Vous êtes sur la page 1sur 292


Concrete Dams

Kenneth D. Hansen, P.E.

Denver, Colorado

William G. Reinhardt
Westfield, New Jersey

McGraw-Hill, Inc.
New York St. Louis San Francisco Auckland Bogoti
Caracas Hamburg Liabon London Madrid
Mexico Milan Montreal New Delhi Paris
San Juan S&o Paul0 Singapore
Sydney Tokyo Toronto
Library of Congress Cataloging-in-Publication Data

Hansen, Kenneth D.
Roller-compacted concrete dams / Kenneth D. Hansen, William G.
p. cm.
ISBN o-07-026072-9
1. Concrete dams. 2. Roller compacted concrete. I. Reinhardt,
William G. II. Title.
TC547.H26 1 9 9 1
624.2-dc20 go-41276

Copyright 0 1991 by McGraw-Hill, 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.

1 2 3 4 5 6 7 8 9 0 DOC/DOC 9 5 4 3 2 1 0

I S B N 0-07-026072-9

The sponsoring editor for this book was Joel Stein, the editing
supervisor was Dennis Gleason, the designer was Naomi Auerbach,
and the production supervisor was Suzanne W. Babeuf. This book was
set in Century Schoolbook. It was composed by McGraw-Hills
Professional Publishing composition unit.

Printed and bound by R. R. Donnelley & Sons Company.


Information contained in this work has been obtained by McGraw-

Hill, Inc. from sources believed to be reliable. However, neither
McGraw-Hill nor its authors guarantees the accuracy or complete-
ness 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 au-
thors 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

For more information about other McGraw-Hill materials,

call 1-800-2-MCGRAW in the United States. In other
countries, call your nearest McGraw-Hill office.

Preface ix

Chapter 1. Development of a New Type of Dam 1

1 .l Background 1
1.2 Definition and Scope 1
1.3 Development of the RCC Dam 2
1.4 Three Paths Taken in RCC Dam Design 0
Bibliography 14

Chapter 2. Roller-Compacted Concrete as a Material 15

2.1 Two Philosophies 15
2.2 Materials for RCC Mixtures 21
2.3 Laboratory Tests 26
2.4 Mixture Proportioning 39
2.5 Properties of RCC 47
Bibliography 62

Chapter 3. Design of RCC Gravity Dams 65

3.1 Background 65
3.2 Site Selection and Foundation Considerations 65
3.3 RCC Dam Design Concepts 72
3.4 Structural Stability and Dam Configuration 75
3.5 Design Details 07
3 . 6 River Diversion and RCC Cofferdams 104
3.7 Appurtenant Structures 106
3.6 Monitoring Performance of Completed Dams 113
3 . 9 Dams Curved in Plan, Including Arch Dams 115
Btbliography 117

Chapter 4. Design and Construction of Lean RCC Dams 121

4.1 Background 121
4.2 Design Concept and Section 122
4 . 3 Mtx Design and Material Properties 129
4.4 Construction Methods 135
Vi Contents

4.5 Construction Control 136

Bibliography 137

Chapter 5. Design and Construction Concepts of a

High-Paste-Content RCC Dam 139

5.1 Background 139

5.2 Design Concepts 140

5.3 Construction Control 152

5.4 cost 153

5.5 Conclusions 153

Bibliography 154

Chapter 6. RCD: The Japanese Approach 157

6.1 Background 157
6.2 RCD Design 159

6.3 RCD Construction 162

6.4 RCD Concrete Mixtures 166
6.5 Benefits 170
Bibliography 171

Chapter 7. RCC Applications in Embankment Dams 173

7.1 General Considerations 173

7.2 Original Design 173

7.3 Rehabilitation 169
7.4 Replacement 193

Bibliography 197

Chapter 6. RCC as a New Construction Method 199

8.1 General Construction Consideratlons 199

8.2 Aggregate Production and Plant Layout 202

8.3 Mixing RCC 204

8.4 Transporting RCC 206

8.5 Compaction 211

8.6 Bonding Lifts 214

8.7 Cleanup, Curing, and Weather Protection 216

8.8 Temperature Control and Contractlon Joints 216

8.9 Special Construction Features 219

Bibliography 227

Chapter 9. Construction Control and Specifications 229

9.1 General Considerattons 229
9.2 Materials Control 231
Contents vii

9.3 Control of Construction Operations 232

9.4 Frequency of Testing 241
9 . 5 Action Required for Nonconformance with Specifications 241
9.6 Tests of Hardened RCC 243
Bibliography 244

Chapter 10. Preliminary Design and Cost Estimates 245

10.1 General Procedure 245
10.2 Foundation Considerations 245
10.3 Dam Cross Section 246
10.4 Volume Determination 240
10.5 Estimating Cost 249
10.6 Example Problem 254
Bibliography 257

Chapter 11. Data and Performance of Completed RCC Dams 259

11 .l Data on Completed RCC Dams 259

11.2 Performance of Completed RCC Dams 26%
Bibliography 291

index 293

Roller-compacted concrete (RCC!) dams are a significant breakthrough

in the design and construction of dams. RCC was developed in the
1980s and its use for new dams and the rehabilitation of existing dams
could make even greater progress during the 1990s and beyond. This
book covers the development of the RCC dam and progress through
the end of the eighties. It is designed to present state-of-the-art infor-
mation on all aspects of RCC dam planning, design, construction, con-
struction control, and performance.
In writing a book on RCC dams, one soon finds out that there are
varied and diverse opinions on how to design RCC dams, especially
with respect to RCC mixtures. This is due to the fact that the devel-
opment came from two distinct civil engineering disciplines, namely
from soils or geotechnical engineering and also from concrete or struc-
tural engineering. Therefore, one must be able to relate to and under-
stand the principles of both disciplines. In Chap. 2, both the soils and
concrete philosophies are described and an attempt is made to bring
the two approaches together.
Throughout the book, more than one way to approach any specific
item of design and construction is usually presented. It is not our intent
to say one method or solution is the only way to solve any specific
problem. We have attempted to present information that allows the
reader to make decisions based on a site- or function-specific situation.
Our approach mainly is to say this is how a dam was designed or built
and, in some cases, what happened. In the case of less-than-desired
performance, methods are presented which may improve future de-
signs. We believe performance-based design is extremely important
and refer the reader to the data on actual performance of many com-
pleted RCC dams presented in Chap. 11.
Three approaches to RCC dam design, (1) the lean RCC dam, (2) the
high-paste-content RCC dam, and (3) the Japanese RCD approach, are

X Preface

presented in Chaps. 4,5, and 6. The lean RCC dam and the high-paste-
content RCC dam are significantly different in their approach to mix-
ture and dam design. Our thanks go to Ernest K. Schrader and Malcolm
R. H. Dunstan for presenting their concepts for the lean and high-paste
RCC dams. As the development of the RCC dam moves forward, we
are seeing many designs moving in between these two approaches.
These may be termed medium-paste designs and include the Japa-
nese RCD method.

When you start to write an all-inclusive book on a specific subject,
you find there are certain chapters that you can proceed with directly
and others that require extensive research and outside assistance. This
means that you must rely on technical assistance and review from
many of your colleagues and friends.
In addition to thanking Ernest Schrader and Malcolm Dunstan for
contributing their specific chapters, special thanks goes to Francis A.
McLean for his input and editing of Chap. 2 and to Glenn S. Tarbox
for Chap. 10. Others who have made significant contributions from the
United States are Howard L. Boggs, Milton A. Kramer, Stephen B.
Tatro, Eric J. Ditchey, Louis H. Roehm, Garry Tucker, Cliff Schex-
nayder, Jeff Allen, Alvin Vissers, Charles C. Hutton, Alan T. Rich-
ardson, Dennis R. Hopman and J. Fred Burns. From Japan, assistance
came from Takeshi Yamauchi, Kentaro Takahi, and Shuhie Kazusa.
Others providing much needed aid were Frank Hollingworth of South
Africa, Brian A. Forbes of Australia, and Shen Chongang of China.
Others who have been involved with the development, technology, and
application of RCC dams are listed in the bibliography at the end of
each chapter. Our sincere appreciation is extended to all who have
participated in the production of this book. It is possible that some
worthy contributor has not been mentioned and, if this is the case, we

Kenneth D. Hansen
William G. Reinhardt

Development of a
New Type of Dam

1 .l Background
Roller-compacted concrete (RCC) dams emerged as a viable new type
of dam during the 1980s. They have gained acceptance worldwide in a
relatively short time due to their low cost, which is derived in part
from their rapid method of construction.
The RCC method evolved not only from the efforts of some influen-
tial concrete dam designers but also from the work of geotechnical en-
gineers who have traditionally designed earth and rockfill embank-
ments. Their combined efforts have produced a concrete dam built by
methods usually associated with earth dam construction. The product
is a less costly dam with the same inherent safety as a conventionally
placed concrete dam.

1.2 Definition and Scope

RCC is more than a new construction material. It is a new construc-
tion method. RCC differs from soil-cement, which uses similar placing
methods, primarily because RCC contains coarse aggregate greater
than G-in (19-mm) maximum size and because it develops material
properties similar to those of conventionally placed concrete. Soil-
cement generally uses pit-run sands as its aggregate and develops
lower strengths than RCC.
Materials that have been called rollcrete, large-aggregate soil-
cement, rolled concrete, or cement-treated base are all considered to
be RCC in this book, which describes the use of RCC for the construc-
tion of new dams and for the rehabilitation of existing dams. It does
not include information on RCC pavements or the application of soil-

2 Chapter One

cement in dams. Some of the knowledge derived from the use of RCC
in pavements and soil-cement in dams does provide information for
the text, however.

1.3 Development of the RCC Dam

1.3.1 Defining the problem
Of all the dams built throughout the world, excluding China,
through 1950, 38 percent of the structures 50 ft(15 m) or higher had
been built with concrete. From 1951 to 1977, the number of con-
crete dams built had dropped to 25 percent of the total. The world-
wide percentage decreased further to 16.5 percent during the period
from 1978 to 1982.
However, this general and steady decline in the popularity of con-
crete dams came during a period when the use of concrete arch
dams in narrow-valley sites was increasing. Therefore, the greatest
decrease was occurring in wide-valley sites, where concrete gravity
dams were being replaced by less costly earth and rock embank-
ments. Their cost advantage over concrete dams was derived
mainly from the greater efficiency of the equipment and methods
used in construction. The increased popularity of embankment
dams also coincided with the emergence of soil mechanics technol-
ogy. This situation led Engineering News-Record to editorialize in
its March 6, 1969, issue:

The technology of mass concrete construction simply has not kept pace
with the art and science of earthmoving. It is time for a study into ways
of reducing the cost of concrete dams.. . .Dams must be conservatively de-
signed and carefully built. But it does seem that in all the years since
Hoover Dam, there should have been more change in the bucket-by-
bucket method of moving mass concrete into place. Whats needed is a lot
more systems analysis and a bit less grandpa-ism.

Concern for the decline of concrete dams led to the organization of

two important meetings of the dam-building community in the United
States at the Asilomar conference grounds in California. The first, in
1970, was called Rapid Construction of Concrete Dams. The second,
in 1972, was entitled Economical Construction of Concrete Dams.
Similar concerns in Japan led to the establishment of the Committee
on Rationalized Construction of Concrete Dams in 1974 by the Minis-
try of Construction.
In the meantime, while embankment dams were being reduced in
Development of a New Type of Dam 3

cost relative to concrete dams, they were-and are-more prone to

failure. No concrete dam higher than 50 ft (15 ml has failed in the
United States since 1928, when St. Francis Dam, a 205-ft-high (62-m)
curved gravity dam in California, failed due to defective foundation
material. Outside the United States, the most recent concrete dam
failure was in 1959 at Malpasset Dam in France. The 200-ft-high (61-
m) thin concrete arch failed by sliding along a weak seam in the left
By comparison, hundreds of earth embankments of all sizes have
failed during the past 60 years. The primary cause of embankment
dam failures is overtopping and internal erosion of the fill material.
Taking note of the relative vulnerability of fill dams, the experts at
Asilomar and elsewhere were searching for a new type of dam that
would combine the safety advantages of concrete and the efficiencies
of embankment dam construction. By a number of different routes,
their searches in the early 1960s and 1970s led to the development of
roller-compacted concrete dam building.

1.3.2 Early progress

Several projects in the 1960s were designed with the idea of combin-
ing the advantages of concrete and embankment dams. These hybrid
dams were the brainchildren of both structural and geotechnical en-
gineers. Unfortunately, because of a high degree of specialization in
these fields, there was limited communication between the early the-
orists. Hence, there was little incremental understanding gained from
these early efforts.
The most notable and earliest example was the 564-ft-high (172-m)
Alpe Gera Dam in Italy, designed by structural engineer Giulio Gen-
tile and completed in 1964. The concept behind Alpe Gera was to
maintain the traditional cross section of the concrete gravity dam
while reducing the unit cost of placing the mass concrete. Some of the
cost reduction was accomplished by reducing the cement content in
the concrete mix used for the interior of the dam, where stresses are
low and durability requirements are minimal.
Placing this lean concrete using earthfill construction methods was
the greatest step forward, however. Instead of building the concrete
dam in vertical lifts to form cantilever blocks, horizontal placement
was introduced. Dump trucks delivered the interior concrete to the
_ dam rather than buckets moved by crane or cableway. Side forms for
blocks were eliminated, as were cooling coils. The consolidation of the
lean concrete by internal immersion vibration rather than external
roller compaction was about all that kept Alpe Gera from being the
first RCC dam (Fig. 1.1).
4 Chapter One

Figure 1.1 (CL) Ape Gera Dam; (b) Close-up from downstream.
Development of a New Type of Dam 5

Another early hybrid was developed by concrete dam designers at

Hydro Quebec in Montreal. Their ideas were incorporated in two 60-
ft-high (18-m) gravity wing walls at the Manicougan I Dam in Quebec
in 1965. On this project, lean mass concrete was placed by dozers for
the core of the dam and was internally vibrated.
A richer mix was used for the upstream face of the wing dams. The
facing concrete was slip-formed vertically. Joints with waterstops
were spaced at 50-ft (15-m) intervals. Precast blocks were used for the
downstream face. Hydro Quebec estimated that the system saved 20
percent of the cost and two-thirds of the time that would have been
required to build the concrete wing walls using conventional methods.
From geotechnical engineers came several site- and function-
specific hybrid projects. These used ideas for a solid soil-cement dam
that had been proposed by Rocha, Folque and Esteves of Spain in 1961
and Nash, Jardine and Humphries of the United Kingdom in 1965.
Sly Creek Dam in northern California was designed in 1967 as a
60-ft-high (18-m) solid soil-cement dam. The entire section had to be
erosion-resistant because the dam was designed to be overtopped and
ultimately inundated during high-flow conditions. The increased
shear resistance of soil-cement over earthfill construction allowed
both the upstream and downstream slopes of Sly Creek to be steep-
ened to 1 H : 1 V. The design, by California engineering consultants
St. Maurice, Helmkamp and Musser, was approved for construction by
the state Department of Water Resources Division of Safety of Dams.
The project was never built, however, because of a lack of funds.
Erosion resistance also was the key criterion for the design of a 22-
ft-high (6.7-m), 6.2-mi-long (lo-km) embankment to enclose a llOO-
acre (445 hectare1 cooling-water reservoir for the Barney M. Davis
power station in Corpus Christi, Texas. The long, low ring dike had to
withstand wave action from within and erosion by floods and heavy
rain from without.
A solid soil-cement cross section with a 12-ft (3.7-m) crest width and
1.5 H : 1 V slopes on both faces was one of the alternatives considered
by engineering consultants, Sargent & Lundy, in 1971. It proved to be
the most economical option when it was bid two years later as an al-
ternative to a more conventional sandfill embankment with soil-
cement slope protection on both sides.
Besides being the only large dam constructed entirely of soil-
cement, the Barney M. Davis reservoir embankment marked the first
recorded use on a dam of vibratory rollers to compact soil-cement. No
joints were incorporated in the 351,000 yd3 (268,000 m31 of soil-cement
used to construct the long, low dike (Fig. 1.2). Transverse cracks oc-
curred in the soil-cement section as anticipated but, they were not of
sufficient width to allow passage of water.
6 Chapter One


Figure 1.2 Barney M. Davis powerplant. (a) Typical em-

bankment section; (b) plan of the cooling lake.

1.3.3 Concepts presented at the Asilomar

A report on the construction of Alpe Gera by the Italians raised a
number of eyebrows at the 1970 conference of dam designers at
Asilomar. But the seminal paper for RCC development was presented
by Jerome Raphael of the University of California at Berkeley.
Raphaels paper, The Optimum Gravity Dam, presented a number
of ideas that were based in part on soil-cement theory and applica-
tions. He proposed the concept of placement and compaction of an em-
bankment with cement-enriched granular pit-run material using
high-capacity, high-speed earth-moving and compaction equipment.
Raphael noted that the increase in shear strength of a cement-
stabilized material would result in a significant reduction of the cross
section compared with a typical embankment dam. Also, the use of
continuous placement methods similar to those being used for earth
dams would generate savings in time and money compared with tra-
ditional concrete dam construction methods.
Robert W. Cannon of the Tennessee Valley Authority took
Raphaels ideas a step further in his paper, Concrete Dam Construc-
tion Using Earth Compaction Methods, presented at the 1972
Asilomar conference. Cannon had earlier reported on the results of a
lean concrete test section where a no-slump mix containing controlled-
gradation aggregates was transported by truck, spread by a front-end
loader and compacted with a vibratory roller. In his Asilomar paper,
Cannon described how this procedure might best be used to construct
a dam of the same configuration as a conventionally built concrete
Development of a New Type of Dam 7

gravity dam. He proposed placing a richer mix on the upstream and

downstream faces using some form of horizontal slip-form machine.
He also suggested that trucks might not be the best way to transport
the no-slump interior or hearting concrete from the batch plant to the
Also at the 1972 conference, Raphael reported on a major installa-
tion of soil-cement for upstream slope protection at Castaic Dam in
California. A placement rate of 500 yd3fh hour (382 m3/h) was
achieved on that project. Raphael also noted in 1972 that techniques
were available to take the next step of building an economical soil-
cement dam. Apparently, he was not aware of the earlier Sly Creek
Dam design or the solid soil-cement alternative proposed and eventu-
ally built for the Barney M. Davis powerplants cooling-water reser-
voir ring dike.

lr3.4 High production rate for Tarbela Dam

Although John Lowe III of Tippetts-Abbett-McCarthy-Stratton
(TAMS) had used what he termed rollcrete (a contraction for roller-
compacted concrete) for the core of Shihmen Dam in Taiwan in 1960 to
1961 (see Sec. 7.2.31, it was the use of RCC for repairs at Tarbela Dam
in Pakistan that had a major impact on the development of the RCC
dam. The high production rates desired for the construction of an RCC
dam were demonstrated at Tarbela. Lowe, who supervised the reha-
bilitation at the TAMS-designed project, calculated that 3.5 million
yd3 (2.7 million m3) of RCC were placed from 1974 to 1986 at Tarbela,
the worlds largest engineered embankment (see Fig. 1.3).
RCC was used initially to replace rock and embankment material
surrounding one of four large outlet tunnels that had collapsed during
the initial filling of the Tarbela reservoir in August 1974. Rapid con-
struction was imperative because the tunnel repairs had to be com-
pleted before spring snowmelt greatly increased the flow in the Indus
The Pakistani workers and the Italian contractor were up to the
task. Some 460,000 yd3 (350,000 m3) of RCC were placed in 42 work-
ing days at an average rate of almost 11,000 yd3 (8400 m3) per day.
When the contractor, Impregilo, had sufficient room to work, a maxi-
mum daily placement rate of 24,000 yd3 (18,000 m3) was achieved.
That remains the highest placement rate for RCC anywhere in the
Following the tunnel repair, extensive rehabilitation work on both
the auxiliary and service spillways at Tarbela was accomplished with
RCC. Massive groins, cofferdams, and stilling basins were required for
8 Chapter One

Figure 1.3 Tarbela Dam, Pakistan. From 1974 to 1986, 3.5 million yd3 of RCC were

this work. In reporting on the repairs in 1980, TAMS project manager

Paul C. Chao noted, Hopes of using rollcrete to construct a dam as
well as an overflow spillway channel have been greatly raised.
The use of RCC at Tarbela is discussed further in sections 3.6.2,
7.3.1, and 7.3.2.

1.4 Three Paths Taken in RCC Dam Design

RCC dam design was evolving in three different directions during the
1970s. In the United States, a lean-concrete alternative based on soils
technology was being developed by the Army Corps of Engineers and
other investigators. British engineers were focusing on the so-called
high-paste alternative, a hybrid of conventional concrete mix design
and earthfill dam construction methods. The Japanese research team
set up to explore rationalized concrete dam construction methods was
developing the third approach, what it called roller-compacted dam,
concrete method, or RCD. Of the three, RCD is the most cautious de-
parture from traditional concrete gravity dam design and construction

1.4.1 Development of the Lean RCC dam

The United States Army Corps of Engineers began a concerted effort
to develop RCC for use in building concrete dams in the early 1970s.
The Corps built field test sections at Jackson, Mississippi, in 1972 and
Development of a New Type of Dam 9

at the site of Lost Creek Dam in Oregon in 1973. The field tests con-
firmed the basic construction method and provided information on
material properties and the strength of the bond between successive
layers of RCC. In fact, the name roller-compacted concrete may have
been first used by Corps investigators Hall and Houghton in reporting
on the Lost Creek test section.
Based on the data developed in these tests, the Corps Walla Walla
District in 1974 designed an RCC alternative for Zintel Canyon Dam
near Kennewick, Washington. Following up on Raphaels Optimum
Gravity Dam paper, the first dam section investigated in the design
was an embankment with a 45 face (1 H : 1 V) on the upstream and
downstream slopes. It was to be constructed almost entirely of a lean
RCC mixture, enriched only in the exposed exterior zones.
As initially proposed, however, the cross section contained too great
a volume of RCC to economically compete with the Corps original
rockfill embankment design. It was not until the designers reduced
the section to one more typical of a conventional concrete gravity dam
with a vertical upstream face that the RCC option was estimated to be
the most cost-effective solution for the flood control project (Fig. 1.41.
The Corps designers called their final RCC product the optimum
gravity dam.
Zintel Canyon Dam was not funded at the time, but many of its con-
cepts were carried over to Willow Creek Dam, another Corps flood
control project, at Heppner, Oregon. That 16sft-high (52-m) dam was
built in less than five months in 1982 and became the worlds first ma-
jor dam to be built entirely of RCC.

6.0 m 6.0 m
20ftOinT P I P20ftOin

Cementitious vertical faces

cementitious fill random fill material
has been placed
(3) (4)

Figure I A Evolution of the proposed section for Zintel Canyon Dam.

10 Chapter One

Thus, the lean RCC dam evolved from a concept in which a cement-
stabilized, controlled-gradation aggregate is placed and compacted
rapidly from abutment to abutment without forms or joints. In order
to make the idea work economically the slopes had to be steepened,
creating a need for some method of forming the vertical upstream
face. The initial plan at Zintel Canyon was to build an earth berm up-
stream to buttress and form the RCC mass. At Willow Creek, that
method was first specified as precast reinforced earth concrete pan-
els tied back into the RCC. Later, the contractor proposed to use ex-
ternally supported precast panels, combined with two coil rods per
panel set into the RCC, and that was the method adopted for the ac-
tual costruction.
The design and construction of the lean RCC dam alternative are
described by the principal designer of Willow Creek, Ernest K.
Schrader, in Chap. 4.

1.4.2 Development of the high-paste

RCC dam
After some initial work in the early 1970s by the Tennessee Valley
Authority on a concrete mix with a low-Portland-cement and high-fly-
ash content, the development of this so-called high-paste RCC alter-
native shifted to the United Kingdom. The properties of the material
were demonstrated in 1976 following field trials in Cornwall, En-
Concrete materials engineer Malcolm Dunstan received a fellow-
ship in 1977 to come to the United States and interview all those in-
volved with RCC dam building. His report, Rolled Concrete-With
Particular Reference to Its Use as a Hearting Material in Concrete
Dams, remains a good source of information on early RCC dam de-
Dunstan subsequently did extensive laboratory research on the
properties of high-fly-ash-content concrete with respect to its use in
dams. This work was done under the sponsorship of the Construction
Industry Research and Information Association (CIRIA). The research
culminated in a full-scale trial in 1978 done by the Southwest Water
Authority at the site of Wimblehall Dam. A significant development
there was the successful use of a laser-guided horizontal slip form to
produce facing elements for an RCC dam.
The intent of the research and field trials was to develop a high-
paste RCC design for Milton Brook Dam at Devon, England. Various
problems prevented that dams construction, however. Although sty-
mied in the United Kingdom, Dunstans work found its expression in
the United States, where it became the basis for the design by the
Development of a New Type of Dam 11

Figure 1.5Upper Stillwater Dam. (Courtesy of U.S. Department of the Interior, Bureau
of Reclamation, Upper Colorado Region.)

United States Bureau of Reclamation for Upper Stillwater Dam, a

294-ft-high (90-m) water-supply dam completed in 1988 in the moun-
tains of central Utah (see Fig. 1.5).
The high-paste RCC dam evolved along the same lines as the Jap-
anese RCD method, in that both started out with the basic cross sec-
tion of a concrete gravity dam. The volume of RCC at Upper Stillwater
was reduced even further by taking into account the tensile strength
of the bond at the joints between successive lifts of the high-paste RCC
mix, The concepts and methods that make up the high-paste RCC al-
ternative are described by Dunstan in Chap. 5.

1.4.3 Development of RCC in Japan

At the same time that considerable progress was being made toward
the development of RCC for dams in the United States, Japan also was
working independently to develop a rationalized method for building
concrete dams that would speed the placement of concrete and lower
the cost of construction. The product of this research through the
12 Chapter One

1970s is referred to by Japanese engineers as the roller-compacted

dam, concrete method, or RCD.
Most of the dam construction in Japan has occurred in the period
after World War II. More than 80 percent of the existing dams in Ja-
pan are built of concrete, and most proposed dams are planned as con-
crete gravity structures because of their inherent safety. Technologi-
cal progress in the placement of mass concrete in Japan had moved
slowly, however, basically in step with progress in the United States,
Like their American counterparts, Japanese designers saw a need to
improve the economics of concrete dam construction and focused on
RCC as the preferred alternative.
Because of seismic, hydrologic, and topographic problems associated
with most dam sites in Japan, designers there have taken a more con-
servative approach to RCC dam construction, Their aim is for a prod-
uct with the same quality and appearance as that of conventionally
placed, mass concrete gravity dams. Various studies and experiments
led to an RCC method that is described in Chap. 6. Trials were done
using RCC in the foundations of Shimajigawa and Okawa dams start-
ing in 1978. RCC was then used for all but the outermost 10 ft (3 m) of
the 292-ft-high (89-m) Shimajigawa Dam (Fig. 1.6). When completed

Figure 1.6 Completed Shimajigawa Dam, Japan.

Development of a New Type of Dam 13

TABLE 1 .l Ten Significant Developments in RCC for Dams

1964 Alpc Gera Dam, a 564-ft-high (172-m) concrete gravity dam in the
Itallan Alps was constructed like an earth embankment, using
dump trucks, dozers. and tractor-mounted immersion vibrators to
place lean concrete in horizontal lifts.
1970 Jerome Raphael presents a paper, The Optimum Gravity Dam, in
\rhlch hc proposes the concept of an embankment made of cement-
enriched, g-ranular pit-run material placed and compacted with
high-speed earth-moving equipment.
1970-1973 Research in the United States by the Tennessee Valley Authority at
Tims Ford Dam and by the Corps of Engineers at Jackson, Missis-
sippi. and at Lost Creek Dam helped to prove the economic feasi-
bility of RCC and to develop the construction methods for its mass
1975 The emergency repair of a collapsed outlet tunnel at Tarbela Dam
in Pakistan using RCC demonstrated the rapid placement rates
possible: 460,000 yd3 (350,000 m3) of RCC were placed in 42 work-
ing days.
1978 Research started four years earlier by Japans Committee on Ratio-
nalized Construction of Concrete Dams led to the start of RCC
placement for the body of Shimajigawa Dam, a 292-ft-high (89-m)
gravity dam.
1978 A full-scale trial of the use of high-fly-ash-content RCC together
with laser-controlled slip-formed facing elements was successfully
completed at Wimblehall Dam in England. This work on high-
paste RCC contributed significantly to the design in the early
1980s of the U.S. Bureau of Reclamations Upper Stillwater Dam
in Utah.
1982 The placement of 433,000 yd3 (331,000 m) of RCC in less than live
months for the U.S. Army Corps of Engineers Willow Creek Dam
in Heppner, Oregon, confirmed the rapid construction rates and
economic viability of dams built entirely of RCC.
1984 RCC came to the southern hemisphere with the design and con-
struction of Australias 131-ft-high (40-m) Copperfield Dam in only
10 months.
1984 The construction of 70-ft-high (21-m) Winchester Dam in Kentucky
using precast concrete panels and an attached polyvinylchloride
membrane to both form the RCC and provide an impervious up-
stream face initiated a concept that may be called a concrete-faced
RCC dam.
1985 The erosion resistance of exposed RCC was proven in the field when
Kerrville Ponding Dam in Texas, a 20-ft-high (6.1-m) RCC dam,
was overtopped during a flood by 14.4 ft (4.4 ml 30 days after con-
struction was completed. It was overtopped by 16.2 ft (4.9 m) due
to an even greater flood two years later, with no appreciable wear
of the RCC crest and downstream slope.

in 1980, Shimajigawa became the first dam in the world to be built

using RCC for the main portion of the dam.
Table 1.1 lists ten significant developments in the history of RCC
use for dams.
14 Chapter One

American Society of Civil Engineers (ASCE), Rapid Construction of Concrete Dams,
New York, 1970.
Avera, J. K., Jr., and K. D. Hansen: Dams That Never End, Wafer Power and Dam
Construction, March 1978.
Cannon, R. N.: Concrete Dam Construction Using Earth Compaction Methods, Eco-
nomical Construction of Concrete Dams, ASCE, New York, 1972, pp. 143-152.
Chao, P. C.: Tarbela Dam-Problems Solved by Novel Concretes, Civil Engineering,
ASCE, December 1980.
Chao, P. C., and H. A. Johnson: Rollcrete Usage at Tarbela Dam, Concrete
International: Design and Construction, vol. 1, no. 11, November 1979.
Dunstan, M. R. H.: Trial of Lean Rolled Concrete at the Tamar Treatment Works,
Report to the Southwest Water Authority, June 1977.
Dunstan, M. R. H.: Rolled Concrete-With Particular Reference to Its Use as a
Hearting Material in Concrete Dams, The Concrete Society, London, March 1978.
Dunstan, M. R. H.: Rolled Concrete for Dams-Construction Trials Using High-
Flyash-Content Concrete, CIRL4, Technical Note 106, London, May 1981.
Dunstan, M. R. H.: Rolled Concrete for Dams-A Laboratory Study of the Properties of
High-Flyash-Content Concrete, Construction Industry Research and Information
Association (CIRL41, Technical Note 105, London, May 1981.
Engineering News-Record Concrete Gravity Dam Built Like Earthtill, Dec. 24, 1964.
Esteves, V. P., J. Folque, and M. Rocha: The Application of Cement-Stabilised Soil in
the Construction of Earth Dams, Proceedings of the Fifth International Conference
on Soil Mechanics and Foundation Engineering, vol. 2, 1961, pp. 701-707.
Hall, D. J., and D. L. Houghton: Roller-Compacted Concrete Studies at Lost Creek
Dam, United States Army Engineer District, Portland, Ore., June 1974.
Hansen, K. D.: Roller-Compacted Concrete Dams Worldwide, Water Power and Dam
Construction Handbook, 1987.
Hirose, T., and S. Takebayashi: Present State and Problems of Rationalized Construc-
tion of Concrete Dams, Concrete Library of Japan Society of Civil Engineers, no. 2,
December 1983.
Humphreys, T. D., F. M. Jardine, and J. K. T. L. Nash: The Economic and Physical
Feasibility of Soil-Cement Dams, Proceedings of the Sixth International Conference
on Soil Mechanics and Foundation Engineering, vol. 2, 1965, pp. 517-521.
International Commission on Large Dams (ICOLD), World Register of Dams, 1984.
Jansen, Robert B.: Dams and Public Safety, United States Dept. of Interior, Bureau of
Reclamation, revised reprint, Denver, 1983.
Kokubu, M.: Development in Japan of Concrete Dam Construction by the RCD
Method, Technical Lecture at 52nd ICOLD Executive Meeting, Tokyo, 1984.
Portland Cement Association, Fact Sheet on Sly Creek Dam, Denver, 1973.
Raphael, J. M.: The Optimum Gravity Dam, Rapid Construction of Concrete Dams,
ASCE, New York, 1970, pp. 221-244.
Raphael, J. M.: Construction Methods for the Soil-Cement Dam, Economical Con-
struction of Concrete Dams, ASCE, New York, 1972, p. 217.
Sivley, W. E.: Zintel Canyon Optimum Gravity Dam, XII ICOLD Congress, vol. 5,
Mexico City, 1976, pp. 141-145.
Tynes, W. 0.: Feasibility Study of No-Slump Concrete for Mass Concrete Construc-
tion, Miscellaneous Paper C-73-10, United States Army Corps of Engineers Water-
ways Experiment Station, Vicksburg, Miss., October 1973.
Wallingford, V. M.: Proposed New Technique for Construction of Concrete Gravity
Dams, Xth ICOLD Congress, vol. 4, Montreal, 1970.

Concrete as a Material

2.1 Two Philosophies

In the development of RCC technology, two philosophies, or ap-

proaches, have emerged with respect to RCC mix design methods.
They can be termed the soils, or geotechnical, philosophy, and the con-
crete philosophy.
There is no distinct line separating the two philosophies. Basically,
RCC mixtures produced using concrete design methods have a more
fluid consistency as measured by a Vebe or vibratory compaction (VC)
test (see Sec. These mixes may be described as being more
workable than those developed using the soils approach, yet both phi-
losophies will produce a concrete that is termed <zero slump.
As described in Chap. 1, three different paths have been taken in
the development of RCC for dams. Two of them, the high-paste RCC
dam and the Japanese RCD method, may be termed concrete ap-
proaches. Upper Stillwater Dam, Elk Creek Dam, and all Japanese
RCD dams are examples of dams with RCC mixtures designed using a
concrete philosophy. Dams such as Willow Creek, Copperfield, Middle
Fork, and Monksville were built with lean RCC mixtures that fit into
the soils approach. A general comparison of factors for the two philos-
ophies is listed in Fig. 2.1.

2.1.1 The soils philosophy

The soils philosophy considers RCC as a cement-enriched processed
soil, or aggregate, whose mix design is based on moisture-density re-
lationships. For a specified aggregate and cementitious material con-
tent, the goal is to determine an optimum moisture content for a lab

16 Chapter Two

Factor Soils (Geotechnical) Concrete

Basis for mix design Optimum moisture/ Lowering of water/cement

maximum dry density ratio & consolidation

Characteristics of All voids not filled- Voids filled with paste & ex-
voids particle-to-particle con- cess
Percent theoretical Usually less than 98% Greater than 98%
air-free density

Consistency Webe or 45 s or more 45 s or less

VC time)
Other names Rollcrete, stablized fill, RCD method (Japan), high-
large aggregate soil- paste RCC
cement, lean RCC

Materials Graded or pit-run Very well graded to minimize


Aggregates-grading Up to 10% by total Less fines-especially if high

amount of tines (No. weight fly-ash content
200 sieve)
Cement t pozzolan Usually less than 202 Greater than 202 lb/yd3 (120
content lb/yd3 (120 kg/m3) kg/m3)

Cohesion Low value-less than Higher values-greater than

200 lb/in (1.4 MPa) 200 lb/in (1.4 MPa)

Permeability Void content depends on Depends on properties &

mix & construction (de- amount of paste
gree of compaction &

Compressive High per unit weight of Decreases with more water-

strength cement-greater at top greater at bottom of lift
of lift
Seepage control con- Upstream membrane Entire gravity dam

Mixing Mostly pugmill Pugmill or drum

Spreading Segregation more of a Segregation less of a problem

Compaction Vibratory or heavy Vibratory roller
rubber-tired roller

Primary action or Compaction Consolidation

Compacted lift thick- Generally 12 in (0.3 12 to 30 in (0.3 to 0.75 m)
ness m)-more possibility of Possibility of paste at sur-
voids at bottom face & slight chance of voids
at bottom.

Figure 2.1 General comparison of factors for RCC philosophies.

Roller-Compacted Concrete as a Material 17

Material: Crushed limestone, 1.5 in MSA
Cement content: 5.5 96
by dry weight ofsolids
Fly ash content: 2.2 %)
140 -

Compactive effort
=32.5 ft+Ib/in3
(modified proctor-ASTM D15571

Compactive effort
= 13.3 **lb/in3

125 -

120. I I I
0 5 10 15 20
Moisture content, % dry wt. of solids
2.2 Laboratory moisture-density curves for RCC subjected to various
compactive energies. [From Reeves & Yates (1985)/

oratory compactive effort that corresponds to the effort or density

applied by the rollers in the field. In the soils approach, paste (cement,
pozzolanand water) does not generally fill all the voids in the aggre-
gate after compaction.
The principles of compaction developed by Proctor in the early
1930s are applied in the soils approach to the proportioning of RCC
mixtures. Proctor determined that for a given compactive effort there
is an optimum moisture content that produces a maximum dry den-
sity. Increasing the compactive effort results in a greater maximum
dry density and a lower optimum moisture content. For example, Fig.
2.2 shows compaction curves for three compactive efforts, the greatest
being modified Proctor compaction (ASTM D 1557) for a crushed lime-
stone aggregate from Austin, Texas.
Based on these compaction principles, dry density is used as the de-
sign index in the soils approach. Dry density is defined as the dry
weight of solids per unit volume of material, independent of water
content. It can be calculated from wet density, and vice versa, by the

Pd = -
18 Chapter Two

RCC Lobomtuy mix design resuIts

,44, , ;tw p spillyoy , ,2o


t I
strength curve

1 8 0 0 NC

1400 f

1200 g
1000 6
1; Density curve \T i3

:,r?:4 5 6
Moximum oggregolx size: 2.5 in
7 8
content, %

10 6oo

Mold size : 6 in X 12 in
Cement content: 4.0 %* Hommer: 10 lb
Fly-osh content: 3.0 %* Drop: 18in
Loyers: 6
*By dry weight oggreguk Blows: 123/layer

Figure 2.3 Comparison of density and compressive strength for

RCC mix-modified Proctor compaction. [From Yates, unpub-

where P,, = dry density, Ib/ft3 (kg/m31

P, = wet density
w = moisture content of total mix expressed as a decimal

If an optimum moisture content is used that corresponds to the

compactive effort achieved by the rollers in the field, a material at
maximum dry density will be produced. Dry density is directly pro-
portional to the seven-day compressive strength as shown in Fig.
2.3 for the RCC mixture used for Stacy Dam in Texas. For a given
(constant) compactive effort, water content either below or above
optimum would produce a lower dry density, and consequently, a
reduced compressive strength. Correspondingly, a lower compres-
sive strength can be expected from an RCC mix that receives less
compactive effort than what was used to determine the optimum
moisture content.
Roller-Compacted Concrete as a Material 19

2.1.2 The concrete philosophy

For the concrete philosophy, the RCC mix is considered to be a true
concrete whose strength and other properties follow the water-cement
relationship established by Abrams in 1918. That is, assuming sound
and clean aggregates, the strength of fully consolidated concrete is in-
versely proportional to its water-cement ratio. Using less water with a
constant amount of cement produces a concrete with higher compres-
sive strength and related properties.
The concrete approach is based on the concept that there is suffi-
cient paste in the RCC mix to more than fill all the voids in the ag-
gregate, making the mixture a fully compacted, no-slump concrete.
For geotechnical engineers, this would be termed a zero-air-voids ap-
proach. The RCC mix should not, however, contain so much paste that
a measurable slump is produced or excess paste is brought to the sur-
face with only a few passes of a vibratory roller.

