridge inlets

13.36

Scuppers

13.36

Design frequency and spread

13.4

Check storm

13.5

Major considerations

13.5

Minimum design frequency

13.6t

Minimum spread

13.6t

Drainage inlet design

13.27

Classes of inlets

13.28

Combination inlet design

13.33

Curb-opening inlet design

13.32

Efficiency

13.32

Equivalent cross slope

13.32

Length

13.32

Equal-length inlets

13.34

Grate inlet capacities

13.33

Curb-opening inlets without depressions

13.34

Depressed curb-opening inlet

13.33

Slotted inlets

13.34

Orifices

13.33

Weirs

13.33

Grate inlet design

13.35f

13.29

Efficiency

13.29

Frontal flow intercepted

13.31

Frontal flow interception efficiency

13.37f

13.31f

Interception capacity

13.32

Overall efficiency

13.31

Side flow interception efficiency

13.31

Splash-over velocity

13.29

Total frontal flow

13.31

This page has been reformatted by Knovel to provide easier navigation.

I.28

Index terms

Links

Highway drainage (Continued)

Grate inlet efficiencies

13.30t

Inlets in sag

13.33

Inlet locations

13.35

Sample design problems

13.40

Slotted inlet design

13.32

Sweeper inlet

13.33

Geometric and pavement guidelines

13.2

Cross slope

13.3t

Grade

13.2

Pavement

13.2

Hydroplaning

13.2

Pavement design factors

13.2

Sheetflow

13.2

Gutter design

13.14

Depressed flow

13.15

Manning's roughness coefficients

13.15t

Manning's equation

13.14

Typical gutter sections

Hydrology
Avoidance of hydroplaning method

13.14f
13.6
13.12

Hydroplaning design rainfall intensity

13.13

Vehicle speed to initiate hydroplaning

13.12

Design rainfall intensity

13.6

Avoidance of hydroplaning

13.6

Driver vision-impairment method

13.6

Rational method

13.6

Driver vision-impairment method

13.13

Gutter flow velocity

13.10

Intensity-duration-frequency curves

13.8

Overland sheet flow

13.9

13.12

This page has been reformatted by Knovel to provide easier navigation.

I.29

Index terms

Links

Highway drainage (Continued)

Manning's roughness coefficients

13.9t

Rational method

13.6

Runoff coefficients

13.7t

Sheetflow

13.8

Sheetflow travel time

13.8

Time of concentration

13.7

Roadside ditch design

13.16

Average shear stress

13.20

Bend shear stress

13.20

Distribution of shear stress

13.21f

Manning's coefficients, open channels

13.18t

Manning's coefficients, vegetal retardance

13.19t

Manning's equation

13.17

Maximum shear stress

13.20

Permissible shear stress

13.23

Permissible shear stress, cohesive soils

13.25f

Permissible shear stress, non-cohesive soils

13.24f

Riprap, angle of repose

13.27f

Shear stresses

13.20

Stable channel design

13.22

Superelevation in bends

13.18

Vegetal cover classification

13.18t

Highway structures, bridge scour

13.24t

15.48

Aggradation

15.48

Contraction scour

15.48

Degradation

15.48

Design approach

15.48

Example design calculations

15.53

Live-bed scour at abutments

15.61

Local scour at piers

15.58

15.57

This page has been reformatted by Knovel to provide easier navigation.

15.64

I.30

Index terms

Links

Highway structures, bridge scour (Continued)

Pier shapes
Pier nose shape correction factors

15.60f
15.60

Scour at cylindrical piers

15.59f

Skew adjustment for abutment scour

15.64f

Hydraulic structures, stream stability

15.39

Armoring

15.40

Armoring layer

15.41

Channel drop structures

15.46

Check dams

15.46

Countermeasures for stream instability

15.43

Deflector spurs

15.44

Engineering analysis

15.39

Example calculations

15.47

Flow control structures

15.43f

Geomorphic factors

15.40f

Guide banks
Hydraulic factors
Incipient motion analysis
Location factors

15.44
15.41f
15.39
15.40f

Retarder/deflector spurs

15.44

Retarder spurs

15.44

River training works

15.43

Spur dikes

15.44

Spurs

15.44

HYDRAIN

13.8

Hydraulic capacity

2.1

Hydraulic conductivity

4.2

Intrinsic permeability
Relative hydraulic conductivity
Relative permeability

15.47f

4.2
4.16
4.2

This page has been reformatted by Knovel to provide easier navigation.

I.31

Index terms

Links

Hydraulic conductivity (Continued)

Vadose zone measurement
Hydraulic design procedure
Hydraulic gradeline
Hydraulic jump
Nonrectangular channels
Rectangular channels
Hydraulic jump basins

24.28
1.24f
2.14
3.9
3.10
3.9
20.17

Definition sketch

20.17f

Design procedure

20.17

Example design

20.19

Hydraulic mean depth

21.5

Hydraulic radius
Hydraulic scale modeling

3.3
21.8

Kinematic similarity

21.8

Hydrology, for highway drainage

Avoidance of hydroplaning method

24.26
13.6
13.12

Hydroplaning design rainfall intensity

13.13

Vehicle speed to initiate hydroplaning

13.12

Design rainfall intensity

13.6

Avoidance of hydroplaning

13.6

Driver vision-impairment method

13.6

Rational method

13.6

Driver vision-impairment method

13.13

Gutter flow velocity

13.10

Intensity-duration-frequency curves

13.8

Overland sheet flow

13.9

Manning's roughness coefficients

Rational method

21.7

21.8

Geometric similarity
Hydro compaction

22.18

13.12

13.9t
13.6

This page has been reformatted by Knovel to provide easier navigation.

I.32

Index terms

Links

Hydrology, for highway drainage (Continued)

Runoff coefficients

13.7t

Sheetflow

13.8

Sheetflow travel time

13.8

Time of concentration

13.7

Hydroplaning

13.2

I
Impact-type stilling basin, pipes and channel outlets
Basic design
Design guidelines

18.30
18.30f
18.30

Energy losses

18.31f

Width selection

18.31f

Infiltration

24.7

Capillary fringe

24.13

Clogging

24.10

Hydraulic loading rates

24.2

Rates

24.7

Infiltration basins

24.11

Clogging of

24.10

Low-infiltration recharge system

24.26

Micro-clogging

24.25

Pilot basin

24.27

Wash-out wash-in

24.25

Infiltrometers

24.8

Buffered

24.8

Double ring

24.8

Single ring

24.8f

In-situ air sparging (IAS)

23.55

Air injection wells

23.57

Biosparging

23.58

This page has been reformatted by Knovel to provide easier navigation.