2.13 Relating the two philosophies

A plot of water-cement ratio of concrete versus compressive strength
as shown in Fig. 2.4 provides a means of relating the two philosophies.
The two dotted curves labeled insufficently compacted concrete have
the general shape of moisture-density curves. The lower curve corre-
sponds to a lesser compactive effort; the higher to a greater effort with
a resulting lower optimum moisture content. The actual water/cement
ratio curve for fully compacted RCC would be parallel to and within 2

, .
Water/cement rotio

Figure 2.4 General relationship between compressive

strength and water/cement ratio. [From Neville (1981).1
20 Chapter Two

percent of a 100 percent theoretical air-free density curve, or zero-

voids curve, to account for entapped air in the mixture. This high total
wet density condition can be reached with mixtures proportioned us-
ing either the soils or concrete philosophy, but is usually attained
with concrete approach mixes. In both cases, strength, in addition to
being affected by the mixture proportions, is also directly related to
compaction. Greater strength is produced by a higher degree of com-
paction or consolidation as indicated by a greater density.
A possible demarcation point between the two approaches could be
that an RCC mix designed using the concrete approach has a Vebe or
vibratory compaction (VC) time of 45 s or less, while mixes resulting
from the soils approach have a Vebe time of greater than 45 s. A Vebe
time of 45 s indicates that there is sufficient paste to fill all aggregate
voids after 45 s of weighted vibration of the mixture. For most soils
approach mixes, no paste can be produced after one minute of vibra-
tion. While vibration will help to consolidate drier mixes, as shown in
Fig. 2.4, there is a limit to how dry the mix can be before full consol-
idation becomes impossible and compressive strength drops sharply.
Aggregate gradation and the amount of cementitious materials,
mainly fly ash, have a marked effect on the amount of water added to
saturated surface-dry (SSD) agggregates to produce a certain consis-
tency. Therefore, concrete approach mixtures do not necessarily have
a greater amount of water added than those designed by soils princi-
ples. An RCC mix that conforms to the soils philosophy may be trans-
formed to one that falls within the concrete approach either by adding
water or possibly increasing consolidation.
For certain aggregate gradings, this can be accomplished by adding
as little as two U.S. gallons of water per cubic yard (16.7 lb/yd3 or 9.9
kg/m3). This amounts to an increase of about 0.5 percent moisture by
dry weight of aggregate to fill all remaining voids in the mixture, ex-
cept entrapped air. With sufficient paste in the mix, increased vibra-
tory rolling can produce a fully consolidated concrete, i.e., one within
2 percent of the zero-air-voids state.
A number of investigators have plotted compressive strength versus
water content for a constant cement content for RCC (see Fig. 2.5). The
results show a wide scatter of values, due not only to variations in ma-
terials but also to differences in the degree of compaction. The scatter is
particularly noticeable as the RCC mixture departs the fully consoli-
dated state of the concrete approach due to a reduction in water content.
In this area, the mixture now conforms to the moisture-density relation-
ships of the soils approach rather than the water/cement ratio law in
that paste does not fill all the voids. In fact, for very dry mixtures, in-
creased compressive strength can be obtained by adding water.
The plot of compressive strength versus water content may not be as
simple as explained here because of two other factors: aggregate grada-
Roller-Compacted Concrete as a Material 21

C+ FA = 120 kg/m3 (202 Ib/yd3)

I ' ' 0' ' ' ' ' ' I

P. 100,

Figure 2.5 Relationship between

water content and compressive
0- strength of RCD [From Hirose
80 90 100 110 120 130 140 150 160 and Yanagida (1982)J
(135) (1521 (169) (185) (202) (219) (236) (253)(2701
Woter content, kg/m3(lb/yd3)

tion and a delay in compaction, both of which can affect the results. Ag-
gregate gradings that are not uniform or that are deficient of sand may
be fully compacted yet still have coarse aggregate to coarse aggregate
contact, creating rock gaps or voids. Any delay in compaction from when
the water is added to the cement also causes reduced density. A one-hour
delay in compaction has been shown to cause as much as a 20 percent
reduction in seven-day compressive strength in laboratory specimens.
With all other factors being constant, the wetter consistency of the
mixes designed by the concrete approach usually results in somewhat
lower compressive strengths and less abrasion resistance than the drier
soils approach mixes. However, high-paste and RCD mixes usually pro-
duce improved bonding at the horizontal lift interfaces and reduced per-
meability along lift lines because of excess paste. Use of these mixes also
tends to produce fewer voids and less segregation at the bottom of lifts.

2.2 Materials for RCC Mixtures

Materials used for RCC include cementitious materials (Portland ce-
ment and pozzolans such as fly ash), aggregates, water, and admix-
tures. A wide range of materials has been used successfully to produce
RCC mixtures.

2.2.1 Cementitious materials

The type and quantity of portland cement or cement plus pozzolan re-
quired in RCC mixes depend on the volume of the structure, its re-
22 Chapter Two

quired properties, and the exposure conditions. In addition, most RCC

dams are large enough to require consideration of the heat of hydra-
tion of the cementitious materials.
Cementitious contents used in RCC dams have ranged from 100 lb/
yd3 (60 kg/m31 of cement used for Urugua-i Dam in Argentina to 418
lb/yd3 (248 kg/m31 for the predominant mix at Upper Stillwater Dam.
The high-paste mix at Upper Stillwater contained 129 lb/yd3 (77 kg/
m3) of cement plus 289 lb/yd3 (171 kg/m3) of class F fly ash.
In Japan, the cement-plus-pozzolan content is usually about 202 lb/
yd3 (120 kg/m3). The pozzolan amounts to 20 to 30 percent by weight
of the cementitious material.

2.2.1 .l Portland cement. For massive RCC dams, engineers should

consider cements with lower heat-generation properties than normal
or ordinary portland cement (ASTM Spec. C150, Type I). They include
moderate-heat cement (ASTM C150, Type II), Portland-pozzolan ce-
ment (ASTM C595, Type IP), and Portland-blast furnace slag cement
(ASTM C595, Type IS).
Strength development for the lower-heat cements is generally
slower than for Type I, particularly at early ages. Beyond 90 days,
however, the lower-heat cements generally produce higher strengths
than Type I. The U.S. Army Corps of Engineers and U.S. Bureau of
Reclamation usually specify Type II cement for dams. Outside the
United States, a blended Portland-pozzolan cement was used for
Brazils Saco de Nova Olinda Dam, while a Portland-blast furnace
slag cement produced the desired strengths and heat-generation con-
ditions for Les Olivettes Dam in France.
The choice of cement types for exposure to aggressive chemicals or
aggregate reactivity should follow standard practice for convention-
ally placed concrete. Sulfate-resisting cement (ASTM C150, Type V)
was used for Lower Chase Creek Dam, for example, because of the po-
tential for exposure to acidic waters. Pozzolans. The selection of pozzolans suitable for RCC should

be based on conformance with applicable standards (such as ASTM
C618), the pozzolans past perfomance in concrete, and its cost and
availability in the volumes required. ASTM C618 defines a pozzolan
as a siliceous or a siliceous and aluminous material, which in itself
possesses little or no cementitious value, but will, in finely divided
form in the presence of moisture, chemically react with calcium hy-
droxide at ordinary temperatures to form compounds possessing
cementitious properties.
Pozzolans reauire moisture to react chemicallv with the calcium hv-
Roller-Compacted Concrete as a Material 23

droxide released during the cement hydration process. Therefore,

pozzolans perform better in the wetter mixtures than they do in the
drier consistency mixes.
Most RCC mixtures that have included a pozzolan to date have used
a Class F (low-lime) fly ash. Because adequate air entrainment is gen-
erally not achievable in.RCC, limits on the carbon content of the fly
ash (as measured by percent lost on ignition) may be raised as long as
strength properties are not adversely affected.
Several completed projects have used a pozzolan other than a class F
fly ash. They include a Class C (high-lime) fly ash for the RCC in the
Stacy Dam spillway. Arabie and Zaaihoek dams in South Africa used
a ground blast-furnace slag called slagment as pozzolan in the RCC
mix. Calcined clay, a natural pozzolan, was used in the RCC mix for
the navigation lock at Tucurui Dam in Brazil.
For large-volume RCC dams, the pozzolan is invariably stored,
weighed, and introduced into the mixer as a separate material. For
small projects where a pozzolan is desired, it may be more cost-
effective to use preblended pozzolan-cements than to store, test, and
handle both pozzolan and cement separately1

2.2.2 Aggregates
For RCC, like conventionally placed concrete, aggregate quality and
gradation are important factors influencing the final product. Slight
differences have occurred among designers in the selection of maxi-
mum size aggregate (MSA), the proportion of sand in the mix, and the
percentage of fines passing a No. 200 (0.75mm) sieve for RCC mix-
tures when compared with conventional concrete mixtures.
The segregation of coarse aggregate at the bottom of RCC lifts has
led to decisions to reduce the MSA in some cases or to increase the
proportion of sand in the mix in other cases. Most soils-approach RCC
mixes have a greater percentage of fines than conventional concrete
mixes. This is particularly so if the fines are nonplastic, fill voids in
the aggregate, and lead to decreased water demand and improved
In some instances, aggregate produced for applications other than
concrete can be used in RCC. Among them, an available free-draining
gravel produced for a zone in an embankment dam and a gravel base
course and an aggregate produced for asphalt paving have been used
for some small projects. The function of the completed structure and
the minimum required properties of the RCC should determine what
aggregate is chosen.
Rounded river gravels and crushed aggregates have been used for
RCC. At Copperfield Dam, difficulty was encountered maintaining an
24 Chapter Two

unformed 0.8 H : 1 V downstream slope using a mix containing

rounded gravel. No problem was encountered in constructing the
same slope at Willow Creek and Galesville dams using crushed aggre-
gate in the RCC mix. Quality. The quality of the aggregate required depends on the

desired properties of the RCC, primarily its strength. For high-strength
RCC, high-quality aggregate is needed. Standard tests to determine the
characteristics and quality of aggregates are listed in Fig. 2.6. Informa-
tion on grading and a means of determining the quality of the aggregate
are required in the early stages of project design. Past experience with an
aggregate source provides an indication of its quality.
For RCC that is not highly stressed or not exposed to freeze-thaw
conditions while wet, lower-quality aggregates may be used. This ap-
plies mainly to the interior concrete.
Suitable aggregate for RCC can come from a variety of sources, but
the material closest to the dam site should be investigated first. At
Middle Fork Dam, an on-site marlstone (oil shale) was successfully
used rather than importing a known, acceptable river gravel from a
source 20 mi (32 km) away. It has been assumed that concrete made
using marlstone aggregate would not be as strong or durable as con-
crete made using granite, basalt, or sandstone. Tests done for Middle
Fork proved those assumptions wrong. The RCC had a high compres-
sive strength versus cement content, released heat slowly and had a
high ratio of tensile strength to compressive strength. Durability was
not a consideration because the RCC was capped with conventional,
air-entrained concrete using the river gravel as aggregate. Grading. Grading for both coarse and fine aggregate (less
than 4.75 mm) and the proportions used have an important effect on
the properties of RCC. Specifications for grading of aggregate have
varied considerably.
Because the goal for concrete approach mixes is to fill all aggregate
voids with paste, a well-graded aggregate designed to produce mini-
mum voids is more important for these mixtures than for soils ap-
proach mixtures. For many of those, the goal has been to use an ac-
ceptable grading that results in the lowest overall cost of the RCC.
The difference in mix design philosophy has produced some differ-
ing trends with respect to specifying aggregates for RCC. This is es-
pecially true with respect to maximum size aggregate (MSA), percent-
age of sand and fines desired, and the number of separate sizes
processed and then combined to produce the desired grading.
For concrete-approach RCC mixes, aggregate requirements are very
Figure 2.6 Characteristics and tests for aggregates for RCC dams.

Test designation
Characteristic Significance (U.S.) Test name
Grading Consistency, compactability ASTM Cl36 Sieve Analysis of Fine and Coarse
economy Aggregate
ASTM Cl17 Materials Finer than 75 km (No.
266) in Mineral Aggregates by
Resistance to abrasion Aggregate quality, wear re- ASTM Cl31 Resistance to Degradation of Small
sistance of surface Coarse Aggregate by Abrasion and
Impact in the Los Angeles Ma-
ASTM C535 Resistance to Degradation of large
Coarse Aggregate by Abrasion and
Impact in the Los Angeles Ma-
ASTM C295 Petrographic Examination of Ag-
gregates for Concrete
Specific gravity-absorption Mix design calculations ASTM Cl27 Specific Gravity and Absorption of
Coarse Aggregate
ASTM Cl28 Specific Gravity and Absorption of
Fine Aggregate
Bulk unit weight or density Mix design calculations ASTM C29 Unit Weight and Voids in Aggre-
Sulfate resistance Soundness against weather- ASTM C88 Soundness of Aggregates by So-
ing and chemical attack dium Sulfate or Magnesium Sul-
Organic impurities Strength gain ASTM C40 Organic Impurities in Fine Aggre-
gate for Concrete
26 Chapter Two

similar to that required for conventional mass concrete. The most

common MSA has been about 3 in (75 to 80 mm), although Tamagawa
Dam used 6-in (150-mm) MSA, and Upper Stillwater Dam required
2-in (50-mm) MSA. For a 3-in MSA, four aggregate sizes have tradi-
tionally been produced, especially for large-volume RCC dams. The
ranges are 1.5 to 3 in (38 to 75 mm), 3/4 to 1.5 in. (19 to 38 mm), U.S.
No. 4 sieve size to % in. (4.75 to 19 mm), plus sand (less than 4.75
mm). The four aggregate sizes are then blended to produce the desired
grading. Another possible method of producing similar results at
lower cost is to start with a crushed river-run aggregate (MSA to
fines) and add one or more of the sizes to meet grading requirements.
Sand percentages have generally been between 30 to 35 percent of
total aggreggate. The percentage of fines passing the No. 200 sieve
(0.075 mm1 has usually been limited to 3 percent of the total weight of
aggregate, especially if a high percentage of pozzolan is used in the
mix. At Elk Creek Dam, the range of minus 200 fines required was 10
to 18 percent of the fine aggregate, which calculates to be about 3 to 6
percent of the total weight of the aggregate.
Soils approach mixes specified for many early RCC dams required
3-in (75-mm) MSA and 30 to 35 percent sand. However, with these
drier-consistency mixtures, there is a greater tendency for the larger
particles to segregate during transport, deposition, and spreading.
Segregation can be minimized by reducing the MSA and by increasing
the percentage of sand. There is a trend toward 2-in (50-mm) MSA and
sand percentages in the 35 to 40 percent range, primarily for mixtures
that conform to the soils philosophy.
Many soils approach mixes have used a single combined aggregate
grading from the MSA to minus 200 fines. Two of these continuous
aggregate grading bands are shown in Fig. 2.7. The shapes of bands
are similar, although the allowable limits for each sieve size for the
2-in (50-mm) MSA for the Stacy Dam spillway are slightly more open
than the grading band for the 3-in (75-mm) MSA used for Galesville
Dam. If two sizes of aggregate are used, they are separated in piles of
plus and -% in (19 mm).
The amount and type of minus 200 fines allowed have varied con-
siderably. It has ranged from 0 to 3 percent of total aggregate for
some concrete approach mixes to 8 to 16 percent for the small North
Loop detention dams at Austin, Texas. In many cases, unwashed ag-
gregate is appropriate for RCC.
As noted in Fig. 2.7, the allowable range for nonplastic fines was 1.5
to 10 percent at Galesville, while Stacy had a 4 to 12 percent range
with the plasticity index (PI) limited to 4. The aggregate actually used
at Stacy contained about 10 percent nonplastic fines. In the U.S. Army
Corps of Engineers manual for RCC, an allowable range of 4 to 7 per-
Roller-Compacted Concrete as a Material 27

90 SIX % Filssln~
4 I 100
80 3 I 98 - 100
2 in 86- 96
I-1/2in 72- 92
1 I 56- 66
3/4 I 49- 5 9
3/6 I 38-48
No. 4 30- 4 0
No 6 23- 3 3
No. 16 16- 26
Na 3 0 13- 2 3
No. 50 lo- 16
No. 100 7- 1 4
No. 200 i5- 1 0
Gram SILOS. mm
GV.lVd I Sand Slit or clay
Cobbles Fine
Coarse Fins Medium


US. stondord sieve sizes

= _!
Size % Passing

2R I 100

2 in 95 - 100
?E 7 0 3 in 60 - 85
% 60 shin 42- 5 4
h 50 No 4 30-42
F 40 No. f3 23 - 35
; 30 No. 30 10-21
a No. 200 4 - 12

100 50 10 5 1.0 0.5 0.1 005 001
Gram sizes. mm
GWVd I Sand
coarse 1 Fine 1 coarse 1 Medium Fine

Figure 2.7 Two RCC aggregate gradation bands. (a) Galesville Dam; (b) Stacy
Dam. [From Oberholzer (1985); Lemons (1988j.J
28 Chapter Two

cent of nonplastic fines is proposed. The allowable percentage is re-

duced for more plastic fine aggregates, which are defined as those hav-
ing a liquid limit (LL) greater than 25 and a PI greater than 5.
RCC mixes made with excessive amounts of cIay fines have shown a
higher water demand due to the surface activity of the clay minerals.
The increased water content increases shrinkage in the RCC and cre-
ates a greater potential for cracking and reduced strength. The gen-
eral consensus is that fines should be nonplastic and allowed only to
the extent that they fill voids to reduce water requirements and im-
prove compactibility.
Pit-run or as-dug aggregates produced with little processing except
screening of oversize rock have been used in RCC mixes, particularly
for dam modification projects. The largest of these was the 3.5million-
yd3 (2.7-million-m3) mix of RCC used at Tarbela Dam in Pakistan.
This type of aggregate would be appropriate for RCC cofferdams. Cost
and resistance to overtopping are major considerations for these struc-
tures which have a relatively short life.

2.2.3 Water
The only requirement for water in RCC mixes is that it be free from
excessive amounts of alkalis, acids, or organic matter that might in-
hibit proper strength gain. Most RCC mixes require 150 to 200 lb of
water per cubic yard (89 to 119 kg/m31 for MSA greater than 2 in (50

2.2.4 Admixtures
With little success, air-entraining, as well as water-reducing and set-
retarding, admixtures have been tried in RCC mixtures, with propor-
tions based on soils principles. Due primarily to the dry consistency
and fines content of these mixes, a proper air-void system has not been
established at any application rate using normal batching or propor-
tioning procedures.
There is a better chance for admixtures to be effective in wetter-
consistency mixes associated with the concrete approach. Air-
entraining and water-reducing admixtures are introduced into all
Japanese RCD mixtures. A water-reducing, set-retarding admixture
was used in the RCC for Elk Creek Dam. A relatively high dosage of
14 to 21 oz per hundredweight (0.87 to 1.3 kg per 100 kg) of portland
cement showed good results. Batch water was reduced by 27.5 lb/yd3
(16.3 kg/m31 and the initial design Vebe time was reduced from 20 to
10 s. Mix design investigations revealed that the use of the water re-
Roller-Compacted Concrete as a Material 29

ducer retarded compressive strength development at early ages from 7

to 55 days and enhanced the strength gain after 55 days.

2.3 Laboratory Tests

Most of the laboratory tests used for RCC to date are standard tests
developed for conventional concrete or, in some cases, soils. They in-
clude tests for acceptance and properties of materials, as well as meth-
ods for handling and breaking specimens, Two procedures that are
done differently in testing RCC, however, are specimen preparation
and consistency tests.

2.3.1 Specimen preparation

At present, there are no generally recognized standards for the prep-
aration of laboratory specimens to determine properties of RCC mix-
tures. The problem is in preparing lab specimens that produce prop-
erties nearly the same as RCC placed in the field, where trucks,
dozers, and vibratory rollers compact and consolidate the mix.
The task of the lab technician is to produce a lab specimen that has
the same density as the final product in the field. This properly as-
sumes that density is proportional to another desired property of the
mix, most often compressive strength for the same mix.
Most specimens have been prepared in standard 6-in-diameter, 12-
in-long (152 by 304-mm) concrete cylinders, which are readily avail-
able and can accommodate MSAs up to 2 in (50 mm). For RCC mixes
using larger MSAs, aggregates larger than 2 in (50 mm) can be wet-
screened out or larger, custom-fabricated molds can be used. Split-
cylinder molds are usually used to allow easier removal of the com-
pacted, dry RCC mixes. Plastic molds inserted in an oversized steel
cylinder have also been used to facilitate specimen removal.
Cylinder preparation falls into three basic methods: impact compac-
tion, vibration, and tamping.

2.3.1 .l Impact compaction specimens. Impact compaction methods for

the preparation of specimens have been limited to the drier-con-
sistency soils approach mixes. The number of layers used in preparing
test cylinders has varied from three 4-in (loo-mm) layers to six 2-in
(50-mm) layers. The number of layers is not as important as the type
of hammer or rammer used and the amount of compactive effort or en-
ergy applied to the material in the rigid cylinder.
Most of the procedures for preparing impact compaction specimens
have been done with the same type of equipment that is used for the
30 Chapter Two

modified Proctor test procedure for soils (ASTM D1557). The test em-
ploys a lo-lb (4.5-kg) hammer that drops 18 in (450 mm) before strik-
ing the surface of the test material. The actual modified compactive
effort equates to 32.55 ft-lb/in3 (2693 kJ/m3).
Lowe used modified Proctor compaction in 1960 to determine the op-
timum moisture content for the rollcrete core of the cofferdam at
Shihmen Dam in Taiwan. A ll-in-diameter (356-mm) compaction
mold was used to perform the tests on the RCC, which was made with
a 3-in (75mm) MSA.
Reeves and Yates later determined that lab specimens compacted
with modified Proctor compactive effort correlated very well with
measured dry densities for the North Loop detention dams at Austin,
Texas, for an RCC made with 1.5-in (38-mm) MAS limestone. The
RCC was placed in the cylinder in six 2-in (50-mm) layers and re-
ceived 122 blows per layer.
More recent tests by Casias, Goldsmith, and Benavidez of the U.S.
Bureau of Reclamation indicate a lower compactive effort may be
more appropriatae than the modified Proctor test procedure. They
found that the modified Proctor compactive effort using the standard
hammer caused too much fracturing of the coarse aggregate. The
amount of aggregate fracturing or gradation change during impact
compaction tests is a function of the particular aggregate. Hard, sound
aggregate should be able to absorb more energy without breakdown.
They concluded that a compactive efort of 13.3 ft-lb/in3 (1100 kJ/m3)
provided better test speciments by impact compaction. This degree of
compaction was obtained by placing the RCC in six 2-in (50-mm) lay-
ers and giving each layer 50 blows.
Wong et al. also noted the breakage of aggregate using the small
sector-shaped hammer associated with the modified Proctor test pro-
cedure. They therefore developed a new rammer contact face consist-
ing of a 5.5-in-diameter (140-mm) flat disk with rounded edges. The
total rammer weight remained at 10 lb (4.5 kg). There was a %-in.
(6-mm) clearance between the disk and the mold. A higher density
was achieved with the modified rammer compared with the standard
rammer for the same 93 blows given to each of three 4-in (lOO-mm)
layers. The tests were done without cementitious materials and the
difference amounted to 1.3 lb/ft3 (20.8 kg/m31 greater density. The
compactive effort used was equal to 12.2 ft-lb/in3 (1,009 kJ/m3). Vibrated specimens. Vibrated test specimens are used prima-

rily for concrete approach RCC mixtures designed to have more paste
than aggregate voids. A 6 by 12-in (152 by 304 mm) steel cylinder is
Roller-Compacted Concrete as a Material 31

rigidly clamped to the same vibrating table used in the Vebe test (see
Sec. and filled in three equal layers. A 20-lb (9.1-kg) weight is
placed on top of each layer and the cylinder is vibrated until paste
forms around the edge of the surcharge. After the third repetition, the
excess concrete is struck off and the cylinder is capped for later test-
ing. This method is described in the U.S. Bureau of Reclamation stan-
dard 4906-86. A similar method is used for the RCD method in Japan
except the size of the specimen is approximately 9 in (240 mm) in di-
ameter with a height of 8 in (200 mm). Tamped specimens. Another method for producing RCC cylin-

ders involves tamping, which can be accomplished by two distinct
methods: (1) the use of a pneumatic pole tamper, and (2) the use of an
electric-powered vibrating rammer. The latter apparatus is known as
a Kango hammer in Europe and elsewhere and a Hilti gun in the
United States. For RCC cylinder preparation, both use a 5- to
5%in-diameter (127- to 146-mm) flat face for the tamper.
There is a marked difference between the amplitude or stroke and
frequency of the two tampers. The pole tampers, also called pogo
sticks or jumping jacks, have a stroke of about 6 in (150 mm), while
the vibrating rammers have a very low amplitude that is impercepti-
ble to the naked eye. The frequency of the rammers varies from 2000
to 2400 impacts per minute, while the frequency of strokes for the pole
tamper is less than 600 impacts per minute. The actual frequency de-
pends in part on the efficiency of the units air compressor. The pole
tamper may be described as more of an impact compaction method
than a tamping method for producing RCC specimens in the lab or in
the field.
Both pole tampers and vibrating rammers have been used exten-
sively for cylinder preparation of RCC mixes that conform to the
soils approach. It appears the pole tamper cannot be used effec-
tively with concrete approach mixes since paste would come to the
surface quickly and adhere to the tamping face. In most cases, the
traditional 6 by 12 in (152 by 304 mm) cylinders have been used
with either the pole tamper or vibrating rammer. For either
method, large aggregate above 2 in (50 mm) is removed from the
mix by wet-screening prior to placement in the cylinder. The use of
either apparatus usually requires a cylinder extension in order to
compact the top lift properly.
With the hand-held pole tamper shown in Fig. 2.8, the tamping con-
tinues until the hammer begins to rebound off the aggregate, indicat-
ing no greater density can be achieved. The time required to reach
32 Chapter Two

Figure 2.8 Pneumatic pole tamper to prepare RCC cylinders.

this condition has varied from 8 to 10 s at Stagecoach Dam to 15 to 20

s per layer at Copperfield Dam.
The Kango hammer (Fig. 2.9) and Hilti gun can be either hand-held
or frame-mounted. More consistent results are obtained if the appara-
tus is frame-mounted in the laboratory and if a constant load is ap-
plied that is concentric with the shaft. Several models have been
used-ranging from a 77-lb (35-kg) unit, the Kango Model 950, to a
lighter Hilti Model TP-400, which weighs 14.3 lb (6.5 kg) with or with-
out surcharge-as engineers have attempted to select a model that ap-
proximates the pressure and frequency of the vibratory rollers to be
used. The tamping time for each of the three or four layers to achieve
no further increase in density of the RCC in the mold has been 20 s or
Neither method produces a specified or constant amount of
compactive energy to the material being compacted, and compaction
of pole-tamped specimens is operator-dependent. Once operators are
experienced with the procedures, the resulting cylinders have proven
to be reasonably uniform. Then, if the density of the cylinders can be
equated to the density produced in a test section, representative spec-
imens may be expected.
Roller-Compacted Concrete as a Material 33

(a) (b)
Figure 2.9 (a) Kango hammer and (b) Hilti gun to prepare RCC cylinders. Comparison of laboratory test specimens and project cores.Lab

cylinders are used primarily to evaluate various mixtures to deter-
mine if a given mixture can be expected to produce the strength re-
quired by the designer at a specific age. Cylinders prepared in the
field at the time of construction have been used mainly for record-
keeping purposes rather than for construction control. That is due
mainly to the rapid construction of RCC dams.
Figure 2.10 presents a comparison of the compressive strength of
laboratory cylinders and cores at various ages. For the tamping im-
pact compaction methods of preparation, the cores had greater com-
pressive strength than the cylinders at comparable ages with the ex-
ception of Willow Creek Dam. Most of the cylinders were prepared
using a pneumatic pole tamper.
In Australia, Forbes first used the pole tamper for cylinder prepa-
ration at Copperfield and Craigbourne dams and then changed to
modified Proctor compaction (ASTM 1557) for Bucca Weir. He deter-
mined that modified Proctor, which has a specified amount of
compactive effort, produced more consistent laboratory results.
34 Chapter Two

Cylinders Cores (without jonts?

Camp. Camp.
strength, Core strength, Greater
lb/ins dia., in lb/in* strength at
Dam Mix* Age (MPal Age (mm1 (MPal equal age
A. Pole Tamper for Cylinder Preparation
1 . Willow Creek A Iint.1 1 yr 2623 1 yr 8 2300 Cylinders
B (upstr.1 1 yr 3779 1Yr 8 2150 Cylinders
C (down) 1 yr 4146 1Yr 8 3220 Cylinders

2. Middle Fork 28 days 1270 42 days 6 2016 Cores

90 days 1650
3. Copperfield A (est.) 28 days 667 28 days 6 946 Cores
90 days 1040 56 days 6 1400 Cores
13 tint.) 90 days 870 172 days 6 1247 Cores (est.)
4 C~alesvllle A (lnt.) 1 yr 1561 14 mo 6 2170 Cores
B iext.1 1 Y 1881 14 mo 6 2000 Cores (est.)
5 Stageconcll 1 yr 1250 1 yr 6 1920 Cores

B. Vebe Apparatus for Cylinder Preparation

1 . Upper A (int.1 1 yr 6174 lYr 6 5171 Cylinders
2. Castilhlanco de (upper) 91 days 2910 90 days 6 2770 Cylinders
10s Arroyos (lower) 91 days 3650 90 days 6 3350 Cylinders

C. VC Apparatus for Cylinder Prepapation

6. Asahi Ogawa

D. Modified Proctor Impact Compaction for Cylinder Preparation

1 . Bucca Weir 90 days 2900 1Yr 4 3800 Inconclusive
2. Stacy Spillway 28 days 2615 28 days 6 2089 Cylinders
90 days 3095 90 days 6 2576 Cylinders

6. by 1%in size (1 in = 25.4 mm; 1 lb/in = 6.9 x 10 MPa.1

12 in (150 mm) long except Japan 8 in (100 mm) long.
+Int. = interior, upstr. = upstream, down. = downstream, ext. = exterior, low. =
lower,up. = upper.

Figure2.10 Comparison of compressive strength of RCC cores and laboratory cylinders.

(See Fig. 2.14 for mixture proportions.)
Roller-Compacted Concrete as a Material 35

A standard deviation of 10.5 percent was achieved for go-day com-

pressive strength breaks based on 55 tests at Bucca Weir as compared
to a standard deviation of 24.7 percent for the earlier Copperfield Dam
Laboratory-prepared specimens using modified Proctor impact com-
paction energy of 32.55 ft-lb/in3 (2693 kJ/m3) have compared more fa-
vorably with in-place densities and strengths than have specimens
prepared with any lesser compactive effort. Tayabji and Okamoto did,
however, report excellent correlation in one case. They compared 60-
day-old cylinders prepared with a compactive effort of 14.2 ft-lb/in3
(1175 kJ/m3) to the strength of 6-in (150-mm) cores drilled from a test
section. The cores were tested between 65 and 82 days after place-
ment. Eighty blows of the modified Proctor hammer on four 3-in (75-
mm) layers produced the compactive effort compared to 122 blows of
the same lo-lb (4.5-kg) hammer on six 2-in (50-mm) layers for modi-
fied Proctor compaction energy.
Cores have generally had less average compressive strength than
cylinders prepared by vibrating laboratory specimens. There are no
available comparisons between tamped cylinders prepared with a
Kango hammer or Hilti gun and core samples at the present time.
However, at Tarbela Dam, Johnson and Chao reported the vibratory
frequency of the Kango hammer more nearly approximated the action
of the vibratory rollers than the modified Proctor test procedure.
Mageroy also reported the Kango hammer specimens produced more
exact average results when compared to field wet densities than mod-
ified Proctor using optimum moisture content at an RCC-paved tunnel
project in Norway. The Kango hammer cylinders did, however, have
more spread in the test results than modified Proctor.
From the results shown in Fig. 2.10, there is considerable difference
in the compressive strength of cores and cylinders. There are many
variables that can account for the strength differences, including ma-
terial, temperature, curing, and delay in compaction variations. Also,
there may be possible variations in cylinder manufacture, handling,
and breaking as well as construction variations such as segregation at
the bottom of the lifts. Still the greatest variations in strength proba-
bly occur due to differences in moisture content and compactive effort
between lab cylinders and the cores, resulting in differences in den-
sity. It should also be noted that the mixture in the cylinders usually
had material greater than 1% or 2 in (38 or 50 mm) screened out and
therefore was not identical to the mixture sampled from the dam.
The difference in strength between lab-prepared cylinders and
project cores, together with the variation in methods to prepare cylin-
ders, indicates that standard methods for RCC cylinder preparation
36 Chapter Two

are needed. The American Society for Testing and Materials estab-
lished a committee to develop such standards in 1988.

2.3.2 Consistency tests

The main purpose of consistency tests is to determine how much water
is required to achieve the desired strength properties and to produce a
mix suitable for compaction by external rolling. For the soils ap-
proach, the desired water content is determined by moisture-density
relationships. For the concrete approach, the water content of the mix
is determined by using a vibrating table to achieve the desired time
for the paste to start appearing on the surface of the RCC mixture. For
both approaches, an RCC mixture is used inwhich the only variable is
the water content. Moisture-density relationship. Specimens for moisture-density

tests are prepared as noted in Sec., which describes impact
compaction specimens. RCC mixes with varying moisture contents are
compacted into the 6 by 12 in (152 by 304 mm) cylinder at modified
Proctor compactive effort or some other selected energy level. The op-
timum moisture is the level that produces a maximum dry density. A
five-point curve is general practice.
Optimum moisture content is thus determined for construction and
should be in the mix at the time of compaction, not at the time of mix-
ing. It therefore may be necessary to introduce more water than opti-
mum during mixing to account for moisture loss due to handling,
evaporation, and early hydration of cement. Also, field adjustments
may have to be made to produce a more compactible mix as deter-
mined by construction of a test section.

The purpose of vibration tests is to establish a Vibration tests.
water content that corresponds to a desired consistency. While there
are a number of similar tests to measure this consistency, they all fol-
low three basic steps:

1. A container or pot is filled loosely with uncompacted concreie, lev-

eled off, and a surcharge is applied to the RCC.
2. The cylinder is attached to a vibrating table which has a constant
frequency and amplitude. The specimen is then vibrated with the
surcharge on the surface until it is fully consolidated.
3. The time in seconds is noted when a ring of paste has formed com-
pletely around the inside edge of the cylinder. This time is the mea-
sure of the consistency or workability of the mix.
Roller-Compacted Concrete as a Material 37

The basic consistency test for no-slump RCC was conceived during
the 1970s by Cannon of the Tennessee Valley Authority. He used a
nonstandard vibrating table, a cylinder filled to the top, and no sur-
charge. Most agencies now use a modified Vebe test. It makes use of
the same %-ft3 (0.0094-m3) container and vibrating table (made by
Dynapac Maskin AB of Sweden) used in the Cannon test. The modifi-
cation is the addition of a surcharge to the loose, leveled concrete. A
weighted Vebe apparatus is shown in Fig. 2.11.
The total weight of the surcharge used varies considerably. In Ja-
pan it is 44 lb (20 kg). The Corps of Engineers uses 27.5 lb (12.5 kg),
and the Bureau of Reclamation uses 50 lb (22.7 kg). As the surcharge
weight increases, the tendency is for the Vebe or VC time to decrease
for the same mix. In the mix design investigations for Upper
Stillwater Dam, a Vebe time of 35 to 45 s was obtained with no sur-
charge. That compared to 25 to 35 s with a 50-lb surcharge for the
same mix.
RCD tests in Japan are done using tables that vibrate at 4000 cycles
per minute (pm) (versus 3600 cpm for the tables initially manufac-
tured in Sweden). Two sizes of containers are used in Japan. The stan-
dard container for RCD has about the same volume as the standard
Vebe container but is about 8 in (200 mm) deep and 9.4 in (240 mm) in
diameter. It is used for preliminary mixture proportioning studies and
for quality control during construction. For tests on mixtures with

Figure 2.11 Modified Vebe appa-

ratus. [From EM 1110-2-2006
U.S. Army Corps of Engineers,
(I 985)J
38 Chapter Two

MSAs up to 6 in (150 mm), a larger container is used, measuring 19 in

(480 mm) in diameter and 16 in (400 mm) deep.
The desired VC time is selected by the designer and is confirmed or
modified in a test section. The desired value in Japan is 20 s, plus or
minus 10 s. At Upper Stillwater Dam, the average Vebe time was 29
s in 1985. After a change in the mix design in 1986 to increase its
workability, the Vebe time dropped to an average of 17 s for the pre-
dominant mix.
A number of mix proportion factors affect the Vebe time. Any
factor that tends to stiffen the mix, such as a higher sand content
and higher temperature, increases the Vebe time. Still, the primary
factor is the amount of water in the mix. Lower Vebe times indicate
a greater water content, thus producing a more fluid consistency.
The relationship between water content and VC value is shown in
Fig, 2.12, which includes data from both the standard and larger
Japanese cylinders. A Vebe time of about 5 s would equate to the
highest water content for the no-slump consistency required for

C + FA = 120 kg/m3 (202 lb/yd3)

Large container

5 0 -
Standard container

110 120
(186.3) (202.5)
Watercontent, kg/m3 (Ib/yd3)

Figure 2.12Relationship between water content and VC

time using large and standard-size containers. [From
Hirosc and Yanagida (1984,J.I
Roller-Compacted Concrete as a Material 39

2.4 Mixture Proportioning

The basic objective in proportioning RCC mixtures is to produce a con-
crete that satisfies the performance requirements using the most eco-
nomical combination of readily available materials that can be placed by
roller compaction methods. The desired physical properties of the mix de-
pend on the function, location, and design chosen for the structure. For
lean RCC dams designed with a relatively impermeable upstream face,
the primary mix design requirement is compressive or shear strength.
The strength is dictated by minimum structural safety requirements
with some overdesign factor to account for variability in the mix together
with appropriate factors of safety. See Figure 3.6 for a list of minimum
RCC design-strength requirements used for various dams.
In the design of high-paste RCC dams, where the entire section is
considered to be the water barrier, cohesion between successive lifts of
RCC and the permeability of the concrete are the controlling mix de-
sign requirements. At Upper Stillwater Dam, the RCC mixes were de-
signed to achieve a 180-lb/in2 (1.24-MPa) direct tensile strength at one
year. The design of Pamo Dam required a 350-lb/in (2.41-MPa) dy-
namic splitting tensile strength at one year to satisfy seismic criteria.
RCC that is exposed and subjected to severe climatic conditions and
high seismic velocities must be designed for durability. A compressive
strength as high as 5000 lb/in2 (34.5 MPa) at 28 days could be re-
Five methods for determining RCC mixture proportions for dams
are cited in the American Concrete Institute (AC11 report 207.5R on
roller-compacted mass concrete. Three mix design methods are de-
scribed in detail in the report and referenced publications are given
for two more methods.
The three methods described are called: (1) proportioning RCC to
meet specific limits of consistency, (2) trial mixture proportioning for
the most economical aggregate-cementitious materials combination,
and (3) proportioning using soils compaction concepts. The AC1 207 re-
port also references the methods used by the Japanese Ministry of
Construction and the U.S. Army Corps of Engineers.
Methods 2 and 3 fall under the soils approach to RCC, while the oth-
ers are concrete approaches because they depend on vibration tests to
determine the water content of the mix. The Corps of Engineers
method also can be used to produce a mix with a water content that
would place it in the soils approach category.
All of the mix design methods should start with the same two steps:
Select the desired properties of the concrete and then determine the
properties of the materials to be used.
Based on the type of structure, the availability of materials, and
their cost. some of the considerations that need to be addressed ini-
40 Chapter Two

tially include the quality and maximum size of the aggregate to be

used, the type of cement, and whether pozzolan will be used and to
what extent. All of the methods should include the preparation of trial
mixes to confirm that the consistency is suitable for roller compaction.
This is usually confirmed in a test section using the placing methods
and equipment that are planned for use on the dam. If the laboratory-
determined mix proves unsuitable for construction, the mix must be
adjusted accordingly.

2.4.1 Soils approach mixture-proportioning

The two mix design methods that fall within the soils approach will be
called the lean RCC method and the simplified soils method in this
text. They are quite similar. They both start with a desired grading
for the aggregates and involve the preparation of cylinders with vary-
ing cementitious contents to determine strength or other properties.
Differences in the two methods center on how the moisture or water
content for the mix is determined and the method for preparation of
laboratory test specimens. The lean XC method. This method is advocated by Schrader

and has been used for most lean RCC dams. It starts with a fmed ag-
gregate grading, varies cementitious contents, and compares results,
primarily compressive strength, with project requirements.
The continuous aggregate-grading band as shown in Fig. 2.7 for
Galesville Dam is typical of this method. A 3-in (75mm) MSA is usu-
ally selected for what is termed the most economical usable gradation.
The amount of water used for laboratory trial mixes is determined by
observing the consistency of mixes of varying water contents and by
relying on past experience. The water content is set somewhere be-
tween the point on the dry side where voids are no longer visible on
the side of laboratory cylinders, and, on the wet side, before the mix
has a rubbery appearance.
With the aggregate grading and water content now fmed, labora-
tory cylinders are prepared with varying cementitious contents using
a pneumatic pole tamper as described in Sec. Most mixes that
have resulted from this method have varied from 100 to 175 lb of ce-
ment per cubic yard (60 to 104 kg/in31 except for special mixes. This
level of cement content provides a good starting point for laboratory
mix design investigations.
If use of a pozzolan is desired, another set of specimens should be pre-
pared using a set percentage of pozzolan with respect to total
cementitious content. This percentage usually varies from 25 to 50 per-
cent. Typically, the laboratory mix design program consists of two cylin-
Roller-Compacted Concrete as a Material 41

ders prepared for testing at 3, 7, 14, 28,90, 180, and 365 days. If there is
insufficient time to obtain results at later ages, the compressive
strengths can be estimated based on the curve shape for the early ages,
test results from previous projects using the same method , or by accel-
erated test methods. Initial work on accelerated laboratory tests for RCC
mixtures were accomplished by Forbes for Bucca Weir in Australia and
Logie and Oliverson for the not yet constructed Pamo Dam.
The mix design program thus provides a family of curves that indi-
cate the effects of various cementitious contents on compressive
strength at various ages. The cement content can be selected to meet
project requirements with consideration of factors of safety and coeffi-
cients of variation. Once a cement content is selected, additional tests
may be run with varying aggregate types or gradings, especially the
percentage of fines passing the No. 200 (0.075-mm) sieve.
The coefficient of variation of compressive strength of cylinders
from six projects using the lean RCC mix design method and prepared
using pneumatic pole tampers has ranged from 22 to 28 percent. In
addition to material variations, a visual means of determining water
content and the nonstandard means of cylinder preparation are major
factors in producing coefficients of variation of this magnitude. The simplified soils method. RCC mixtures have been propor-
tioned since 1960 using soil compaction principles. The same funda-
mentals have been the basis for determining cement content for soil-
cement mixes for more than 50 years. The basic method is quite
similar to the lean RCC method in that it starts with a fixed aggre-
gate grading and involves a test program of varying cementitious con-
tents and comparing results once a water content is determined. An
aggregate grading band for Stacy Dam in which the mix was designed
using this method is shown in Fig. 2.7.
Rather than a visual determination of water content, the opti-
mum moisture content is determined by the moisture-density princi-
ples described in Sec. 2.1.1, using impact compaction with a standard
hammer or rammer dropped a prescribed number of times. A modified
Proctor compactive effort of 32.55 ft-lb/in3 (2693 kJ/m3) has been used
for most actual projects, although some lesser compactive effort has
been suggested by some researchers (see Sets. and The
number of blows from a lo-lb (4.5-kg) hammer dropping 18 in (450
mm) per unit volume defines the compactive effort.
To solve the aggregate breakage problems encountered by some re-
searchers with the standard modified Proctor rammer, the larger, 5.5-
in-diameter (140-mm) flat disk with rounded edges may be used (see
Sec. The remaining steps in the mix design method are then
identical to those described for the lean RCC method (Sec.
42 Chapter Two

The simplified soils method has been used with a fixed conservative
cementitious content for some relatively small volume dams where
heat generation was not a concern and where time for a longer, more
extensive mix design program was not available. The North Loop de-
tention dams in Austin, Texas (see Sec., used a set 200-lb/yd3
(119 kg/m3) of cement and 80 lb/yd3 (47 kg/m3) of pozzolan in the RCC
Likewise, this mix design method has been used extensively for the
modification or rehabilitation of existing dams where the primary de-
sign consideration for the exposed RCC has been the durability and/or
erosion resistance of the material. In these cases, the laboratory in-
vestigation can involve varying cementitious contents to achieve a de-
sired level of durability as measured by the loss of weight after 12 or
more cycles of freezing and thawing similar to that developed for soil-
cement. The design mix durability can also be based on the loss of
weight of specimens subjected to high-velocity water jets or on a min-
imum compressive strength. For many of the mixes designed by the
simplified soils approach, the cement and pozzolan contents are ex-
pressed as a percentage of dry weight of aggregate. The percentages
are helpful in the volumetric proportioning of RCC associated with
most pugmill mixing operations.

2.4.2 Concrete approach proportioning

As noted in Sec. 2.4, proportioning RCC mixes using vibration consis-
tency tests can be accomplished by a number of methods. The method
called proportioning RCC to meet specified limits of consistency by
AC1 is referred to here as the high-paste method. It will be described
below in addition to the Japanese RCD method and the Corps of En-
gineers method.
Because all of them are based on a Vebe time (or VC valuej-indi-
eating full consolidation of the RCC-the basic premise of these meth-
ods is that the volume of paste must exceed the voids in the aggregate.
Therefore, there is a greater need to closely control the aggregate
grading to minimize voids and the amount of paste required. All in-
volve proportioning mixes using absolute volume concepts in which
the weights and specific gravities of all materials are used to calculate
a unit volume of concrete. The final mix, therefore, consists of batch
weights to produce a cubic yard or cubic meter of RCC.
Concrete approach mix design methods usually involve fixing all
but one of the basic materials (cementitious materials, water, or ag-
gregate content) and then varying that component until the desired
consistency or required properties are acheived. Each variable can be
adjusted this way to optimize all mix components.
Roller-Compacted Concrete as a Material 43 The high-paste method. The high-paste mix design method

was developed by Dunstan and modified by the U.S. Bureau of Recla-
mation for its design of Upper Stillwater Dam. The method is closely
aligned to the design concepts for the high-paste RCC dam described
in Chap. 5 in which the entire structure is considered to be imperme-
able and bonding between lifts is achieved due to the character of the
mix. In order to satisfy these criteria, a relatively high cemetitious
content mix (see Fig. 5.1) producing a high-paste content RCC is re-
In designing a mix for a high-paste-content RCC dam, two conflict-
ing requirements must be resolved. Sufficient cementitious material
is needed to acheive a low permeability and assure the bond (cohesion)
between successive lifts of RCC. At the same time, the volume
changes produced by heat generated by the cementitious materials
must be minimized. The problem has been solved with liberal substi-
tutions of pozzolan for cement, assuming a suitable pozzolan is avail-
able at a reasonable cost.
The steps in the mix design procedure used by the Bureau of Recla-
mation are

1. In addition to determining densities and thereby specific gravities

of the cement (Cl, fly ash (FA), coarse aggregate (CA), water (WI,
and sand, the void ratio of the total aggregate is determined. The
CA and sand conform to standard Bureau of Reclamation grada-
tions for conventional concrete.
2. Determine a required W/C t FA ratio by weight based on the de-
sign compressive strength requirements at a certain age. For 4300
lb/in (29.7 MPa) at one year, a W/C t FA of 0.5 is required,
whereas for 2300 lb/in2 (15.9 MPa) at one year the W/C t FA is 0.7.
3. Determine a relationship of C to FA that will produce the desired
compressive strength within a specified time. For a one-year
strength, the Bureau of Reclamation uses 25 percent cement and 75
percent fly ash. Now, the proportions of cement, fly ash, and water
can be calculated for a unit-paste (C! t FA t W) volume.
4. Depending on the time allowed for an exposed lift in the dam, a
paste/mortar (p/m) ratio is selected. For a lift age between 12 and
24 hours, a p/m ratio of 0.39 is used.
5. Determine a mortar percentage based on the requirement that the
volume of mortar should exceed the volume of voids by 5 to 10 per-
cent-7 percent is a good starting point.
6. The coarse aggregate percentage can now be calculated by sub-
tracting the mortar percentage from 1.0.
7. Assuming an entrapped air volume of 1.5 percent, all the necessary
44 Chapter Two

values have been determined to calculate batch weights for 1 yd3 of

RCC based on saturated surface dry @SD) condition of the aggre-
8. A trial mix is proportioned in the laboratory and a Vebe time is
measured. If the Vebe time is not in the desired range, adjustments
are made in the mix, mainly in water content. A water content
change initiates revisions in other material proportions, and the
mix is adjusted until all basic requirements, including consistency,
are satisfied.
9. The mix can be further refined by more testing. In order to study
various combinations of components such as FAK, W/C t FA,
C t FA/sand, or various sand gradations, the Bureau of Reclama-
tion laboratory uses 2-in (51-mm) square mortar cubes while
changing one variable and keeping others constant. The Japanese RCD method. Criteria for mixes designed for the
RCD method include:

1. Cement content should be as low as possible while being consistent

with strength requirements. Some fly ash should be used as an ad-
mixture to reduce heat of hydration and mixing water require-
2. A sand/aggregate ratio higher than for conventional mass concrete
should be used to reduce segregation and to facilitate compaction
by a vibratory roller.

Cementitious content is governed by strength requirements. Figure

2.13 shows the relationship between cementitious content and com-
pressive strength. A cement-plus-fly ash content is selected that will
produce 2840 lb/in2 (200 kg/cm21 compressive strength at 91 days. For
all RCD projects, this value has been either 202 or 219 lb/yd3 (120 or
130 kg/m3). The percentage of fly ash to total cementitious content has
been either 20 or 30 percent, A good starting point is 219 lb/yd3 (130
kg/m31 for total cemetitious content and 30 percent for fly ash substi-
tution. This produces a mix with 153 lb of cement and 66 lb of fly ash
per cubic yard (91 and 39 kg/m3).
The water content is selected after performing laboratory tests to
determine: (1) the unit weight of mortar, (2) the compressive strength
of the mix with varying water contents, (3) the VC values with vary-
ing water contents. Using a compressive strength versus water con-
tent curve together with the VC value versus water content curve, a
water content that corresponds to a VC value of about 20 s is selected.
This water content should be close to and on the wet side of the com-
pressive strength peak, as shown in Fig. 2.5.
Roller-Compacted Concrete as a Material 45

F i g u r e 2 . 1 3 Fklationship b e -
tween cementitious content and
compressive strength with age
f o r T a m a g a m a D a m . [From
Yamauchi et a1.M
Age, days

The grading of coarse aggregate for maximum unit weight is deter-

mined by unit weight tests using a vibrating table with varying per-
centages of the particle sizes of coarse aggregate. Next, with cement,
fly ash, and water contents constant, several mixtures are made with
varying sand/aggregate ratios. VC values are determined for each
mixture. The sand/aggregate ratio producing the lowest VC value is
selected. Tests in Japan indicate that there is a sand/aggregate ratio
that produces a minimum VC value using a large container. This ratio
is in the range of 30 to 32 percent. A cement content versus compres-
sive strength test series then is used to determine the final
cementitious content, and the mix is ready for field trials. The U.S. Army Corps of Engineers method. This method is

based on experience with mix designs for seven RCC projects and is
described in engineer manual 1110-2-2006. It basically follows AC1
Standard 211.3, Standard Practice for Selecting Proportions for No-
Slump Concrete. The manual includes several tables developed from
the Corps experience with RCC. The mix-proportioning method can
be used with a wide range of materials and project requirements.
The steps to be taken after determining the required concrete prop-
erties and the properties of the proposed materials are

1. Evaluate the aggregate. If plastic fines are suspected, determine

the liquid limit (LL) and plasticity index (PI) of the fine aggregate
and compare LL and PI with the maximum allowable percentages
of minus No. 200 (0.075mm) sieve fines, usually 4 to 7 percent for
nonplastic fines.
The proposed aggregate gradings are compared with gradings
46 Chapter Two

presented in tables for coarse aggregate and fine aggregate (sand)

or with AC1 idealized grading curves that may be computed for
nonstandard MSAs.
Proportions of coarse aggregate and fine aggregate are either
determined by comparison with tables or by computation if the
RCC is to be made with conventionally graded aggregates.
2. Estimate the water requirements for a certain MSA and modified
Vebe time from a table. For a 3-in (75-mm) MSA and a Vebe time
of 20 to 24 s, the mixing water range is 160 to 200 lb/yd3 (94 to 119
kg/m3) with an average of 184 Ib/yd3 (109 kg/m31.
3. Entrapped air content is assumed to be 1 percent of total volume.
4. Determine the maximum water/cement (W/C) ratio for the partic-
ular exposure or other conditions from the project document, the
Corps, or AC1 tables. The W/C ratio can be converted to a WI
C t fly ash ratio if required using the AC1 211.1-89 report.
5. Compute the required weight of cement using the water require-
ment previously determined and the selected W/C ratio.
6. The absolute volumes of cement, pozzolan, and water content can
now be computed.
7. Determine the absolute volume of total aggregate by subtracting
the absolute volumes of materials calculated in the previous step.
If the aggregate is pit-run or an all-inclusive gradation produced
by minimal processing, trial batches are required.
8. For well-controlled aggregate gradings, the coarse aggregate (CA)
percentage is determined from a table based on the MSA and fine-
ness modulus (FM) of the fine aggregate, or sand. For a sand FM
of 2.60 and 3-in MSA, 67 percent CA is suggested.
9. The fine aggregate percentage is then computed by subtracting
the coarse aggregate from 100 percent. Absolute volumes of both
the CA and FA can now be computed.
10. From the absolute volumes previously computed, the mortar
(C t FA t air t sand) can be computed and compared with a ta-
ble. For a 3-in (75-mm) MSA, the mortar content should range be-
tween 11.2 and 12.6 ft3/yd3 (0.415 and 0.467 of absolute volume)
with an average of 12 ft31yd3 (0.444 of absolute volume).
11. Compute the volume of paste and the paste/mortar ratio. In this
calculation, fines passing the No. 200 (0.075-mm) sieve are con-
sidered to be paste. The minimum paste/mortar ratio according to
this method should be about 0.42 to ensure that all voids in the
fine aggregate are filled.
Roller-Compacted Concrete as a Material 47

12. Convert all absolute volumes to batch weights and prepare trial
batches to determine Vebe-time consistency and measured air
13. Adjust the mix as necessary to produce the desired consistency.

2.4.3 Project mixture proportions

The mixture proportions for RCC dams more than 50 ft (15 m) high or
which contain more than 13,000 yd3 (10,000 m3) of RCC are shown in
Fig. 2.14. Minor adjustments to the proportions noted were made in
the field for some dams to improve compactibility or to maintain a de-
sired consistency. In some cases, the water content was changed to ac-
commodate temperature variation at the site.
A column W/C t P or waterlcementitious material ratio is included
in Fig. 2.14. Any conclusions drawn using this ratio must be viewed
with caution because cement and pozzolan do not produce equal prop-
erties in the RCC. The different classes of cement and pozzolan also
will produce different properties.

2.5 Properties of RCC

The properties of RCC in place depend on the quality of materials

used, mixture proportions, and the degree of compaction or consolida-
tion. Because a wide range of materials and mixes have been used,
there are no typical values for RCC properties that fall within a nar-
row range. RCC properties that are aggregate-dependent, such as
elastic and thermal properties, are similar to conventional concrete
made from the same aggregate.
Because RCC mixes that conform to the soils approach usually con-
tain more than 2 percent air voids, the degree of compaction plays a
greater role in producing strength for these mixtures. The voids are
produced by particle-to-particle contact of the aggregate without suf-
ficient fines or paste to fill all the voids. Increased compaction tends to
decrease these voids, producing a denser RCC with a corresponding in-
crease in strength. Poorly graded aggregates or those with a high per-
centage of coarse aggregate may have an aggregate matrix that is
fully compacted and yet have a relatively high percentage of voids re-
sulting in lower density and strength. Even though there may be
greater volume of voids in a soils approach mix, all aggregate contacts
are cemented together.
The compaction of a no-slump RCC mixture by rolling produces a
material that is anisotropic for many of its properties. This is espe-
cially true for watertightness, where the permeability in the verti-
cally compacted direction can be appreciably less than in the horizon-
b Figure 2.14 Rcc mixture proportions for dams.

Vebe or Cement* Pozzolan Water, MM, Sand CA %

Mix v c lb/yd3 lb/yd3 lb/yd3 Admixtyfe, P*
W lb/yd3 lb/yd s a n d
Dam designation time, s (kg/m? (kg/m? (kg/n?) lb&-d3 C + P C + P (II% (kg/m? (kg/d ( b y wt.)



153 1 66 [ 160 1 0.55WFtA ( 0.73

(91) (39) (95) (0.325) + air

3. Mano 20 162 40 174 0.51 WRA 0.88

I (96) (24) (103) (0.30) t air

142 61 152 0 . 5 5 WRA 0.75

(84) (36) (90) (0.325) t air

162 40 172 0 . 4 2 WRA 0.85

(96) (24) (102) (0.25) t air

5. Asahi Ogawa 20 162 40 155 0.51 WRA 0.77

(96) (24) (92) (0.30) t air

7. Nunome 20 142 77 165 0 . 4 2 WRA 0.75

I (84.5) (45.5) (98) (0.26) t air

1 . Kengkou 15 101 135 165 0 . 4 7 WRA 0.70 0.57 3.1 1345 2309 36.6
(60) (80) (98) (0.28) (80) (798) (1370)

2 . Tianshengqiao #2 93 143 0 0.61

(55) (85)

3 . Longmentan 91 145 0.61

(54) (86)

United States

1 . Willow Creek A (interior mass) - 80 II 32 180 0 1.61 0.29 3 total 3956 (1 size) -
(47) (19) (107) (75) (2347)
B (upstream) - 175 II 0 185 0 1.06 0 3 total 3902 (1 size) -
(104) - (110) (7.5) (2315)
C (downstream) - 175 II 80 185 0 0.73 0.31 3 total 3826 (1 size) -
(104) (47) (110) (75) (2270)

2. Winchester (now - 175 0 175 0 1.00 0 3 total 3790 (1 size) -

Carroll E. E&on) (104) - (104) (75) (2248)

3. Middle Fork 112 II 0 160 0 1.43 0 3 1152 2138 35.0

(66) (95) (75) (683) (1268)

4. Galesville A (interior) - 89 86 190 0 1.09 0.49 3 1310 2560 33.9

(53) (51) (113) (75) (777) (1519)
B (downstream) 115 110 190 0 0.84 0.49 3 1290 2520 33.9
(68) (65) (113) (75) (765) (1495)

5. Grindstone Canyon (upper portion) - 130 0 200 0 1.54 0 3.5 1361 2500 33.8
(77) - (119) (89) (781) (1531)

6. Monksville 105 0 230 0 2.15 0 3 1500 2313 39.3

(62) - (119) (75) (890) (1372

7. Lower Chase Creek 20 108 v 67 180 0 1.03 0.38 2.5 1203 2445 33.0
(64) (40) (107) (63) (714) (1451)

8. Upper Stillwater A (interior) 19 134 II 290 171 17 oz. wR4 0.40 0.68 2 1148 2213 34.1
179) (172) (658 cm? (50) (681) (1313)
B (upstream) 15 155 II 343 169 21 oz. wR4 0.34 0.69 2 1162 2128 35.3
(92) (203) (100) (812 cm? (50) (689) (1262)

9. Elk Creek Spring 1987 10 118 II 56 174 41 oz. WRA 1.00 0.32 3.0 1227 2422 33.6
17-21 (70) (33) (103) t set (75) (728) (1439)
( 1586 cm?

I t
Figure 2.14 RCC mixture proportions for dams. (Continued)
Vebe or Cement* Pozzolan Water, MSA, Sand CA 5%
Mix v c lb/yd3 lb/yd3 lb/yd3 Admixtyre, w 2-t in lb/yda Ib/yda sand
Dam designation t i m e , e (kg/m? (kg/m? (kg/ma) lb/yd3 C t P 1C t P (mm) (kg/m? (kg/m31 (by wt.1


United States
10. Stagecoach - 120 130 233 0 0.93 0.52 2.0 1156 2459 32.0
(711 (77) (1381 (501 (6861 (14591
11. Stacy Spillway 210 105c 259 0 0.82 0.33 1.5 total 3500 (1 size) -
(now S.W. Freese) (125) (621 (1541 (381 (20761
12. Marmot- lo 120 180 175 4 1 oz. WRA 0.58 0.60 3 1270 2060 38.1
Replacement (71) (107) (104) t set (751 (7531 (12221

(1586 cm 1
I I I I retardin5 II II I I

1. La Manzanilla 6% by 3
wt. (751

1 . Copperfield A (exterior1 185 II 0 219 0 1.18 0 2 toted 3 5 7 3 (1 size) 4 0 . 0
(1101 - (130) (501 (21201
B (interior) 135 II 51 219 0 1.18 0.27 2 total 3573 (1 s i z e ) 4 0 . 0
(801 (301 (130) (501 (21201
2. Craigbourne 118 II 101 197 0 0.90 0.46 2 total 3 8 3 5 (1 size) 36.0
(70) (601 (1171 (531 (22751
3. Bucca Weir - 152 II 152 185 0 0.61 0.50 1.6 total 3523 38.0
(90) (90) (110) (401 (2090)

4. Wrights Basin - 244 122 266 0 1.00 0.33 1.6 totd 3329 (1 size) -
(145) (72.5) (158) (40) (1975)



1. Castilblanco de 10s (upper) 38 147 II P 158 185 0 0.56 0.61 1.6 1126 2421 31.7
Arroyos (87) (94) (101) 140) (668) (1440)
(lower) 46 172 II P 144 172 0 0.54 0.57 1.6 1334 2447 31.6
(142) (85.5) (102) (40) (672.5) (1462)

2. Erizana - dike - 152 IP 152 193 0 0.64 0.60 4 930 2798 24.9
(90) (90) (115) (100) (552) (1660)

3. Los Morales (upper) 116 258 170 0 0.45 0.69 1.6 1104 2394 31.6
(69) (153) (101) (40) (655) (1420)
(middle) 121 214 165 0 0.49 0.64 3.1 944 2562 26.9
(72) (127) (98) (80) (560) (1520)

4. Santa Eugenia HC-1

HC-2 / :: 1 :;j% 1 ;; / fo; 1 : 1 1:: 1 1:: ;;) 1 f;; 1 f!: 1 1::


1. Les Olivettes 148 R 79 211 0 0.93 0.35 2.5 total 3756 (1 size) -
(88) (47) (125) (63) (2250)


1. Tashkumy (lower) 20-30 202 IPN - 177 0.51 WRA 0.875 0.25 2 total 3708 30.4
(120) (105) (0.30) (50) (2200)
(upper) 20-30 169 - OAZWRA 0.25 2 total 3708
(100) (0.25) (50) (2200)

Figure 2.14 RCC mixture proportions for dams. (Continued)

Vebe or Cement* Pozzolan Water, MSA, Sand* CA lo

Mix v c lb/yd3 lb/yd3 lb/yd3 AdmixtIre, w 2, i n lb/yd3 lb/yd3 sands
Dam designation time, s (kg/m31 (kg/m? (kg//m? lb/yd3 C t P C t P (mm) (kg/ma) (kg/ma) (by wt.)


South Africa

1. De Mist Kraal - 98 98 177 2.0 oz air 0.91 0.50 3 1241 2918 29.8
Diversion (58) (58) (105) (78 cm3) (75) (736) (1731)

2. Arabie (now - 61 125 S 190 2 0 1.02 0.67 3 1473 2210 40.0

Mokgama Matlala) (36) (74) (113) (75) (874) (1311)
3. Zaaihoek 61 142 S 179 2.02 OS air 0.89 0.70 3 1019 2933 26.9
(36) (84) (106) (84 cd (75) (640) (1740)

4. Knellpoort 103 239 172 0 0.50 0.70 2.0 1175 2820 29.4
(gravity arch) (61) (142) (102) (50) (697) (1673)

5. Wolwedans 98 229 140 0 0.43 0.70 2.1 1145 2569 30.8

(gravity arch) (58) (136) (83) (53) (679) (1524)


1. Ain Al Koreima upstream 337 0 202 0 0.60 0 2.5 40.0

(200) (120)
downstream 169 0 202 0 1.20 0 (63)
(100) (120)

2. Rwedat 169 0 202 0 1.20 0 2 38.0

(100) (120) (50)



1. Saco de Nova (lower) - 12711P - 284 0 2.24 approx. 3 694 2935 19.1
Olinda (75) (168) 0.20 (75) (410) (1735)
1. Urugua-i - 101 0 111 0 1.75 0 3.0 - - 49.0
IbJ) 1 (0) (105) (75)
NOTE: M S A = m a x i m u m 3
*I = ASTM Type I or ordinary ortland cement (OPC) unless noted; IP = ASTM Type IP
Portland-pozzolan cement; II = A 8 TM Type II; V = ASTM Type V.
F = ASTM Class F (low-lime); unless noted C = ASTM Class C (high-lime); N = ASTM Class N
(natural); S = Ground blast-furnace slag (called sla ent in South Africa); unless noted
R = ROLAC (a mixture of 60%. OPC, 35% S & 5%%estone dust).
Includes proportion of pozzolan in portlanh-pozzalan cement.
Passing 4.75mm U.S. or 5.0-mm sieve unless noted.

54 Chapter Two

tal direction. Most testing has been performed on cores and cylinders
in only one direction, however, so few data are currently available on
the anisotropic properties of hardened RCC.

2.51 Strength Properties

Results obtained from the testing of cores from RCC dams are more
indicative of actual properties of the material in the structure than re-
sults obtained from laboratory specimens. Strength properties de-
scribed using core samples will be emphasized in this section.

2.5.1 .I Compressive strength. Compressive strength is relatively

easy to determine. Many other properties are directly related to the
concretes unconfined compressive strength at a certain age.
The relationship between unit water content and compressive
strength at 91 days is shown in Fig. 2.5 for a Japanese RCD mix con-
taining 142 lb/yd3 (84 kg/m31 of cement plus 61 lb/yd3 (36 kg/m31 of fly
ash. As noted in Sec. 2.1.3, the compressive strength increases with a
reduction in water content as long as the RCC is fully compacted. The
maximum compressive strength for a certain mix is obtained at the
optimum water content consistent with a specified compactive effort
for the material. Water contents less than optimum produce lower
compressive strength. This indicates that the presence of voids in the
mix has greater negative effect on strength than the positive effect of
water reduction.
For most RCC dams, the designer establishes a relatively fixed wa-
ter content which is based on a Vebe time or VC value for concrete
approach mixtures or on maximum dry density for mixes designed us-
ing the soils approach. The designer for a soils approach RCC mix may
decide to specify a water content slightly wet of optimum in order to
obtain improved workability. However, once a water content and a
compactive effort are established, the concrete compressive strength
depends on the cement or the cement plus pozzolan content. Compres-
sive strength increases with time and the amount of cementitious ma-
terials in the mix. Figure 2.13 shows the increase in compressive
strength at ages up to 91 days with varying cementitious contents.
For an RCC mixture that has a consistency that corresponds to the
soils approach, the effect of compaction on strength is illustrated in
Fig. 2.15. The mixture contains 150 lb/yd3 (89 kg/m3) of cement, no fly
ash, and 2 in (50 mm) MSA with 40 percent sand and 4 percent pass-
ing the No. 200 (0.05-mm) sieve. The laboratory specimens subjected
to a compactive effort of 14.2 ft-lb/in3 (1175 kJ/m3) showed a consid-
erably higher seven-day compressive strength than those that were
vibrated per Bureau of Reclamation procedure 4906-86.
In this test, there appears to be a convergence of the compacted and
Roller-Compacted Concrete as a Material 55

0 I I I I I
2 3 4 5 6 7 8
Calculated moisture carrtent(SSD),%

Figure 2.15 Comparison between seven-day compres-

sive strength of compacted and vibrated 6- by 12-in
(152- by 304-mm) cylinders. [C = 150 lb/yd3 (89 k8/
m?.l [From Tczyabji and Okamoto (1987)./

vibrated strength curves at greater than 6 percent total moisture con-

tent. This indicates that wet of this value the moisture fills all the
voids following consolidation and the mixture then conforms more
closely to the concrete approach.
Compressive strengths obtained from both cores and cylinders are
shown in Fig. 2.10 for various RCC dams. Tensile strength. Tensile strength of RCC for dams can either
be determined by tests to measure direct tension or splitting (indirect)
tension. The splitting tension test is also known as the Brazilian test.
For conventional concrete dams, splitting tension tests of cores show
strengths averaging about 10 percent of compressive strengths. Direct
tension tests of the same cores show tensile strengths at about 5 per-
cent of compressive strengths or about half that of splitting tension.
An analysis of RCC mixes for nine dams, after discarding the high
and low values to obtain a more representative average, shows split-
ting tension to average 13 percent of compressive strength. The range
was from 11.9 to 14.3 percent.
Because of its steep downstream slope, Upper Stillwater is the only
RCC dam constructed to date for which the direct tensile strength of
the RCC was the primary design criteria. A minimum direct tension
of 180 lb/in (1.24 MPa) at one year was required. Laboratory tests of
the proposed high-paste mix using sandstone aggregate produced 220
lb/in2 (1.52 MPa) or 4.4 percent of its compressive strength of 5000 lb/
in2 (34.5 MPa) at one year. Shear strength. Construction of a concrete dam using RCC

methods produces a structure with lift lines every 1 or 2 ft (0.3 to 0.6
56 Chapter Two

m) vertically. The shear strength at the compacted lift lines is more

important to the designer than the shear strength of the parent ma-
The total shear strength can be determined using Coulombs
where s = unit shear stress
c = unit cohesion
p = unit normal stress
6 = the angle of internal friction

The cohesion is also called the bond stress, whilep tan + defines the
sliding friction resistance. A direct shear test is the usual method for
obtaining cohesion and angle of friction data using various normal
loads. McLean and Pierce have summarized shear strength data for
the RCC parent material and bonded and unbonded RCC joints, and
compared the RCC results with conventional concrete shear values
(see Fig. 2.16). They note that there is significant scatter in the values
because test data are obtained from a variety of mixtures and materi-
als, information sources, and test procedures.
The break bond shear strength may also be called the peak
strength, and the sliding friction values referred to by McLean and
Pierce denote the residual shear strengths.

2.5.2 Elastic properties Modulus of elasticity. The modulus of elasticity E, also known
as Youngs modulus, is the ratio of the normal stress to its correspond-
ing strain for compressive or tensile stresses below the proportional
limit of the material. The modulus of elasticity of conventional con-
crete is proportional to its unit weight and compressive strength. An
average value of E can be obtained by the formula

where w is the unit weight in pounds per cubic foot.

Based on limited data, the value of E for RCC produced from normal-
weight, high-quality aggregates is similar to conventional concrete or
slightly greater due to the higher unit weight of RCC.
It has been shown that the type of aggregate used in RCC is a major
factor in determining the elastic modulus. Lower-unit-weight aggre-
gates such as the marlstone used for Middle Fork Dam and the sand-
stone for Upper Stillwater Dam produced considerably lower values
for E than predicted by the formula above. In dam design, a low mod-
Figure 2.16 Summary of RCC shear strength data and comparison with conventional
concrete. [From McLean and Pierce (1981J.I

Range of values
Break-bond Sliding friction Statistical Evaluation

No. of No. of $, $7 c, lb/ Mean 4, Std. dev. Mean e, Std. dev.

Subset projects tests deg. c, lb/in* deg. in deg. 0 deg. lb/in c, lb/in
Parent Material
Conventional 2 18 40-76 269-573 - - 58 14 364 134
RCC 3 40 33-76 74-641 - - 49 16 437 179
Bonded Joints
Conventional 6 62 25-78 205-527 - - 55 13 350 98
RCC 8 142 24-73 9-622 - - 48 9 280 172
Lean 6 76 24-67 9-441 - - 47 10 189 106
Rich 2 55 35-73 165-622 - - 49 9 384 175
Unbonded Joints
Conventional 10 146 - 35-51 18-216 47 3 86 50
RCC 4 163 - - 34-52 14-83 43 4 35 18
Lean 2 58 - - 34-47 14-83 40 4 46 24
Rich 2 105 - - 38-52 16-48 45 4 30 8
Conventional Concrete
Reference Value
Triaxinl 4 226 40-48 605-1360 - - 44 2 1075 217
Tension-compression 2 95 52-69 300-615 - - 59 3 480 90

*Apparent cohesion (c,).

Includes sliding friction results of break-bond tests.
*Harboe E.M. Pro erties of Mass Concrete in Bureau of Reclamation Dams Bureau of
Reclamatibn Report 80. C-1009, December 1961.
50 Chapter Two

ulus of elasticity is desirable in order to reduce the potential for crack-

ing at a certain stress level. Values of E for a number of RCC mixes
are shown in Fig. 2.17. Poissons ratio. This value is the ratio of transverse (lateral)

strain to the corresponding axial (longitudinal) strain resulting from
uniformly distributed axial stress below the proportional limit of the
material. For RCC this value has usually ranged from 0.17 to 0.22,
with 0.20 being typical. This is about the same as for conventional
concrete. Values for Poissons ratio are also listed in Fig. 2.17.

2.5.3 Permeability
The total seepage through an RCC dam is the sum of the water pass-
ing through the material itself plus that through any cracks or joints
in the structure. Measured seepage through RCC dams in service is
discussed in detail in Chap. 11.
Permeability values for RCC mixtures have ranged from 2 x 10m2

Compressive Creep
Age, strength, E Poissons coefficient
Dam Mix days lb/in (MPa) (X 10s) ratio (X 10-Y
Willow Creek A 7 577C4.0) 1.20 - 1.97
28 1172(8.1) 1.59 0.14 1.09
90 1730(11.9) 1.91 0.17 0.52
B 7 997C6.9) 2.20 0.48
28 1845c12.7) 2.67 0.19 0.34
90 2649c18.3) 2.78 0.18
c 7 1147(7.9) 2.40 - 0.58
28 2056c14.2) 2.91 0.21 0.39
90 3961f27.3) 3.25 0.21 0.31
Middle Fork 28 127Oc8.8) 0.98 0.16
Galesville A 14 mo.* 2095C14.4) 3.23 0.19
B 1 4 mo.* 2000(13.8) 3.29 0.22
Monksville 28 745(5.1) 0.90
Upper Stillwater A 28 0.66
105 3925C27.1) 1.96 0.23
365 5171(35.7) 2.07 0.29 0.53
Les Olivettes 28 1905(12.7)
90 2130(14.7) 2.85
Saco de Nova 7 1.04 2.15
Olinda 28 1.79 1.28
360 - 0.32
*Properties from cores extracted from dam-all other properties obtained from 6 x 12
in (152 x 304 mm) cylinders prepared by pnuematic tamping.

Figure 2.17 Elastic and creep properties for RCC mixtures.

Roller-Compacted Concrete as a Material 59

cm/s for the lean mix A at Willow Creek Dam to 4 x lo- lo cm/s for the
high-paste mix A used at Upper Stillwater Dam, as shown in Fig. 5.1.
Dunstan proposes in Chap. 5 that impermeability is the single most
important property of the RCC mix, and impermeability of RCC can
be directly related to its cementitious content, also as shown in Fig.
5.1. This fact is especially applicable to RCC mixtures than conform to
the concrete approach where the paste exceeds the voids in the aggre-
gate. Therefore, greater cementitious contents produce a more water-
tight paste, which controls the permeability of the RCC material. For
soils approach mixes, greater impermeability can be achieved by a
combination of increased cementitious content, greater compaction,
and sufficient well-graded fine aggregate, all of which reduce voids in
the material.

2.5.4 Durability
The durability of RCC is especially important if the material is ex-
posed to weather or severe hydraulic forces. Its durability has been
documented by both laboratory tests and case studies in the field. Freeze-thaw resistance. Because proper air entrainment in

RCC is generally not attainable with admixtures, freeze-thaw resis-
tance must come from its strength and impermeability. High-strength
RCC mixes with low permeabilities have greater freeze-thaw resis-
tance than lean, low-strength mixes.
Freeze-thaw resistance in a dam is of greatest concern for horizontal
surfaces or other surfaces exposed to freeze-thaw cycles while wet.
High-strength exposed RCC pavements have shown very good freeze-
thaw resistance after being in service in British Columbia, Canada,
for more than 10 years. Also, the relatively lean RCC on the unformed
downstream faces of Willow Creek and Galesville dams have been
continuously damp in places due to seepage and have shown little no-
ticeable freeze-thaw deterioration.
The freeze-thaw resistance of actual RCC pavements and dams is
better than would be predicted by laboratory tests or test sections.
Most RCC mixes have shown poor freeze-thaw resistance in the labo-
ratory when subjected to the severe ASTM C566 rapid freezing and
thawing test. Similarly, large blocks of RCC from the 1973 test section
at Lost Creek Dam in Oregon totally deteriorated at the U.S. Army
Corps of Engineers cold weather test facility at Treat Island, off the
coast of Maine. The blocks were subjected to the combined effect of
saltwater, waves, and alternate freeze-thaw and wet-dry cycles.
If RCC mixes are designed for durability using freeze-thaw weight
loss tests and criteria as developed for soil-cement, acceptable freeze-
60 Chapter Two

thaw durability can be expected. The amount of cement to produce a

sufficiently durable RCC mix may be greater than that required to
achieve other properties such as compressive strength. Little or no
pozzolan replacement for cement is advised where horizontal RCC sur-
faces will be exposed to early freeze-thaw cycles while wet because
high early strength is required under these conditions. Erosion or abrasion resistance. The erosion resistance of RCC
is proportional to its compressive strength and the abrasion resistance
of the aggregate used in the mix. RCC has shown excellent resistance
to erosion and abrasion both in the laboratory and in the field. In con-
junction with the design for Willow Creek Dam, lean, large-aggregate
RCC panels performed well when subjected to high-velocity water jets
at the Corps of Engineers test flume at Detroit Dam, Oregon. When
mass sections of RCC are to be subjected to high-magnitude hydraulic
forces such as in a stilling basin or plunge pool, the RCC lifts should
be well bonded to prevent delamination.
The erosion-resistance properties of RCC have been demonstrated
on many projects. The most notable are the spillway rehabilitation at
Tarbela Dam, the spillway for the North Fork of the Toutle River de-
bris retention dam, and the Kerrville ponding dam. The performance
of these RCC structures is described in Chap. 7.

2.5.5 Volume changes

In any massive concrete structure, the understanding of and the de-
sign for volume changes is necessary to minimize uncontrolled crack
ing. The reduction of volume due to thermal or drying shrinkage is of
concern in the design of RCC dams as described in Sec. 3.51. Thermal properties. The RCC properties that may be needed

in a thermal analysis include specific heat, diffusivity, conductivity,
and coefficient of thermal expansion, together with tensile-strain ca-
pacity. Typical thermal properties are shown in Fig. 2.18. Thermal
properties are mainly aggregate dependent, so RCC and conventional
concrete made from the same aggregate source have similar values. Drying shrinkage. Increases in moisture cause concrete to ex-

pand and decreases in moisture cause it to shrink. In the cement hy-
dration process, water combines with the cement so the basic process
is one of moisture loss or shrinkage.
In any concrete mix, it is only the paste that shrinks. So for a con-
stant cemetitious content, the drying shrinkage rate depends prima-
rily on the amount of water in the mix. Because RCC requires less wa-
Figure 2.18 Thermal properties for RCC mixtures.

Adiabatic rise, F (0
Specific heat, Diffusivity, Conductivity, Coeff. of exp.
BtulIb . F ftzlb Btulftlh . F in/in.F x 10e6 InitiaI
Dam Mix (k Cal/kg. C) (m/h) x 10m3 (k &/m/b . 0 m m / m m . C x lo- T e m p . 3-day 7-day 2%day

Willow Creek (basalt) A 0.22 0.03 1.05 3.9 53 13 - 22

(0.22) (2.8) (1.56) (7.0) (12) (7) - (12)
B 0.22 0.03 1.05 4.0 55 23 29 36
(0.22) (2.8) (1.56) (7.2) (13) (13) (16) (20)
C 0.22 0.03 1.05 4.0 52 23 29 36
(0.22) (2.8) (1.56) (7.2) (11) (13) (16) (20)
Middle Fork 0.22 0.027 - 3.5 - - - 16
(marlstone) (0.22) ( - - - 9)
Upper Stillwater A 0.06 4.9 45 4 341
(quartzite/sandstone) (5.6) (8.8) (7) (2) (11)
2o (19)
A 0.06 49 16 t
(5.6) (C, (9) (9)
Saco de Nova Olinda 0.22 0.016 5.9 46 18 23
(0.22) (1.71) (10.57) (8) (10) (12)
22 (13)
Elk Creek 0.18 43 17
(basalt/sandstone) (0.18) (6) (9)
Pam0 (granite) 0.2322 0.0332 1.183 3.31
(0.2322) (3.55) (1.76) (5.93)
*ASTM C494 Type D water-reducing and set-retarding admixture.
I ASTM C494 Type A water-reducing admixture. I
62 Chapter Two

ter and cement than conventional concrete, RCC shrinks less in the
hydration process. Creep. When concrete is subjected to a load, the deformation

caused can be divided into an immediate deformation such as an elas-
tic strain and a time-dependent compressive deformation called creep.
Creep begins immediately and continues at a decreasing rate for as
long as the load remains on the concrete.
Creep in RCC is a function of the aggregate and RCC strength. Ag-
gregates that produce a low modulus of elasticity concrete will pro-
duce a high creep value. Higher-strength mixes have a more rigid ce-
ment paste and, therefore, a lower deformation or creep value when
loaded. For RCC dams, a high creep value or the ability to relieve a
sustained stress is usually desireable. Compressive stresses in gravity
dams are generally low, however, and creep is not usually a major con-
cern to dam designers. Creep values for various RCC mixes are given
in Figure 2.17.

2.5.6 Unit weight

The unit weight or density of concrete depends primarily on the spe-
cific gravity of the aggregate and the amount of voids in the RCC
mass. There are few entrained-air voids in RCC, and compaction re-
duces that number further. This means that there are more solids in a
unit of volume of RCC and, therefore, its unit weight is generally
greater than conventional concrete made with aggregate of the same
specific gravity. Unit weights greater than 150 lb/ft3 (2400 kg/m3) are
common for RCC.