I.33

Index terms

Links

In-situ air sparging (IAS (Continued)

Blowers

23.58

Compressors

23.58

Design parameters

23.57t

Design principles

23.56

Pilot tests

23.58

Process principles

23.56

Interception wells
Intrinsic soil permeability

4.10
4.2

J
Jet

5.28

Jet hydraulics, see plume hydraulics

5.28

Prime factors
Buoyancy flux

5.28
5.28

Plane jet

5.29

Simple jets

5.29

Mass flux

5.28

Momentum flux

5.28

Zone of established flow

5.29

Turbulent jet properties

5.30t

Joukowsky equation

12.2

Joukousky head rise

12.3

K
Kariz

1.7

Keulegan's resistance relation for rough flow

6.4

Kinematic wave modeling

Overland flow

14.64

14.99

14.64

Equilibrium peak discharge time

14.65

Linear approximation

14.64

This page has been reformatted by Knovel to provide easier navigation.

I.34

Index terms

Links

Kinematic wave modeling (Continued)

Nonlinear approximation

14.64

Overland surface length

14.65

Storm sewer flow models

14.99

ILSD-B2, 14.100t

14.101

ILSD-B3, 14.100t

14.103

SWMM-TRANSPORT, 14.100t

14.102

USGS, 14.100t

14.101

WASSP-SIM, 14.100t

14.103

Kinetic energy correction factor

3.3t

Kinetic similarity

21.8

L
Labyrinth spillway

17.30

Chute slope

17.30

Design procedure

17.30

Nappe oscillation

17.30

Subatmospheric pressures

17.30

Laminar flow

2.16

Liquid column separation

12.2

Logarithmic law of the wall

6.3

M
Macrodispersive flux

4.5

Dispersive mass flux

4.6

Equation for

4.6

Longitudinal macrodispersivity

4.6

Manifold
Manning's conveyance

8.28
19.45

This page has been reformatted by Knovel to provide easier navigation.

I.35

Index terms
Manning's equation

Gutter flow
Manning's resistance coefficient

Links
3.11

3.15

10.11

13.10
16.4

13.14
21.7

14.5

13.14
3.11

Composite sections

16.39

Compound channels

3.15

Equation for overland flow

Equivalent roughness parameter

3.13
19.35t

Floodplains

19.34t
19.36t

Minor natural streams

19.35t

Open channels

13.18t

Overland sheet flow

13.9t

Pipes and channels

13.11t

Street and pavement gutters

13.15t
8.15t

Vegetal degree of retardance

13.19t

Vegetation

16.21

Manning-Strickler power form

Meandering channel, slowly
Mechanical dispersion

3.14t

8.1t

Major natural streams

Tunnels

16.39

14.62

Excavated channels
Linings

19.34

16.22

6.4
5.10
4.5

Coefficient of longitudinal-hydrodynamic dispersion

4.5

Formation-scale dispersion

4.5

Longitudinal dispersivity

4.5

Macrodispersive flux

4.5

Spreading

4.5

Transverse dispersivity

4.5

Mixing, see dispersion

This page has been reformatted by Knovel to provide easier navigation.

I.36

Index terms
Molecular diffusion

Links
4.4

Effective diffusion coefficient

4.5

Fisk's law

4.4

In porous medium

4.5

Millington-Quirk equation

4.5

Momentum correction coefficient

3.3t

Momentum equation

2.8

Monad kinetics

4.7

Montuori number
Morning glory spillway
Circular crest coefficients

3.28
17.22
17.24f

Coordinates

17.26

Design procedure

17.22

Flow conditions

17.23f

Nappe-shaped profile elements

17.24f

Throat control
Muncipal water systems

17.25f

17.22
11.1

Muskingum-Cunge routing

14.104

Muskingum routing formula

14.103

N
Net positive suction head

10.20

Available

10.20

Cavitation

10.22

Criteria from pump tests

10.22f

Incipient cavitation

10.23

Major considerations

10.22

Required

10.21

Safety factor considerations

10.22

Suspress visible cavitation

10.23f

NETWORK

19.26

This page has been reformatted by Knovel to provide easier navigation.

I.37

Index terms

Links

Newton-Rhapson method

9.8

Newton's second law

2.8

Nikuradse equivalent roughness size

6.5t

Non-Newtonian flow

22.80

Apparent viscosity

22.80

Critical velocities

22.81

Flow in suspension

22.81

Headloss computation

22.85

Examples
Turbulent flow

22.85
22.88f

Plastic flow

22.81

Plastic flow/laminar stage

22.82

Plastic flow/turbulent stage

22.82

Reynold's number

22.81

Rigidity

22.81

Specific gravity of sludge

22.83f

Suspension/laminar stage

22.82

Suspension/turbulent stage

22.82

Yield stress

22.81

Sewage sludges

22.84f

O
Open-channel flow computer programs

19.25

DAMBRK

19.26

FESWMS-2DH

19.25

FLDWAV

19.26

HEC-2

19.25

HEC-RAS

19.25

NETWORK

19.26

RMA-2

19.25

SWMM

19.26

This page has been reformatted by Knovel to provide easier navigation.

I.38

Index terms

Links

Open-channel flow computer programs (Continued)

TABS-MD

19.26

UNET

19.26

WSPRO

19.25

WSP-2

19.25

Open channel flow concepts

3.1

Alternate depths

3.5

Bed shear stress

6.2

Bernoulli equation
Boundary shear stress
Mean
Bousinesq momentum flux correction coefficient
Celerity

14.8
16.2
16.24

6.4
16.4
14.19

Critical slope

3.1

Critical state

3.1

Effective roughness height

Energy gradeline
Exner equation
Flow area

19.34

14.10

Chezy's law

Eddy loss

16.7

14.5
3.2

Courant criterion

16.4f

16.4

Channel section geometric properties

Conveyance factor

6.39

3.23
16.21
3.4
6.27
3.1

Flow resistance

16.3

Friction slope

3.22

14.5

3.1

21.4

Froude number
Full-dynamic wave equation

14.6

Gradually varied flow

3.16

Hydraulic depth

3.3

21.4

This page has been reformatted by Knovel to provide easier navigation.