Abrams, D. A.: Design of Concrete Mixtures, Bulletin No. 1, Structural Materials Re-
search Laboratory, Lewis Institute, Chicago, 1918.
American Concrete Institute, Roller-Compacted Mass Concrete, Report 207.5R 1988.
American Concrete Institute Standard Practice for Selecting Proportions for Normal,
Heavyweight and Mass Concrete, Report 211.1, 1989.
Bahrner, V.: New Swedish Consistency Test Apparatus and Method, Betong
(Stockholm), no. 1, 1940, pp. 27-38.
Bouyge, B., A. P. Langois, and J. P. Martin, Quality of Works in RCC in France: A
Contractors Solution, Roller-Compacted Concrete II, ASCE, New York, February
1988, pp. 309-322.
Casias, T. J., V. D. Goldsmith, and A. A. Benavidez: Soil Laboratory Compaction Meth-
ods Applied to RCC, Roller-Compacted Concrete II, ASCE, New York, February
1988, pp. 107-122.
Crow, R. D., T. P. Dolen, J. E. Oliverson, and C. D. Prusia, Mix Design Investigation-
Roller-Compacted Concrete Construction, Upper Stillwater Dam, Utah, Bureau of
Reclamation Report REC-ERC-84-15, June 1984.
Dolen, T. P., A. T. Richardson, and W. R. White, Quality ControllInspection-Upper
Roller-Compacted Concrete as a Material 63

Stillwater Dam, Roller-Compacted Concrete 11, ASCE, New York, February 1988,
pp. 277-293.
Dunstan, M. R. H.: A Method of Design for the Mix Proportions of Roller-Compacted
Concrete To Be Used in Dams, Transactions, 15th ZnternationaZ Congress on Large
Dams, (Lausanne 1985), International Commission on Large Dams, Paris, vol. 2, pp.
Forbes, B. A., The Development and Testing of Roller-Compacted Concrete for Dams in
Australia, Transactions 16th International Congress on Large Dams (San Francisco,
1988), International Commission on Large Dams, Paris, vol. 3, pp. 89-117.
Hirose, T., and 1. Yanagida, Some Experiences Gained in Construction of Shimajigawa
and Okawa Dams. Proceedines. CZRZA International Conference on Rolled Concrete
for Dams, Construction Industry Research and Information Association, London,
June 1981.
Hirose, T., and T. Yanagida: Burst of Growth Demands Speed, Economy, Concrete ln-
ternational, May 1984.
Hollingworth, F., and F. H. W. M. Druyts, Rollcrete: Some Applications to Dams in
South Africa, Water Power & Dam Construction, January 1986, pp. 13-16.
Hopman, D. R., and D. R. Chambers, Construction of Elk Creek Dam, Roller-
Compacted Concrete 11, ASCE, New York, February 1988, pp. 251-266.
Johnson, H. A., and P. C. Chao: Rollcrete Usage at Tarbela Dam, Concrete
International: Design and Construction, vol. 1, no. 11, November 1979.
Lemons, R. M.: A Combined RCC and Reinforced Concrete Spillway, Roller-
Compacted Concrete 11, ASCE, New York, February 1988, pp. 51-60.
Logie, C.V., and Oliverson, J.E., Roller Compacted Concrete Mix Design for Pamo
Dam, Roller-Compacted Concrete 11, AXE, New York, February 1988, pp. 187-202.
Lorenzo, A., G. L. Oberholtzer, and E. K. Schrader: Roller-Compacted Concrete Design
for Urugua-i Dam, Roller-Compacted Concrete 11, ASCE, New York, February 1988,
pp. 357-368.
Lowe, J., III: Use of Rollcrete in Earth Dams, discussion of Utilization of Soil-Cement
as Slope Protection for Earth Dams by W. G. Holtz and F. C. Walker; First ASCE
Water Resources Engineering Conference, Omaha, Nebraska, 1962 (not published in
McLean, F. G., and J. S. Pierce: Comparison of Joint Strengths for Conventional and
Roller-Compacted Concrete, Roller-Compacted Concrete 11, ASCE, New York, Feb-
ruary 1988, pp. 151-169.
Mageroy, H., RCC Pavement Developments in Norway, Draft paper for publication in
Journal of Construction Division of ASCE.
Neville, A. M.: Properties of Concrete, 3rd ed., Pitman, London, 1981.
Oberholtzer, G. L.: Galesville Project Materials Engineering for Roller-Compacted
Concrete, Roller-Compacted Concrete Seminar, ASCE Southern Idaho Section,
April 1985.
Parent, W. F., W. A. Moler, and R. W. Southard, Construction of Middle Fork Dam,
Roller-Compacted Concrete, ASCE, New York, May 1985, pp. 71-89.
Portland Cement Association: Soil Cement for Water Control: Laboratorv Tests.
Skokie, Ill., 1976, 26 pp.
Portland Cement Association: Design and Control of Concrete Mixtures, 13th ed.,
Skokie, Ill., 1988.
Proctor, R. R.: Fundamental Principles of Soil Compaction, Engineering News-Record,
vol. III, Aug. 31-Sept. 28, 1933, pp. 245-248, 286-289, 348-351, and 372-316.
Quin, J. T., S. P. Rezende, and E. K. Schrader, Saco Dam-South Americas First RCC
Dam, Roller-Compacted Concrete 11, ASCE, New York, February 1988, pp. 340-356.
Reeves, G. N., and L. B. Yates, Jr.: Simplified Design and Construction Control for
Roller-Compacted Concrete, Roller-Compacted Concrete, American Society of Civil
Engineers (ASCE), New York, May 1985, pp. 48-61.
Schrader, E. K., and D. Namikas, Performance of Roller-Compacted Concrete Dams,
Transactions 16th International Congress on Large Dams @an Francisco, 19881, In-
ternational Commission on Large Dams, Paris, vol. 3, pp. 339-364.
Tayabji, S. D., and A. S. Okamoto: Bonding of Successive Layers of Roller-Compacted
Concrete, Construction Technology Laboratories, Skokie, Ill., March 1987.
64 Chapter Two

US. Army Corps of Engineers: Willow Creek Dam Concrete Report, vol. 1 and 2,
Walla Walla. Wash.. October 1984.
U.S. Army Corps of Engineers: Roller-Compacted Concrete, EM 1110-2-2006, August
U.S. Bureau of Reclamation: Guidelines for Designing and Constructing Roller-
Compacted Concrete Dams, June 1987.
U.S. Bureau of Reclamation: Procedure for Casting No-Slump Concrete in Cylinder
Molds Using Vibratory Table, procedure 490686.
Urlich, C. M., J. D. Deatheridge, J. H. Rahe, and R. E. Dunne, Design and Construc-
tion of Lower Chase Creek Dam, Roller-Compacted Concrete II, AXE, New York,
February 1988, pp. 236-250.
Wong, N. C., J. A. Bischoff, and D. H. Johnson: Strengthening and Raising Gibraltar
Dam with RCC, Roller-Compacted Concrete IZ, AXE, New York, February 1988,
pp. 92-106.
Yamauchi, T., J. Harada, T. Okada, and S. Shimada, Construction of Tamagawa Dam
by the RCD Method, Transactions, 15th International Congress on Large Dams
(Lausanne, 19851, International Commission on Large Dams, Paris, vol. 2, pp. 39-

Design of
RCC Gravity Dams

3.1 Background
Roller-compacted concrete has been used primarily for gravity-type
dams to date, and the design of gravity dams will be emphasized in
this chapter. The design of a gravity dam using RCC is fundamentally
no different from the design of a gravity dam constructed of conven-
tional concrete. Therefore, many of the principles and formulas for
gravity dam design will not be repeated here. A good reference for ba-
sic dam design is the text Design of Gravity Dams, prepared by the
U.S. Bureau of Reclamation WSBR).
This chapter will focus on the differences between RCC dams and
conventional concrete gravity dams in the design of their sections and
appurtenant structures. Philosophies and details for the lean RCC,
high-paste RCC and Japanese RCD approaches to gravity dam design
are described in Chaps. 4, 5, and 6, respectively. Therefore, this chap-
ter will concentrate on the overall design of RCC dams, keeping in
mind those different design approaches.
References are made to low, moderate-height, high, and very high
dams in this and other chapters in the book. A low dam is defined as
one that has a structural height of less than 100 ft (30 ml, a moderate-
height dam is one ranging from 100 to 300 ft (30 to 90 ml, a high dam
is one from 300 to 500 ft (90 to 150 m), and a very high dam is greater
than 500 ft (150 ml.

3.2 Site Selection and Foundation

3.2.1 Site selection
Site selection and foundation requirements for RCC dams are basi-
cally the same as those for conventional concrete gravity dams. How-

66 Chapter Three

ever, because RCC costs less per unit volume than conventional mass
concrete, designers have more freedom in optimizing the selection of
the site. They are no longer tied as strongly to a site that minimizes
the volume of concrete in the dam structure.
Designers can now investigate other sites where a larger-volume
RCC dam may be required to maximize the benefits of the entire
project. Those benefits could include increased pool storage, greater
power head, and shorter penstocks for a hydroelectric project.
In preparing preliminary designs and cost estimates for alternative
dam types, the optimum location of an RCC dam may be different
from the optimum location of an embankment dam. This will depend
primarily on the topography and geology of the site. Such was the case
in studies for Pamo Dam, where the optimum location for the RCC
dam was 0.8 mi (1.3 km) downstream from the optimum earthfill dam
site. The downstream RCC site was selected to reduce foundation
preparation and pipeline transmission costs. Also, there was little
need for the RCC site to be located close to the borrow pits for the
earthfill alternative.

3.2.2 Rock foundations

Sound rock foundations are considered the most suitable for concrete
dams because they possess high bearing capacity and have a high de-
gree of erosion and seepage resistance. The RCC dams completed to
date have been founded on many different rock types, such as basalt
(Willow Creek), limestone (Winchester), marlstone (Middle Fork),
granite (Copperfield), meta-andesite (Galesville), siltstone (Bucca
Weir), quartzose sandstone (Upper Stillwater), and quartzite (Saco).
Assuming an adequate foundation, there is no structural height re-
striction for an RCC dam. The tallest RCC dam constructed to date is
the 328-ft (100-m) Tamagawa Dam, but higher dams are being
planned. In Japan, the 377-ft (115-m) Sakaigawa Dam is under con-
struction and the 410-ft (125-m) Gatssan Dam and 509-ft (155-m)
Miyagase Dam are in design.
Rock foundations that lack major faults and shears are most suit-
able for RCC dams. Faults and shears do not necessarily eliminate a
site from consideration, but it may be expensive to treat them to en-
sure an adequately safe foundation.
Because each site is unique, engineers experienced in evaluating
foundations should investigate the site and determine possible
treatments. A foundation investigation program, usually involving
drilling, is considered essential in order to evaluate foundation con-
ditions properly. The conditions from the rock surface down to 30 to
Design of RCC Gravity Dams 67

60 ft (10 to 20 m) are considered to be of the greatest importance, as

they have the greatest effect on the ability of the foundation to
withstand loads without unacceptable short-term and long-term de-
Foundation investigation and design is probably more important
than the design of the dam section itself. History has shown that the
potential for failure of a concrete gravity dam, while extremely re-
mote, is more likely in the foundation than in the dam itself. Special
attention should be given to identifying potential sliding planes in the
foundation rock. Foundation rock properties. In the investigation program, five

properties of the foundation rock needs to be determined, namely (1)
compressive strength, (2) shear strength, (3) deformation modulus, (4)
Poissons ratio, and (5) permeability.
The compressive strength of the foundation rock is an important
consideration in determining the base thickness of the dam. De-
signers should calculate a minimum base thickness that reduces
the maximum allowable bearing stress to allowable values as de-
termined by dividing the compressive strength of the foundation
material by an appropriate factor of safety, usually 4.0 according to
USBR criteria for the usual loading combination.
The shear strength of the foundation rock, including any dis-
continuities, depends upon the cohesion and internal friction proper-
ties of the rock, together with the applied normal load. The total
strength of the foundation rock can be determined using Coulombs
equation in the same way it is used for the RCC material within the
dam. A linear relationship between shear resistance and normal load
for intact rock is shown in Fig. 3.1~.

Unit norm01 stress p Unit norm01 stress p

(a) (b)
Figure 3.1 Relationship between shear strength and normal stress for rock.
(a) Shear strength of intact rock; (b) shear strength of an existing joint in
66 Chapter Three

Because joints, shears, and faults possess little or no cohesion, the

shear strength of a discontinuous rock mass is essentially derived
from sliding friction. The shear strength of an existing joint in rock is
nonlinear, as shown in Fig. 3.lb. Shearing resistance of rock with
discontinuities should be based on physical tests of the material in or-
der to plot a shear strength versus normal stress relationship. A unit
shear strength for design can be determined once a range of normal
stresses imposed by the dam is calculated. Factors of safety are then
applied to the sliding friction shear strength depending upon the par-
ticular load combination being investigated.
When adequate shear strength data are not available for the foun-
dation rock, a conservative approach for preliminary design is to as-
sume zero cohesion and a conservative value for the sliding friction
shear strength for the type of rock. A minimum shear friction factor of
safety of 1.5 can then be applied for the usual loading combination. A
minimum shear friction factor of safety of 1.5 in the case of zero cohe-
sion is the criterion for the design of concrete dams in Spain for the
usual loading combination.
The magnitude of the foundation deformation modulus is often not
as critical as the variation in modulus across the foundation. Conven-
tional gravity dams have been successfully constructed on low-
modulus materials such as siltstone, claystone, gravel, and sand. Mid-
dle Fork, on marlstone, and Bucca Weir, on siltstone, are examples of
RCC dams founded on rock with a relatively low deformation modu-
lus. RCC dams on very low modulus rock or on nonrock foundations
are described in Sec. 3.2.3.
A determination of deformation patterns for complex foundation
conditions may be determined with a joint-shear index and shear cat-
alog. Poissons ratio-the value of transverse strain to its correspond-
ing axial strain-is needed for more rigorous mathematical analyses
of the dam and its foundation.
Abrupt changes in deformation modulus can result in differential
settlements that can cause cracking in an RCC dam. Therefore, the
designer should identify low-modulus zones and plan improvement
measures such as grouting the weak material or excavating and re-
placing it with conventional concrete or RCC.
Although permeability of the foundation rock is the main factor in
determining whether a grout curtain is required, most designs for ma-
jor dams include an upstream grout curtain to reduce seepage under
the dam as a matter of course.

3.2.i.2 Foundation excavation guidelines. The amount of excavation

required depends on the depth of the overburden and weathered rock.
All overburden covering a rock foundation, such as soil, alluvium, and
Design of RCC Gravity Dams 69

talus, should be removed. In addition, highly weathered rock should

be removed for low dams to provide a foundation of slightly weathered
rock. For moderate-height to high dams, excavation to fresh rock sur-
faces is usually required.
The actual depth of excavation in rock is highly site-dependent and
can vary considerably, from as little as 5 ft (1.5 ml to more than 100 ft
(30 ml. It is usually in the 5 to 50 ft (1.5 to 15 m) range. The use of a
method to evaluate rock quality such as the RQD (rock quality desig-
nation) may be appropriate in determining the depth of rock excava-
tion. The designer should keep in mind that unless the quality of the
rock improves with depth, additional excavation and subsequent
building back up only increases the cost of the dam. Cross-canyon
shaping of the foundation should avoid major protrusions which can
produce cracking.
Not all discontinuities in the foundation rock are discovered in the
foundation investigation program. Such was the case at Upper
Stillwater and Elk Creek dams, where unanticipated shear zones were
uncovered during excavation. They required additional excavation
and replacement with conventional concrete. Contract documents
should be structured so that payment for unanticipated foundation
improvements can be resolved quickly and fairly in order to prevent
costly change-of-conditions claims by the contractor. Foundation improvement and drainage. The usual methods of

foundation improvement include consolidation grouting, curtain
grouting, and replacement of weak zones and discontinuities with con-
ventional concrete or RCC. After loose material on the foundation
rock is cleaned, low-pressure consolidation grouting of the foundation
rock helps fill naturally occurring voids, fracture zones, and cracks. In
addition, consolidation grouting helps repair rock damage that may
have been produced by blasting and by relaxation during rock exca-
Consolidation grouting can be accomplished by drilling holes with a
minimum diameter of 1% in. (38 mm) spaced 10 to 20 ft (3 to 6 m)
apart and filling them with cement grout. The holes can vary in depth
from 20 to 50 ft (6 to 15 m), depending on the foundation conditions
and dam height.
Curtain grouting is commonly used to control seepage beneath
dams, even those founded on tight, low-permeability rock. The con-
struction of the grout curtain is usually accomplished by drilling a
single line of 11/2 to 3 in. (38 to 75 mm) holes spaced 10 ft (3 m) on
center near the upstream face of the dam. The spacing can vary de-
pending on the condition of the rock. The depth of the holes also de-
pends on hydrostatic head and foundation rock conditions. The depth
70 Chapter Three

of the grout curtain can vary from 40 percent of the head for dense
foundations to 70 percent in poorer-quality rock foundations according
to USBR criteria.
Another guideline suggested by Simmonds is that:


where D = depth of grout curtain

H = height of dam
C = a constant varying with the foundation quality, the size of
the dam, and the significance of leakage. C can vary from
25 to 80 ft (8 to 25 m), with 50 ft (15 m) being a reasonable
average for major dams.

Drain holes are usually provided in rock foundations for all but
some low to moderate-height dams to reduce uplift pressures in the
foundation and improve the stability of the dam. They are usually lo-
cated immediately downstream of the grout curtain and consist of 3-
in-diameter (75-mm) holes spaced 10 ft (3 m) on center. The function
of the drain holes is to intercept and remove any seepage water by-
passing the grout curtain and thus reduce the buildup of hydraulic
pressure (uplift) beneath the dam.
At Upper Stillwater Dam, the grout curtain and drain holes were
drilled from the same line in the gallery, 2 ft (0.6 m) from the
gallerys upstream face. The grout curtain was drilled 5 from ver-
tical upstream, while the drain holes slanted 10 from vertical
In addition to using RCC to replace weak foundation material, RCC
or shotcrete can be used for surface foundation improvements to pro-
tect rock that is susceptible to air slaking. It is good practice to exca-
vate the final 1 to 2 ft (0.5 m) of rock just prior to placement of con-
ventional concrete, shotcrete, or RCC to minimize the surface
deterioration of the rock.
Because RCC is invariably less costly per unit volume than conven-
tional concrete, the designer should maximize the use of RCC in the
foundation. However, conventional concrete surfaces generally are
built up first to facilitate the start of RCC placement. The size of the
starting pad for RCC placement can be as small as one roller width by
two roller lengths.
Handheld compactors or small rollers may be used for compacting
RCC or conventional concrete into areas that are inaccessible to the
large rollers used for the remainder of the dam. Thin lifts usually are
required when using such equipment in order to achieve roughly the
same density as the RCC for the body of the dam. Conventional con-
Design of RCC Gravity Dams 71

Crete should be used at the dam-abutment interface in order to assure

bonding to the rock surface and to minimize seepage at this contact.

3.2.3 Very low modulus or Nonrock

Nonrock foundations such as silt, sand, gravel, and clay may be ac-
ceptable for an RCC gravity dam. Such foundations are usually con-
sidered only for low dams, and a number of factors must be studied
first. The principal considerations for low-modulus rock or nonrock
foundations are differential settlement, seepage, piping, and erosion
at the downstream toe. Foundations of this type may require special
measures such as upstream and downstream aprons, cutoff walls, and
a drainage system.
Lower Chase Creek is an example of an RCC dam placed on a very
low modulus foundation. The Cedar Falls and Dryden replacement
dams in the State of Washington are examples of low RCC gravity
dams constructed on nonrock foundations. The site for Lower Chase
Creek consists of a conglomerate overlaid by alluvium. The designers
decided to construct the 64-ft (20-m) dam atop an RCC foundation mat
that extended through the alluvium and onto the conglomerate. Alter-
native designs consisting of a cutoff wall to conglomerate or building
the dam on compacted alluvium also were considered.
It was determined from plate bearing tests that the conglomerate
had a deformation modulus of 18,000 lb/in2 (0.125 GPa). The analysis
concluded that a dam founded on the conglomerate would be more
stable, experience less settlement, have reduced seepage potential,
and be no more costly than the cutoff-wall alternative. The RCC
foundation mat is as deep as 25 ft (7.6 ml. It extends out 10 ft (3 m)
upstream and downstream from the dam section on a 1 H : 1 V slope
and then extends vertically down to the conglomerate surface. The
completed dam is shown in Fig. 3.2.
Cedar Falls Dam is located in a valley where glacial sand and
gravel deposits are as much as 600 ft (180 m) deep. The foundation
design solution included replacing the upper 16 ft (4.9 m) of low-
density sand with compacted soil to reduce the potential for
earthquake-induced liquefaction, to decrease seepage below the RCC
dam, and to reduce settlement of the structure. A 20-ft-deep (6.1-m)
sheet pile cutoff at the heel (upstream base) of the dam and a concrete
upstream blanket were provided to reduce foundation seepage and to
lengthen the seepage path. A downstream filter and drain system con-
sisting of a uniformly graded gravel with a geotextile fabric was de-
signed to collect seepage and to control uplift and piping. A down-
stream slope of 0.8 H : 1 V was selected to help balance bearing
72 Chapter Three

Figure 3.2 Lower Chase Creek Dam (completed dam, downstream view).

pressures at the heel and toe under normal maximum design load and
thereby reduce the potential for tilting or differential settlement. The
dams section is shown in Fig. 3.3.

3.3 RCC Dam Design Concepts

3.3.1 General
Although it is difficult to generalize about RCC dam designs, they
may be divided into two basic concepts: (1) the conventional concrete

1 &in concrete

Glacial outwash
Figure 3.3 Section of Cedar Falls Dam.
Design of RCC Gravity Dams 73

gravity dam concept, and (2) the upstream membrane concept. The
high-paste RCC dam described in Chap. 5 and the Japanese RCD
structure described in Chap. 6 are dams designed using conventional
concrete gravity dam concepts and criteria. The lean RCC dam de-
scribed in Chapter 4 fits primarily into the upstream-membrane ap-
proach. The differences are primarily in the design philosophies
rather than the construction methods.
Concrete philosophy RCC mixes are generally used in dams de-
signed using the conventional concrete gravity dam concept, while
soils philosophy mixes are usually specified for RCC dam designs us-
ing the upstream-membrane concept, Some recent designs indicate
that certain aspects of each concept are being used to produce designs
that fall in between the two concepts as well.
Before proceeding with the final design of an RCC dam, the de-
signer should have project goals clearly in mind and should be famil-
iar with all aspects of the site. Factors that can affect the general con-
cept to be used in the design of the dam include the owners
requirements for cost, speed of construction, appearance, watertight-
ness, and operations and maintenance.
Upstream-membrane concept dams tend to be more economical, es-
pecially for low to moderate-height dams, than those that follow con-
ventional concrete gravity dam concepts. Depending on design details,
however, the cost advantage may be counteracted by disadvantages in
appearance, watertightness, and maintenance. In addition, the in-
creased shear or tensile properties of a concrete approach RCC mix
may be used to produce a steeper downstream slope and, therefore,
less volume in the dam.
The purpose of the dam could encourage a certain design approach
as well. A conventional concrete concept would be favored for a com-
plex hydroelectric dam, for example, while the upstream-membrane
concept seems more logical for a flood-control dam where watertight-
ness may not be a prime consideration. Dams for water storage or
multipurpose reservoirs can be designed in accordance with either
The location of the dam is another factor affecting the choice of
design concept. If the site is located in a seismically active area, the
greater tensile capacity of a concrete approach mix may be re-
Whatever design approach is employed, the designer should select
design details and an overall layout that considers the RCC construc-
tion method. The overall design should be as simple as possible in or-
der to take full advantage of rapid construction using RCC. Sugges-
tions about the location and configuration of spillways and conduits
through the dam are noted in Sec. 3.7.
74 Chapter Three

3.3.2 Conventional concrete gravity dam

A design that conforms to the concrete gravity dam concept assumes
that an RCC dam should have the same appearance, watertightness,
and performance as a conventional concrete gravity dam but cost less.
In general, the entire dam is considered as the water barrier. Japanese RCD method. The design of a dam for construction

by the RCD method is identical to that of a gravity dam to be con-
structed entirely of conventional concrete. The RCC is used for the
large volumes in the interior portions of the dam and replaces more
costly conventional concrete. Unique mix design and construction
methods, including transverse joints spaced at 49 ft (15 m), were de-
veloped in Japan to achieve the goals described in Chaps. 2 and 6. High-paste RCC dam. The design of a high-paste RCC dam dif-
fers from the RCD method in that the permeability of the RCC mix is
tied to the dam height as proposed by Dunstan in Chap. 5. The per-
meability requirements define an RCC mix with a minimum
cementitious content, a high percentage of which is fly ash. The
strength of the mix can then be used to define a minimum down-
stream slope for the dam. Design details such as exterior slipformed
facing elements that may lack vertical joints provide simplicity and
the opportunity for rapid construction rates for this type of RCC dam.

3.3.3 Upstream-membrane concept

This concept is unique to the RCC concrete gravity dam, producing a
structure somewhat similar to a concrete-faced rockfill dam. Lowe
felt, however, that a concrete-faced RCC dam should be a much better
structure than a concrete-faced rockfill dam because of its low com-
pressibility, which is several orders of magnitude less than the com-
pressibility of a rockfill embankment.
The upstream-membrane concept considers the upstream face,
rather than the entire dam, as the primary water barrier. It also rec-
ognizes that the complete bonding of each lift is doubtful for most RCC
dams, especially lean RCC dams. It should be noted that there are nu-
merous examples of unbonded lifts in dams constructed of conven-
tional concrete that have performed well for many years. Upstream
facing methods are described in Sec. 3.5.2, while factors involved in
bonding successful lifts are noted in Sec. 3.5.3. Because no credit is
given for the tensile capacity of either the face concrete or the RCC,
especially at the lift lines, the entire dam must remain in compression
under usual loading conditions, including uplift.
Design of RCC Gravity Dams 75

3.4 Structural Stability and Dam

Before starting to design the dam section and details, the designer
should have the site selected and an approximate height and design
concept in mind. The geometry of the RCC dam will be affected by
both structural stability and construction considerations. In most
cases, the designer selects a dam configuration and then checks it to
assure adequate structural stability, rather designing to produce a
certain shape.

3.4.1 Design criteria, loads, and factors of

safety Design criteria. The basic criteria for the design of a gravity
dam are

1. The dam must be safe against sliding on any horizontal plane

within the dam, at its contact with the foundation, or along any
plane or discontinuity within the foundation materials.
2. The dams cross section must be proportioned so that allowable
stresses in either the concrete or a foundation are not exceeded.
3. The dam must be safe against overturning of any section within the
dam, at the contact with the foundation, or within the foundation.
The USBR no longer calculates a factor of safety for overturning.
Before a gravity dam can actually overturn, other failures will oc-
cur, such as crushing of the material at the toe and cracking of the
upstream face, with an accompanying increase in uplift pressure
creating a reduced shear resistance. Overturning is controlled by
limiting tensile stresses on the upstream face. Earthquake forces
are not considered as contributing to overturning because of their
oscillatory nature. The overturning factor of safety, if desired, is
simply the total resisting moments divided by the total moments
tending to cause overturning about the toe. Loads. The loads that may act on a pavity dam are shown in
Fig. 3.4 and include the following.
1. Horizontal loadings

H, Hydrostatic pressure of reservoir on the upstream face. The unit

water pressure is the product of w, the unit weight of water (usually
62.4 lb/ft3 or 1000 kg/m31 and h, the distance from the water surface
to the point under consideration. The resultant water load
H, = LO x hl2.
76 Chapter Three

Maximum ,

i I
I b

Figure 3.4 Possible loads for a gravity dam.

H, Horizontal silt pressure. Including the effect of water, it is as-

sumed equivalent to the pressure created by a fluid weighing 85 lb/
ft3 (1360 kg/m).
H, Ice load on the upstream face. If the dam is located in an area
where ice may form, ice pressure may be estimated at 10,000 lb per
linear foot (14,900 kg per linear meter) of contact with the dam for
ice depths of 2 ft (0.6 ml or greater.
H, Impact of waves or seiches against the upstream face. This in-
cludes the effect of a potential landslide into the reservoir.
H, Hydrostatic pressure of tailwater against the downstream face.
H, Inertial force of the reservoir water against the dam during an
H, Inertial force of the mass of the dam during an earthquake.
2. Vertical loadings

V, Gravity dead load of the dam and appurtenances such as gates

and bridges. It should be remembered that RCC weighs more than
conventional concrete made with the same aggregates and can ex-
ceed 150 lb/ft3 (2403 kg/m31.
V, Vertical pressure of water on the upstream face, if that face is
Design of RCC Gravity Dams 77

inclined. Vertical silt pressure (including the effect of water) may be

assumed equivalent to the pressure of a soil with a wet density of
120 lb/ft3 (1925 kg/m3).
V, Uplift pressures on any horizontal plane-an internal hydro-
static pressure.
V, Inertial force of the mass of the dam during an earthquake.

The structure is also subjected to thermal stresses generated by hy-

dration of the cementitious materials and subsequent cooling, as de-
scribed in Sec. 3.5.1.
The presence and effect of hydrostatic uplift pressures within a dam
and its foundation have been the subject of controversy for many
years. For design purposes, it is now accepted that uplift pressures act
upon the full area of the gravity section and vary from full reservoir
pressure at the upstream face to zero or tailwater pressure at the
downstream toe for dams without drains. Where effective drains are
located near the upstream face, the uplift pressure is assumed by the
U.S. Rureau of Reclamation to be reduced to tailwater pressure plus
one-third of the difference between reservoir and tailwater hydrostatic
pressure at the line of the drains (see Fig. 3.4). Loading combinations. An RCC dam should be designed for all

reasonable combinations of the loads previously listed. As the proba-
bility decreases that a certain combination of loads will occur at the
same time, the required factor of safety for design decreases accord-
ingly, as noted in Sec. Most combinations of loads can be cat-
egorized as usual, unusual, or extreme, as described below.

1. Usual loading combination: The usual loading combination con-

sists of hydrostatic pressure of the reservoir water at the normal
operating elevation, gravity dead loads, uplift, silt pressure, ice
pressure, and tailwater pressure, if applicable, together with appro-
priate temperature effects occurring at the same time.
2. Unusual loading combination (overtopping): The unusual loading
combination consists of hydrostatic pressure of the reservoir water
at the maximum design elevation, gravity dead loads, uplift, silt
pressure, ice pressure, and tailwater pressure, if applicable, to-
gether with the appropriate temperature effects occurring at that
3. Extreme loading combination (earthquake): The extreme loading
combination consists of hydrostatic pressure of the reservoir water
at the normal operating elevation, gravity dead loads, uplift, silt
pressure, ice pressure, temperature effects, and tailwater pressure,
78 Chapter Three

if applicable, plus the effects of the maximum credible earthquake


The dam should also be analyzed for any other loading combination
the designer feels appropriate, including uplift with drains inopera-
tive. Factors of safety. The factors of safety to be used for the de-
sign of a gravity dam are usually determined by the governmental
agency responsible for regulating the safety of the dam . One pub-
lished guideline is USBR Monograph No. 19, Design Criteria for Con-
crete Arch and Gravity Dams.
In the LJSBR criteria, minimum factors of safety required for foun-
dation stresses are 4.0, 2.7, and 1.3, for the usual, unusual, and ex-
treme combinations of loads, respectively. Similarly, the minimum
factors of safety for stresses applied to concrete within the dam are
3.0, 2.0, and 1.0, for the usual, unusual, and extreme loading combi-
nations, respectively.
Limits are also applied to maximum allowable compressive stresses.
In no case shall the allowable compressive stress exceed 1500 lb/in2
(10.3 MPa) for the usual load combinations. The maximum allowable
compressive stress is increased by 50 percent to 2250 lb/in2 (15.5 MPa)
for the unusual load combinations. For the extreme loading condition,
the maximum allowable compressive stress can be determined in the
same manner, using a factor of safety greater than 1.0.
Concrete is assumed to crack if its tensile strength is exceeded by
the extreme combination of loads, including the loads induced by a
maximum credible earthquake. The dam is then analyzed with crack-
ing included, and if structural stability is assured, the dam can be con-
sidered safe against a sudden release of the reservoir despite the sus-
tained damage.
Not all designers and agencies agree on design criteria and fac-
tors of safety. The two items that have attracted the greatest atten-
tion. and with it some controversy, are whether to take credit for
the tensile resistance of the RCC at lift lines and what factor of
safety to apply for shear resistance. A redefined factor of safety for
shear resistance based only on sliding friction resistance is de-
scribed in Sec. 3.3.3.
The USBR, following its own criteria, designed Upper Stillwater
Dam for a minimum tensile strength of 180 lb/in (1.24 MPa) in the
high-paste RCC and between successive lifts. Most designers do not
allow for the tensile strength of the RCC under usual loading condi-
tions. This may be due to the use of leaner mixes or the possibility
Design of RCC Gravity Dams 79

that poor construction control could lead to a lift surface with no bond
and, therefore, no tensile capability.

3.4.2 Method of analysis

There is little difference in the structural analysis of an RCC dam
built with few or no transverse joints and a concrete gravity dam built
in blocks. Restraint at the abutments and temperature effects usually
cause vertical transverse cracks in an RCC dam. In some cases, the
RCC dam design calls for vertical joints. Either way, most RCC dams
can be analyzed using a two-dimensional gravity method.
Absent cracks or joints, an RCC dam can be analyzed structurally in
three dimensions. In such a case, the dam would be restrained at the
abutments and should be considered fixed at those points. The dam
can then develop tensile stresses at the upstream face similar to a
fixed beam loaded horizontally.
For RCC dams constructed across wide canyons, the major portion of
the structure may be considered to be two-dimensional and analyzed
by the gravity dam method. If an RCC dam is built across a narrow
canyon, the edge conditions will have more effect on its structural ac-
tion. More complex methods of analysis will be required, such as a
trial-load twist analysis, the beam and cantilever method, or a finite-
element analysis. All three are described in the USBRs Design of
Gravity Dams.
A static analysis should be applied to all gravity dams. If the dam is
located in a seismically active area, a dynamic stability analysis is
also required. In this case, the selected dam configuration may be
evaluated using a two-dimensional finite-element static and dynamic
analytical model such as the EAGD-84 computer program developed
at the University of California at Berkeley. The minimum dam sec-
tion is analyzed using the ground motion time-history accelerogram
representing the design earthquake. The program performs an earth-
quake analysis of gravity dams, including the effects of dam-water-
rock interaction and the effects of materials such as alluvium and sed-
iments at the bottom of the reservoir. The two-dimensional finite-
element analysis results are used to evaluate the stress distribution
within the dam and to calculate the shear-friction factor of safety at
the base of the dam under both static and dynamic conditions. Sliding analysis. The sliding or shear-friction factor of safety

(SFF) is a measure of the stability of the dam against sliding or shear-
ing. It can be used to define a minimum base width for the gravity
dam and the slope above the base, or to check the SFF for an already
80 Chapter Three

determined section. The shear-friction factor of safety is defined as


CA t ( W - lJ)tan&

where c = unit cohesion

A = area of section
W = vertical weight above section
U = uplift at section
4 = angle of internal friction
H = horizontal driving force

W, V, and H are loads on the dam, while c and + are concrete prop-
erties determined from direct shear tests of specimens prepared from
the design mix or cores extracted from a test section using the same
mix. Figure 2.16 provides a summary of values of c and + derived from
RCC mixes.
The minimum SFF required is determined by the criteria selected.
For the usual combination of loads and USBR criteria, the minimum
SFF within the dam is 3.0. Based on these criteria with a vertical
downstream face, the steepest downstream slope for variable cohesion
c divided by structural height h and coefficients of internal friction
tan 6 are shown later in Fig. 5.5, omitting the influence of drains.
A realization that cohesion on some RCC lift surfaces may be zero
has led some designers to redefine the shear factor of safety. Instead of
taking into account both an average cohesion and a sliding friction re-
sistance on the lift surface, the cohesion resistance is not considered.
The design for shear is then based strictly on the sliding friction factor
or the residual shear resistance of the RCC in conjunction with a re-
defined lower factor of safety (FS).
It may appear that this redefined lower FS is compromising the
overall safety of the structure. This is not the case, however, because
the ultimate shear resistance of the concrete is equal to its residual
strength once the bond (cohesion) is broken, as shown in Fig. 3.5. Sug-
gested values of the redefined FS have ranged from 1.5 to 2.0, which
may be the approximate ratio of the peak to the residual shear
Where zero cohesion at lift lines and 100 percent efficient drains are
assumed, the downstream slope for various shear-friction factors of
safety and coefficients of friction (tan 4) may be determined by using
Fig. 10.2.
Even with the reduced SFF, some positive cohesion may be required
Design of RCC Gravity Dams 81

Figure 3.5 Ultimate and resid-

/ ual shear strength of concrete.

high in a spillway section when overtopping is considered, because of

the small volume of concrete (low vertical weight) above the horizon-
tal plane being investigated. A bonding mortar over the entire lift
could be required in this case to develop sufficient cohesion to ensure
adequate shear capacity.
The zero-cohesion concept was used for the design of Copperfield
and Craigbourne dams. Most soils approach dams have, however, been
designed with some relatively low value of cohesion and the tradi-
tional factor of safety for usual loading conditions. Stress analysis. For a gravity dam subjected to the usual load
combinations, the maximum compressive stress occurs at the down-
stream toe and the minimum compressive stress or maximum tensile
stress at the upstream heel. Most designers will allow no tensile stress
in an RCC dam under usual load conditions, including uplift pres-
Once the maximum stresses within the dam are calculated, the con-
crete mixture is designed to provide sufficient strength to withstand
the maximum stress multiplied by an appropriate factor of safety. Fig-
ure 3.6 indicates minimum RCC strength requirements used for the
design of a number of RCC dams.
The maximum compressive stress is usually not a major factor in
design, except for very high dams. Depending on the base width, the
maximum compressive stress is roughly equal to only 1 lb/in2 . ft
(0.025 MPaim) of dam height. Because tests to determine compressive
strength are quite reliable and relatively easy to perform, this param-
eter is usually specified, since other properties can be determined by
their relationship to the concretes compressive strength.
For a dam with a vertical upstream face, Fig. 5.4 can be used to de-
termine a downstream slope for various dam heights without drains
as a function of compressive strength divided by a factor of safety.
The RCC mixture for Upper Stillwater Dam, as noted in Sec.
82 Chapter Three

Compression Shear Other

Age, strength, strength, criteria,
Dam days lb/in (MPa) lb/in (MPa) lb/ins (MPa)

Shimajigawa 90 2844 (19.6)

Willow Creek A 90 1100 (7.6) c = lll(O.77)
(interior) Q = 63
Middle Pork 90 1650 (13.4) c = lOO(O.69)
4 = 45
Copperfield 90 3175 (15.0) c=o FS
c#l = 45 (shear) = 2.0
Craigbourne 90 1450 (10.0) c=o FS
c$ = 45 (shear) = 2.0
Upper 365 3000 (20.7) c = 300 (2.07) Static
Stillwater tension = 180
Elk Creek 365 2000 (13.8) c = 50 (0.35)
Pam0 365 3000 (20.7) Dynamic
tension = 350

Figure 3 . 6 Minimum RCC Design Strength (See Fig. 2.14 for RCC mixture proportions
used.), was designed to resist a minimum direct tensile stress of 180

lb/in (1.24 MPa) at the lift lines, although there is actually no tension
in the dam under any condition. The following paragraphs explain
how the USBR takes advantage of the tensile strength of the concrete.
A stress analysis is performed for the usual combination of loads,
excluding uplift. Under this condition all stresses in the dam must be
compressive, with a greater compressive stress at the toe than at the
In 1895, Levy theorized that once the hydrostatic reservoir pressure
at a certain depth below the water surface (wh) exceeded the compres-
sive stress in the concrete at the upstream face, the concrete would
crack. No one has ever observed or recorded this imaginary cracking.
Two reasons why this cracking has never occurred in a properly pro-
portioned concrete gravity dam is that the concrete is capable of re-
sisting some tensile stress and that 100 percent uplift or hydrostatic
pressure does not actually occur at the upstream face.
Based on tests on concrete and masonry specimens to determine
their porosity and uplift coefficient, Leliavsky proposed that a factor of
0.85 be applied to the maximum hydrostatic uplift for design purposes.
The USBR does not apply the 0.85 reduction factor but does use the
RCCs tensile strength to reduce the minimum required compressive
stress (C,,,i,,) at the upstream face.
Design of RCC Gravity Dams 83

The following equation basically says that the concrete will not
crack until the hydrostatic pressure exceeds the compressive stress in
the dam plus the tensile capability of the concrete. The equation also
takes into account the effect of drains in reducing hydrostatic pres-
sure. Therefore, the calculated compressive stress at the upstream
face for the usual combination of loads excluding uplift must exceed
Cmin as calculated from the equation.

Cmin = whp - f,IFS (but not less than zero)

where w = unit weight of water

h = depth of water to plane being considered
p = a drain effect factor (see below)
f, = tensile strength of RCC at lift lines
FS = factor of safety

The value ofp is 1.0 if no drains are included in the design, and 0.4
if drains are used. The 0.4 represents the ordinate of a triangular
stress distribution that is equivalent to the uplift stress distribution
where the pressure is reduced to 0.33 of the maximum stress at the
line of the drains, assuming no tailwater. (See uplift pressure distri-
bution with drains in Fig. 3.4).
In the case of a dam with no drains, where p = 1.0, f, = 180 lb/ins
(1.24 MPal, and FS = 3.0, the equation indicates that for the first 138
ft (42 m) of water depth, no additional compressive strength is needed
to withstand full hydrostatic uplift pressure. For greater depths, the
whp factor is greater than f,IFS (60 lb/in2 = 0.41 MPa) and the initial
stress analysis must produce a compression on the upstream face
equal to or greater than the Cmin computed by the equation. With
drains included in the design and the 0.4 drain reduction factor ap-
plied, tensile strength of RCC of 180 lb/in2 (1.24 MPa) is sufficient to
withstand the hydrostatic uplift pressure for all but very high dams.

3.4.3 Gravity dam section configuration

In order to produce a dam that is structurally efficient and economi-
cal, gravity dams are generally triangular in cross section. They are
generally located straight across the valley, but may be curved up-
stream in plan to better fit the topography of the site, such as for Saco
Dam. See Sec. 3.9 for a discussion of RCC dams curved in plan, includ-
ing arch dams.
The crest elevation of an RCC dam is normally identical with the
maximum water surface. A parapet wall is usually added to take care
of freeboard. This is different from embankment dam design, where
freeboard and settlement considerations raise the elevation of the dam
04 Chapter Three

crest. Some details pertaining to the design of the dam section are dis-
cussed in the following sections. Upstream face. The upstream face of a gravity dam is usually

vertical in order to concentrate the weight of concrete upstream to bet-
ter resist the reservoir water loading and to simplify construction. It is
common practice in Japan and elsewhere to add a batter to the lower
part of the upstream face to increase the base thickness and thereby
improve sliding stability at the base, as shown in the cross section for
Shimajigawa Dam, Fig. 3.7~. If a batter is used, stability and stresses
should be checked at the elevation where the batter intersects the ver-
tical upstream face.
Where construction considerations dictate a flatter downstream
slope than what is required for stability, such as in the case of an
unformed downstream face, sufficient base width for sliding is invari-
ably provided. See Fig. 3.7b for the cross section of Galesville Dam. Downstream slope. The downstream slope is generally con-

stant from the base to near the crest of the dam. A constant slope is
considered efficient from the point of view of both structural design
and construction. A curved downstream slope may be considered for a
high RCC gravity dam in order to minimize volume. It can be con-
structed rather easily
The point of intersection of the slope with the upstream face should
preferably be at the crest elevation in order to minimize the dams vol-
ume without sacrificing structural stability. It may be higher for other
reasons, such as to accommodate the spillway or to start the slope at
the downstream edge of the crest. If the point of intersection is higher
than the crest elevation, the effective slope that defines the base width
is flatter than the actual slope. At Upper Stillwater Dam, where the
actual slope is 0.6 H : 1 V, the effective slope is 0.68 H : 1 V.
It is poor practice in seismic areas to locate the point of intersection
of the downstream slope with the vertical upstream face below the
crest elevation. The obtuse angle formed by the intersection of a down-
stream vertical face with the slope will create an undesirable stress
concentration when the dam is subjected to seismic shaking, as was
proven at Koyna Dam in India. Crest detail. The crest width is usually determined either by

construction considerations or by the width of a roadway across the
dam. For low dams, the minimum crest width should be one RCC
equipment width plus the width of the upstream face, or about 10 ft (3
m). For moderate-height dams, a minimum crest width of 16 ft (5 ml is
recommended in order to better accommodate two widths of construc-
tion equipment. A crest width of nearly 30 ft (9 ml was used for the
294-ft-high (90-m) Upper Stillwater Dam.
\ 8.J 25 m
Desian flood level el. 300.00 m

Surcharge water level el. 297.10 m

Normal high-water level el., 286.50 m [

Sedimentation level el. 247.50 n

el. 242.50 n
* tl. 232.00 m
conduit, = 1200 mm
8 El. 224.00 m(L)
h.rj 1 _--------- - - - - -
El. 233.00 m(R)
El. 214.00 m
h t - - -
0.56 m 4 34.4,
9.15 m L 69.040 m I I 10.500 m
8:012 in


Figure 3.7 Cross sections for three RCC dams through spillway. (a) Shimajigawa Dam.


Design of RCC Gravity Dams 87

Once a horizontal crest is established, there are a number of ways of

connecting it with the downstream slope. The most common method is
to use a downstream vertical face to intersect with the slope. There
also may be an intermediate steep slope near the crest, as shown in
Fig. 3.7~ for Upper Stillwater Dam. Another solution is to determine
the base width using an effective slope and then connect the toe posi-
tion so determined with the downstream edge of the horizontal crest. Overflow spillway. When an overflow spillway is incorporated

into the dam, the spillway section should be similar in shape to the
nonoverflow section. In particular, the downstream slope of the spill-
way section should be parallel to the slope of the nonoverflow section.
Curves describing the spillway crest and also the intersection of the
spillway slope with the stilling basin or other energy dissipator are
described in the USBRs Design of Gravity Dams. Spillway configura-
tion is discussed further in Sec. Other dam profiles. Although the traditional triangular grav-

ity dam profile produced by a vertical upstream face and downstream
slope is an efficient section, other shapes may be used, especially for
low dams. Other dam profiles generally have sloped faces both up-
stream and downstream.
Such was the case for Bucca Weir in Australia, as shown in Fig. 7.3.
The 39-ft-high (12-m) weir, which would be submerged under the l-
in-lo-year flood, was considered too low to properly incorporate a gal-
lery in the section and drain holes in the foundation properly. There-
fore, for stability against overturning under maximum uplift
conditions, a vertical upstream face section would have required ei-
ther a relatively flat downstream slope or posttensioned anchors ex-
tending into the foundation. Instead, by sloping the upstream face at,
the angle of repose of the RCC mixture (1 H : 1 V), no forms were nec-
essary and the downstream slope could be steepened to 0.5 H : 1 V.
The steepened downstream face was justified as forms were necessary
to construct the stepped conventional concrete spillway facing.
Composite sections consisting of an earth embankment upstream
and a concrete gravity downstream include the North Loop detention
dams at Austin, Texas, where the embankment was needed to support
a four-lane roadway across the dam (Sec., and the Sir Coffer-
dam in Turkey, where the upstream fill material was used to form the
vertical upstream face of the RCC dam.

3.5 Design details

3.5.1 Thermal study and cracking analysis
Volume changes caused by temperature and moisture variations have
long been a concern during the design of mass structures of conventional
as Chapter Three

concrete. The same concern applies to RCC dams. However, dams

constructed of RCC have two advantages with respect to thermal
control over dams constructed using conventional concrete. The ad-
vantages are: (1) the lower cementitious contents of RCC mixtures,
and (2) the rapid placement of RCC, which minimizes surface expo-
sure to radiant heat and what are usually higher air temperatures.
Both of these conditions create a reduced temperature rise within
the RCC dam, therefore making it less susceptible to cracking when
compared to a similar dam constructed of conventional concrete.
Concrete expands with an increase in temperature and contracts with
a temperature drop. Similarly, concrete expands with an increase in
moisture and contracts with a moisture loss. Because concrete is strong
in compression and weak in tension, the decrease in volume caused by
decreases in temperature or moisture are of greatest concern. Thermal
tensile strains that can cause cracking deep into or through the dam are
of much greater importance to designers than drying shrinkage, which is
usually limited to the exposed surface of the mass structure.
If the RCC in the dam were free of any restraints, the decrease in
volume caused by a uniform temperature drop across the section
would present no problem. RCC dams are bonded to rock foundations,
however, and are thus subjected to an external restraint. Restraint
can also be initiated when the cool exterior surface restrains a hot in-
terior. The two types of restraint lead to two different temperature
analyses, which may be called external restraint in the first case and
internal restraint in the second.
External restraint is greatest (fully restrained) at the dam-
foundation contact and nearly zero (unrestrained) at the crest near the
center of the dam. Internal restraint is greatest in the zones where the
temperature change is occurring at the slowest rate. For the external
restraint condition, the temperature of the mass peaks and then cools
over a period of time to reach the average ambient air or reservoir wa-
ter temperature at the site. When the volume reduction due to this
relatively long-term temperature drop combined with foundation re-
straint exceeds the tensile strain capacity of the RCC, transverse
cracking occurs at the section of least resistance. Cracks produced by
the external restraint condition are usually vertical or near vertical
and can extend through the entire dam section.
In the interior restraint condition, the temperature at the center of
the concrete mass is higher than at the exposed dam faces. As the sur-
face cools and wants to contract, compressive stresses remain in the
warm interior while tensile stresses develop at the outer surface.
Surface cracking occurs when the tensile stress, due to nonuniform
temperature change, together with internal restraint, causes a strain
that exceeds the capacity of the concrete. The resulting surface cracks
are generally vertical, transverse to the dams axis, and do not extend
Design of RCC Gravity Dams 89

through the entire dam. Once cracks are initiated, the energy re-
quired to propagate these cracks is less, so both restraint conditions
can combine to produce a deeper crack.

3.5.1 .l Thermal studies. Thermal computational methods can range

from a sophisticated computer-aided finite element analysis to hard com-
putation methods. Most major RCC dams have employed some form of a
finite element analysis to determine temperature distributions within
the concrete structure from which strains and stresses are determined.
Finite element models can be detailed three-dimensional grids of
the entire structures. However, two-dimensional finite-element-based
computer programs are usually employed. Two of the proprietary pro-
grams available for thermal analysis are ADINAT and ABACUS. A
nonproprietary program called THERM or DOT/DETECT developed
by the University of California at Berkeley for the U.S. Army Corps of
Engineers Walla Walla district was used for Willow Creek and Elk
Creek dams.
The heat-flow program simulates temperature distribution with
time within the RCC dam as the dam is constructed in thin horizontal
lifts. The temperature within the structure depends upon the placing
temperature, the heat-generating characteristics of the RCC mixture,
and the boundary conditions at the various exterior surfaces. The
boundary temperatures include that of the air or water and absorbed
radiant heat on the exposed upper RCC surface, upstream and down-
stream faces, and the temperature of the foundation rock.
In conducting a thermal analysis, it is necessary to determine cer-
tain properties of the RCC mixture and to develop a detailed construc-
tion schedule. The properties can include specific heat, diffusivity,
conductivity, coefficient of thermal expansion, adiabatic temperature
rise at various ages, tensile strength at various ages, as well as mod-
ulus of elasticity and creep coefficients at various ages.
The planned construction schedule is required because expected air
temperatures are an input variable to the computer program. If there
is a delay in the start of construction, or construction is not progress-
ing as fast as anticipated, or both, it may be necessary to recalculate
the temperatures within the mass.
A delay in the construction start usually means the ambient tem-
peratures are warmer and that higher peak temperatures than pre-
dieted will occur. In addition, a slower rate of placement indicates ma-
terial properties that are time-dependent, such as tensile strength and
modulus of elasticity, may differ from predicted values. Also, the plac-
ing temperature of the RCC is a required input for the analysis and
this value is directly affected by ambient temperatures.
The revised analysis may predict higher tensile stresses in the
structure, which may then dictate that more joints be placed in the
90 Chapter Three

upper portion of the dam. Such was the case for Monksville Dam. If
there is a possibility that there will be design changes related to ther-
mal conditions after the start of construction, provision for payment
should be made in the original contract. A modified version of the Uni-
versity of California at Berkeleys thermal analysis program was used
for Saco Dam to analyze surface temperatures and to assess the prop-
erties of both the conventional concrete facing and the lean interior
RCC mix. Drying shrinkage stresses were added to thermal stresses to
predict cracking of the upstream face more accurately. Controlling temperature rise. The maximum internal tempera-

ture rise can be limited either by reducing the temperature rise of the
RCC mixture or by reducing the placing temperature. Consistent with
strength and permeability requirements, the use of low-heat cement,
a reduction in total cementitious material, and an increase in pozzolan
percentage will reduce the temperature rise of the mixture. Precooling
aggregates by winter production or water spray, introducing ice to sat-
isfy mixing water requirements, or placing the RCC at night are
methods that have been used to reduce the placing termperature of
the RCC mixture. Night placement also aids with minimizing radiant
heat effects on the exposed upper surface.
Dam sites that have low annual average temperatures, such as high
mountain locations, provide a greater potential temperature drop. De-
signers of RCC dams at such locations may desire to limit concrete
placing temperatures to as low as 50F (10C). This was the case at
Upper Stillwater Dam, located at about 8000 ft (2450 m) above sea
level, where the annual average temperature is only 37F (3C). The
50F (10C) RCC placing limitation was met almost all the time by us-
ing ice as the mixing water. With no-slump RCC there is less mixing
water than in conventional concrete, so expensive liquid nitrogen had
to be introduced into the mixers to cool the RCC on only two occasions.
In a thermal study done for Miyagase Dam in northern Japan, four
different concrete placement starts were studied--January, April,
July, and October. The maximum expected temperature in the RCD
for one year was plotted and compared with the allowable maximum
temperatures of the material. It was determined that starting RCC
placement in October was best. Cracking analysis. Assuming that the computer-generated

thermal study produces a reasonably accurate prediction of the mag-
nitude and location of temperature peaks, the next step facing the de-
signer is to determine whether the subsequent temperature decrease
will produce cracking. If cracking is indicated, a greater challenge is
to determine the location and width of the cracks and what, if any-
thing, to do about them.
Design of RCC Gravity Dams 91

Tensile stresses which are proportional to strains within the elastic

range can be estimated from the temperature study by using a finite-
element computer program or by manual calculations, with restrain
factors taken into account. A thorough analysis can be quite complex
if all time-dependent variables, such as creep (or relaxation) and the
changing properties of every RCC lift are considered.
Manual calculation of tensile stress can be estimated by the follow-
ing formula:
where CT = temperature stress
R = degree of restraint
E = modulus of elasticity of the RCC
p = coefficient of thermal expansion of the RCC
At = temperature drop of the RCC

The degree of external and internal restraint R may be taken from

the charts in Fig. 3.8a and b, which were produced by Fijisawa and
Nagayama. The charts indicate that if the elastic moduli of the con-
crete E, to that of the foundation rock Ef is constant and the temper-
ature change in the concrete is uniform, then the degree of external
restraint is determined by the ration of the height H to length L of the
concrete block, or the distance between transverse joints. In cases
where the concrete can be assumed to be fully restrained (R = 11, such
as adjacent to the foundation, the stress is simply the reduction in
temperature multiplied by the coefficient of thermal expansion and
modulus of elasticity of the material.
The tensile strain capacity of concrete can vary considerably de-
pending on its mix proportions and the rate of loading. When strain is
applied slowly, the tensile strain capacity of the concrete can be far
greater than when it is applied rapidly. A slow loading rate is gener-
ally the case under most temperature conditions for RCC dams. How-
ever, thermal shock may occur when there is a large and rapid tem-
perature differential from day to night, such as at the time of change
between seasons. In this case, a more rapid loading rate should be as-
sumed in determining the lower tensile strain capacity of the concrete.
The two main factors governing the tensile strain capacity of RCC
are its tensile strength and the modulus of elasticity. High strain ca-
pacities are produced by concretes with high tensile strength and low
modulus of elasticity E. It should also be noted that while a richer
RCC mix generates more heat, it also has greater tensile strain capac-
ity. Tensile strain capacity also increases with time, as do most con-
crete properties.
Aggregates can have a major effect on the tensile strength, modulus
of elasticity E, and coefficient of thermal expansion of the RCC.
0 01 0 2 0 3 04




Figure 3.8 Degrees of restraint for thermal cracking
analysis. (a) External restraint by foundation rock.
(1 = Degree of external restraint at base of concrete,
2 = ratio of height to length of concrete; 3 = ratio of elas-
tic moduli of concrete to foundation rock.) (b) Internal re-
straint. (1 = Degree internal restraint at concrete sur-
face; 2 = ratio of depth where temperature drop occurs in
concrete to length of concrete; 3 = uniform temperature
drop, 4 = transular temperature drop; 5 = parabolic
temperature drop.)
Design of RCC Gravity Dams 93

Crushed aggregates usually produce greater tensile strength than

natural aggregates, all other factors being the same. Aggregates with
a low coefficient of thermal expansion and a low E reduce the poten-
tial for cracking. Locating and importing aggregates with better ther-
mal properties from sources more than a few miles away is usually not
a practical, economical solution, however. Aggregates available at the
site are generally used for RCC dams, and the thermal properties of
those aggregates are included in the cracking analysis.
Tensile strain capacity can be measured directly on RCC cylinders
loaded in tension. It may be calculated approximately by dividing the
tensile strength by the modulus of elasticity of the RCC in tension if
little or no plastic strain occurs during loading.

351.4 Design for crack control. After it has been established that the
dam will crack and the approximate spacing or location of the cracks
has been identified, the designer must now determine a crack control
solution. It is the authors opinion that it is virtually impossible to de-
sign a straight RCC gravity dam without some cracks occurring. One
solution is do nothing about the predicted cracks initially.
At Upper Stillwater Dam where cracks at lOO-ft (30-m) spacing
were predicted, it was decided to provide no crack control in the de-
sign. It was reasoned that if the anticipated cracks proved to be a
problem at a later date, sufficient funds would have been saved on
construction to pay for future crack repairs. Another solution is to in-
stall water-stopped and drained joints partially or completely through
the dam. Upstream facing methods, including those with joints, are
described in Sec. 3.5.2.
It should be remembered that a concrete dam will crack at points of
least resistance and greatest restraint. Cracks will form in notches or
crack inducers placed in the upstream face and at points of sudden
change in foundation contour. By applying these principles, joints
may be located at a few strategic locations, such as where the dam sec-
tion changes.
Vertical transverse cracks do not reduce the stability of an RCC
gravity dam. The problem associated with cracks is the potential for
seepage, which could cause loss of water and unsightliness. The
greater the degree of watertightness required for an RCC dam, the
greater the need for an impermeable upstream face, which may in-
clude waterstopped joints. Another factor in achieving the overall de-
sired watertightness is the permeability of the RCC material as de-
scribed in Sec. 2.5.3.
Thermally induced longitudinal cracks are not considered to be a prob-
lem in an RCC gravity dam. In the thermal study for the 509-ft-high
(155-m) Miyagase Dam, which has a base width of 591 ft (197 m), it was
deterimined that longitudinal contraction joints could be eliminated.
94 Chapter Three

3.5.2 Seepage Control and Upstream

Facing Methods
The various methods chosen for reducing or controlling seepage have
produced the greatest variation in design solutions for RCC dams. The
basic form of seepage reduction can be divided into two categories: (1)
those solutions that rely upon the entire interior RCC mass for the
dams impermeability, and (2) those that rely on an impermeable or
relatively impermeable upstream face or membrane as the primary
water barrier. For secondary seepage control, the upstream facing de-
signs may also include partial or full bedding mixes between lifts, and
some form of drainage collection system downstream from the face.
Willow Creek, which was the initial example of a lean RCC dam,
and Upper Stillwater, the first high-paste-content RCC dam, are both
examples where the entire RCC section was considered in design as
the water-retaining element.
Precast concrete panels were used as forms for the vertical RCC up-
stream face at Willow Creek, while horizontal slip-formed elements of
conventional concrete served both as forms and as the durable exterior
skin for Upper Stillwater. (See Fig. 8.7 for a photograph of construc-
tion of the facing elements at Upper Stillwater.)
While the initial seepage that occurred through Willow Creek may
be considered as acceptable for a flood-control dam, it is not considered
acceptable by most designers for other reservoir purposes.
Based on Willow Creeks performance, designers have developed a
number of more impermeable solutions. Nearly all of these later
seepage-reduction solutions have been of the upstream-face water-
barrier type.
The selection of the upstream face will depend primarily on water-
tightness requirements and cost considerations. Some information on
cost of the various facing methods is contained in Fig. 10.6, and seep-
age performance in Sec. including that of Willow Creek. Membrane-faced concrete panels. Precast concrete upstream

facing panels with a 65mil-thick (1.65-mm) polyvinyl chloride (PVC)
liner bonded to the downstream side in the manufacturing process
were used at Winchester Dam. The joints in both directions were heat-
welded in the field to form an impermeable upstream membrane and
were backed by an average of 18 in. (0.46 m) of conventional concrete
(see Fig. 3.9). A bedding mix atop each 1-ft-thick (0.3-m) RCC lift ex-
tended another 1.5 ft (0.46 m) downstream. The PVC membrane was
embedded horizontally in conventional concrete in the foundation key-
way and wrapped into a trench cut into the abutments to complete the
impermeable dam facing. In the entire construction process, care must
be taken so as not to puncture the membrane.
Design of RCC Gravity Dams 95

Figure 3.9 Winchester Dam (membrance-faced concrete panels).

This method was also used for the 253-ft-high (77-m) Urugua-i Dam
in Argentina. For Urugua-i, the thickness of the conventional concrete
downstream panels averaged 2.3 ft (0.7 m). The width of the bedding
mix atop each 16-in-thick (0.4-m) RCC layer varied from 16 ft (5 m)
near the top to 39 ft (12 m) at lower elevations.
Two patents, one for the membrane-lined panels and another for the
method of installing the panels, have been issued in the United
States. Royalty payments are, therefore, required to use this method
in areas where the patent is in-force. Reinforced and jointed concrete face. The spillway for Stacy
Dam was constructed with an M-in (0.46-m) constant-thickness con-
crete face that was reinforced in both directions. Vertical galvanized
steel sheet water-stopped joints were spaced horizontally at 50 to 60 ft
(15 to 18 m) with reinforcing extending across the joint. Construction
joints were spaced at 13 ft (4 m) vertically. See Fig. 3.10 for a detail of
the face, which is attached to the RCC by steel anchor bars placed atop
RCC lifts as construction proceeded upward.
The reinforced upstream concrete face was cast after the interior
RCC. Placement was delayed until internal RCC temperature started
to decrease as a precaution against potential thermal cracking in the
face. The reinforced and jointed concrete face is very similar to the up-
stream face design for a concrete-faced rockfill dam, Unreinforced concrete face with waterstopped joints. This basic

design has had considerable variation with respect to face thickness,
joint spacing, and depth of the contraction joint into the RCC mass.
Dams constructed by the Japanese RCD method, such as Shimajigawa
and Tamagawa, have lo-ft-wide (3-m) conventional concrete upstream
faces with vertical joints spaced 49 ft (15 m) on center. At Elk Creek
Dam, the average face thickness was 3 ft (0.9 m), with a horizontal
96 Chapter Three

3/4 in diameter threaded anchors

Roller- compacted 3.5ft(l.O7m) long@ 2ft(0.6m)V,

c/c each way,each face

Figure 3.10 Section plan of upstream facing-Stacy Dam.

joint spacing up to 300 ft (91 m) as a means of controlling major trans-

verse cracking. The waterstopped and drained joint used at Elk Creek
is shown in Fig. 3.11. For the Japanese dams and at Elk Creek, crack
inducers in line with the waterstopped joints were produced by insert-
ing galvanized steel sheeting into the RCC prior to compaction across
the dam for its entire height.
The minimum concrete face thickness for Monksville Dam was 1.3
ft (0.4 m) with a designed joint spacing in the face of 20 ft (6.1 m).
During construction, the joint spacing was increased to 40 ft (12.2 m)
except near the abutments, thereby eliminating nearly one-half of the

Figure 3.11 Elk Creek Dam (waterstopped joint).

Design of RCC Gravity Dams 97

waterstops. No provision for contraction joints through the RCC mass

was made in the original design. However, a delay in construction
start to warmer weather caused the designers to revise their thermal
analysis. The higher predicted peak temperatures caused a redesign
to incorporate monolith joints through the RCC at 120-ft (36.6-m) in-
tervals for the upper 40 ft (12.2 ml of the dam.
A bedding mortar was placed over the entire surface of each 2-ft
(0.6-m) lift for the Japanese dams and at Elk Creek. The bedding mix
at Monksville extended 8 ft (2.4 m) from the upstream face atop the
1-ft-thick (0.3-m) lifts. Unreinforced concrete face without waterstopped joints. Unrein-

forced upstream faces constructed of conventional concrete have been
used on quite a few low to moderate-height RCC dams. Examples of
this type of design include Middle Fork, Galesville, and Grindstone
Canyon. For these dams, the face thickness has averaged between 18
in (0.46 m) and 2 ft (0.6 m) with a partial bedding mix extending up to
8 ft (2.4 m) downstream.
Mix design and construction control for unreinforced concrete faces
are important in order to produce a high degree of impermeability.
The conventional concrete upstream-face mix should primarily be de-
signed to meet permeability and durability requirements. A high-
range water-reducing admixture has been used to reduce mixing wa-
ter requirements. This tends to minimize drying shrinkage of the face.
More importantly, a retarding admixture should be considered in or-
der to keep the facing concrete alive until the next lift is placed. In no
case should the top surface of the facing concrete be allowed to dry out
to produce a cold joint and a potential seepage path. Applying a bond-
ing mortar mix to any surface that has the potential to become a cold
joint is one solution.
There is some disagreement as to whether the RCC or conventional
concrete should be placed first. Those who believe the conventional
concrete should precede the RCC as shown in Fig. 3.12 feel this
method produces fewer voids at the interface of the two materials.
Others feel it is easier to construct the face by placing the RCC first.
They claim this method would have less potential negative effect on
the construction schedule and may not require a high-range water re-
ducing admixture (a superplasticizer) in the facing concrete. If the
RCC is placed first, it should be limited to a single lift and any large
aggregate rolling to the upstream form should be removed.
In either case, the material placed first must be alive while the
other is placed. Vibrators or other external methods of consolidating
the material must be used to minimize voids at the interface.
98 Chapter Three

Varies 20 in (0.5 m) min. to 35 in (0.9 m) max.

Facing panels
with imcwvious
(2.2 X 5.0m)-


(2 2m)

37ft (12.0)m ot lower elevations

Figure 3.12 Upstream face detail-Urugua-i Dam.

At Middle Fork and Grindstone Canyon dams, blockouts in the up-

stream wood forms produced notches in the facing concrete that acted
as crack inducers at lo- and 16-ft (3.7- and 4.9-m) horizontal spacing,
respectively. The unreinforced concrete face will crack at the location
of the weakened planes formed by the crack inducers. The cracks in
the notches can then be sealed prior to reservoir filling.
The design for Galesville Dam included no crack inducers in the
face concrete. A sprayed-on elastomeric membrane was placed on the
upstream face. The 40-mil-thick membrane, applied in two coats, did
seal the face but was unable to span several thermally induced cracks
that occurred. Other upstream facing methods. In China, two unique methods

have been used to increase the watertightness of the upstream face of an
RCC dam. For Kengkou Dam, a 2.4-in-thick (60-mm) asphalt-mortar
membrane was placed between the RCC surface and a 2.4-in-thick (60-
mm) precast concrete panel upstream form. At Longmentang Dam, an
expansive-cement-concrete mix was used for waterproofing. No trans-
verse joints were placed in either dam.
Design of RCC Gravity Dams 99

3.5.3 Lift thickness and bonding

successive lifts Lift thickness. The design lift thickness depends primarily on
the construction equipment available and the consistency of the RCC
mixture, The lift thickness is defined as the thickness of RCC that is
compacted at one time, although it may have been spread in a number
of layers and preconsolidated by the treads of the spreading dozer.
In determining a lift thickness, the goal is to provide a thickness
that can be compacted to the required density uniformly throughout
the lift with readily available equipment considering the consistency
of the RCC mixture.
The thickness of RCC lifts has ranged from 9 in (0.23 m), for
overtopping protection for the Brownwood Country Club Dam in
Texas, to 30 in (0.75 ml at Tamagawa Dam in Japan. The thin lift was
chosen for the Brownwood Dam rehabilitation project because only
1400 yd3 (1070 m3) of RCC was required and the designer figured this
thickness could be easily placed by a small contractor using available
spreading and compaction equipment. The thicker lifts at Tamagawa
are typical of the Japanese RCD method wherein a wetter consistency
mix is spread in three or more layers and then compacted. The 30-in
(0.75-m) lifts at Tamagawa were used for the upper portion of the dam
while 20-in (0.50-m) lifts were used for the lower portion adjacent to
the foundation rock.
The most typical RCC lift thickness to date has been 12 in (0.30 ml.
This includes nearly all completed lean RCC dams as well as the high-
paste RCC used for Upper Stillwater Dam. Elk Creek Dam was built
using a 2-ft (0.60-m) lift thickness with a layer of bedding mortar
spread over the entire surface of each lift.
Thicker lifts are desirable as there are fewer lift lines which form
potential seepage paths and planes of shear weakness. Caution
should be applied in selecting lifts greater than 12 in (0.30 ml, how-
ever, particularly for dams constructed of lean, drier-consistency
mixes. Thicker lifts put on in a single layer are more difficult to
spread and compact and take longer to place than thin lifts. The
construction complexity creates the potential for voids at the bot-
tom of the lift and reduced bonding. If thicker lifts are considered,
they should be well-investigated in a test section prior to the start
of construction.
Design details for many RCC dams are independent of the lift thick-
ness. However, if facing elements or precast concrete panels are used
to form the exterior faces or a stepped spillway is incorporated into the
design, the height of the elements, panels, or steps should be an even
multiple of the lift thickness, mainly to simplify construction.
In areas where handheld compactors or small rollers are necessary
100 Chapter Three

due to space limitations, the lift thickness should be reduced to ensure

adequate compaction.
RCC is generally placed in horizontal lifts that are sloped slightly to
allow for drainage of rainwater. Middle Fork Dam had a 2 percent
downward slope toward the unstream face. An upstream to down-
stream slope may work better in removing rainwater and keeping it
from collecting behind the critical upstream facing system. An up-
stream dip improves the shear friction factor between successive lifts,
however. Bonding successive lifts. Because RCC dams are constructed

in a series of compacted lifts, bonding of the successive lifts is impor-
tant both from a stability and performance standpoint. Poorly bonded
lifts have lower shear resistance due to low or no cohesion at the in-
terface, have less tensile resistance for seismic loading, and offer a
path for horizontal seepage. Therefore, the designer should consider
methods for improving or assuring adequate bond between successive
lifts where needed. Bonding methods fall into two basic categories: (1)
RCC to RCC bonding and (2) bonding RCC with bedding mixes. RCC to RCC bonding. Research on the factors affecting the

bond between successive lifts of RCC has been going on in the field
and in the laboratory since 1970. Cores extracted from RCC dams cor-
related with construction records have provided additional valuable
The basic factors affecting bond have been identified, but have not
been completely quantified. Whether there is some bond or no bond
between lifts is probably more important that the degree of bond or
shear resistance. The principal factors that affect bonding are

1. Condition of the lower RCC lift surface

2. Time delay between placement of RCC lifts
3. Consistency of the covering RCC
4. Compaction or consolidation of the covering RCC

The lower RCC surface must be kept continuously moist but with-
out ponding water to assure bond. Excessive surface moisture is det-
rimental to bond development, but drying of the surface may lead to
no bond.
All loose and dry material should be removed from the RCC prior to
placement of the next lift. Additional cleaning, such as by sand-
blasting or washing, produces little or no benefit. At Willow Creek
Dam, wheeled haul vehicles tracked mud and other debris onto the
dam and their turns disturbed the lift surfaces, which contributed to
Design of RCC Gravity Dams 101

the poor bonding and subsequent seepage at the lift lines. Conveyor
transportation of the RCC from the mixing plant to the dam has been
required on some projects to help eliminate some of these surface dis-
turbance problems.
If a compacted RCC lift is covered with the next lift before the lower
lift reaches initial set, satisfactory bond will usually develop. With in-
creased time delay, a cold joint begins to develop with a resulting loss
in bond strength. Many factors, such as RCC mixture proportions, am-
bient temperature, and surface moisture conditions, affect the time at
which the cold joint begins to develop. Set retarders, high-fly-ash-
content mixes, low temperatures and moist curing extend the life of
the lower RCC lift.
One method of expressing the point at which bond strength poten-
tial decreases takes into account two major factors in RCC strength
development, time and temperature. Referred to as joint maturity, it
has been applied to most lean RCC dam designs. Joint maturity is ex-
pressed as the product of the time of exposure and air temperature at
the surface of the lift, the units thereby being degree Fahrenheit-
hours. As an example, a delay of 8 h with an average temperature of
70F is 560 degree-hours. There is no direct metric conversion, except
at a fixed time or temperature.
There has been little agreement among designers on establishing
joint maturity values, which have ranged from 350 to 2000 degree-
hours. For a lean RCC mix and a temperature at the surface of about
70F (21C) a cold joint usually begins to develop at about 4 h and is
close to full development at 6 h. After 6 h, there is no further signifi-
cant decrease in sliding friction shear resistance until after 30 to 48 h
according to Schrader.
Therefore, if a well-bonded joint is desired, a low joint maturity
value should be specified. Similarly, if shear friction resistance is the
main concern, a higher value should be used. When the contractor ex-
ceeds the specified limiting joint maturity value, most specifications
require application of a bedding mix to assure adequate bond or shear
resistance between lifts.
The use of joint maturity values in project specifications encourages
the contractor to place successive lifts rapidly because the application
of beddng mixes is both costly and time consuming. While the bedding
mix is being applied, the RCC-placing equipment and crew usually
are not working, causing additional costs to the contractor.
To aid the contractor in planning construction, especially for small
projects, joint maturity may be expressed strictly in hours. Whatever
value or method is used for joint maturity, it should be based on field or
laboratory studies using the actual RCC mix. In the absence of such
studies, a conservative but not unreasonable value should be specified.
Bond between lifts improves as the volume of paste increases for the
102 Chapter Three

covering RCC as long as the lower compacted lift is still alive and
moist. This applies to all RCC mixes, Since the water content is basi-
cally constant for a certain mix design approach, bond improves with
mixes containing more cement and pozzolan.
Once the lower lift has hardenend, bond depends on the adhesion of
the paste of the covering mix into the pore structure of the lower lift.
In this case, high-paste RCC mixes, those in which there is a greater
volume of paste than voids in the fine aggregate, are required for ad-
equate bond. The joint properties of a high-paste mix can be defined in
terms of paste/mortar ratio. The research by Dunstan indicates that a
minimum paste/mortar ratio of 0.4 by volume for the RCC mix is
needed to fill voids and achieve a good bond between lifts for an expo-
sure time of one day or less.
Compaction of the covering RCC is a factor in achieving bond, at
least to to the extent that high density indicates a reduction of voids
at the lift interface. Poorly compacted or segregated RCC mixtures
produce greater voids at the lift line and, therefore, less potential area
for achieving bond. Bonding RCC with bedding mixes. Satisfactory bond at horizon-

tal lifts can be assured with the use of a bedding mix of either mortar
or concrete. A mortar bedding mix is designed to fill the surface voids
in both the compacted lift below and the covering layer above as well
as to glue the two RCC layers together. A mortar bedding mix over
the entire surface of each 2-ft-thick (0.6-m) lift is typical of the Japa-
nese RCD method and also was used for Elk Creek Dam.
Biaxial shear tests of cores extracted from the Elk Creek test sec-
tion showed that the mortar bedding significantly increased the aver-
age shear strength as compared to areas where no mortar was applied.
The test program also indicated higher shear strength when there was
no cleaning of the lift as compared to lifts that were washed prior to
application of the mortar. Wash water remaining on the lift surface
may have increased the water/cement ratio, thereby weakening the
bedding mortar mix.
The bedding mortar for Elk Creek contained 384 lb of cement, 149
lb of fly ash and 436 lb of water per cubic yard, plus sand, to produce
a slump of 8 to 9 in (230 to 300 mm). The metric mix would be 228 kg
of cement, 88 kg of fly ash, and 259 kg of water per cubic meter. The
mortar bedding mix was spread to a thickness of %I to % in (6 to 13
mm) and was estimated to cost $0.27/yd2 to place.
The mortar mix for Japanese RCD dams uses 470 kg of cement, 270
kg of water, and 1391 kg of sand per cubic meter. The U.S. customary
units are 792 lb of cement, 455 lb of water, and 2345 lb of sand per
cubic yard.
Concrete is used generally for partial bedding mixes that are a
Design of RCC Gravity Dams 103

downstream extension of the upstream concrete facing mix. The bed-

ding mix generally extends an additional 2 to 8 ft (0.6 to 2.4 ml down-
stream from the widest part of the upstream concrete face, is of the
same mixture, and is spread to a thickness of the MSA, usually 3! in
(19 mm). If correctly applied, concrete bedding mixes provide assur-
ance of positive bond and seepage control for the area over which they
are applied.

3.5.4 Galleries

Galleries have traditionally been incorporated in conventional con-

crete dams to provide space for: (1) collection of drainage, (2) drilling
both drain and grout holes into the foundation, (3) inspecting and
monitoring behavior of the dam, and (4) carrying out remedial work.
Galleries also have been used for more specialized functions such as:
(1) access and space for mechanical and electrical equipment for oper-
ation of spillway gates and outlet works, (2) access for electrical power
or control cables, and (3) access routes for visitors to a major dam.
The construction of galleries has consistently proven to be the most
inconvenient phase of the RCC dam construction process. Gallery con-
struction results in decreased RCC placement rates at the elevations
involved and labor-intensive work in forming the gallery. At Elk
Creek Dam, it was estimated that the gallery construction process in
the lower, confined areas of the dam reduced RCC placement rates by
50 percent.
The impediment to RCC placement and lower shear friction factors
in the gallery area area have led designers to question the need for
galleries in some cases. In all cases where galleries are included in the
design, engineers and contractors have strived to produce efficient
gallery construction methods. Some of the more innovative gallery
construction methods developed for RCC dams are described in Sets.
4.2.4 and 8.9.2.
Whether a gallery is required or not depends primarily on the de-
sign requirements for the dam. While most gallery functions can be
accommodated elsewhere, a gallery may be important in providing a
space to drill and reestablish drain holes into the foundation and to
receive and collect drainage from holes drilled from above.
The purpose of drain holes into the foundation is to reduce uplift
pressures, which in turn affect the shear resistance on any horizontal
plane due to a change in the normal stress. The need for a gallery is
thus greatly reduced if foundation drains are eliminated and the dam
is designed for full uplift pressure.
In Chap. 4, Schrader suggests that galleries not be considered for
dams less than 100 ft (30 m) high. Designing for full uplift can result
in greater mass, which may be needed anyway to properly construct
104 Chapter Three

an unformed downstream slope of 0.8 H : 1.0 V or flatter. Winchester

and Lower Chase Creek are examples of low dams constructed without
a gallery. The 103-ft-high (31-m) spillway section for Stacy Dam also
was designed with no gallery.
Galleries are recommended for higher dams by nearly all authori-
ties as internal dam and foundation drainage provide both economic
and stability benefits. The inspection and monitoring performance
function of a gallery is also more critical in higher dams.
The location of a gallery is usually a trade-off between functional
requirements and construction ease. Where possible, the gallery
should be at a single level and as low in the dam as practical. The gal-
lery may be terminated where structural requirements no longer re-
quire uplift pressure reduction such as close to the abutments. This
approach was used in the design for Monksville Dam.
Two equipment widths, or about 16 ft (5 ml, from the upstream face
of the dam to the gallery wall is considered the minimum. The gallery
should be as small as possible, with the height being an even multiple
of and in line with RCC lifts but not more than 8 ft (2.4 m). Three feet
tl 1111 is considered the minimum width, with 6 ft (2 m) being more
common. Structural considerations will determine if and how much
reinforcing is needed in the RCC above the gallery. A gutter is gener-
ally provided along the upstream gallery face to contain seepage flow
and to support flow-measuring weirs.

3.6 River Diversion and RCC Cofferdams

3.6.1 River diversion

The cost of river diversion can be an important benefit of RCC dams
when compared to alternative dam types, Diversion of the river past
an RCC dam while under construction is usually through conduits in
or under the dam, or possibly through a tunnel in one of the abut-
ments. The conduit should be located so as to minimize any negative
effect on the rapid construction of the RCC, such as in a cut in the
foundation rock. Consideration should also be given to incorporating
the diversion conduit into the dam as an outlet conduit. For diversion
of very large rivers, see Sec. 3.6.2.
The design of diversion capacity depends on the frequency of the
flood to be passed. Generally, the design diversion flood is five times
the construction period, noted in years. Thus, if an RCC dam can be
constructed in one year rather than two years for an alternative dam
type, the diversion capacity can be reduced from a once-in-lo-year
flood to a once-in-s-year flood.
Further savings can accrue if the initial construction can be
Design of RCC Gravity Dams 105

planned to coincide with the low-stream-flow season. In this case the

diversion capacity can be considerably less than a 5- to lo-year-
frequency flood. This concept was recognized by the designers of
Galesville Dam, where the diversion conduit capacity was reduced
from 6000 ft3/s (170 m3/s) to 1200 ft3/s (34 m3/s). The diversion was
through a lo-ft-diameter (3-m) conduit encased in conventional con-
crete located in a trench cut into foundation rock at the dams left
Diking the river channel to divert the flow to the conduit is ade-
quate in most situations. If the cofferdam is overtopped by a greater-
than-anticipated flood, damage can be minimal because of the erosion
resistance of RCC. Both Craigbourne Dam and Bucca Weir in Austra-
lia were overtopped during construction with no significant damage to
the recently placed RCC.
At Craigbourne Dam, the once-in-3-year event overtopped both the
cofferdam and the RCC dam, which was 13-ft (4 m) high at the time.
Although the top layer was only 48 h old when the flood occurred,
the 6-in (150-mm) overtopping only plucked out fines from the RCC
Bucca Weir was overtopped on two occasions by flows up to 3 ft (0.9
m). The formwork for the downstream concrete facing contained the
RCC edge, but the resumption of placement was delayed because of
sediment deposited on the slightly sloping top layer. A full cleanup
followed by a layer of bedding concrete was necessary before the next
RCC lift could be placed.