I.39

Index terms

Links

Open channel flow concepts (Continued)

Hydraulic gradeline

3.4

Hydraulic jumps

3.9

Hydraulic radius

3.3

Keulagan's resistance relation for rough flow

6.4

Kinematic-wave approximation

14.6

Kinetic energy correction factor

3.3t

Logarithmic law of the wall

6.3

Manning's law

6.5

Manning's roughness coefficients

Manning-Strickler power form

19.34t
21.6

Momentum correction coefficient

3.3t

Normal depth

21.6

Permissible shear stress

16.2
14.6

Saint-Venant equations

14.6

Shield's diagram
Shield's regime diagram

16.7

16.24

6.19
6.25f
3.3

Specific momentum

3.4

Stage

3.4

Steady flow

3.4

Subcritical flow

16.3

Supercritical flow

16.3

Superelevation

3.4

Surcharge flow

14.8

Tractive force

19.36t

6.2

Specific energy

Top width

19.35t

3.3

Quasi-steady dynamic wave equation

Shear velocity

21.7

6.4

Modular limit

Prismatic channel

21.6

21.4

3.4
16.2

16.7

This page has been reformatted by Knovel to provide easier navigation.

16.24

I.40

Index terms

Links

Open channel flow concepts (Continued)

Turbidity currents

6.97

Uniform flow

3.4

3.11

Unsteady flow

3.4

14.4

Vegetative flow retardance

Wetted perimeter
Optimization

16.5
3.4
1.24

Constraints

1.30

Dynamic programming

1.30

Feasible solution

1.30

Linear programming

1.30

Nonlinear programming

1.30

Objective function

1.30

Orifice plate

21.12

Orifice spillway

17.20

Proportions of

17.20

Overfall spillways

17.13

Bucket exit angle

Diffusing plunging jet

17.15f
17.13

Hydraulic characteristics

17.13

Plunging jet characteristics

17.16t

Plunge pools

17.13
17.14f

Scour hole

17.13

Trajectory

17.13

Overflow spillways
Abutment contraction coefficients
Crest pressures

17.21f

17.13

Flip bucket radius

Pressures for flip-buckets

1.30

17.1
17.2
17.6f

Discharge coefficients

17.3

Hydraulic characteristics

17.1

17.7f

This page has been reformatted by Knovel to provide easier navigation.

I.41

Index terms

Links

Overflow spillways (Continued)

Nappe-shaped crest profiles

17.8f

Pier contraction coefficients

17.2

High-gated overflow crests

Overland flow hydraulics
Kinematic wave modeling

17.5f
14.61
14.64

Equilibrium peak discharge time

14.65

Linear approximation

14.64

Nonlinear approximation

14.64

Overland surface length

14.65

Manning's n, equation for

Steady open-channel plane flow

14.62
14.61f

Time of concentration

14.66

Akan's equation

14.66

Izzard's equation

14.66

Morgali and Linsley's equation

14.66

Overland texture factor

14.68t

Yen and Chow's equation

14.67

P
Penstocks

8.21

Mechanical starting time

8.22

Penstock branches

8.24

Thickness

8.22

Water column starting time

8.23

Penstock manifolds
Perfect gas law
Permeability

8.27f
4.17
4.2

Intrinsic

4.2

Relative

4.2

Piezometric head

8.25f

4.1

9.1

This page has been reformatted by Knovel to provide easier navigation.

8.26f

I.42

Index terms
Pipe flow

Links
2.18

Extended period simulation

9.12

Row in branching systems

9.7

Flow in parallel

2.18f

9.5

Flow in series

2.18f

9.5

Local losses

2.21

9.5

K factors

2.23t

Networks

9.11

Gradient method

9.20

Hardy Cross method

9.11

Linear theory method

9.17

Newton-Rhapson method

9.18

Unsteady flow

9.23

Pipeline models,
Modeling approach

2.4

Numerical models

2.2

Operational models

2.4

Planning models

2.4

Pipe material characteristics

Acoustic velocity

12.3
12.4f

Hoop stress

12.3

Physical properties

12.3

Poisson ratio

12.3

Young's modulous

12.3

Pipe systems
Branching systems
Networks

9.5
9.7
9.11

Parallel systems

9.5

Series systems

9.5

This page has been reformatted by Knovel to provide easier navigation.

22.6

I.43

Index terms
Plume

Links
5.28

Properties of

5.30t

Simple plumes

5.29

Plume hydraulics

5.28

Pore velocity (average)

4.2

Power channel (canal)

8.1

Power intakes

8.4

Gate arrangement

8.13f

Horizontal intake

8.4

8.5f

Inclined intake

8.4

8.7f

Vertical intake

8.4

8.6f

Vortex formation

8.14

Pressure regulating valves

9.4

Pressurized pipeline

2.1

Pump definitions

10.2

Allowable operating range

10.2

Atmospheric head

10.2

Centrifugal point

10.2

Condition points

10.2

Best efficiency points

10.2

Normal condition point

10.2

Rated condition point

10.2

Specified point

10.2

Cutoff head

9.4

Datum

10.2

Elevation head

10.2

Friction head

10.2

Gauge head

10.2

Head

10.2

High-energy pump

10.2

Impeller balancing

10.2

This page has been reformatted by Knovel to provide easier navigation.

I.44

Index terms

Links

Pump definitions (Continued)

Single-plane balancing

10.2

Static balancing

10.2

Two-phase balancing

10.3

Overall efficiency

10.2

Power

10.3

Brake horsepower

10.3

Pump efficiency

10.4

Pump input power

10.3

Pump output power

10.3

Water horsepower

10.3

Pump pressures

10.4

Field test pressure

10.4

Maximum allowable casing working pressure

10.4

Maximum suction pressure

10.4

Shutoff

10.4

Speed

10.4

Suction conditions

10.4

Maximum suction pressure

10.4

Net positive suction head available

10.4

Net positive suction head required

10.5

Static suction lift

10.5

Total discharge head

10.5

Total head

10.5

Total suction head, closed suction tests

10.4

Total suction head, open suction

10.4

Total dynamic head

10.4

Velocity head

10.4

Pump characteristics

12.9

Abnormal (four quadrant) characteristics

12.11

Corrected pump curves

10.24

This page has been reformatted by Knovel to provide easier navigation.

I.45

Index terms

Links

Pump characteristics (Continued)

Example calculation

10.24

Minor losses

10.24t

Design point

12.11

Duty

12.11

Head and torque characteristics, radial flow pump,

Four quadrant

12.17f

Negative rotation

12.16f

Positive rotation

12.15f

Suter diagram

12.17f

Karman-knapp circle diagram

12.14

Nameplate

12.11

Rated conditions

12.11

Pump hydraulics
Hydraulics of valves
Impeller wear, effect of
Operation

10.9
10.13
10.16f
10.14

Affinity laws

10.14

In parallel

10.14

Homologous laws

10.14

Pipeline hydraulics

10.9

Colebrook-White equation

10.12

Darcy-Weisbach equation

10.12

Hazen-Williams equation

10.9

Manning's equation

10.11

Roughness factors

10.12

Pump performance curves

10.18f

9.3

10.8f

10.9

10.14

10.15f

Pump specific speed,

Equation for

10.16

Suction specific speed

10.18

System curves

10.9

This page has been reformatted by Knovel to provide easier navigation.

I.46

Index terms

Links

Pump hydraulics (Continued)

Variable speed pumps
Typical discharge curves

10.14
10.18f

Pump performance
Homologous (affinity) laws

10.14

Dynamic similarity

12.10

Kinematic similarity

12.10

Power coefficient

12.1

Torque coefficient

12.11

Performance parameters

12.9

Flow coefficient

12.10

Flowrate

12.10

Head coefficient

12.10

Impeller diameter

12.9

Power coefficient

12.10

Rotational speed

12.9

Total dynamic head

12.10

12.10

Pump operation for transients

12.11

Abnormal (four quadrant)

12.11

Zones of possible pump operation

12.13f

Pump selection, hydraulic considerations

10.28

Flow range of centrifugal pumps

10.28

Steps

10.30

Pump specific speed,

Equation for

10.16

Suction specific speed

10.18

Pump standards
Pump station design
Hydraulic transients

10.1
10.32
10.37

Pipe material selection

10.37

Valve selection

10.37

This page has been reformatted by Knovel to provide easier navigation.

I.47

Index terms

Links

Pump station design (Continued)

Operating problems

10.34

Piping

10.34

Piping design

10.35

Criteria

10.36

Pressure design

10.35

Vacuum condition

10.36

Pump operating ranges

10.32t

10.32f

1.7

1.8f

2.12

2.31

Pump surge protection devices,

Pump surge control devices,

Q
Qanat
Quasi-steady flow

R
Radiation

5.3

Long-wave back radiation

5.4

Net atmospheric long-wave radiation

5.4

Net shortwave radiation

5.3

Shortwave sollar radiation

5.4

Stephan-Boltzman law

5.4

Rapidly varied unsteady flow

3.26

Montuori number

3.28

Surges

3.26

Vedernikov number

3.28

Rational method

13.6

Assumptions

13.7

Intensity-duration-frequency curves

13.8

Kinematic wave equation

13.8

Rational formula

13.7

This page has been reformatted by Knovel to provide easier navigation.

I.48

Index terms

Links

Rational method (Continued)

Runoff coefficients

13.7t

Sheetflow

13.8

Sheetflow travel time

13.8

Weighted runoff coefficients

13.7

13.8t

Recharge basins,
Artificial

24.2

Induced

24.14

With clogging

24.11f

With high groundwater

24.14f

Relative humidity

5.6

Relative permeability

4.4

Brooks-Corey model

4.4

Mualem model

4.4

Van Genuchten model

4.4

Relative transport

4.6

Degradation

4.7

Equilibrium partitioning

4.6

Monad kinetics

4.7

Partitioning of chemicals

4.6

Sorption

4.6

Reliability
Anticipated load

1.31
7.3

Definition of

7.19

Design load

7.3

Equation for

7.2

Failure probability

7.2

Load-resistance interference
Diagram of
Performance functions
Probabilistic approaches

7.1

7.21
7.23f
7.19
7.3

This page has been reformatted by Knovel to provide easier navigation.

I.49

Index terms

Links

Reliability (Continued)
Reliability formulas

7.22t

Reliability index

7.19

Reliability measures

7.3

Safety factor

7.3

Safety margin

7.3

Resistance
Reliability analysis methods

7.2
7.19

Advanced first-order second-moment method

7.25

Direct integration method

7.21

Mean-value first-order second-moment method

7.24

Monte Carlo simulation methods

7.38

Reservoir sedimentation hydraulics

6.89

Einstein dimensionless bed load transport

6.90

Exner equation

6.94

Governing equations

6.92

Normal flow computations

6.91

Suspended-load sediment routing

6.90

Theoretical considerations

6.89

Retention storage, see detention storage

Reynold's number
Non-Newtonian flow

2.14
22.81

Rigid column analysis

12.7

Risk

1.31

Failure probability
Risk function accounting
Risk-based design
Annual expected flood damage cost

21.4

7.2
7.38
7.38
7.47

Conventional approach

7.47

Hydrologic parameter uncertainty

7.49

Hydrologic inherent/parameter uncertainties

7.49

This page has been reformatted by Knovel to provide easier navigation.

22.80

I.50

Index terms

Links

Risk-based design (Continued)

Hydraulic uncertainty

7.48

Conventional risk-based design

7.46

Optimal risk-based design problem

7.45

Return period design

7.45

Tangible costs

7.46

Total annual expected cost

7.44

U. S. Army Corps of Engineers risk-based analysis

procedure

7.50

Risk-reliability evaluation

1.32

Composite risk

1.33

Failure

1.33

Load

1.32

Optimal risk-based design

1.34

Resistance

1.32

Risk assessment

1.33

Risk-based design

1.33

Safety factor

1.33

Roman aqueducts

1.20f

7.2

1.22f

S
Sadd-el-Kafara dam

1.13

Saint-Venant equations

3.22

Safety factor

1.33

Safety margin

7.3

Safe Water Drinking Act

9.26

Saturated vapor pressure

5.6

Sedimentation computer models

7.3

15.68

BRI-STARS

15.69

GSTARS

15.68

HEC-6

15.68

This page has been reformatted by Knovel to provide easier navigation.