3.6.2 RCC cofferdams

For very large rivers, it may be necessary to divert the river to one
side of the original channel behind a concrete wall or possibly an is-
land or outcropping in the river. Upstream and downstream coffer-
dams then are built to allow construction of a portion of the dam, in-
cluding outlet conduits. In order to complete the remainder of the
dam, the river is diverted through the outlets and dam construction
continues on the other side. This river diversion scheme is not limited
to concrete dams, but does provide a logical opportunity for the use of
RCC for the cofferdams and also the diversion wall.
The use of RCC for cofferdams and mass diversion walls offers the
advantage of rapid construction and erosion resistance. The erosion
resistance is useful in withstanding wave action and allowing safe
overtopping of the cofferdam.
The first RCC cofferdam was used for the rehabilitation of Tarbela
Dam. A composite earthfill-and-RCC cofferdam was constructed down-
stream of the auxiliary spillway plunge pool to allow dewatering of
106 Chapter Three

the construction site. The RCC half of the cofferdam was on the water
side and was subjected to high-velocity flow and wave action from dis-
charges from one of the outlet tunnels. The earthfill on the spillway
plunge pool side was used as an economical forming method for the
vertical face of the RCC section. After the lining of the plunge pool
was completed, the left side of the cofferdam was breached and the
auxiliary spillway was operated. The earthfill portion of the cofferdam
was completely washed away during the first season of spillway oper-
ation but the remaining RCC portion has remained in place even after
being subjected to spillway flows up to 400,000 ft3/s (11,300 m3/s).
RCC has also been used for both the upstream and downstream cof-
ferdams for the conventional-concrete Yantan Dam as well as for ei-
ther an upstream cofferdam or a diversion wall for three other dams in
China (see Fig. 11.2).
At Yantan, a dumped earth-and-rock dike was first constructed in
the Honshui River. Then the RCC cofferdams were quickly placed be-
hind the embankment, allowing the main dam to be constructed dur-
ing the rivers low-flow season. The 172-ft-high (52.5-m) upstream and
132-ft-high (40.2-m) downstream cofferdams for Yantan, requiring a
total of 366,000 yd3 (275,300 m3), are moderately large RCC dams in

3.7 Appurtenant Structures

The appurtenant structures for a dam include the spillway, intake
tower, outlet conduit, associated gate chambers and stilling basins,
and any other waterways through or below the dam. Their basic hy-
draulic design and capacity are the same as for a conventionally
placed concrete gravity dam. Cost savings associated with rapid unin-
terrupted horizontal placement of RCC dams has led to location and
configuration of appurtenant structures to minimize interference with
the construction method.

3.7.1 Spillways
Spillways for any dam must be designed to pass the design flood and
all lesser-capacity flows safely and economically. With ever-in-
creasing hydraulic requirements, the cost of spillways has become a
major economic factor in dam design and construction. Concrete dams
in general and RCC dams in particular can incorporate spillways into
the structure at little additional cost when compared to embankment-
type dams, which usually require separate spillways. A concrete dam,
therefore, shows a greater cost advantage when compared with an em-
bankment dam as the required spillway capacity increases.
Since concrete dams have a high degree of erosion resistance and
structural stability, there is little chance of failure when the dam is
Design of RCC Gravity Dams 107

overtopped. The inherent safety of a concrete dam when overtopped

has been recognized by the Japanese, who require the spillway capac-
ity for an embankment dam be at least 20 percent greater than for a
concrete dam at the same site. Similarly, designers of RCC dams in
the United States have sized their primary spillway for an event less
than the PMF (probable maximum flood) and relied upon overtopping
the entire dam for the infrequent greater-volume flow. At Middle Fork
Dam, the service spillway, which was combined with the outlet con-
duit, was sized to accommodate the once-in-500-year event. The dam
itself would serve as the auxiliary spillway, which would be
overtopped during a major flood.
Because of their ability to be safely overtopped, concrete dams are
often favored in areas where historical stream-flow records are inad-
equate or where a special condition makes determining spillway size
risky. A concrete dam was thus favored in Chile, where the designers
could not predict with any degree of confidence the flood flow resulting
from a potential volcanic eruption upstream and subsequent rapid
snow melt.

3.7.1 .l Spillway configuration. Any type of spillway configuration

that can be functionally justified for either a conventional concrete
gravity dam or an embankment dam can be used for an RCC dam.
Economics and the RCC construction method have, however, favored
long uncontrolled spillway crests incorporated in the dam and aligned
with the streambed. The long spillway notch reduces the head re-
quired to pass the design flow and with it the height of the RCC sec-
tion on either side of the spillway. The nonoverflow sections on either
side of the spillway at the crest are more expensive to construct be-
cause of restrictions in construction space and access. Their height
should, therefore, be minimized.
Reducing the flood pool height by using a long spillway crest rather
than a separate, narrow, side-channel spillway can reduce the re-
quired dam height markedly. A reduced flow depth also requires
shorter side-training walls to contain the flow. A material-efficient de-
tail for downstream training walls for the unsurfaced RCC spillway at
Galesville Dam is shown in Fig. 3.13. Site conditions may require the
downstream portion of the spillway to be tapered to better direct flows
into the river channel, as shown in Fig. 3.14 for Pamo Dam.
High-head gated spillways require support piers and downstream
spillway chutes constructed of conventional concrete. This use of
formed reinforced concrete sections together with higher adjacent dam
sections makes the gated spillway more costly than long uncontrolled
spillway crest designs.
The spillway crest is designed to provide efficient discharge of the
flow without creating excessive negative pressures. An ogee crest can
Flgure 3.13 Galcsv~llc Dam (downstream face and spillway)

Figure 3 . 1 4 Artists conception of Pamo Dam (downstream face and spillway).

Design of RCC Gravity Dams 109

be located at the upstream face for a dam that has a relatively narrow
crest width, such as at Galesville Dam (see Fig. 3.7b), or downstream of
an entrance apron, as at Upper Stillwater Dam. The ogee can be con-
structed of conventional concrete or shotcrete following completion of
the RCC dam construction.
Three basic designs for the downstream spillway surface have been
used for RCC dams. They are: (1) the traditional smooth, conventional
concrete surface, (2) a stepped spillway of conventional concrete, and
(3) an unformed exposed RCC surface. Copperfield Dam (Fig. 3.151,
Upper Stillwater Dam (Fig. 3.161, and Galesville Dam (Fig, 3.13) are
good examples of the three types of surfaces.
With the traditional concrete spillway, the objective is to provide a
smooth flow surface for the prevention or minimization of cavitational
damage. The stepped spillway design is more widely used in RCC
dams. The rough stepped surface produces a highly turbulent, well-
aerated boundary layer that eliminates negative pressures and pre-
vents cavitational damage.
If an exposed RCC face for the spillway is to be used, the steepness
of the downstream slope is usually limited to the angle of repose of the
RCC material because of construction considerations. This limiting
slope is generally 0.8 H : 1 V, but may have to be further flattened if
rounded river gravel is used as the RCC aggregate. An unformed RCC
slope was constructed at 0.75 H : 1 V at Les Olivettes Dam with the

Figure 3.15 Copperfield Dam (spillway).

110 Chapter Three

Figure 3.16 cpp~~ Stlli\\ater km ispIllway facci.

aid of a specially designed vibratory compactor at the end of the dipper

stick of a backhoe. The contractor that developed the equipment be-
lieves it can be used to construct steeper slopes, up to 0.65 H : 1 V. Characteristics of stepped spillways. In the early 1900s rubble

masonry dams were usually constructed with stepped spillway chutes
in order to simplify construction and to dissipate energy. Not much
hydraulic research was accomplished on this type of spillway, and as
conventional mass concrete replaced masonry for the construction of
gravity dams, smooth chutes replaced stepped spillway chutes.
Now, with the advent of the RCC construction method and the rel-
atively easy incorporation of conventional concrete steps concurrent
with horizontal RCC placement, a renewed interest in stepped spill-
ways has developed. The steps improve hydraulic behavior of the flow
and reduce the velocity of the water, leading to less potential for cav-
itation and less-expensive stilling basins when compared to smooth
spillway chutes.
The steps act as roughness elements to minimize flow acceleration
and terminal velocity. Turbulence induced by the steps helps speed
the development of a boundary layer and induces entrained air to bulk
the flow. Cavitation potential is thus reduced by both the reduced ve-
locity and the cushioning effect of the entrained air.
Laboratory hydraulic model studies were completed prior to the con-
Design of RCC Gravity Dams 111

.; 60 -

? 50-
; 40- Downstream
Dam Slope (I-W

3. _ 0 Upper Stillwater 0.60 1.0

o Monksville 0.70: 1.0
20~ D Stagecoach 0.80: 1.0

10 I I I I I I I I I
0 15 30 45 60 75 90 105 120 135 150
Unit discharge,ft3/s/ft
Figure 3.17 Stepped spillway energy dissipation (2-ft-deep steps). (From Houston,

struction of Upper Stillwater, Monksville, and Stagecoach dams. Fig-

ure 3.17 shows the percentage of energy dissipation for a stepped spill-
way plotted versus unit discharge for 2-ft-high (0.6-m) steps. The
lower the depth of flow (lower unit discharge), the greater the energy
dissipation because the steps have the greatest effect on the flow clos-
est to the downstream face.
Physical models are needed primarily to determine crest shape,
evaluate step geometry near the spillway crest, and aid with stilling
basin design. The crest should be designed to match the theoretical
nappe shape for the operating head. The crest shape together with the
initial step configuration should develop early tumbling action of the
flow adjacent to the spillway surface rather than deflecting the flow
outward away from the steps. Model studies have shown that using a
proper crest shape and smaller steps near the spillway crest are essen-
tial in producing the appropriate hydraulic characteristics of the un-
controlled, stepped spillway.
With reduced velocity at the toe of the spillway, the need for a ter-
minal energy dissipator is either reduced or eliminated, depending on
the erosion potential of the downstream surface rock. For Upper
Stillwater Dam, laboratory hydraulic studies indicated that the steps
produced a 70 percent energy reduction over a conventional smooth
spillway for the design discharge. This led to an 85 percent reduction
112 Chapter Three

in the length of the stilling basin. It was thus designed to be only 30 ft

(10 m) long. This length quiets the flow beyond the end sill for low to
medium discharges. Some waves are produced at maximum discharge.
At Monksville, no terminal energy dissipator structure was required
because of the dense foundation rock at the end of the spillway.
A 2-ft-high (0.61-m) step appears to produce a good combination of
hydraulic efficiency and easy constructibility. Training walls are
needed to contain the flow. For Monksville Dam, the designers deter-
mined that the training wall height should be twice that required by
model similitude at maximum discharge to account for freeboard, ac-
commodate bulking of the flow, and reduce spray.
While there have been no hydraulic studies conducted on steep,
unformed, exposed RCC spillways, it appears that they have sufficient
roughness to reduce cavitational damage and energy. Exposed RCC in
a spillway should be designed for a high degree of durability and ero-
sion resistance, At Willow Creek and Galesville dams, a higher
cementitious content than the interior RCC was used for the exposed
downstream zone.

3.7.2 Intake tower and outlet conduit

With the RCC method of construction, the intake tower and outlet
conduit have invariably been located so as not to interfere with effi-
cient RCC placement. A popular way of accomplishing this goal is to
locate the intake structure upstream from the dam, the outlet conduit
on or in the dams foundation, and the control gates and energy
dissipator at the downstream toe.
In many respects, the above arrangement has more similarities to
an embankment dam than a conventional concrete gravity dam in
that these features are located outside of rather than within the dam
cross section. Locating any of these items within the dam complicates
RCC placement and increases the overall cost.
Multilevel intake towers have been attached to and thereby sup-
ported by the upstream face in a number of RCC dams, such as Elk
Creek. This location does not interfere with RCC placement and per-
mits construction of the reinforced concrete tower separate from the
RCC dam. It also is simpler to design than a freestanding tower, es-
pecially in seismically active areas.
The outlet conduit starting from the intake structure is preferably
placed in a trench cut into the foundation and encased in conventional
concrete prior to the start of RCC placement. Locating the gate cham-
ber at the downstream toe requires a pressure conduit to the point of
hydraulic control, but this location ensures sufficient space and easy
access for constructing the required mechanical and structural func-
Design of RCC Gravity Dams 113

Site-specific conditions may dictate that power-plant penstocks and

large-capacity flood outlets cannot easily be routed under the dam. In
this case, consideration should be given to locating all the conduits
through the dam in a single conventional concrete block near one of
the abutments. Constructing this block prior to RCC placement will
help minimize any potential problems resulting from dissimilar ther-
mal properties between the RCC and the conventional concrete.

3.8 Monitoring Performance of Completed

Any dam is subjected to external loads that cause deformations as
well as seepage through the structure and its foundation. Loads and
the dams response to them should be monitored for any sign of abnor-
mal behavior, and corrective action should be taken before the prob-
lem becomes a threat to the dams safety.
In developing a program to monitor the performance of an RCC
dam, primary consideration should be given to obtaining information
with which the dams safety can be evaluated. Of secondary impor-
tance is obtaining information to check initial design assumptions in
order to provide improved design criteria or construction methods for
future RCC dams.
The monitoring of completed RCC dams should consist of measure-
ments obtained from instruments and visual inspections. Neither in-
strumentation nor visual inspection is sufficient on its own.

3.8.1 Instrumentation

The amount of instrumentation provided depends on site and opera-

tional conditions as well as the owners needs. Instrumentation takes
on greater significance with higher dams, poorer or more complex
foundation conditions, and locations where a dam failure would have
significant consequences. A private owner of a single low-budget dam
may not be as interested in the secondary function of instrumentation
as a large governmental agency that constructs many dams at differ-
ing site conditions over a long period of time.
The instrumentation program should be determined by the dams
designer, who best understands the dams purpose, design assump-
tions, and potential problems. The designer should determine which
items should be monitored at all times to help evaluate the safety of
the dam and what information is needed over the long term to check
design assumptions.
Once the desired primary and secondary functions are determined,
the type, number, and location of the instruments can be selected
based on reliability and cost. The location of instruments in an RCC
114 Chapter Three

Item measured Instrument Location Comments

Primary functions (to monitor safety)
1. Seepage or leakage a. Weir a. Drainage tunnel Volume of flow and
outlet, abutment change with time
grains and gallery determined
b. Flume b. Downstream of
c. Calibrated con- dam
tainer and stop- c. At drain holes
2. Uplift pressure Piezometer At dam/foundation
contact and within
3. Crack or joint a. Extensometer or a. Across crack or Install after forma-
width micrometer joint in dam tion of crack or joint
b. Offset points 6. On either side of
4. Water level a,b. Staff gauge, *a. Upstream face of Determines hydraulic
float, pressure dam load on dam (a) up-
balance b. Stilling basin stream and (b)
5. Structural defor- a. Surveying instru- a. Target or brass cap
mation and dis- ments monument or dam
placement crest and abut-
ments downstream
of dams
b. Plumb line or in- b. In hole in dam
Secondary functions (to determine validity of design assumptions1
1. Concrete tempera- Electric thermometer Embedded in dam at Time history of tem-
ture or thermocouple predetermined grid perature desired
2. Foundation distor- Multiple position In foundation at
tion and displace- borehole extenso- points of potential
ment meter or plumb line movement
3. Dam stress and Stress or strain At computed maxi-
strain meter mum stress point3

4. Vibration Seismograph In enclosureon abut- Time history of seis-

ment or top of dam mic event deter-

Figure 3.18 Potential items for instrumentation.

dam can be quite important, because numerous or poorly located in-

struments can slow the placement of the material. Potential instru-
mentation items, types, and locations are listed in Fig. 3.18.

3.8.2 Visual inspection and simple

Visual inspection and simple monitoring can take many forms and
should be performed regularly. Seepage is the single-most-important
Design of RCC Gravity Dams 115

item that can be evaluated in this manner. Some questions that

should be answered in inspecting for seepage are
1. Where is the seepage occurring: through the entire dam or at se-
lected places, such as cracks or horizontal lift lines within the
dams structure, or through the dams foundation?
2. What is the magnitude of the seepage? Is it reducing over time with
a constant reservoir level?
3. What is the color, chemistry, or pH of the seepage water? If the seep-
age water has a high alkalinity, has it produced water quality or ad-
verse visual problems downstream? Does it contain any sediments?
4. Have any calcium carbonate deposits been noticed that clog drain
holes or appear on the downstream face, in the gallery, or in wa-
terways downstream of the dam?

Of special importance is an abrupt increase in seepage. Based on

performance of existing RCC dams, it is assumed that seepage will re-
duce with time and that wet spots will eventually dry up. Therefore, if
a significant increase in seepage is noted without a rise in reservoir
level or if new cracks are noticed that pass some water, there is reason
for concern. The situation should be analyzed to determine the cause
and if any remedial action is required. It should be remembered that
during cold weather, seepage may increase because of thermal con-
traction that opens up existing cracks or creates new ones. Drains
should be reestablished if they become clogged.
Other items that can be visually observed include movements of the
dam, its appurtenances, and abutments; freeze-thaw or other deterio-
ration; erosion in the structure or on downstream surfaces following a
flood; or cracking or other damage from an earthquake.
Cores extracted from the dam can be an excellent tool for post-
construction analysis. In addition to being tested to determine the
properties of the RCC, including bonding between lifts, the cores can
be visually inspected. Inspection of cores for voids can provide an in-
dication of the degree of compaction, segregation, and permeability,
usually at the bottom of the lift.

3.9 Dams Curved in Plan, Including Arch Dams

A properly designed RCC dam that is curved upstream places the en-
tire upstream face in compression under the usual loading condition,
whereas a straight gravity dam can have tensile stresses near the
abutments because of the fixed-beam action of the structure across the
valley. Curving the structure provides additional resistance to over-
turning, a potential for a reduction in volume, and the compression of
the upstream face can help produce a more impermeable structure.
116 Chapter Three

The use of RCC for a dam curved in plan started with Saco de Nova
Olinda Dam, completed in 1986 in Brazil. The dam was curved near
the right abutment strictly to accommodate local site conditions. No
structural credit was taken for the curvature as the 0.8 H : 1 V down-
stream slope was maintained throughout the gravity structure.
Several curved upstream cofferdams have been constructed using
RCC in China since early 1988 to provide greater stability during
overtopping, as noted in Sec. 3.6.2.
The first RCC gravity arch dam was the 164-ft-high (50-m)
Knellpoort Dam completed in South Africa in 1988. (Fig. 3.9) It was
followed by the 230-ft-high (70-m) Wolwedans Dam completed 1 year
later, Both gravity arch dams were designed by the Republic of South
Africas Department of Water Affairs primarily as a means of reduc-
ing the cost of the structures through reduced volume.
Both gravity arch dams have a vertical upstream face and a 0.5 H :
1.0 V downstream slope. Knellpoort has a crest length of 656 ft (200
m) while Wolwedans is 879 (268 m) long at its crest, thus producing a
crest length/height ratio of 4.0 for Knellpoort and 3.8 for Wolwedans.
While it may be possible to construct an RCC arch dam that will not
crack in the moderate climate of South Africa, provision was made in
the design to control cracks should they occur. To maintain arch ac-
tion, provision for possible cracking was made as follows: Crack induc-
ers and waterstops were placed in both conventional concrete faces.

Figure 3.19 Knellpoort Uam (nearing completion of gravity arch).

Design of RCC Gravity Dams 117

The inducers were installed at about 33-ft (10-m) spacing at the

curved upstream face and radially opposite them in the downstream
face. At Knellpoort, the crack inducers were formed voids, while sheet
bond breakers were used for Wolwedans.
Crack directors in the form of metal cavity forms were placed at 3-ft
(l-m) height intervals between the crack inducers and waterstops to
help ensure that any cracking occurred radially. If wide cracks occur,
grout will be injected into the voids formed by the metal cavities and
from there into the cracks. The dams were divided into 16-ft (5-m)
grout lifts. In the year since Knellpoort was completed no grouting of
cracks has been required.
The design of an arch dam using RCC is similar to the design of a
dam constructed of conventional concrete. As RCC is used in thinner
and higher arch dams, there is greater need for higher strength and
more uniform concrete to withstand the higher stresses imposed upon
the structure. Shear resistance, especially cohesion between succes-
sive lifts, takes on added significance in order to maintain proper can-
tilever action.
It is anticipated that gravity arch or arch dams using RCC will be
designed with no full section vertical joints, because joints would have
to be grouted to endure proper arch action. Cracking must be mini-
mized and preferably eliminated altogether. Special attention must
thus be given to thermal aspects of mix design and construction. It is
very likely that construction will have to be limited to the cooler
months of the year and that the materials will have to be cooled to
produce the low RCC placing temperatures needed.
The RCC construction method requires sufficient working room to
economically place and compact the material. Therefore, RCC should
be considered only for dams with true arch action above a certain
height and with a crest thickness no less than 10 ft (3.0 ml and pref-
erably at least 16 ft (5.0 m). A minimum height for RCC to be consid-
ered for a single-curvature arch dam might be 100 ft (30 m), while 200
ft (60 m) might be the lower limit for a doubly curved arch dam.

AC1 Committee 224, Control of Cracking in Concrete Structures (AC1 224R-80) Amer-
ican Concrete Institute, Detroit, 1980, 42 pp.
Benson, S. A., Verigin, W. M., and Carney, M. J., Cedar Falls Roller Compacted Dam,
in Roller-Compacted Concrete II, ASCE, New York, February 1988, pp. 39-50.
Boggs, H. L., and Richardson, A. T., USBR Design Consideration for Roller Com-
pacted Concrete in Roller Compacted Concrete, ASCE, New York, May 1985, pp.
Bouyge, B., Langois, A. P., and Martin, J. P., Quality of Works in RCC in France: A
Contractors Solution in Roller Compacted Concrete II, ASCE, New York, February
1988, pp. 309-322.
118 Chapter Three

Bureau of Reclamation, Design Criteria for Concrete Arch and Gravity Dams, Mono-
graph No. 19, September 1974.
Bureau of Reclamation, Design of Gravity Dams, Denver, Colorado, 1976.
Campbell, D. B., and Johnson, P. C., RCC Dam Incorporates Innovative Hydraulic Fea-
tures, Proceedings of the Conference Water for Resource Development, Hydraulics Di-
vision, AXE, Coeur dAlene, Idaho, 1984, pp. 138-142.
Cannon, R. W., Design Considerations for Roller Compacted Concrete and Rollcrete
Dams, Concrete International, American Concrete Institute, Detroit, December
Chonggang, S., RCC Dams in China, ICOLD Sixteenth Congress, San Francisco, June
1988, vol. V, pp. 387-392.
Chugoku Regional Construction Bureau, Construction of Shimajigawa Dam with
Roller Compacted Dam Concrete, Ministry of Construction, Japan, 1981.
Copen, M. D., Lindholm, E. A., and Tarbox, G. S. Design of Concrete Dams, Chap. 8 in
Handbook of Dam Engineering, Alfred R. Golze (ed.), Van Nostrand Reinhold, New
York, 1977, pp. 385-488.
Ditchey, E. J., and Schrader, E. K., Monksville Dam Temperature Studies ICOLD
Sixteenth Congress, San Francisco, June 1988, vol. III, 662, pp. 379-396.
Dunstan, M. R. H., Rolled Concrete for Dams-Construction Trials Using High Fly
Ash Content Concrete, CIRIA Technical Note 106, Construction Industry Research
and Information Information Association (CIRIA), London, May 1981.
Forbes, B. A,, RCC in Dams in Australia, in Roller-Compacted Concrete II, AXE,
New York, February 1988, pp. 323-339.
Fujisawa, T., and Nagayama, J., Cause and Control of Cracks by Thermal Stress in
Concrete Dams, ICOLD Fifteenth Congress, vol. II, Q57, Lausanne, 1985.
Gilbrough, N., Roller Compacted Concrete and Tunnel Boring Machines-Two
Tools for Designers and Planners, Water Power 1985, ASCE, New York, Septem-
ber 1985.
Hirose, T., Nagayama, I., Takemura, K., and Sato, H., A Study of Control of Temper-
ature Cracks in Large Roller Compacted Dams, ICOLD Sixteenth Congress, San
Francisco, June 1988, vol. III, Q62, pp. 119-135.
Hollingworth, F., Druyts, F. H. W. H., and Maartens, W. W., Some South African Ex-
periences in the Design and Construction of Rollcrete Dams, ICOLD Sixteenth Con-
gress, San Francisco, June 1988, vol. III, Q62, pp. 33-51.
Hollingworth, F., Hooper, D. J., and Geringer, J. J., Roller Compacted Concrete Arched
Dams, Water Power & Dam Construction, November 1989.
Hopman, D. R., and Chambers, D. R., Construction of Elk Creek Dam, Roller-
Compacted Concrete II, ASCE, New York, February 1988, pp. 251-266.
Houston, K. K., Hydraulic Model Studies of Upper Stillwater Dam Stepped Spillway
and Outlet Works, Bureau of Reclamation, REC-ERC-87-6, October, 1987.
Jackson, H., The Construction of the Middle Fork and Galesville RCC Dams, Water
Power & Dam Construction, January 1986.
Leliavsky, Serge, Uplift in Gravity Dams-Part Four, Water Power, January 1960,
pp. 244-332.
Lemons, R. M., A Combined RCC and Reinforced Concrete Spillway, Roller-
Compacted Concrete II, ASCE, New York, February 1988, pp. 51-60.
Levy, Maurice, Communication to the French Academy of Sciences, August 1895.
Logie, C. V., Economic Considerations in Selection of a Roller Compacted Concrete
Dam, ASCE, New York, May 1985, pp. 111-122.
Lowe, John III, Roller-Compacted Concrete Dams-An Overview, Roller-Compacted
Concrete II, ASCE, New York, February 1988, pp. l-20.
Membrane-Lined Panels Face New RCC Dam, Highway & Heavy Construction, Feb-
ruary 1985, pp. 64-65.
Molcr, W. A., and Moore, J. F., Design of Seepage Control Systems for RCC Dams,
Roller-Compacted Concrete II, ASCE, New York, February 1988.
Oberholtzer, G. L., Lorenzo, A., and Schrader, E. K., Roller-Compacted Concrete De-
sign for Urugua-i Dam, Roller-Compacted Concrete II, ASCE, New York, February
Design of RCC Gravity Dams 119

Portland Cement Association, Bonding Roller Compacted Concrete Layers IS 231W,

Skokie, Illinois, 1987, 8 pp.
Portland Cement Association, Small Concrete Dams, EBOOPW, Skokie, Illinois, 1980.
Quin, J. T., Rezende, S., and Schrader, E. K., Sac0 Dam-South Americas First
RCC Dam, Roller-Compacted Concrete II, ASCE, New York, February 1988, pp.
Richardson, A. T., Upper Stillwater Dam Roller-Compacted Concrete Design and Con-
struction Concepts, ZCOLD Fifteenth Congress, Lausanne 1985, vol. II, Q57, pp. 143-
Saucier, K. L., No-Slump Roller Compacted Concrete (RCC) for Use in Mass Concrete
Construction, Technical Report SL-87-17, US. Army Corps of Engineers, Water-
ways Experiment Station, Vicksburg, Mississippi, October 1984.
Schrader, E. K., The First Concrete Gravity Dam Designed and Built for Roller-
Compacted Construction Methods, Concrete International, American Concrete Insti-
tute, Detroit, October 1962, pp. 16-24.
Simmonds, A. W., Lippold, F. H., and Keim, R. E., Treatment of Foundations for Large
Dams by Grouting, Transactions ASCE, vol. 116, 1951, pp. 548-572.
Tarbox, G. S., and Hansen, K. D., Planning, Design, and Cost Estimates for RCC
Dams, Roller-Compacted Concrete II, ASCE, New York, February 1988, pp. 21-
Tayaabji, S. D., and Okamoto, P. A., Bonding of Successive Layers of Roller Compacted
Concrete, Report to U.S. Bureau of Reclamation, Denver, March 1987.
Thomas, H. H., The Engineering ofLarge Dams, John Wiley & Sons, London, 1976.
Urlich, D. M., Deatherage, J. D., Rahe, J. H., and Dunne, R. E., Design and Construc-
tion of Lower Chase Creek Dam, Roller-Compacted Concrete II, ASCE, New York,
February 1988, pp. 236250.
Von Thun, J. L., and Tarbox, G. S., Deformation Moduli Determined by Joint-Shear
Index and Shear Catalog, Symposium Societie International Mechnique des Roches,
Nancy, France 1971.
Wang S., Kengkou Roller Compacted Concrete Dam, ICOLD Sixteenth Congress, San
Francisco, June 1988, vol. III, Q62, pp. 689-696.
Woodward Clyde Consultants, Stagecoach Dam Spillway Modeling Study, June 1985.

Design and
Construction of
Lean KC Dams

4.1 Background
Many of the RCC dams completed in the world to date have used low-
cementitious-content mixes. These lean mixes have tended to be dry,
but wetter mixes have also been used. By minimizing cement and/or
pozzolan contents, a direct and considerable material cost savings is
achieved. This is complemented by indirect savings that can be sub-
stantial. If no pozzolan or admixtures are used, storage, proportioning,
mixing, material delivery, and coordination are simplified. Lean
mixes also produce concretes with low internal temperatures from
hydration, low elastic moduli, and high creep rates. The combination
of these characteristics results in few or no vertical joints being re-
quired. This is a major indirect benefit of lean RCC because of the
avoided cost and construction complexity of installing contraction
The trade-off for lean mixes as compared to high-cementitious-
content RCC mixes is a reduction in the quality of the bond along the
lift-to-lift interface of RCC surfaces. Normally, there is adequate mass
and strength for sliding stability and compressive loads in lean RCC
dams. But seepage should be expected unless special measures are
taken to control it.

This chapter was contributed by Ernest K. Schrader.

122 Chapter Four

Wdiow Creek Dcm, lean RCC dam.

4.2 Design Concept and Section

Typically, RCC dams have been most economically, rapidly, and eas-
ily built by using the dry, lean approach and by keeping construction
simple. Each dam should be evaluated to determine what approach is
best for its location. Early experience has shown that a single RCC
mix for the entire project (or at least for all concrete between two se-
lected elevations) is the most practical approach.
Dams built with leaner mixes require more mass for stability. They
also realize a savings, however, because no downstream forming is
necessary for the typically flat (0.8 H : 1 V) downstream slopes. With
crushed aggregate and careful placing, a slightly steeper slope is pos-
sible without using forms. Use of rounded aggregate and a less dili-
gent placing method requires a cross section with a slightly flatter
downstream slope.
The quality of the layer-to-layer bond at the interface of RCC lifts in-
fluences the design cross section. If no special controls are employed to
minimize seepage and improve shear resistance, a thicker dam is usually
necessary to provide sliding stability and to offset uplift pressures. A
common condition with lean RCC dams is that they have extra mass,
producing thick sections. Thus, seepage is less of a stability factor than it
is in higher-strength RCC dams with narrower cross sections.
Design and Construction of Lean RCC Dams 123

Typically, cohesion and friction factors for lift joints in lean RCC
dams are influenced mostly by the character of the aggregate and the
degree of care used to reduce segregation during construction. Increas-
ing cement contents in lean mixes usually has little effect on the over-
all sliding stability of the completed dam.

4.2.1 Facings and seepage control

Properly designed lean RCC mixes with adequate nonplastic fines will
result in low permeability (Fig. 4.1) within the compacted mass. Seep-
age will occur along the interfaces of each lift of compacted material
unless measures are taken to control it. Even if lift lines are not wa-
tertight, adequate shear strengths and sliding resistance can usually
be achieved. This is especially true if uplift pressures are reduced
through internal drainage.
A variety of techniques can be used to minimize or eliminate seep-
age. These range from relatively expensive upstream membrane lin-
ers which provide total watertightness to less costly methods which
provide good but not absolute watertightness. Permeability and seep-
age control methods have been discussed in detail by the author (see
Bibliography) and in Sec. 3.5.2.
A common method of improving watertightness involves placing a
highly sanded, high-slump, retarded conventional concrete bedding
between layers of RCC. The mix is spread about 1 in (25 mm) thick on
a clean lift surface. The next layer of RCC is then spread over the bed-
ding and compacted into it. The bedding zone only needs to extend
downstream from the upstream face for a distance equal to about 8
percent of the hydraulic head at any elevation.
Commercially available elastic strips that are impregnated with
chemical grout and expand when they come into contact with water

Project Cement, Ib/yd3 Fly ash, lb/yd3 ftis, E - 12

Willow Creek 80 52 61
Willow Creek 175 00 312
Willow Creek 175 80 66
Willow Creek 315 135 220
Lost Creek Test Section 94 75 370
Lost Creek Test Section 120 140 42
Lost Creek Test Section 235 00 128
Waterways Experiment 517 00 10
Station Test
NOTE : 1 lb/y@ = 0.59 kg/m3; 1 ft/s = 0.3 m/s.

Figure 4.1 RCC permeability (cores).

124 Chapter Four

can also be placed between RCC layers near the upstream face. They
act as waterstops when wetted.
Another approach to seepage is to simply collect and drain it inter-
nally. This can be done with vertical drain holes, gravel drains, or po-
rous concrete placed along the lifts in question. Flexible, hollow tub-
ing that has filter fabric forming the tubes wall has also been used as
a drain between RCC lifts and at the RCC-to-abutment contact.
The most effective seepage and uplift control method used for lean
RCC dams has been to use a heavy-duty PVC or high-density polyeth-
ylene sheet as a liner behind the upstream facing panels. This method
was first used at Winchester Dam and has since been included in the
design of other RCC dams.
The liner is placed between the RCC and precast facing panels,
which act as stay-in-place forms. The liner is attached to the rear face
of the panels when they are cast. Seams are heat-welded during erec-
tion. The liner is keyed into the foundation and abutments and be-
comes a water-stop extending from abutment to abutment. At Win-
chester, this design has resulted in an attractive dam with no seepage
or maintenance to date. The 74-ft-high (23-m) dam was completed in
1984 and typically maintains a full reservoir for water supply.
Precast panels used for the upstream face at Willow Creek Dam did
not have a liner or employ other seepage controls. RCC lift lines were not
drained or effectively bedded to prevent seepage. While the upstream
face is attractive, substantial but tolerable levels of seepage occurred.
Figure 11.4 shows how the Willow Creek seepage compares with other
dams. Construction of Willow Creek Dam is shown in Fig. 4.2.
RCC was placed directly against wooden forms at Copperfield Dam.
This resulted in an economical, effective, and essentially crack-free
upstream face. But it was not esthetically pleasing (see Fig. 11.6).
Dry, lean RCC mixes honeycomb when they are compacted against a
vertical form. This has little significance below normal reservoir lev-
els where the surface is not visible and is not subjected to alternate
wet-dry cycles, rapid temperature changes, freezing, and wave action.
Above the reservoir level, this shortcoming can be overcome with a
conventional concrete facing that is placed monolithically with the
RCC. This approach has been used on a number of dams, but it must
be done carefully and with a thorough knowledge of all the technical
issues involved. Strength and mass are much less important, for ex-
ample, than strain capacity, creep, low elastic modulus, minimal dry-
ing shrinkage, and minimal adiabatic temperature rise.
If the conventional concrete face is too thick [more than about 15 in
(0.4 m) or average distance from the upstream form to the RCC], it can
develop significant thermal stresses and result in cracks that would
not appear with a thinner facing. In this case, thicker is more expen-
Design and Construction of Lean RCC Dams 125

Figure 4.2 Construction at Willow Creek Dam.

sive and technically inapprorpriate. Similarly, a mix with less cement

and a lower elastic modulus is more desirable than a rich, high-
strength mix with more heat, stiffness, and cost.
Unless special precautions are taken to minimize drying shrinkage
as well as thermal contraction, internally developed stresses can also
result in cracking when none would otherwise occur in the conven-
tional concrete. Controlled shrinkage is accomplished by careful selec-
tion of the aggregates, use of superplasticizers, placing RCC at low in-
ternal temperatures, using a minimum-cement-content mix and
thoroughly curing the compacted RCC, and by minimizing the volume
of conventional concrete.
Even with careful controls, it may still be necessary to provide ver-
tical contraction jointing in the upstream face. The need for joints will
depend on the severity of the climate, the construction schedule, and
the quality of concrete possible with available materials.
The design and construction aspects of a monolithically placed facing
must be considered at the same time. When done properly, no reinforcing
or anchorage is necessary or desirable. The most common approach has
been to place a superplasticized conventional facing mix first, spread the
RCC layer into it, and compact the the dry, lean RCC into the facing as
the conventional concrete rapidly loses slump but before it reaches an
initial set.
A second approach has been to use a wetter and more responsive
RCC mix which allows it to be consolidated into the facing mix with
large high-energy mass-concrete vibrators, The wetter RCC is needed
126 Chapter Four

only near the face. But for simplicity of construction and better bond-
ing of lifts, it may be used for the entire dam. Due to the loss of
strength and/or increase in cement content and cost, the added water
results in a less efficient RCC mix, however.
In both approaches it is imperative that the facing concrete and
RCC be thoroughly compacted or consolidated together so that they
form a monolithic mass.
It is important to note that conventional facing concrete does not
provide seepage control unless it has a thickness of about 8 percent of
the hydraulic head, no cracks, and waterstopped jointing. In fact, al-
though the facing may be attractive, it may aggravate seepage prob-
lems by causing cracks in the RCC that would not have occurred oth-
erwise. The lift lines between RCC lifts are continuous with the lift
lines in the facing concrete. These lines are the source of seepage. At a
desired maximum facing thickness of about 15 in (0.4 ml, the length of
the flow path along the conventional concrete lift line has no signifi-
cant impact on watertightness.

4.2.2 Lift thickness and bonding

Early in the development of RCC, lifts of about 9 in (230 mm) were
considered appropriate. With additional experience and testing,
thicker layers were found to be fully compactible. Twelve-inch (300-
mm) lifts have become the norm because they are easy to control and
spread with earthmoving equipment. Lifts of 15 to 18 in (0.38 to 0.46
m) are also possible, however. When wetter mixes are used, it is pos-
sible to spread several layers of about 9 to 12 in (230 to 300 mm) and
compact them simultaneously as one lift 2 to 3 ft thick (0.6 to 0.9 ml.
Using thicker lifts is desirable when practical because doing so re-
duces the number of lift joints and consequently the number of poten-
tial seepage paths and potential shear planes.
The same concerns and precautions discussed earlier to minimize
seepage between lifts also apply to bond strength. Principally, they in-
volve keeping surfaces clean and having sufficient fresh paste at the
interface to fill void space and provide strength. The age or maturity
of the joint at the time that the next layer is placed was initially
thought to be very important. As experience has shown, it is an im-
portant consideration but not as critical as originally thought. For ex-
ample, within practical placing schedules wherein a subsequent layer
of RCC is placed within 12 to 24 h of the previous layer, the bond
strength at the lower and upper time limits will be about the same.
In broad terms, lean, dry RCC mixes made with reasonably suitable
aggregates typically result in shear values along joints on the order of
Design and Construction of Lean RCC Dams 127

100 lb/in (0.7 MPa) with a friction angle of 45. Poor aggregates such
a friable sandstones and granites can result in lower strengths. Typi-
cal tensile strengths are on the order of 50 to 100 lb/in2 (0.35 to 0.7
MPa), but zones with little or no reliable tensile strength are also com-
mon. When increased strength is needed for tension or shear, a bed-
ding mix between layers will provide the increased shear resistance-
in most cases creating a bond equal to the strength of the compacted
RCC. This can be double the strength at the interface without bed-

4.2.3 Thermal concepts and joints

Thermal stresses in RCC dams require careful analysis. Because of
the large surface area compared to the mass of each lift, more detailed
mathematical modeling must be done than for conventional mass con-
crete dams. The thermal study should include a comprehensive anal-
ysis of weather conditions and construction scheduling.
By incorporating the material property advantages of lean RCC into
both the design and construction schedule, it is possible to eliminate
most, if not all, vertical monolith joints. The critical aspects are: plac-
ing cool concrete (during the coolest months, when possible); minimiz-
ing the adiabatic temperature rise by minimizing cement content; ob-
taining a high creep rate, low thermal expansion coefficient, and low
elastic modulus by minimizing cement and using appropriate aggre-
gates. One of the more effective and economical ways of providing a
precooled RCC mix has been to produce large stockpiles of aggregate
in the winter and using the naturally cooled aggregate during the
summer months.
When joints are necessary to accommodate thermal contraction
within the RCC mass, they should be provided with waterstops placed
in the conventional concrete at the upstream face. It is not necessary
for the joints to be carefully formed or to go through the entire RCC
mass. Providing a weakened plane near where cracks are expected to
form is sufficient. Also, as long as the joint is defined at the crest, up-
stream face, downstream face, and base of the dam, it does not need to
be built into the interior mass. It will eventually self-propagate to the
interior if the joint is placed in the correct place and if it is necessary.

4.2.4 Galleries
Galleries can be incorporated into RCC dams when appropriate, but
they should be technically necessary and justifiable. Most structures
less than 100 ft (30 m) high can be efficiently and safely built with
128 Chapter Four

lean RCC by providing slightly more mass and eliminating the gal-
lery. Winchester Dam in a good example of an RCC dam with no gal-
A gallery will decrease the mass, provide a weakened zone where
longitudinal cracking could start, may require troublesome and ex-
pensive reinforcing steel, and will slow production by 15 to 50 percent
in the region of the gallery. It also will restrict the placing area be-
tween the gallery and upstream face, often resulting in lower-quality
RCC than otherwise would have been achieved there. The resulting
short seepage path between the gallery and upstream face will require
special treatment such as bedding each layer or the area will become
a source of considerable seepage.
A gallery is justified in higher dams primarily because internal
drains to the gallery allow an economical reduction of the mass and
because of the access provided for foundation drains. When a gallery is
appropriate, its location should be well conceived and coordinated
with the practical aspects of construction. As much as possible, it
should be located at a single level. Multiple galleries should be
avoided if possible. Sufficient space [25 to 30 ft (7.6 to 9.1 m)] should
be provided between the upstream face of the dam and the gallery to
allow equipment to operate.
Conventional gallery construction methods such as wood forming or
precast elements are expensive, time consuming, and disrupt the plac-
ing operation by blocking the access of hauling vehicles. An effective
construction method has been to replace RCC with a noncemented fill
in the area of the gallery as each layer of RCC is compacted. Later, the
loose material is excavated, leaving the void as the gallery. This al-
lows direct inspection of the rough interior RCC mass. Using precast
concrete, slipformed concrete or forms with conventional concrete as
the gallery walls hides the RCC from direct inspection.

4.2.5 Instrumentation
Instrumentation in lean RCC dams should generally be the same as
would be provided for a conventionally placed dam of similar size,
risk, and purpose. When monolith joints are eliminated or greatly re-
duced, it is prudent to place extra instruments for measuring temper-
ature peaks and rates of drop so that predictions used as the basis for
design can be verified. Also, when temperatures are dependent on the
construction schedule and placing rates do not meet that schedule,
temperature monitoring should be provided.
Uplift pressures along lift lines should be monitored if they are an
important aspect of stability.
Seepage and its change with time are especially critical. Typically,
Design and Construction of Lean RCC Dams 129

there is a substantial and continued reduction in seepage with time

for lean RCC dams (see Fig. 11.4). Seepage should be monitored care-
fully on a continuous basis to assure that leaching and/or chemical at-
tack does not deteriorate the lean RCC with time.
Internal crack meters (extensometersl may be desirable across some
joints and/or in areas judged to be potential crack locations. Examples
are above foundation joints and at the spillway-to-nonoverflow section
As routine procedure, precise level and alignment surveys should be
made across the top of large dams on a seasonal or annual basis using
permanent monuments in the structure.

4.2.6 Appurtenant structures

Appurtenant structures such as valve houses should be constructed
independently of the RCC or be totally encased or supported in it. Pre-
fabricated structural steel multiple-level intakes have been anchored
to RCC dams, and cast-in-place intake towers have been built away
from dams. In the case of a separate intake tower, hinged bracing to
the dam with struts can provide lateral stiffness and result in sub-
stantial cost savings as compared to a tall, unbraced tower.

4.3 Mix Design and Material Properties

4.3.1 Mix Design

Lean RCC mix designs include cement, aggregate, and water. In some
cases, pozzolans and admixtures may also be included. Typical cement
contents are on the order of 80 to 125 lb/yd3 (47 to 74 kg/m3), but
higher cement contents have been used in select portions of some
structures or when only poor-quality aggregate was available.
Pozzolans usually are used only if a high cement content is truly
needed and if pozzolan is economically available.
The amount of pozzolan that is optimal depends on each project. In
some cases, adding silt or rock dust has had the same effect as adding
pozzolan. In other cases, replacement of up to 50 percent pozzolan for
cement has resulted in essentially the same strength as the cement-
only mix. In still other cases, the addition of pozzolan has resulted in
a loss of long-term strength below what was achieved with the same
cement content and no pozzolan. Each case is evaluated individually.
The common approach to mix designs for lean RCC has been first to
obtain the most economical smooth aggregate gradation within the
broad overall limits. Two examples of aggregate gradation bands were
shown in Fig. 2.7. The key is to have a good blend with enough sand
and fines to separate the coarse aggregate so that they do not contact
130 Chapter Four

each other or nest together. Also important is maintaining an ade-

quate content of nonplastic fine material smaller than 75 lo (#200
For lean mass mixes the amount of fines ideally is in the range of 5
to 8 percent. More fines can be tolerated if they have zero plasticity,
whereas the amount of fines must be reduced if they exhibit some
plasticity, which causes difficulty with mixing and compaction.
After the most economical gradation has been established for a
project, the aggregate is mixed with cement at various contents span-
ning the probable ultimate cement factor. Typical cement factors used
in trials are 75, 100, 125, 150, and either 50 or 200 lb/yd3 (44, 59, 74,
89 and either 30 or 119 kg/m3). These normally result in mixes with
strengths from less than 1000 lb/in2 (6.9 MPa) to more than 2500 lb/
in2 (17.2 MPa) at one year. Two cylinders are broken in compression
at each cement factor at ages of 3, 7, 14, 28, 56, 90, 180, and 365 days.
Enough early data (through 28 days) can be accumulated to allow pre-
dictions of long-term strengths and to provide a good basis for design
and cost estimates. As later-age strength data become available, it is
used to verify predicted strengths or to make minor adjustments.
The moisture content should be great enough to allow full compac-
tion with a pneumatic tamper or large vibrating roller, but not so
much that an operating roller sinks into the compacted mix. The mois-
ture content may vary with different aggregates, but it will be very
consistent for a specific aggregate regardless of cement content. In its
fresh state, the mix behaves more like cement-treated soil with limits
on handling time than it does fresh concrete.
Based on results of visual compactability and specific tests of the
hardened mixes at various cement contents, the minimum cement
content needed for the aggregate source can then be selected. If the
selected mix results in excessive heat from hydration or if pozzolan is
readily available at a reasonable cost, another set of mixes is tested
with pozzolan.

4.3.2 Material Properties

Testing usually includes compressive and indirect (split-cylinder) ten-
sile strengths, density and air content, and modulus of elasticity.
From the data and from relationships established with other RCC ma-
terials, properties such as adiabatic temperature rise, specific heat,
Poissons ratio, shear strength, direct tensile strength, thermal expan-
sion, creep, and strain capacity can be reasonably well predicted. On
large projects or those with some peculiarity in the mix, special tests
for these properties may be appropriate.
Typical properties of some RCC mixes are given in Fig. 4.3. Some of
Figure 4.3 RCC mix designs and properties.

W Split Modulus of
Cement, Pozzolan, Water, CtP Age, Compression, tensile, elasticity E, Poissons
Dam lb/yd3 lb/yd3 lb/yd3 (by wt) days lb/in lb/in lb/in x lo6 ratio

80 32 180 1.61 3 419 - - -

Willow Creek
7 571 - 1.20 -
28 1172 90 1.59 0.14
90 1731 165 1.91 0.17
365 2623 - - -
Willow Creek 175 0 185 1.06 3 656 - -
I 991 2.20 -
28 1845 225 2.67 0.19
90 2641 270 2.78 0.18
365 3779 - -
Willow Creek 175 80 185 0.72 3 789 - -
7 1147 - 2.40 -
28 2056 2.91 0.21
90 3961 340 3.25 0.21
365 4146 - -
Willow Creek 315 135 200 0.44 3 1372 - -
7 2021 - -
28 3408 - -
90 4469 - -
365 5790 - -
Zintel Canyon 50 0 200 4.00 3 120 - - -
7 160 - - -
28 350 - - -
90 530 - -
365 740 - -


Figure 4.3 RCC mix designs and properties. (Continued)

W Split Modulus of
Cement, Pozzolan, Water, ctp Age, Compression, tensile, elasticity E, Poissons
Dam lb/yda Iblyd lblyda (by W days lb/in lb/ix? lb/in x 10 ratio
Zintel Canyon 100 0 200 2.00 3 190 - 0.31 -
7 280 - 0.68 -
28 630 90 1.54 -
90 1090 165 2.15 0.21
365 1550 - 2.57 -
Zintel Canyon 150 0 200 1.33 3 410 - - -
I 620 - - -
28 1100 - - -
90 1550 - - -
365 2530 - - -
Zintel Canyon 200 0 200 1.00 3 680 - 1.35 -
7 990 - 1.54 -
28 1620 200 2.39 -
90 2130 255 2.47 0.20
365 3100 - 3.28 -
Elk Creek 94 38 170 1.29 3 210 - - -
7 270 - - -
28 410 60 - -
90 820 115 1.69 0.19
Tarbela 200 0 - - 3 - - - -
(1982) 7 1110 - - -
with silt 28 1490 - - -
90 1980 - - -
(2 yr) 2760 - - -
Tarbela 235 0 - - 3 547 - - -
(pre-1982) 7 662 - - -
no silt 28 951 - - -
Middle Fork 100 0 1 3 5 (60) 1.35 3 490 - - -
7 540 - - -
28 868 - - -
90 1290 - - -
365 - - - -
Middle Fork 120 0 1 3 5 (180) 1.13 3 565 - - -
7 797 - - -
28 1655 - - -
90 2000 - - -
365 - - - -
Monksville 105 0 205 (122) 1.95 3 487 37 - -
7 650 64 0.35 -
28 745 95 0.90 -
90 - - - -
365 - - - -
Upper 182 210 183 0.47 3 - - - -
Stillwater 7 1360 - - -
28 2130 110 - -
90 3510 150 - -
365 5220 205 1.10 0.13
Upper 121 269 175 0.45 3 - - - -
Stillwater 7 770 50 - -
28 1220 80 - -
90 2150 130 - -
365 4780 200 1.43 0.14


Figure 4.3 RCC mix designs and properties. fContinue&

W Split Modulus of
Cement, Pozzolan, Water, ctp Age, Compression, tensile, elasticity E, Poissons
Dam lbiyd3 lb/yd3 lb/yd3 (by wt) days lb/in lb/in lb/in x 10 ratio
Upper 129 236 180 0.43 3 - - - -
Stillwater 7 1100 55 - -
28 1620 110 - -
90 2770 130 - -
365 4960 220 1.60 0.19
Ncnes: Strengths are for 6 x 12-in. cylinders with t 1%in aggregate removed. The data are
from laboratory mix design studies and/or field cylinders during construction. Actual water
contents used during construction may vary due to field conditions. The water content is based
on saturated surface dry aggregates. 1 lbiyd = 0.59 kg/ma; 1 lb/in3 = 6.9 x lo3 Mpa.
Design and Construction of Lean RCC Dams 135

the values for the same property vary considerably. That is mainly
caused by the broad range of aggregates used to make RCC. This fac-
tor can be used to great advantage if the structural design and mix
design (material property) aspects of a project are mutually under-
stood and coordinated with the economies and practicalities of con-
struction. For example, a dam that can be built with a very economi-
cal low-cement-content mix by slightly increasing the mass can often
result in substantial overall project savings. It can also result in a con-
crete than is much less likely to crack than a more expensive, higher-
strength mix with more cement (or cement plus pozzolan) and greater
laboratory strengths.

4.4 Construction Methods

Specific construction methods pertinent to bonding, lift thickness, and
facing methods that are an inherent part of design have already been
discussed. Most of the basic RCC placing methods for lean RCC are
the same as for high-paste RCC (see Chap. 5) or the RCD mixes used
by the Japanese (see Chap. 6). The main difference is that the lean
mixes usually tend to be drier and have less paste. They tend to be
more susceptible to segregation, and therefore require more attentive
operators. Lean-mix dams do not require more equipment, time, or
personnel than other types of RCC dams, however.
Lean mixes are often drier and so will not respond as well as wetter,
slower-setting mixes to retarders or admixtures. Consequently, lean
mixes usually must be delivered, spread, and compacted more rapidly
than other types of RCC-typically within 30 min of the start of mix-
A critical aspect of lean RCC is mixing. Exact proportions of ce-
ment, aggregate, and water must be used and they must be thor-
oughly, uniformly, and consistently blended. This has been done most
effectively with specifically designed continuous-mix pugmills and
continuous volumetric or weight controlled feeders.
Minimizing the number of aggregate stockpiles (preferably to two)
and preblending the sand and fines with the l-in (25-mm) minus ag-
gregate during its production has been very beneficial. Mixing prob-
lems can be further eased by using large-aggregate stockpiles that
contain a uniform moisture level near the moisture necessary in the
final mix. These steps leave little for the mixer to do except blend one
or two aggregates with the cement and a small amount of added wa-
Because the lean mixes inherently have low cement contents, they
must be metered and mixed with special care. Most commercial mix-
ing equipment is designed for higher cement volumes where the
136 Chapter Four

criticality of blending is not so great. For example, if two portions out

of the same batch of mixed RCC contain 10 lb (4.5 kg) more and less
cement than the design requires, the average (the amount fed into the
mixer) is correct. However, having 10 lb (4.5 kg) more and 10 lb less
than 100 lb (45 kg) for a hypothetical lean mix design is a variation of
20 percent. If the mix design had called for 400 lb (180 kg) of cement,
the variation would have been only 5 percent.

4.5 Construction Control

Lean RCC dams have typically been controlled by method rather
than by laboratory tests of the RCC. By the time laboratory cylinder
results are obtained, the lean RCC dam may be substantially com-
pleted. Alternatively, if the aggregate source and gradation are as
specified, if proportioning is correct, if the mixer is pretested to ensure
adequate blending, if the mix is delivered and compacted within al-
lowed time constraints, if the field density is achieved by using
prequalified equipment and is verified by in-place nuclear densities,
then the end product should have adequate strength, watertightness,
mass, and stability.
The moisture content of most lean RCC mixes has been controlled in
the field by the placing supervisor, who continually monitors how it
behaves under the roller. If the mix becomes slightly wet, the roller
sinks. If it becomes slightly dry, the mix will begin to segregate. The
modified Vebe method of controlling moisture has not worked well on
lean RCC unless the mix has been wetter than normally used.
The main tests which control quality during RCC production are
gradation, in-place freshly compacted density, and tests of the profi-
ciency of the mixer and its variability.
A combined gradation test is recommended for the aggregate being
used during each shift of placement. If the results are out of specifica-
tion limits but the mix continues to place well and achieve normal
densities, production can usually be allowed to continue while the
problem is being corrected. Only if placement or density problems de-
velop or if the gradation problem persists through a third placement
shift is a shutdown of production necessary to alter the aggregate gra-
Nuclear density testing of the freshly compacted mix should be ac-
complished on a regular basis during placement. Densities within 98
to 100 percent of the theoretical air-free density provide similar
strengths. For each mix and structure, the designer needs to establish
permissible densities or compaction values. A typical criterion may re-
quire no test results less than 90 percent of the theoretical air-free
density; allow 2 percent of the tests to be less than 95 percent if the
Design and Construction of Lean WC Dams 137

samples are from isolated areas in noncritical zones; and require that
98 percent of all tests to be above 95 percent density with the average
of all tests at 97 percent.
If low densities are the result of inadequate compaction, rerolling
lifts within about 15 min of placement can improve the test results.
However, rerolling after the mix reaches initial set should be
avoided because it can damage the concrete even though it increases
the density. Overrolling should also be avoided because it can reduce
density in much the same way that overrolling affects some compacted
Mixer test methods, batch or feed-rate accuracies, allowable mixer
variability, and the amount of overdesign needed to account for mixer
variability have been discussed in detail by the author (see Bibliogra-
phy, Schrader 19871. Basically, more cement and higher average
strengths are needed when using plants that produce more variable
mixes. Overall, coefficients of variation for lean RCC projects have
typically ranged from 20 to 28 percent.

Ditchey, E., and Schrader, E. K., Monksville Dam Temperature Studies, ZCOLD Six-
teenth Congress, San Francisco, June 1988, vol. III, Q62, pp. 379-396.
Quin, J., Rezende, S., and Schrader, E. K., Saco Dam-South Americas first RCC
Dam, Roller-Compacted Concrete II, ASCE, New York, February 1988, pp.340-356.
Schradcr, E. K., Behavior of Completed RCC Dams, Roller-Compacted Concrete II,
ASCE, New York, February 1988, pp. 76-91.
- Compaction of Roller-Compacted Concrete, Proceedings of the American Con-
cretk Institute Symposium on Consolidation of Concrete, San Francisco, March 1986.
---, Design for Strength Variability: Testing and Effects on Cracking in RCC and
Conventional Concretes, Lewis Tuthill Symposium, American Concrete Institute,
October, 1987.
-, Roller-Compacted Concrete, The Military Engineer, September-October 1977.
- Watertightness and Seepage Control in Roller-Compacted Concrete Dams,
Roller-Compacted Concrete, ASCE, New York, May, 1985, pp. 11-30.
Schrader, E. K., and Namikas, D., Performance of Roller Compacted-Concrete Dams,
ICOLD Sixteenth Congress, San Francisco, June 1988, vol. III, Q62, pp. 339-364.
Snider, S., and Schrader, E. K., Monksville Dam: Design Evolution and Construction,
Roller-Compacted Concrete II, ASCE, New York, February 1988, pp. 220-235.
Tatro. S. B., and Schrader, E. K.. Thermal Considerations for Roller-Compacted Con-
crete, Journal of the American Concrete Institute, March-April 1985. _
US. Army Corps of Engineers, Willow Creek Dam Concrete Report, Walla Walla,
Washington, 1984.

Design and Construction

Concepts of a High-Paste-
Content RCC Dam

5.1 Background
Roller compaction of concrete for dams has many advantages over the
conventional method of dam construction. However, designers of RCC
dams should still aim to achieve the excellent properties that have
been obtained in conventionally placed concrete gravity dams. Some
designers have expressed concern about seepage between lifts of Wil-
low Creek Dam. This has led to the suggestion that RCC dams should
be designed as embankment dams comparable to rockfill dams with
an impervious upstream membrane. It has even been suggested that
advantage should be made of the permeability between lifts to reduce
internal uplift. Unless this concern about permeability can be re-
moved, some of the potential economy of the RCC method of construc-
tion may be lost.
The central philosophy of the high-paste-content RCC dam includes
the following:

m It should perform as well as a conventionally placed concrete dam.

9 The method of construction should be kept as simple as possible.
. Full advantage should be made of the speed of construction.

The latter is one of the main advantages of the new method of dam
construction and it can lead to a significant proportion of the potential
economy of RCC over competing methods.
Within these overall concepts, the author proposes that the interior

This chapter was contributed by Malcolm R. H. Dunstan.

140 Chapter Five

Upper Stillwater Dam, a high-paste-content RCC dam.

RCC should be considered to be the watertight barrier and that the

faces of the dam should be considered to be only a durable skin. Also,
the design of the mixture proportions cannot be considered outside the
context of the design of the whole structure. The performance of the
concrete, and thus the mixture proportions, should be varied to take
into account the specifics of the site. The mix should be optimized to
produce the most economical solution in terms of the volume of the
dam and the overall cost.

5.2 Design Concepts

Concrete gravity dams have acquired a reputation for being extremely
safe. Very few dams of this type have failed, even though they have
been subjected to overtopping, to extreme seismic activity, and to
many other loads outside their normal design criteria. Dams cannot
be considered to be temporary structures, particularly concrete dams,
which are extremely difficult to remove once constructed.
Design and Construction Concepts of a High-Paste-Content RCC Dam 141

It is central to the design philosophy of the high-paste-content RCC

dam, therefore, that the performance of the completed dam should be at
least as good, and preferably better, than that of a conventional concrete
gravity dam. This should be the first design criterion for this type of dam.
The design philosophy of conventionally placed concrete dams is
that the whole structure should be considered to be impermeable. Ver-
tical joints were introduced in concrete dams initially as a construc-
tion feature. Over time, joints have been employed to reduce the po-
tential for cracking due to thermal movement, one of the main
concerns in the placement of large concrete masses. The use of con-
traction joints and other measures to reduce thermal cracking have
made the construction of concrete dams more and more complicated,
however. These complications have increased costs and reduced the
economic competitiveness of conventionally placed concrete dams.
A second design criterion for the high-paste-content RCC dam,
therefore, should be to keep the method of construction as simple as
possible. With simplicity will come speed, which is one of the main ad-
vantages of the overall method of construction.
In order to achieve these two criteria, the third criterion should be
that the design of a high-paste-content RCC dam must be considered
as a whole. This includes the cross section of the structure; the details,
such as spillways and galleries; and the design of the mixture propor-
tions. All will interact and it should be the objective of the design to
arrive at the optimum solution for the particular site and the partic-
ular purpose for which the dam is being designed.

5.2.1 RCC mixture design

The choice of the optimum mixture proportions is only one of the fac-
tors to consider within the concept of the design of an RCC dam. For
some sites it may be necessary to design a concrete with particular
properties, for example, because of an extreme temperature regime. It
may then be beneficial to use these properties in order to refine the
cross-section of the structure. The majority of designs will be site-
specific, but there are common factors which have to be considered
during the design of the RCC mixture proportions. These include:

1. The height and size of the dam

2. The purpose for which the dam is designed
3. The balance between the cross section of the dam and the proper-
ties of the concrete.
4. The temperature regime at the site
5. The availability of cementitious materials
142 Chapter Five

6. The labor, plant, and material cost relationship

The height, and to a lesser extent the size, of the dam together with
its intended use will dictate a range of acceptable in situ
permeabilities. It is not the permeability of the RCC itself but the po-
tential leakage path along the joints between the layers of RCC that
can cause problems. Willow Creek had unacceptable leakage along
these joints. Because the leakage is localized along lift lines, the po-
tential for deterioration is enhanced.
A number of tests of the overall in situ permeability have been car-
ried out on RCC structures. The results of some of these tests are
shown in Fig. 5.1, which contains data from eight different countries.
The data are derived from published literature and personal corre-
spondence and include some test sections and RCC structures that are
not gravity dams. The permeability measurements were carried out in
a variety of ways-by use of packer tests, by falling-head permeability
tests, by measuring the flow through the whole dam, and by testing of
cores. In spite of the different approaches, there seems to be a rela-
tionship between low in situ permeability and high cementitious con-
tent. This is to be expected, as conventional concrete follows a similar

Milton Brook++

Holbeom Wood

Willow Creek

0 I I I I I
0 100 200 300
Cementitious coti. kg/m3
Figure 5.1 Relationship between permeability and cementitious content.
Design and Construction Concepts of a High-Paste-Content RCC Dam 143

Treatment of surface
of horizontal joint Bedding mix
Willow Creek None Partial
Holbeam Wood None None
Copperfield None Partial
Ohkawa Green cut Full
Shimajigawa Green cut Full
Tamagawa Green cut Full
Upper Stillwater None None
Bishops Stortford None None
Milton Brook None None

Figure 5.2 Lift treatment for various RCC structures.

The treatment of the lift surface and the use of bedding mixes do not
seem to have a significant effect upon the overall permeability. Some
of those structures shown in Fig. 5.1 used full treatment of the lifts
together with a bedding mix spread over the entire lift surface. Some
used no treatment of the lifts and partial bedding mixes, while others
used no treatment and no bedding mixes, as shown in Fig. 5.2.
It is not possible to differentiate between the RCC lift surface treat-
ment methods based on the permeability test results. The reason may
be that any modification to the method of construction, such as
treatment of the Iift surface or application of bedding mixes, can slow
the rate of construction. Any slowing of the pace of RCC placement
can lead to a reduction in the performance of the bonding between the
layers. Thus the modifications may only counteract the intended effect
of their introduction.
This does not mean that treatment of the RCC surface may not be
necessary in special cases, but it seems unlikely that the general use
of bedding mixes and lift surface treatment improves the performance
of RCC structures in terms of in situ permeability.

5.2.2 Permeability

The author believes that any dam which is to store water should have
a minimum permeability of 10e6 cm/s. Because the total flow through
structures is dependent upon head, a dam which is 165 ft (50 m) high
may need a minimum permeability of 10m7 cm/s; 330 ft (100 m) high,
10-s cm/s; 650 ft (200 m) high, lo- cm/s; and for a dam higher than
that, 10-l cm/s.
These design criteria may be too low. The U.S. Bureau of Reclama-
tion has defined a limit of 1.5 x lo- cm/s as acceptable for conven-
tional concrete dams. Completed concrete dams have generally been
found to have permeabilities ranging from 5 to 50 x 10-l cm/s.
144 Chapter Five

Because of the large number of lifts in an RCC dam, it is important

to consider the overall permeability of the lift interfaces during the
design. For certain structures more stringent limits may need to be
applied, particularly because of the relative infancy of the method of
Recognizing the general correlation between permeability and ce-
ment content, the use of the design criterion for in situ permeability
will therefore define a minimum cementitious content. Permeability,
therefore, is the first design criterion to be used for the design of the
mixture proportions.

5.2.3 Density
The density of the RCC will also be an important factor in the final
design of the structure. It has been found that the entrapped air con-
tent of RCC placed for existing dams has varied between 8.5 percent to
less than 0.5 percent. A relationship has been proven between the in
situ density [as a proportion of the theoretical air-free (taf) density]
and a factor called the paste/mortar ratio. This relationship is shown
in Fig. 5.3.

h laximum t h e o r e t i c a l 1/
d ensity with fine
a ggregate having a
oid ratio of 0.32

Maximum theoretical
density with fine
aggregate having a
void ratio of 0.4

0.2 0.3 0.4 0.5

Paste/mortar ratio

Figure 5.3 Relationship between in situ density and paste/mortar ratio.

Design and Construction Concepts of a High-Paste-Content RCC Dam 145

The paste/mortar ratio is the ratio of the volume of paste

(cementitious material, water, and, if used, entrained air) to the vol-
ume of mortar (paste and fine aggregate). The density decreases rap-
idly below a paste/mortar ratio of between 0.35 and 0.40. The reason
for this decline is that below this level there is insufficient paste to fill
the voids in the fine aggregate. A typical compacted aggregate will
have a void ratio of between 0.32 and 0.40. Therefore, unless there is
sufficient paste to fill these voids, there will be a volume of entrapped
air which cannot be removed.
Also shown in Fig. 5.3 are the maximum theoretical densities that
can be achieved with fine aggregates having void ratios of 0.32 and
0.40. It can be seen that the relationship for the actual results is par-
allel to the theoretical lines. The scatter of results is caused by the
range of void ratios of the fine aggregate used. There are in reality a
wide range of parallel relationships, depending upon the void ratios of
the fine aggregates.
It is possible to reduce the void ratio of a fine aggregate by use of
nonplastic fines. However, there will be an optimum content above
which the addition of more fines of this type will increase the void ra-
tio and thus the need for further paste. A high fines content can also
lead to high water demand.
A relatively high paste/mortar ratio is required to obtain a density
having a high percentage of theoretical air-free density, which leads
to good strength properties. Density should therefore be the second
criterion to be used in the design of the mixture proportions.

5.2.4 Shear strength

In addition to permeability and density, the shear strength (cohe-
sion and coefficient of friction) and direct tensile strength are fac-
tors which should be considered during the design of an RCC dam.
(Although compressive strength is usually used, a relationship be-
tween direct tensile strength and compressive strength is assumed
Using the Bureau of Reclamation criteria, two relationships have
been developed between the downstream slope and the height of the
dam: First, compressive strength and factor of safety (see Fig. 5.4) and
second, coefficient of friction and cohesion, assuming a factor of safety
of 3 (see Fig. 5.5).
A balance has to be made between the properties of the concrete
that can be economically obtained and the downstream slope of the
dam. Any improvement in the properties can lead to a steepening of
the downstream slope and thus a decrease in the volume of the dam.
This steepening of the downstream face will also lead to a reduction in
146 Chapter Five

0 100
200 300 400
Height, ft
Figure 5.4 Relationship between compressive strength
and downstream slope. (From: Boggs and Richardson,


S=O.9 0.8 0.7 0.6

0 0.25 0.50 0.75 1.0
Cohesion C I b/in*
---I-, -
Height H ft
Figure 5.5 Relationship between cohesion, coeffi-
cient of friction and downstream slope. (From: Boggs
and Richardson, 1985.)
Design and Construction Concepts of a High-Paste-Content RCC Dam 147

the volume of excavation and a reduction in the length of the diver-

sion works. Thus compressive strength and in-situ shear strength at
the joints are additional criteria to be considered during the design of
the mixture proportions.

5.2.5 Thermal volume changes

Another consideration is the potential temperature generated dur-
ing construction and the consequent thermal volume change and
tensile strain induced by the movement. The temperature condi-
tions at the site are an important consideration. The heat from so-
lar gain can be a significant factor in the total temperature rise of
the concrete. Because the RCC is placed in thin layers, a large area
is exposed to the ambient conditions. This can be beneficial if the
air temperatures are less than that of the surface of the concrete. It
can also be a major problem if the air temperatures are higher than
the RCC. For each project, a study needs to be made of the potential
heat gain, taking into consideration the conditions at the site.
Thus, the criteria to be used during the design of the mixture pro-
portions (and the shape of the RCC dam) are permeability, density,
compressive (tensile) and shear properties and thermal volume
The cost of the cementitious materials available at the site of an
RCC dam can have a significant impact on the design of the dam. If
pozzolans (natural pozzolan, blast-furnace slag or fly ash) are readily
available at a reasonable cost, a high proportion of the cementitious
content can be formed of pozzolan. It has been found in laboratory
tests that the proportion can be up to 80 percent of the total volume of
cementitious material. Such a high proportion would increase the
paste content and thus lead to a high density, as a percentage of the-
oretical air-free density. It also results in a relatively high
cementitious content (and thus reduced permeability), and a high fac-
tor of safety against the potential of cracking due to the thermal
movement of the concrete as described below.
A study has been made in the United Kingdom of the factor of
safety against thermal cracking. This is defined in the study as the
ratio of the tensile strain capacity (which will define the resistance
against cracking) divided by the potential temperature rise (which
will be related to the thermal volume change and thus the tensile
strain induced in the long term). The study indicated that as the fly-
ash proportion of the cementitious content was increased up to approx-
imately 70 to 80 percent, there was little, if any, decrease in the ten-
sile strain capacity in the long term but that the potential
temperature rise dropped.
148 Chapter Five

However, in order to be able to obtain the best properties from the

use of pozzolans in concrete, it has been found that a low water/
cementitious ratio is required. Above a certain level, varying between
a water/cementitious ratio (by weight) of between 0.6 to 0.8, there will
be little, if any, contribution to strength from fly ash, or natural
pozzolan for ages up to 91 days. Thus the use of a high proportion of fly
ash, or natural pozzolan, will only be beneficial if there is a reasonable
level of cementitious content. This will, in turn, lead to a relatively
low watericementitious ratio (the water content being relatively fured
by the desired consistency for roller compaction).
There is no single solution to the design of an RCC mixture. Many
factors must be considered and the design of the concrete is but one
factor within the context of the design of the entire structure.

,5.2.6 Properties of compacted RCC

During the design of the early RCC dams there was little data on
the hardened properties of RCC. Although RCC has been used as a
road base since the 1930s the purpose for which these concretes
were designed was different from that for dams. However, there is
now sufficient data available for engineers to be able to design an
I RCC concrete for a wide range of properties. The most important

1. Permeability
2. Tensile strength and tensile strain capacity
3. Shear strength at lift surfaces (cohesion and coefficient of friction)

Other less important properties which have been measured are:

4. Compressive strength
5. Poissons ratio
6. Coefficient of thermal expansion
7. Creep

It is the in situ properties, as measured on cores or the body of the

structure itself, which are important. The properties of laboratory
specimens may give misleading results in certain cases. Generally,
concrete properties at ages of 91 to 365 days are used, as the dam will
not be put into service until after this time. Concrete will continue to
increase in strength for many years, with greater gains if it contains
The range of total in situ permeabilities of RCC that have been
Design and Construction Concepts of a High-Paste-Content RCC Dam 149

achieved to date vary from 2 x 10- cm/s (Willow Creek) to 1 x lo-l1

cm/s (Milton Brook trials) (see Fig. 5.1). Generally, concretes designed
using the concepts of a minimum paste content (for density) will have
a permeability in the region of 10e7 to lo-l1 cm/s. This range is ade-
quate for the majority of conditions for which RCC dams are presently
being designed. For the full range of properties of RCC used for dams,
including high-paste dams, see Sec. 2.5.
Cores extracted from high-paste-content roller-compacted con-
cretes have been found to be generally well bonded at the joints.
The variation of density from the top to the bottom of each layer has
been found to be minimal. Figure 5.6 shows cores extracted from
the Milton Brook trials Concretes designed in this fashion can al-
most be considered to be totally monolithic with only a small de-
crease in strength at the lift lines and little or no decrease in per-

Figure 5.6 Cores taken from Milton Brook Dam trials

150 Chapter Five

5.2.7 Design of a Cross Section

Because the design of each dam will be site-specific and dependent
upon many factors, it is difficult to define more than a range of possi-
ble mixture proportions and cross sections that may be feasible in any
particular situation.
Generally, relationships similar to those shown in Figs. 5.4 and 5.5
can be used to obtain the range of downstream slopes possible in any
particular situation. A balance has to be made between the cost of the
additional cementitious material needed to obtain the desired shear
strength properties (and thus a steeper downstream slope) and the
savings possible due to the smaller volume and decreased base width.
Using the design criteria explained in Sec. 5.2 for a 160-ft-high (50-
m) dam which is to retain water for the majority of its life, a minimum
permeability of 10e7 cm/s might be considered prudent. For a normal
range of material properties, a concrete containing a cementitious
content of approximately 250 lb/yd3 (150 kg/m31 might be used (see
Fig. 5.1). If fly ash is being used, one might expect a portland cement
content of 85 lb/yd3 (50 kg/m3) and a fly-ash content of 165 lb/yd3 (100
kg/m). The ratio will vary if blast-furnace slag or natural pozzolan is
used. The expected performance of a concrete with these mixture pro-
portions would allow an average downstream slope of 0.7 H : 1 V in a
dam of this height.
For a 500-ft-high (150-m) dam, a minimum permeability of 7 x
lo- cm/s would be prudent. To achieve this, a cementious content of
320 lb/yd3 (190 kg/m3) might be required. The range could be 50 lb/yd3
(30 kg/m3) in either direction depending upon the quality control of
the materials available. Such a concrete might have a portland ce-
ment content of 85 lb/yd3 (50 kg/m31 and a fly-ash content of 235 lb/yd3
(140 kgim3). The average properties of such a concrete could allow an
average downstream slope of approximately 0.77 H : 1 V with this
height of dam.
However, by increasing the cementitious content to 390 lb/yd3 (230
kgim3) [Portland cement content of 120 lb/yd3 (70 kg/m31 and fly-ash
content of 270 lb/yd3 (160 kg/m311 it might be possible to reduce the
average downstream slope to 0.725 H : 1 V. Thus a balance can be made
between these two options or any design chosen between the two.
The in situ shear properties achieved with high-paste-content RCC
could be adequate for a dam over 650 ft (200 ml high with a shear fric-
tion factor of safety of 3. With the knowledge of the performance of
RCC and the associated experience with the design of the mixture pro-
portions, it will be possible to envision conditions under which RCC
dams significantly above that height could be built.
Design and Construction Concepts of a High-Paste-Content RCC Dam 151

5.2.8 Detailed design

One of the main criteria for the design of an RCC dam has been simpli-
fication of the method of construction so that there is no reduction in the
placement rate. Any contraction joints between the upstream and down-
stream faces, as are used in the Japanese RCD method, introduces com-
plications that may reduce the speed of placement. It has been shown
that it is feasible to place RCC for considerable lengths without detri-
mental cracking as long as the material is protected from rapid changes
in temperature. Unreinforced concrete that is subject to rapid changes in
air temperature of 36F (20C) or more (depending upon the coefficient of
expansion of the concrete) will generally suffer from thermally induced
cracks, particularly if the concrete is restrained.
For example, cracks between 10 and 16 ft (3 to 5 m) deep are pre-
dicted to penetrate both faces of Upper Stillwater Dam due to the ex-
treme changes in air temperature at the high-elevation site. Most ex-
ternal faces of RCC dams will eventually crack unless the climatic
conditions are very mild.
Therefore, unless some other form of noncementitious upstream
membrane is applied to an RCC dam, the watertight membrane is
best designed to be in the center of the dam where it is protected from
changes in temperature by the facing concrete. It is this concept which
is used for the design of the high-paste-content RCC dam. The faces
are considered to be purely a durable skin protecting the interior
RCC, which is a watertight barrier.
The faces of an RCC dam can be formed by a number of methods.
The main criterion is that construction of the faces should not inter-
fere with the placement of the interior RCC. The method that seems to
be most suited to the larger RCC dams is slipforming of interlocking
facing elements. This was first conceived for the Milton Brook Dam in
the United Kingdom and was used for Upper Stillwater Dam. By
roller-compacting the interior concrete against the facing element as
soon as possible (the minimum can be as little as 4 h) the interior con-
crete bonds well to the facing element. The whole structure can be
considered to be monolithic.
Other methods of forming the face, as described in Chap. 8, are also
perfectly feasible with the use of a high-paste-content concrete. It is
not necessary to design the external face to perform as a watertight
Generally, it is desirable to provide drains in all but the smallest
concrete dams. Drains can be provided in a high-paste-content RCC
dam by drilling from a gallery and/or the top of the dam (to intersect
with a gallery) after the dam is completed.
152 Chapter Five

The galleries can be formed by a number of methods. If slipformed

facing elements are used, the same machines can be used to form the
walls of the gallery. This was done successfully at Upper Stillwater,
where the soffit was formed using half-sections of corrugated metal
pipe. An alternative method is to use an all-precast gallery. Again,
the main criterion should be the noninterference with the rate of
placement of the interior RCC.
High-paste-content RCC is able to bond to rock in the same way as
it bonds to previously placed concrete. As such, it is possible to place
the concrete directly on an approximatley horizontal foundation sur-
face. If the foundation of the dam is particularly rough, leveling con-
crete with mixture proportions as near to the RCC as possible should
be used to provide a platform from which the RCC placement can
start. Similarly, when the RCC is to be placed against a steeply sloped
rock abutment, it is necessary to provide concrete compacted by con-
ventional immersion vibrators into which the high-paste-content con-
crete can be roller-compacted.
The thickness of each layer of RCC will be determined by the need
for good bonding across the horizontal joints and by construction needs
rather than by the requirement for high density throughout the layer
thickness. High-paste-content concrete, because of its mobility and co-
hesiveness, can be spread and compacted in considerable thicknesses
without difficulty. However, in order to obtain a good bond between
successive layers, it is necessary to have considerable compactive ef-
fort available at the interface. Because at least one layer of RCC
should be placed each day, the output of the RCC batching-mixing
plant will probably limit the thickness of each layer. Generally, a 1-ft
(30-cm) layer thickness has been found to be the optimum.

5.3 Construction Control

Because of the emphasis on speed of placement with RCC, it is very
important that the materials and proportions be as specified before
the concrete is placed. After the concrete has been compacted, it is ex-
tremely difficult and often counterproductive to remove the material.
Consequently, good quality control is required at the batching and
mixing plant. This is the point at which the in situ properties of a
high-paste-content concrete will be defined. Unless the RCC is mixed
in the correct proportions, it will not achieve the desired in situ prop-
Some form of control is required immediately after mixing which
will define whether the mixture proportions are correct. There are a
number of methods that can be used. High-paste-content concrete has
Design and Construction Concepts of a High-Paste-Content RCC Dam 153

one advantage over the other forms of RCC, in that it is more plastic.
After compaction the concrete will feel like a stiff jelly under foot. This
is because the concrete will consist of aggregate particles suspended in
a stiff paste. The cohesion of the concrete rather than aggregate-to-
aggregate interlock will support the roller. The feel of a lean RCC, and
to a lesser extent the concrete used in the RCD dams, is completely
different. After roller compaction there is no movement in these con-
cretes. Essentially, they are compacted aggregate with paste filling
most of the voids.
This difference is also noticeable during the manufacture of speci-
mens. High-paste-content concrete can be compacted by use of a vi-
brating table alone, although some form of surcharge load is advanta-
Two tests have been used to control the placement of high-paste-
content RCC: the Cannon test and the loaded Vebe. Both enable mea-
surement of the workability (by measuring the time taken to reach a
certain state) and also the fresh density. If the density and consistency
fall within a definable range, the concrete will be satisfactory.
A relationship can be derived between the water content of the mix
and the Cannon or Vebe, time which measures workability. This rela-
tionship enables the operator of the batching and mixing plant to modify
minutely the water content of the mix to obtain a consistent workability.
Control of the fresh density will also enable the density of the concrete
after roller compaction to be maintained above the desired level.