I.51

Index terms

Links

Sedimentation computer models (Continued)

HIVEL-RD

15.69

HY9

15.69

RMA

15.69

SMS

15.69

Sediment bedforms

6.34

Antidunes

6.36

Bars

6.36

Bedform charts

6.42

Bedform classification

6.41

Bedform criteria

6.41

Bedform effects on flow classification

6.37

Bedform height and steepness

6.45

Bedform predictor

6.40

Bedform progression

6.37

Bedform types

6.39

Bed shear stress

6.39f

Dimensionless characteristics of bedforms

6.39

Dunes

6.34

Flow resistance due to bedforms

6.44

6.52f

Form drag

6.47

Ripples

6.36

River stage

6.46

Schematic of bedforms

6.35

Shear stress partition

6.47

Einstein partition

6.47

Example of Nelson-Smith partition

6.50

Nelson-Smith partition

6.49

Skin friction

6.47

Stage-discharge relation

6.51

Brownlie method

6.53

This page has been reformatted by Knovel to provide easier navigation.

I.52

Index terms

Links

Sediment bedforms (Continued)

Einstein-Barbarossa method

6.51

Englund-Hansen method

6.52

Sediment bedload transport

6.26

ACRONYM

6.31

Bedload transport functions

6.26

Bedload transport relations

6.27

Ashida and Michiue

6.27

Einstein

6.28

Englund and Fredsoe

6.28

Fernandez Luque and van Beek

6.28

Meyer-Peter and Muller

6.28

Parker

6.28

Wilson

6.28

Yalin

6.28

Deposition from suspension

6.26

Dimensionless Shield's stress

6.30

Einstein number

6.27

Erosion into suspension

6.26

Exner equation

6.27

Mixtures

6.27

Sediment motion threshold condition

Angle of repose
Forces on particle
Granular sedimentation

6.29f

6.16
6.20
6.17f
6.17

Bank

6.20

Sloping bed

6.20

Streambed

6.17

Mobility of grain

6.16

Sediment mixtures

6.21

Shield's diagram

6.19

This page has been reformatted by Knovel to provide easier navigation.

I.53

Index terms
Sediment properties

Links
6.6

Armor

6.14

Fall velocity

6.12

Fall velocity diagram

6.13f

Geometric mean diameter

6.10

Geometric standard deviation

6.10

Porosity

6.11

Rock types

6.6

Sediment grade scale

6.9t

Sedimentation scale

6.7

Shape

6.12

Size

6.7

Size distribution

6.8

6.14f

6.16f
Specific gravity of sediments

6.7

Subpavement

6.14

Terminal fall velocity

6.12

Sediment suspended load

6.55

Boundary conditions

6.55

Depth-averaged volume suspended-sediment

concentration

6.60

Eddy diffusivity

6.57

Entrainment relation for sediment mixture

6.66

Equilibrium suspension

6.57

6.7t

Example computation of sediment load and rating

curve
Garcia-parker function

6.68
6.66f

Kinematic eddy viscosity

6.58

Mass conservation

6.55

Nominal "near bed" elevation

6.59

Prandtl analogy

6.57

Rousean distribution of suspended sediment

6.59

This page has been reformatted by Knovel to provide easier navigation.

6.15f

I.54

Index terms

Links

Sediment suspended load (Continued)

Rousean relation
Sediment entrainment and deposition

6.58
6.57f

Sediment entrainment relations

6.63

Vertical averaged concentration

6.60

Sediment total bed material load

6.64t

6.65t

6.74

Ackers-White relation

6.79

6.85f

6.86f

Brownlie relations

6.78

6.83f

6.84f

6.81f

6.82f

6.87f

6.88

Gradually varied flow

6.79

Hydraulic resistance

6.78

Normal flow

6.79

Sediment transport

6.78

Dimensionless relations

6.74

Englund-Hansen relation

6.77

Gradually varied flow

6.77

Hydraulic resistance

6.77

Normal flow

6.77

Sediment transport

6.77

Yang relation
Sediment transport

6.80
6.22

Critical condition for suspension

6.24

Shield's regime diagram

6.24

Transport modes

6.22

Sediment transport hydraulics

6.2

Bed load layer

6.6

Bed shear stress

6.2

Channel resistance equations

6.4

Chezy's law

6.4

Keulegan's resistance relation for rough flow

6.4

Manning's law

6.5

Manning-Strickler power form

6.4

This page has been reformatted by Knovel to provide easier navigation.

I.55

Index terms

Links

Sediment transport hydraulics (Continued)

Fixed-bed roughness

6.5

Logarithmic law of the wall

6.3

Logarithmic velocity profile

6.3

Movable-bed roughness

6.5

Nikuradse equivalent roughness

6.5t

Shear velocity

6.2

Velocity distribution

6.2

See page trenches

24.30

Shaduf

1.4

1.6f

Shear

5.9

13.20

Shear stress
Bed shear stress
Boundary shear stress

13.20
6.2
16.29

Average shear stress

16.29

Bend adjustment factor

16.32

Local boundary shear adjustments

16.29

Mean shear stress

16.29

Critical bed shear stress

15.50

Shear velocity
Sheetflow

6.2
3.16

Shield's diagram

6.19f

Shield's regime diagram

6.25f

Side-channel spillways

17.18

Hydraulic characteristics
Siphon spillway

13.2

17.18
17.32

Air-regulated siphon

17.34

Blackwater siphon

17.32

Design procedure

17.32

Standard siphon

17.32

Soil aquifer treatment

24.34

17.33f

This page has been reformatted by Knovel to provide easier navigation.

13.8

I.56

Index terms
Soil matrix suction
Soil vapor extraction

Links
4.3
23.34

Efficiency

23.39

Equation for flow rates

23.35

Example site data for design

23.50

Example design worksheet

23.49

Extraction pump/blower

23.45

Feasibility assessment

23.35

Injection blower

23.45

Instrumentation

23.46

Macroscopic mass transfer limiting scenarios

23.39

Manifolding

23.46

Maximum achievable mass removal rates

23.36

Minimum total vapor flowrate

23.38

Pilot tests

23.40

Treatment system schematic

23.34

System design

23.42

Vapor treatment

23.45

Well construction

23.43

Soil-water retention curve

4.3

Specific energy

3.3

Specific momentum

3.9

Spillway aeration ramps

17.40

Cavitation index

17.40

Cavitation characteristics

17.43

Design procedure

17.44

Spillway basins

18.24

Baffled apron

18.33

Pier heights

23.44f

17.41f

18.3

Baffle-block basin
Design procedure

23.41f

18.34f

18.34
18.35f

This page has been reformatted by Knovel to provide easier navigation.

I.57

Index terms

Links

Spillway basins (Continued)

Baffle-sill basin

18.24

Basin I

18.7

Basin II design guidelines

18.3

18.8f

Basin III design guidelines

18.9

18.10f

Basin IV design guidelines

18.13

18.13f

Design examples

18.4

Expanding stilling basins

18.24

Impact-type for pipes and channel outlets

18.30

Modified basin IV design guidelines

18.17

Negative-step basin

18.24

Positive-step basin

18.24

Riprap

18.35

Saint Anthony Falls basin design guidelines

18.11

18.12f

Sloping aprons

18.19

18.20f

Design guidelines

18.19

Examples

18.22f

Jump lengths

18.21f

Structural arrangements

18.25

Submerged deflector buckets

18.40

Velocities entering
Spillway chutes

18.17f

18.23f

18.6
17.38

Side-wall freeboard

17.38

Smooth chutes

17.38

Stepped chutes

17.38

Design procedure

17.38

Headless

17.40

Spillway sample design

17.49

17.39f

Spillway types,
Labyrinth

17.30

Morning glory

17.22

This page has been reformatted by Knovel to provide easier navigation.

I.58

Index terms

Links

Spillway types (Continued)

Orifice

17.20

Overall

17.13

Side-channel

17.18

Siphon

17.32

Air-regulated

17.34

Standard siphon

17.32

Spillway chute

17.38

Smooth chute

17.38

Stepped chute

17.38

Tunnel spillway

17.36

Flat-tunnel

17.36

Flip-bucket

17.38

Inclined tunnel

17.36

Steady flow
Quasi-steady flow
Stephan-Boltzman law
Storm sewer design

2.11

2.13

2.12

2.31

5.4
14.22

Constraints and assumptions

14.22

Example design

14.24

Rational formula

14.24

Rational method

14.22

Storm sewer hydraulic simulation models

14.91

Classification of models

14.91

Dynamic-wave models

14.93

CARESDAS

14.97

DAGVL-A

14.94

HYDROWORKS

14.97

ISS

14.96

MOUSE

14.95

SURDYN

14.97

This page has been reformatted by Knovel to provide easier navigation.

I.59

Index terms

Links

Storm sewer hydraulic simulation models (Continued)

Summary of properties

14.94t

SWMM-EXTRAN

14.93

UNSTDY

14.97

Noninertia models

14.98t

DAGVL-DIFF

14.98T

DHI

14.99

HVM

14.98T

MOUSE

14.98T

NISN

14.98T

Nonlinear kinematic wave models

14.99

ILSD-B2, 14.100t

14.101

ILSD-B3, 14.100t

14.103

SWMM-TRANSPORT, 14.100t

14.102

USGS, 14.100t

14.101

WASSP-SIM, 14.100t

14.103

Storm sewer junction hydraulics

Control volume
Hydraulic equations
Junction benching

14.30
14.33f
14.32
14.31f

Junction classification

14.30

Point junctions

14.44

Storage junctions

14.42

Flow equations

14.99

14.43t

Three-way sewer junctions

14.34

Experimental studies

14.35t

14.38t

Headloss coefficients

14.36f

14.37f

Loss coefficient equations

Two-way sewer junction

14.34
14.37

Benching

14.42f

Experimental studies

14.40t

This page has been reformatted by Knovel to provide easier navigation.

I.60

Index terms

Links

Storm sewer junction hydraulics (Continued)

Headloss coefficients

14.41

Storm sewer network hydraulics

14.44

Bottleneck determination

14.56

Channel system capacity determination

14.56

Hydraulic performance graph

14.50

Major characteristics

14.50

Procedure

14.52

Hydraulic simulation models

14.91

Junction continuity equation

14.45

Network solution methods

14.47

Cascade method

14.47

One-sweep explicit method

14.47

Overlapping segment method

14.48

Simultaneous solution method

14.48

Stratification of reservoirs
Annual cycle

5.7
5.14

Effects and causes

5.7

Epiliminion

5.7

Hypoliminio

5.7

Metaliminion

5.12

Thermocline

5.7

Stream system types

5.12

19.23

Dendritic channel systems

19.23

First-order streams

19.23

Higher-order streams

19.23

Network channel systems

19.23

19.25f

Simple channels

19.23

19.24f

Subcritical flow

13.17

Subcritical state

3.1

Surge

19.24f

2.12
This page has been reformatted by Knovel to provide easier navigation.

I.61

Index terms
Surge tanks

Links
8.17

Hydraulic stability

8.17

Maximum down surge

8.19

Maximum upsurge

8.19

Thoma formula

8.17

Surging

12.2

Supercritical flow

13.17

Supercritical state

3.1

Superelevation
Swape

3.16

8.2

1.4

T
TABS-MD

19.26

Tailrace channels

8.29

Tail-tunnels

8.28

Manifold

8.28

Outlet structures

8.29

Surge tanks

8.28

Tepidarium

8.32f

1.17

Thermal budget

5.3

Thoma formula

8.17

Tractive force

2.22

Time of concentration

13.7

Transients

2.11

Bulk modulus of water

2.36

Causes

2.34

Joukowsky relations

2.39

Kinetic energy

2.35

Physical nature

2.35

Wavespeed

2.38

Valves

2.40

This page has been reformatted by Knovel to provide easier navigation.

13.18

I.62

Index terms
Transitions
Local loss coefficients

Links
2.23t
2.23

Transmissivity

4.8

Trashracks

8.9

Tower structures

8.9

Multilevel intake tower

Tunnels

8.10f
8.14

Manning's n values
Tunnel spillway

8.15
17.36

Entrance structures

17.36

Flat tunnel section

17.36

Flip bucket

17.38

Inclined tunnels

17.36

Turbidity currents
Bed friction coefficient

6.97
6.102f

Densimetric Froude number at plunging

6.100

Downslope through a quiescent body of water

6.99f

Internal hydraulic jump

6.104

Lake Superior

6.105

Plunging flow

6.100

Sediment entrainment coefficient

6.103f

Shape factor

6.103f

Water entrainment coefficient

6.101f

Turbulent diffusion (mixing)

Transverse turbulent diffusion
Vertical turbulent diffusion
Turbulent flow

5.9
5.10
5.9
2.15

This page has been reformatted by Knovel to provide easier navigation.

I.63

Index terms

Links

U
Uncertainties

1.31

Data uncertainties

1.31

Economic uncertainties

1.31

7.1

Hydraulic uncertainties

1.32

7.1

Hydrologic uncertainties

1.32

7.1

Model uncertainties

1.31

7.1

Parameter

1.31

Structural

1.31

Natural uncertainties

1.31

Structural uncertainties

1.31

Uncertainty analysis techniques

Analytical techniques

7.4
7.4

Exponential transforms

7.4

Fourier transforms

7.4

Mellin transform

7.6

Approximate techniques

7.9

First order variance estimation method

Hair's probabilistic point estimation method

7.1

7.7t

7.9
7.16

Rosenblueth's probabilistic point estimation

method
UNET

7.12
19.26

Uniform flow

3.4

Chezy equation

3.11

Manning's equation

3.11

Manning's resistance coefficient

3.11

Sheetflow

3.16

Special cases

3.15

Superelevation

3.16

Universal soil loss equation (USLE)

3.11

15.39

This page has been reformatted by Knovel to provide easier navigation.

I.64

Index terms
Unsteady flow
Saint-Venant equations

Links
2.11

2.32

3.25

Urban drainage hydraulics,

Drainage channel hydraulics

14.3

Boussinesq momentum flux

correction coefficient

14.5

Darcy-Weisbach formula

14.5

Friction slope

14.5

Manning's formula

14.5

Preissman slot

14.8

Sewer pipe flow geometry

14.6

Surcharge flow

14.8
14.11

Unsteady flow

14.4

Storm sewer pipe flow

14.13

Boundary conditions

14.20

Classification of flow

14.13f

Courant criteria

14.19

Discretization

14.19

Entrance flow types

14.14f

Exit flow types

14.15f

Initial conditions

14.20

Moving water surface discontinuity

14.21

Pipe flow conditioning

14.6f

Rating curves

14.18f

Single sewer

14.13

Time variation of flow

14.16

U.S.G.S. National Aerial Photography Project

14.8

14.10

Pressurized transient pipe flow

Surge speed

14.8

14.21t

19.7

This page has been reformatted by Knovel to provide easier navigation.

3.24

I.65

Index terms

Links

V
Valves

11.11

Air release valves

11.47

Air vacuum valves

11.47

Altitude valves

11.47

Angle valve

22.4

Ball valves

12.5f

Blow-offs

11.47

Butterfly valves

11.45

Check valves

22.4
12.5f

22.4

Lift check

22.4

Swing check

22.4

Spring check

22.4
22.4

Control valves

11.46

Flow control valves

11.47

12.8

Gate valves

11.44

12.5f

Geometric characteristics

12.5

12.8f

Cross-sections of

12.5f

Globe valve

12.5f

Headlosses

12.6

Equation for

22.4

12.7

Hydraulic characteristics

12.8f

Isolation valves

11.44

Pinch/diaphragm

22.4

Plug valve

22.4

Pressure reducing valves

11.46

Pressure relief valves

11.47

Pressure sustaining valves

11.46

Manning's equation

22.3

22.4

Ball check

Cone valves

22.3

3.11

This page has been reformatted by Knovel to provide easier navigation.

22.3

I.66

Index terms

Links

Valves (Continued)
Manning's resistance coefficient
Needle valve

3.11
12.5f

Throttling valves

22.4

Typical valves and applications

22.4t

Valve closure
Classification of

12.7
12.9t

Valve operation

12.8

Vedernikov number

3.28

Venturi condition
Wastewater treatment plant

2.22f
22.1

Distribution boxes

22.2

Distribution channels

22.2

Flowmeters

22.4

Gates

22.3

Seating head

22.3

Unseating head

22.3

Pipe manifolds
Process flow diagram
Valves

22.4

22.2
22.40
22.3

Butterfly valves

22.3

Check valves

22.4

Gate valves

22.3

Table of valves and applications

22.4

Throttling valves

22.4

W
Wastewater treatment plant hydraulic design
Aeration tanks
Cascade system
Configuration

22.36
22.53
22.77f
22.55

This page has been reformatted by Knovel to provide easier navigation.

I.67

Index terms

Links

Wastewater treatment plant hydraulic design (Continued)

Design example

22.65

Distribution box

22.58

Effluent channel

22.64

Effluent weir

22.64

Inlet baffles

22.64

Inlet channel

22.58

Inlet flow distribution

22.58

Mean cell residence time (MCRT)

22.53

Mixed-liquor suspended solids (MLSS)

22.53

Process configuration

22.53

Return activated sludge (RAS)

22.53

Schematic diagram

22.65

Solids residence time (SRT)

22.53

Waste activated sludge (WAS)

22.53

Bar screens

22.41

Approach channel

22.41

Design example

22.43

Effective bar opening

22.43

Operating headloss

22.43

Schematic diagram

22.44

Bases for design

22.67f

22.45t

22.38

Contact chambers (see mixing chambers)

Granular media filter

22.65

Backwash requirements

22.66

Collection systems

22.65

Design example

22.71

Filter operation

22.66

Grit tanks

22.74t

22.43

Aerated grit chambers

22.43

Design example

22.47

22.49t

This page has been reformatted by Knovel to provide easier navigation.

I.68

Index terms

Links

Wastewater treatment plant hydraulic design (Continued)

Detritus tank

22.43

Effluent weir

22.47

Horizontal flow type

22.43

Hydraulic design parameters

22.45

Hydroclone

22.43

Inlet baffle

22.45

Inlet channel

22.45

Vortex type

22.43

22.49f

Hydraulic profile

22.41

22.42f

Mixing (contact) chambers

22.71t

Design example

22.72

Design parameters

22.71

Diffusers

22.73

Submerged discharge

22.73

Planning phases

22.36

Plant layout

22.39

Process selection

22.38

Pumping, slurry, and chemicals

22.77

Sedimentation tanks

22.47

22.78t

Design example, primary

22.53

22.55t

Design example, secondary

22.53

22.60t

Effluent launder

22.52

Hydraulic design parameters

22.48

Inlet baffle

22.48

Inlet channels

22.48

Inlet flow distribution

22.48

Outlet conditions

22.48

Overflow weir

22.52

Primary tanks

22.48

22.54f

Secondary (final) tank

22.48

22.59f

This page has been reformatted by Knovel to provide easier navigation.

I.69

Index terms
Water budget

Links
5.1

5.2f

Water column separation,

Water density, function of

5.8

Dissolved solids

5.8

Salinity and suspended solids

5.8

Temperature

5.8

Water distribution hydraulics, pipe flow

Extended period simulation

2.18
9.22

Flow in parallel

2.18f

9.5

Flow in series

2.18f

9.5

Local losses

2.21

9.5

Networks

9.11

Gradient method

9.20

Hardy Cross method

9.11

Linear theory method

9.17

Newton-Rhapson method

9.18

Unsteady flow

9.23

Water distribution, modeling

9.31

Application

9.32

Calibration

9.32

Calibration process

9.32

EPANET

9.33

History of

11.9

Model development

11.10

Model selection

9.32

Network representation

9.32

Problem definition

9.32

Software packages

11.10

Verification

9.32

Water quality modeling

9.26

Conservation of energy

11.9

11.11t

9.3

This page has been reformatted by Knovel to provide easier navigation.

I.70

Index terms

Links

Water distribution, modeling (Continued)

Conservation of mass

9.3

Fixed grade node

9.2

Hydraulic head

9.1

Loading condition

9.1

Hardy-Cross method

9.11

Piezometric head

9.1

Pressure regulating valves

9.4

Pump curve

9.4

Water distribution pipeline design,

Pipeline design

11.34

Flexible pipe

11.37

Internal pressures

11.34

Loads

11.34

Rigid pipes

11.36

Thrust restraints

11.38

Preliminary design

11.12

Alignment

11.12

Rights-of-way

11.13

Subsurface conflicts

11.12

Water distribution piping materials

11.13

Asbestos-cement pipe (ACP)

11.31

Ductile iron pipe (DIP)

11.13

High-density polyethylene (HDPE) pipe

11.29

Poly vinyl chloride (PVC) pipe

11.18

Reinforced concrete pressure pipe (RCPP)

11.25

Steel pipe

11.20

Water distribution systems

11.1

Average day demand

11.2

Fire demands

11.7

Maximum day demand

11.2

This page has been reformatted by Knovel to provide easier navigation.

I.71

Index terms

Links

Water distribution systems (Continued)

Peaking coefficients

11.8

Peaking factors

11.2

Peak hour demand

11.2

Planning and design criteria

11.4

Service pressures

11.8

Storage

11.7

Emergency storage

11.7

Fire storage

11.7

Operational storage

11.7

Supply

11.7

Water demands

11.3t

11.9t

11.9t

Water duties

11.2

11.5t

Water hammer

2.12

12.1

Case studies

12.24

Definition of

12.2

Entrapped air

12.2

Joukowsky equation

12.2

Joukowski head rise

2.39

Liquid column separation

12.2

Mitigation of

12.1

Time constants

12.24

Elastic time constant

12.24

Flow time constant

12.24

Pump and motor inertia time constant

12.24

Surge tank oscillation inelastic time constant

12.24

Water properties

12.3

2.5t

Water reuse,
Constructed aquifers

24.39

Potable reuse

24.40

Restricted irrigation

24.33

This page has been reformatted by Knovel to provide easier navigation.

I.72

Index terms

Links

Water reuse (Continued)

Role of recharge

24.33

Water treatment plant

22.1

Distribution boxes

22.2

Distribution channels

22.2

Flowmeters

22.4

Gates

22.3

Seating head

22.3

Unseating head

22.3

Pipe manifolds
Process flow diagram
Valves

22.2
22.40
22.3

Butterfly valves

22.3

Check valves

22.4

Gate valves

22.3

Table of valves and applications

22.4

Throttling valves

22.4

Water treatment plant hydraulic design

22.10

Bases for design

22.12

Design flows

22.12

Annual average demand

22.13

Maximum demand

22.13

Minimum demand

22.13

Hydraulic profile

22.3f

Hydraulic treatment capacity

22.13

Plant hydraulic design

22.13

Plant layout

22.12

Plant operating modes

22.15

Plant siting

22.12

Process selection considerations

22.10

Head available

22.18

22.10

This page has been reformatted by Knovel to provide easier navigation.

I.73

Index terms

Links

Water treatment plant hydraulic design (Continued)

Unit process head requirements

22.10t

Rated treatment capacity

22.13

Treatment process bases for design

22.13

Water treatment plant process hydraulics

22.15

Coagulation

22.19

Design example

22.21

Flocculation

22.20

Flocculation/sedimentation basin

22.16f

Hydraulic mixing

22.20

Pneumatic mixing

22.19

Rapid mixing

22.19

Sedimentation

22.20

Filtration

22.26

Example hydraulic calculation

22.28t

Hydraulic design parameters

22.26

Process criteria

22.26

Membranes
Biological fouling

22.32t

22.34
22.35

Configuration

22.35f

Hyperfiltration

22.34

Hyperfilters

22.34

Microfilters

22.34

Modified fouling index (MFI)

22.34

Nanofilters

22.34

Reverse osmosis

22.34

Scaling control

22.34

Silt density index (SDI)

22.34

Transmembrane pressure (TMP)

22.36

Ultrafilters

22.34

Water use

22.14t

11.5t

This page has been reformatted by Knovel to provide easier navigation.

I.74

Index terms
Wavespeed
Wave suppressors
Raft-type

Links
2.37
18.26
18.27f

Underpass-type

18.28

Weisbach coefficient

14.66f

Well duplets

4.12

Wind shear velocity

5.14

WSP-2

19.35

WSPRO

15.39

18.29f

19.25

Z
Zangar's equation

24.30

This page has been reformatted by Knovel to provide easier navigation.