5.4 cost
The cost of any RCC dam will be very site-specific, depending upon the
local material costs. Nevertheless, although the cementitious content
of a high-paste-content concrete is higher than other forms of RCC, it
has been found that the overall costs of the structure can frequently be
less than the equivalent lean RCC dam because of optimization of the
properties relative to the volume of the dam.

5.5 Conclusions
The overriding criteria for the design of an RCC dam should be to ob-
tain in situ properties at least equivalent to those of a conventional
concrete dam and to simplify the method of construction as much as
possible so that advantage can be taken of the very rapid method of
construction. The economy of the construction of conventional con-
crete dams floundered when complications were introduced in order to
overcome problems such as the heat of hydration. It is important that
154 Chapter Five

RCC dam designers do not follow the same path. It is also important
that the RCC mixture proportions be optimized to obtain the least-
expensive overall solution for any particular dam rather than to ob-
tain the greatest possible materials cost saving.
A relatively high paste content is required if a density near the the-
oretical air-free density is to be obtained. In addition, a certain
cementitious content is necessary in order to obtain a minimum in
situ permeability on a par with that obtained in conventional concrete
dams. It is the permeability between the lifts rather than the parent
material that can constitute a problem. Because there are many more
horizontal lifts in an RCC dam than in a conventionally placed con-
crete dam, considerable care needs to be taken to minimize the risk of
any leakage along lift interfaces.
If the interior concrete is to be a watertight barrier-and it is ar-
gued that this is the only feasible solution for any but the smallest
RCC dams- then a minimum cementitious content probably exceed-
ing 250 lb/yd3 (150 kg/m31 will be needed. By taking advantage of the
properties that can be obtained by use of an RCC with a cementitious
content of this order, it is possible to reduce the cross section of the
dam so that the volume of concrete to be placed can be reduced. Other
advantages such as a reduction in the quantity of excavation and a
reduction in the length of diversion works follow because of the nar-
rower base width.
The cost of the optimized solution can frequently be less than that of
a dam designed to minimize the material costs. Additionally, the over-
all properties of the concrete can be significantly better.
There will be a few cases where pozzolans are not available in any
form near a dam site and an alternative solution will be required.
However, in the majority of cases a high-paste-content RCC with a
high proportion of pozzolan in the cementitious content will be feasi-
ble. It is therefore suggested that for an RCC dam of any reasonable
height, the use of a high-paste-content concrete will usually be the
best solution.

Back, P. A. A., Report of an informal meeting (Dec. 1, 1986), BNCOLD Newsletter, Sum-
mer Edition, London, July, 1987.
Boggs, H. L., and Richardson, A.T., USBR Design Considerations for Roller-Compacted
Concrete Dams, in Roller-Compacted Concrete, AWE, New York, 1985.
Dunstan, M. R. H., Rolled Concrete for Dams: Construction Trials Using High-Flyash-
Content Concrete, Technical Note 106. CIRIA. London. May 1981.
- Rolled Concrete for Dams-A Laboratory Study of the Properties of High-
Fly&h-Content Concrete, Technical Note 105, CIRL4, London, May 1981.
- Design Considerations of Roller-Compacted Concrete Dams, Water Power and
Dani Construction (London), January 1986.
Design and Construction Concepts of a High-Paste-Content RCC Dam 155

Lowe, John, III, Roller Compacted Concrete Dams-An Overview, Roller Compacted
Concrete II, AXE, New York, February 1988, pp. l-20.
McConnell, A. D., Discussion to Question 57, ICOLD Congr., 15th, vol. 5, Lausanne,

The Japanese Approach

6.1 Background
Japanese engineers have taken a conservative approach to roller-
compacted concrete (RCC) dam building in order to meet the difficult
challenges posed by the topography, hydrology, and geology of the vol-
canic island nation. Although RCC is used primarily for only the in-
terior of Japanese dams, the design criteria adopted impose the same
strict demands for watertightness, strength, and durability on the
RCC as it does for traditional mass concrete dams.
The intent of Japanese research and development of RCC is to en-
hance the speed and economy of dam construction. The conservative
design approach and the painstaking execution of RCC placement are
more time-consuming and costly than the methods developed for RCC
dams in the United States. Japanese dams employing RCC so far have
proven to be more watertight than RCC dams built in the United
States and other countries.
Production and economy are enhanced somewhat in the Japanese
approach by layering RCC into thicker lifts than are common in the
United States and elsewhere. However, to prevent thermal stress
cracking, Japanese designers require transverse contraction joints ev-
ery 49 ft (15 m) in all RCC dams. They also restrict RCC placement
rates to avoid heat buildup from the hydration of cement. Completed
lifts are cured for as long as four days. All lift surfaces are treated as
cold joints and bedded with mortar before the next lift of RCC is
placed. Inspection galleries are required in all RCC dams in Japan,
and all RCC dams are encased in conventional concrete for water-
A research team organized in 1974 by Japans Ministry of Construc-

158 Chapter Six

IICD tcchquc was refined at Tamagawa Dam.

tion claims to have originated the roller-compaction method as part of

its independent search for rationalized concrete dam construction
techniques. The name given to the method then-roller-compacted
clam, concrete method, or RCD-has been adopted by Japanese engi-
neers to signify Japans pioneering role in the new technology.
Masatane Kokubu, Professor Emeritus at the University of Tokyo,
chaired the Research Committee on Rationalized Execution of Con-
crete Dam Construction. Its 20 members were commissioned to study
various methods of rapid placement of concrete, including the growing
RCD: The Japanese Approach 159

body of international literature and tests done on methods of placing

lean, superstiff concrete mixes compacted with vibratory rollers. The
primary aim of the early research and testing was cost reduction in
concrete dam construction, As test and actual construction data have
accumulated, however, the focus of the Ministry of Construction ap-
pears to have shifted so that the principal objective now is shortening
the construction duration of RCD projects.
A secondary goal of RCD development in Japan is to expand the
number of sites where concrete gravity dams can be built. That would
be accomplished by using an RCD foundation mat to disperse stresses
and increase shear resistance of inferior foundations.
Both goals stem from the urgent need to build more and higher
dams with multiple functions at increasingly difficult and
earthquake-prone sites. As Japans population increases and grows
more prosperous, the demand for flood control and new water supplies
also is increasing. To meet that demand, as many as 300 new dams
will be required to be built in the decade ending in 1995. Most of those
over 100 ft (30 m) high will be concrete gravity dams.
Steep river gradients; frequent, high-volume floods of short dura-
tion; and high population densities along rivers impose severe restric-
tions on dam designers. River diversions are often difficult and costly.
Spillways are generally large for all types of dams. Because of the fear
of overtopping, the safety criteria for fill dams require that they have
spillway capacities 1.2 times greater than that of concrete dams.
These conditions impose significant cost penalties on fill dams, espe-
cially if the spillway is built independently of the embankment. As a
result, while most of the dams being built in the world are fill dams, in
Japan about 70 percent of the 291 dams under construction in 1982
were concrete dams.

6.2 RCD Design

The harsh consequences of a dam failure in Japan-both for the de-
signers and for the cities and towns downstream-have led Japanese
engineers to move very cautiously away from traditional mass con-
crete dam design. Speed and economy are important concerns, but
Japanese dam designers have not been willing to depart significantly
from traditional concrete dam design practice to take full advantage of
the potential savings offered by RCC. The cross section of RCD dams,
the properties of their surfaces, and their watertightness are based on
the same conservative criteria used for conventional concrete gravity
The criteria used in the development of RCD call for the concrete
mix to be plastic and workable for optimum compactibility, yet stiff
160 Chapter Six

enough to support a number of passes with vibratory rollers. Com-

pared to soils approach mixes, RCD is a relatively wet, high-fines-
content material that fits into the concrete approach category. It has a
vibratory compaction WC) time of 20 ? 10 s.
Structural design criteria require that the strength properties of
RCD concrete must be close to those of conventional mass concrete.
Yet the material has been used only for the interior portions of dams,
where stresses are low and durability requirements are slight. In all
of the completed RCD projects and others that are under construction
or design in Japan, the interior mix is encased on all sides in conven-
tional mass concrete cast against forms and vibrated to make it mono-
lithic with the RCD. Upstream facings are thickest, typically 10 ft (3
m) thick. Downstream facings range from 6.6 to 10 ft (2 to 3 m) thick,
and conventional concrete foundation mats are about 5 to 10 ft (1.5 to
3 m) thick. Transverse contraction joints and waterstops are required
for the upstream and downstream faces.

6.2.1 Sliding stability

Sliding stability is ensured by careful attention to lift surface treatment
and bonding of all lift interfaces with bedding mortar. The shear strength
of properly bedded lift joints in RCD dams has been found to be between
one-fifth and one-seventh that of the compressive strength of the parent
material. Japanese dam designers consider that to be sufficient to allow
construction of RCD dams to heights of 500 ft (150 m) and greater.

6.2.2 Uplift
Uplift pressures in RCD dams are controlled to the same degree as
conventional concrete dams by use of a thick upstream facing and by
bedding all lifts. The cross sections of all RCD dams completed to date
have vertical upstream faces and downstream slopes varying from
0.76 to 0.81 H : 1 V with 0.80 H : 1 V being typical. Combined with the
high unit weight of RCD, this section provides sufficient weight to re-
sist overturning forces.

6.2.3 Temperature control

Temperature control measures to prevent cracking from thermal
stress are the most important element in the design of RCD dams.
Temperature control is achieved by restricting the type and amount of
cement used in the mix, precooling mix constituents prior to place-
ment, and adjusting the lift thickness and placement schedule to ac-
commodate temperature extremes. Designers also require transverse
contraction joints spaced at 49 ft (15 m) in all RCD dams.
RCD: The Japanese Approach 161 Tamagawa Dam. In the design of Tamagawa Dam, Japans

largest completed RCD dam, the maximum temperature was limited
to 86F (30C) at the base where the external restraint was greatest,
and to 95F (35C) for the rest of the 328-ft-high (100-m) structure.
The temperature drop at exposed lift surfaces was limited to 9F (5C).
Other temperature control measures taken at Tamagawa Dam

. Replacing 30 percent of the cement with fly ash

n Restricting the cement-fly ash content to 219 lb/yd3 (130 kg/m31
. Restricting lift thickness near the base of the dam to 30 in (75 cm)
as opposed to 39 in (100 cm) for the lifts at higher elevations
n Restricting the placement rate for concrete that was in contact with
rock to two days per lift during cool months and as long as five days
between lifts from July to early September. For interior concrete,
the placing rate was restricted to from two to four days per lift, in-
creasing as the months got warmer (see Fig. 6.1);
H Water cooled to 59F (15C) was sprayed on fresh lifts and mixing
water was cooled to 39F (4C) during the summer season to reduce
the temperature rise in the RCD
. Completed lifts were insulated during winter shutdowns and when
outlet works were being installed in the dam body to prevent rapid

711-8131 3 3 14

9/l-11/30 2 2 10

Figure 6.1 Upper and lower limits for concrete placement at Tamagawa Dam.
162 Chapter Six

To help compensate for the time and expense of bedding mortar

treatment, lift thickness on RCD dams has been steadily increasing.
Lifts as thick as 39 in (100 cm) have been placed. Adequate compac-
tion of the material is achieved by spreading lifts in successive layers
using tracked dozers and by proportioning RCD mixes for the opti-
mum workability suitable for compaction. Specific mixture propor-
tioning methods have been developed by the Ministry of Construction.

6.3 RCD Construction

Research and development work by the Ministry of Construction bore
fruit in the late 1970s with the construction of two RCD demonstra-
tion projects. The first was Shimajigawa, a 292-ft-high (89-m) gravity
dam, where RCD was used for the interior concrete. The second was at
Ohkawa Dam, where an RCD base mat was built to improve the foun-
dation for a 78-m-high (256-ft) concrete gravity dam, its stilling basin
and downstream energy dissipation works.
The 347,000 yd3 (265,000 m3) RCD foundation mat for Ohkawa
Dam was begun in July 1979. The mat, over 1310 ft (400 m) long and
66 ft (20 m) deep, was completed in nine working months at a peak
monthly placement rate of 52,970 yd3 (40,500 m31. The foundation for
a new stilling basin at Shin-Nakano Dam was built in 1980 using
about 14,800 yd3 (11,300 m3) of RCD. After Okawa, however, no new
RCD base mats for conventional gravity dams have been mentioned in
any of the literature on Japanese dams available in English.
Placement of 216,000 yd3 (165,000 m3) of RCD for the body of
Shimajigawa Dam started in October 1978. All of the concrete, includ-
ing the 199,000 yd3 (152,000 m3) of conventional concrete used for the
foundation rock anchorage and for the upstream and downstream
faces, was in place before June 1980. The entire project was completed
in March 1981. That date is the basis for the Japanese claim that
Shimajigawa is the worlds first dam built with RCC.
Experience gained at Shimajigawa and Ohkawa was used for the
design and construction of two other multipurpose dams, Pirica and
Tamagawa. Some 944,000 yd (722,000 m3) of RCD was used for the
body of Tamagawa, the larger of the two. Construction started on that
328.ft-high (100-m), 1448-ft-long (441.5-m) gravity dam in 1980. Con-
creting of the dam began in September 1983 and was completed in
1987. Placement of the total 1.51 million yd3 (1.15 million m3) of con-
crete was accomplished over a total of 24 working months. Concrete
work was halted for four to five months each winter.
RCD for four other dams was completed between 1987 and 1989. RCD
also is being used as the body concrete for four dams under construction
and four others under design in Japan. Among the latter, Miyagase Dam
RCD: The Japanese Approach 163

will be 509 ft (155 m) high and Gassan Dam will by 410 ft (125 ml high.
Figure 6.2 lists all RCD dams built, under construction or design.

6.3.1 Uniform methods

Based on field trials at Ohkawa Dam directed by the Ministry of Con-
struction and on research conducted at the Institute for Construction
Equipment, a draft Guide to Design and Construction of RCD Con-
crete was completed in 1978. It has been updated a number of times
since then. The RCD method that has evolved to satisfy the conserva-
tive design criteria includes these characteristics:

1. The batched concrete is delivered to the dam surface with cranes,

inclines, or other suitable means. The RCD is then loaded into end
dump trucks which are used to place the material in piles on the dam.
End dumps have been used exclusively because of their availability
and the restricted working areas at the generally tight sites.
2. The lifts are placed in lanes from abutment to abutment. Tempo-
rary forms are placed between lanes to keep the working surface level.
At Tamagawa, the 33-ft-wide (10-m) lanes were placed starting from
the upstream side of the dam.
3. The dumped piles are spread by bulldozers in three or five suc-
cessive layers to form one lift. Lift thickness is 20 in (500 mm) near
the base of the dam and 30 to 39 in (750 mm to 1 m) at higher eleva-
tions. Extensive working of the concrete with the dozers is the key to
preventing segregation. The tamping effect of the dozer tracks also
aids the compaction. For these reasons, Kokubu calls the spreading of
the material with bulldozers the secret to the RCD method.
4. Before compaction, transverse contraction joints are cut into the
lifts at 49-ft (15-m) centers in each lane using a backhoe-mounted vi-
bratory cutter. Joints are cut after the material is spread by bulldoz-
ers. Sheets made of galvanized iron or polyvinyl chloride plastic are
inserted as a crack-inducer (Fig. 6.3).
5. Lifts are smoothed by two passes of a self-propelled vibratory
roller operating without vibration followed by 6 to 13 passes with vi-
bration. Final passes are.,made with a rubber-tired roller. For the 27.
in (700-mm) lifts at Shimajigawa, tandem vibratory rollers were op-
erated at 0.62 mi/h (1 km/h). Two passes were made without vibration
and 13 were made with vibration on each lift.
At Tamagawa, the thickest lifts got 12 passes by a tandem vibratory
roller with a deadweight of 7.7 tons (7 metric tons) and a frequency of
2600 cycles per minute. Six passes with a 28.6-ton (26-metric-ton)
rubber-tired roller finished the compaction. Use of the tired roller
Figure 6.2 RCD projects in Japan.

RCC Height Total dam volume C t FA

qo. (Planned) Name of dam OWIW n-l (ft) In3 (yd? kg/m3 lb/yd3

1 1980 Shimajigawa MCJG 89 292 317,000 (415,000) 91 t 39 = 130 152 t 66 = 219

Uppe1 84 t 36 = 120 141 t 61 = 202
2 1982 Shin-Nakano Hokknido LG 75 246 11,300 (14,800)* 84 t 36 = 120 141 t 61 = 202
3 1984 Ohkawa MCJG 78 256 265,000 (347,000) 96 t 24 = 120 162 t 40 = 202
4 1986 Tamagawa MCJG 100 328 1,154,ooo (1,509,000) 91 t 39 = 130 153 t 66 = 219
5 1987 Mano Fukushima LG 69 226 219,000 (286,000) 96 t 24 = 120 162 t 40 = 202
6 1988 Pirika Hokkaido DA 40 131 360,000 (471,000) 84 t 36 = 120 141 t 61 = 202
7 1988 Shiromizugnwa Yamagata LG 55 179 315,000 (412,000) 96 t 24 = 120 162 t 40 = 202
8 1988 Asahiogawa Toyama LG 84 276 361,000 (472,000) 96 t 24 = 120 162 t 40 = 202
9 1988 Nunome WRDPC 72 236 370,000 (484,000) 78 t 42 = 120 131 t 71 = 202

Under construction (end of 1989)

LO (1990) Dodairagawa Gunma LG 70 229 346,000 (453,000) 96 t 24 = 120 162 t 40 = 202

11 (1991) Kamuro Yamagata LG 61 199 298,000 (390,000) 96 t 24 = 120 162 t 40 = 202
12 (1991) Sakaigawa Toyama LG 115 377 712,000 (932,000) 91 t 39 = 130 153 t 66 = 219
13 (1992) Asari Hokkaido LG 74 242 484,000 (633,000) 96 t 24 = 120 162 t 40 = 202

In final design

L4 (1991) Ryumon MCJG 99.5 326 844,000 (1,104,OOO) 91 t 39 = 130 153 t 66 = 219
15 (1991) GaSSaIl MCJG 125 401 1,450 (1,897) 91 t 39 = 130 153 t 66 = 219
L6 (1991) Hattabara MCJG 85 279 500,000 (654,000) 84 t 36 = 120 141 t 61 = 202
17 (1994) Miyagase MCJG 155 509 1,914,ooo (2,503,OOO) 91 t 39 = 130 153 t 66 = 215

*Stilling basin.
Base m a t .
MCJG = Ministry of Construction, Japanese government
DA = development agency
LG = local government
WRDPC = Water Resource Development Public Corporation
RCD: The Japanese Approach 165

Figure 6.3 tialvanizcd crack inducers are vihrated in place at Tamagawa Dam.

eliminated the problem of small cracks discovered near the top of lifts
that were compacted with only vibratory rollers. It also made cleanup
of the lift surfaces easier because there was less loose material to re-
6. After curing, lift surfaces are green-cut with mortar brushes and
water jets. Excess mortar and loose aggregate are removed with motor
sweepers and hand labor. Water jets are used for a final cleaning. The
water is removed and a 0.6-in (15mm) layer of high-slump mortar is
spread over the cleaned lift surface by wheel loader or by hand to bond
it to the next lift of concrete (Fig. 6.4). The bedding also helps to re-
duce segregation at the construction joints. The placing, cutting, and
cleaning of each lane of RCD at Tamagawa took an average of three
days. To prevent loosening of coarse aggregate and washing out of
mortar during green-cutting, the concrete was allowed to cure from 24
to 36 h before water jetting started.
7. Upstream and downstream faces of the dam are formed with con-
ventional immersion-vibrated concrete placed against formwork. To
eliminate construction joints at the interface, the plastic facing mix is
(1) immersion-vibrated with ganged vibrators mounted on a backhoe;
(2) the RCD lane adjacent to the face concrete is placed to its full lift
height and the interface between the two is immersion-vibrated; (3)
the interface area is compacted with six passes of a vibratory roller
166 Chapter Six

Flgure 6.4 Bedding mortar is spread on all compacted lifts

operating in a transverse direction; (4) the RCD is compacted with 12

passes of a roller operating in a longitudinaal direction, Waterstops
and drains are installed at the transverse contraction joints in the face

6.4 RCD Concrete Mixtures

Lift thickness at Tamagawa was increased from earlier projects in or-
der to make the RCD method more economical. The trend is likely to
continue. Effectively compacting the lower portion of deep lifts can be
difficult, however, because of segregation caused by dumping the RCD
from end dumps. Working the mix from piles into layers with dozers
helps to avoid segregation. Nevertheless, Japanese designers consider
the compactibility of the mix to be extremely important in the RCD
A relatively wet mix with a high fines content is used to enhance
compactibility. In general, a water/cement ratio of 70 to 85 percent
and unit cement content of 202 to 219 lb/yd3 (120 to 130 kg/m3) are
being used for RCD concrete (Fig. 6.5). The unit amount of water is
150 to 180 lb/yd3 (90 to 110 kg/m3). As much as 30 percent of the unit
cement content is fly ash. The fly ash retards the set, helps to reduce
thermal stress from the heat of hydration, and enhances the long-term
strength of the concrete.
Figure 6.5 RCD mix designs.

WI Fl Basic Contents (kg/m)

MSA Air, Sla, C + F, C + F, =ct
Dam mm percent percent percent percent A, ma W C F S kgfhm
Ohkawa 80 1.5 2 1 32 85 20 0.573 0.340 102 96 24 691 123
Shimajigawa (upper) 80 1.5 I? 1 34 87.5 30 0.549 0.343 105 84 36 752 130
Shin-Nakano 80 1.5 f 1 34 79 30 0.551 0.330 95 84 36 723 -
Tamagawa 150 1.5 ? 1 30 73 30 0.596 0.356 95 91 39 657 261
Pirika 80 1.5 2 1 30 75 30 0.597 0.396 90 84 36 668 -

Conventional concrete RCD (Tamagawa)

Maximum size coarse aggregate, mm (in) 150 (6) 1 5 0 (6)
Slump, cm (in) 4 r 1 (1.6 + 0.4) 0
Vibrating compaction (VC) value, s - 20 2 10
Water/cement ratio, % 68 73
Sand/aggregate ratio weight, % 25 30
Unit content
Water, kg/m3 (lb/yd) 115 (194) 95 (160)
Cement, kg/m (lb/yd3) 170 (287) 130 (219)
Fine aggregate, kg/m3 (lb/yd? 523 (882) 657 (1107)
Coarse aggregate, kg/m (lb/yd) 1570 (2646) 1544 (2603)
Lift thickness, cm (in) - 75 (30)
a = aggregate,
A = absolute volume of aggregate,
C = cement,
F = fly ash,
p = cement paste/mortar (ratio by volume),
S = sand,
W = water,
o = compressive strength, at 91 days (cylinders)
166 Chapter Six

Well-graded aggregates made from processed rock are generally

used in RCD. To enhance workability, the sand content is high-30 to
34 percent. Coarse aggregate in the RCD mix for Tamagawa was sized
at a maximum of 6 in (150 mm). The maximum at other completed
RCD dams was 3 in (80 mm).

6.4.1 Vibrating compaction value

RCD consistency is measured using an apparatus similar to that used
in a Vebe test. The result is called the vibrating compaction, or VC,
value. It represents the time required for mortar to fill the voids be-
tween coarse aggregate particles and for the paste to rise to the sur-
face of the concrete as a result of compaction, or consolidation, by ex-
ternal vibration.
Mix proportions of RCD and the number of passes with vibratory
rollers are largely determined using this test. The optimum VC value
differs depending on the quality of the aggregates, the mix design of
the RCD, the performance of the vibratory roller, the size of the dam,
air temperature, and other conditions. Low VC values provide con-
crete that can be easily compacted and the VC value drops as the unit
water content rises. Bleeding becomes a problem, however, if the VC
value is too low.
Specifically, for designing mixes with large aggregates in the labo-
ratory a steel cylinder 18.9 in (48 cm) in diameter and 15.7 in (40 cm)
high is filled with RCD. A clear plastic plate is placed on top of the
cylinder and a 44-lb (20-kg) weight is set on top of the plate. The con-
crete is vibrated at 3000 cycles per minute at an amplitude of 0.039 in
(1 mm). The optimum value is obtained in about 60 s using this cyl-
For preliminary studies and quality control testing during construc-
tion, a smaller cylinder, 9.4 in (240 mm) in diameter and 7.9 in (200
mm) high, is used to determine the VC value of concrete passing a 1.6-
in (40-mm) screen. All other procedures are the same as for the large
cylinders. The optimum VC value is obtained in roughly 20 s using
the smaller, standard-size containers. See Fig. 2.12 for a comparison of
VC values using both the large and smaller standard container with
varying water contents.

6.4.2 Proportioning
Because of the desire for a low heat rise and good workability, the unit
cement and water content of the mix are also important. Typically,
the RCD mix is designed by keeping the unit binder content-the ce-
ment plus fly ash-and the unit water content constant. Trials are
RCD: The Japanese Approach 169

A Consistency + S t r e n g t h + Mix
meter test
(Unit water vs. (Sand aggregate (Unit cement
ratio vs. VC content vs.
valve) compressive
I strength)

(Unit water content

I vs.
VC valve

Determine unit
, water content
Figure 6.6 RCD mix proportioning flow diagram.

done with various sand-aggregate ratios until the optimum VC values

are reached. A test section is then built using the same equipment and
procedures as are specified for construction and the mix is adjusted ac-
cording to permeability and strength test results. The flowchart for
the steps used in proportioning RCD is shown in Fig. 6.6.

6.4.3 Compressive strength

Laboratory and field tests of the compressive strengths of RCD at com-
pleted projects indicate good results in all cases (see Fig. 6.7).
Strengths achieved were more than adequate to meet design require-
ments. Long-term compressive strength gains were higher than ex-
pected-at Tamagawa the 36.5day strength was 1.4 times that of the
91-day figure.

6.4.4 Watertightness
Permeability tests done in boreholes at Ohkawa indicate a high de-
gree of watertightness in the lifts and at the bedded joints between
lifts. The results were essentially the same-within a range of 1.7 to
7.4 x 10m6 cm/s.
170 Chapter Six

.$2 2 0 0
3 2
E 100
0 2 8 91 182 365
Age, days

Figure 6.7 Strength gains for RCD mixtures with varying

water contents. (100 k&cm = 1422 lb/in.)

No leakage has been observed on the downstream face of

Shimajigawa Dam since construction was completed in 1980. Seepage
into foundation drains has not exceeded 8 ga.l/min (30 L/min), and
minimal leakage has been observed from contraction joints.

6.5 Benefits
The potential benefits of the RCD method over conventional concrete
dam-building techniques in Japan are lower construction costs; faster
construction; greater worker safety because of uncongested, level
working surfaces; and, because of the possibility of developing more
sites by using an RCD base mat for a conventional dam, greater con-
crete gravity dam design diversity.
Very general comparisons put the overall cost advantage for RCD
over the monolith method in Japan at 10 to 15 percent in ordinary
conditions and at 40 percent for conditions very favorable to RCD, i.e.,
dams in wide valleys with few embedded structures.
Cost savings are realized in labor and materials. Among the inher-
ent advantages of RCD, the number and skill level of crews required
for RCD construction are lower than for conventional concrete dams.
The road-building equipment required to transport, place, and com-
pact the concrete is less costly, more flexibly deployed, and easier to
operate than the specialized materials conveyance systems used for
dams built by the monolith method. In addition, there is a reduction in
the amount of cement used, a large reduction in the total quantity of
forms and no cooling pipes required in the RCD method. At
Tamagawa, cement savings were roughly 10 percent and form savings
were close to 32 percent.

6.5.1 Faster placement

Japanese engineers report that the RCD method is significantly faster
than conventional techniques in Japan. At Shimajigawa, the daily av-
RCD: The Japanese Approach 171

erage RCD placement rate was computed to be 24 percent faster than

the rate would have been if Shimajigawa had been a conventional
dam. The peak monthly placement rate at Ohkawa was 52,970 yd3
(40,500 m3) as compared to an estimated 29,430 yd3 (22,500 m3) per
month for placing the base mat in 6.6-ft (2-m) blocks of conventional
mass concrete.
The time savings at Tamagawa over conventional placement meth-
ods were five to seven months, depending on the type of cable crane
selected in the hypothetical case used for making the comparison.
Work was shut down for as long as five months each winter during the
seven-year construction period. By eliminating one more winter shut-
down, the expedited RCD work saved one year on the overall schedule.
Placement of the 944,000 yd3 (722,000 m3) of RCD used for the body
of Tamagawa Dam took 24 working months spread out over four
years. By comparison, 433,000 yd3 (331,000 m3) of RCC was placed by
crews working two g-hour shifts, six days a week at Willow Creek
Dam in less than five months in 1982.
Problems with shutting down for the winter and restarting the RCD
work in the spring account for some of the long construction duration
at Tamagawa. But the RCD method is inherently slower than other
methods of placing RCC. The long curing times, green-cutting and
cleaning of each lift, placement of a bedding mix over the entire sur-
face of each lift, cutting of contraction joints and the consolidation of
the RCD and facing concrete at the interfaces also helped to stretch
out the schedule. Funding delays due to the dependence on annual ap-
propriations from the Finance Ministry may also have contributed to
the relatively slow progress at Tamagawa.

6.5.2 Research and development

For future projects, Japans formidable construction research estab-
lishment is aiming at improving the efficiency of joint treatment, RCD
placement around internal structures, concrete conveyance, and the
bonding of conventional facing concrete with the RCD body mix. A
method of controlling concrete temperature and the compaction mech-
anism of vibratory rollers also are being studied.

Development Division, River Bureau, Ministry of Construction, Development in Japan
of Concrete Dam Construction by the RCD Method, 1984.
Dunstan, M. R. H., Design Considerations for Roller-Compacted Concrete Dams, Wa-
ter Pouw & Dam Construction, January 1986.
Harada, J., T. Okada, S. Shimada, and T. Yamaguchi, Construction of Tamagawa Dam
by the RCD Method, ICOLD Congr., 15th, Lausanne, 1985
Hirose, T., Research and Practice Concerning RCD Method, ICOLD Congr., 13th, Rio
de Janeiro, 1982.
172 Chapter Six

Hirose, T., and S. Takebayashi, Present State and Problems of Rationalized Construc-
tion of Concrete Dams, Concrete Library of Japan Society of Civil Engineers, no. 2,
December 1983. Hirose, T., and T. Yanagida, Dam Construction in Japan: Burst of
Growth Demands Speed, Economy, Concrete I n t e r n a t i o n a l , May 1984.
International Commission on Large Dams (ICOLD), Roller-Compacted Concrete for
Gravity Dams-State-of-Art, Bulletin 75, 1989.
Kokubu. M.. Develonment in Jaoan of Concrete Dam Construction bv the RCD
Methbd, text of speech, ICOLD hongr., 14th, Tokyo, 1984.
Nagataki, S., T. Yanagida, and T. Okumura, Construction of Recent RCD-Concrete
Dam Projects in Japan, Roller-Compacted Concrete, ASCE, New York, May 1985,
pp. 90-101.
Research Committee on Rationalized Execution of Concrete Dam Construction, Tech-
nical Guide to RCD Construction Method, (Draft, 1981), Technology Center for Na-
tional Land Development, August 1981.
Shimizu, S., S. Jojima, and T. Yanagida, The Development of the RCD Method in Ja-
pan, Water Power & Dam Construction, January 1986.
Takahi, K., The Outline of Design of RCD Dams, presented at Roller-Compacted Con-
crete II, ASCE, San Diego, March 1988 (not published in proceedings).

RCC Applications
in Embankment Dams

7.1 General Considerations

Roller-compacted concrete (RCC) has many potential applications in
all types of dams. The material and method are not limited to the con-
struction of concrete dams in general or to gravity dams in particular.
The basic properties of RCC-high compressive and shear strength,
low permeability, and high erosion resistance when compared to
nonstabilized materials-together with its rapid placement and rela-
tively low cost have led to many applications in embankment dams.
RCC has been used for various functions in the original design of
embankments as well as for repair, modification or replacement of ex-
isting fill dams. Use of RCC for all the above applications is discussed
in this chapter.
Many of the early applications of RCC were part of the design for an
embankment dam. While these early applications were called
rollcrete, soil-cement, or cement-stabilized fill at the time, they can
now be termed roller-compacted concrete. Although many of the fol-
lowing examples of RCC use in embankments are site-specific, the de-
sign fundamentals upon which the solutions were based are more gen-
erally applicable. Figure 7.1 provides a list of RCC applications in
embankment dams.

7.2 Original Design

The varied uses of RCC in the original design of an embankment dam
illustrate the versatility of the material and construction method.
RCC has been used for foundation improvement, upstream slope pro-
tection, the central core, spillways, and downstream overtopping pro-
Figure 7.1 RCC applications in embankment dams.

Year RCC Max. height V&me, MSA, in C t FA,
completed Name of dam country, state OWW/engilWX dam, ft. (ml yd3(m3) (mm) lb/yd3/(kg/m3)

Original design

Foundation improvement

1968 Cochiti U.S.A., U.S. Army Corps of Engineers, 255 58,500 3 167 + 0
New Mexico Albuquerque District (78) (44,700) (75) (107 t 0)
1988 Mount St. Helens U.S.A., U.S. Army Corps of Engineers, 184 15,000 3 320 t 80
sediment retention Washington Portland District (56) (11,500) (75) (190 t 47)

Upstream slope protection

1963 Ute U.S.A., State of New Mexico/ 121.5 22,600 3 245 t 0

New Mexico Bechtel Corp. (37) (17,300) (75) (145 t 0)
1983 Ute (raising) U.S.A., State of New Mexico/ 132.5 16,000 1 290 t 0
New Mexico U.S. Bureau of Reclamation (40) (12,200) (25) (172 t 0)
1988 Jackson Lake U.S.A., U.S. Bureau of Reclamation 39 45,000 1YZ 308 t 0
(rebuilding) Wyoming U.S. Bureau of Reclamation (12) (34,400) (38) (183 t 0)
1989 Mount St. Helens U.S.A., U.S. Army Corps of Engineers, 184 64,000 3 320 t 80
sediment retention Washington Portland District (50) (48,900) (75) (190 t 47)

Central Core

1961 Shihmen cofferdam Taiwan Shihmen Reservoir Administration 210 20,000 3 142 t 89
Authorityflippetts-Abbett- (64) (15,300) (75) (84 t 531
McCarthy-Stratton (TAMS)
1970 Karun cofferdam Iran Ministry of Energy/ 164 13,000 4 169 t 0
Campenon Bernard (50) (10,000) (100) (100 t 0)
1975 Al Massira cofferdam Morocco LEtaWCampenon Bernard 66 2490 4 169 t 0
(20) (1900) (100) (100 t 0)
1982 Holbeam W o o d U.K., Southwest Water Authority/ 39 5360 1% t 160 t 0
Devon SWW Land Drainage Group (12) (4100) (40) (95 t 0)
Spillway--gravity section
1984 Dolet Hills U.S.A., Southwestern Electric Power Co./ 35 26,100 350 + 175
Louisiana Freese & Nichols (11) (20,000) (208 t 104)
1986 Bucca Weir Australia, Queensland Water Resources 39 24,300 lY? t 152 t 152
Queensland Commission/Gutteridge, (12) (18,600) (40) (90 t 90)
Haskins & Davey
1988 Stacy (now S.W. U.S.A., Colorado River Municipal Water 156 116,700 1% 210 t 105
Freese) Texas Dist./Freese & Nichols (48) (89,200) (38) (125 t 52)
Overtopping protection
1979 Revelstoke cofferdam Canada, B.C. Hydro 150 9900 1Ys 284 t 0
British B.C. Hydra (48) (7600) (38) (169 t 0)
1984 North Loop (2 dams) U.S.A., Trammel1 Crow Co./ 35 & 32 20,700 3 200 t 80
Texas Freese & Nichols (11 & 10) (15,800) (75) (119 + 47)
1985 Great Hills U.S.A., Trammel1 Crow Co./ 41 13,000 lY2 246 t 98
Texas Camp Dresser & McKee (12) (9,900) (38) (146 t 52)
1986 Bucca Weir Australia, Queensland Water Resources 39 7,100 1Ya 152 t 152
Queensland Commission/Gutteridge, (12) (5400) (40) (90 t 90)
Haskins & Davey
Composite darn
1986 Arabie (now Mokgoma South Africa, Lebowa Dept. of Development/ 118 132,000 3 61 t 125
Matlala) Lebowa Theron, Prinsloo, & van Tender (36) (101,000) (75) (36 + 74)

Outlet tunnel restoration

1975 Tarbela (abutment Pakistan, Pakistan Water & Power Develop- 470 460,000 9 187 t 0
restoration-tunnel 2) Abbottabad ment Authority/TAMS (143) (350,000) (230) (111 t 0)
Spillway plunge pool, stilling basin repairs, etc.
1980 Tarbela (lining of Pakistan Pakistan Water & Power Develop- 470 1,180,OOO 6 250 t 0
service spillway ment Authority/TAMS (143) (902,000) (150) aver.
plunge pool) (148 t 0)

Figure 7.1 Roller-Compacted Concrete Applicators in Embankment Dams. 6htinued.J

Year RCC Max. height V&me, MSA, in C + FA, lb/
completed Name of dam country, state Owner/engineer dam, It. (In) yd3tm3 (mm) yd3/Ck@m31
1980 Tarbela (cofferdam for P&iStan Pakistan Water & Power Develop- 470 60,000 6 250 + 0
auxiliary spillway merit Authority/TAMS (143) (36,060) (160) aver.
plunge pool (148 t 0)
1981 Tarbela (lining of Pakistan Pakistan Water k Power Develop- 470 1,230,OOO 6 260 t 0
auxiliary spillway ment Authority/TAMS (143) (940,000) (150) wer.
plunge pool) (148 t 0)
1982 Tarbela (backfill of PakiStan Pakistan Water & Power Develop- 470 110,060 6 260 t 0
tunnel 4 stilling ment Authority/TAMS (143) @4,000) (150) aver.
basin) (148 t 0)
1983 Tarbela (lining of PEkiStan Pakistan Water & Power Develop- 470 96,000 6 250 t 0
tunnel 4 plunge pool) ment AuthorityflAMS (143) (73,000) (160) aver.
(148 t 0)
1983 Tarbela gravity wall- Pakistan Pakistan Water & Power Develop- 470 175,000 6 250 t 0
powerhouse area) merit Authority/TAMS (143) (134,000) (150) aver.
(148 t 0)
1986 Tarbela (gravity wall Pakistan Pakistan Water & Power Develop- 470 200,000 6 250 t 0
extension-power- ment Authority/Charles T. Main (143) (153,000) (150) aver.
house area) (148 t 0)
Spillway Replacement
1980 North Fork Toutle River U.S.A., U.S. Army Corps of Engineers, 38 18,660 1Ya 600 t 0
debris retention struc- Washington Portland District (12) (13,800) (38) (297 t 0)
Overtopping protection

1980 Ocoee 2 U.S.A., Tennessee Valley Authority 30 4450 % -

Tennessee (9) (34001 (19)
1984 Brownwood Country U.S.A., Brownwood Country Club/ 19 1400 1% 310 - type
Club Texas Freese & Nichols (61 (10701 (381 IP
(184 - IP)
1986 Spring Creek U.S.A., Colorado Division of Wildlife/ 50 4840 1% 225 + 0
Colorado Morrison-Knudson Engineers (15) (37001 (38) (133 t 0)
1986 Harris Park 1 U.S.A., Harris Park Water & Sanitation 18 2300 1% 285 t 0
Colorado District/Edward Shaw (5.51 (17601 (381 (169 t 0)
1988 Addicks & Barker U.S.A., US. Army Corps of Engineers, 48.5 & 36.5 56,700 1% 292 t 244
Texas Galveston District I;15 82 111 (43,350) (381 (173 t 1451
1988 Commanche Trail U.S.A., City of Big Spring/ 30 6500 1% 248 t 63
Texas Freese & Nichols (9) (48701 (38) (147 t 371
1989 Goose Lake U.S.A., City of Boulder/ 28 4600 3 360 t 0
Colorado Harza Engineering (8.51 (35201 (751 (214 t 0)
1989 Boney Falls U.S.A., Mead Paper Co./ 22 4850 % 217 t 165
Michigan Harzo Engineering (71 (37101 (191 (129 t 981

I Replacement
1985 Kerrville Ponding U.S.A., Upper Guadalupe River Authority/ 21 22,000 3% 200 t 0
Texas Espey-Huston & Assoc. (6.41 (16,800) pit run (119 t 01
1986 Dryden (2 dams) U.S.A., Chelan Public Utility District I/ 14 4,200 1 400 t 100
Washington CHPM-Hill (4.31 (3210) (251 (237 t 59)
1986 Cedar Falls U.S.A., Seattle City Light/ 34 5500 2 185 t 155
Washington R. W. Beck & Assoc. (10.41 (42001 (501 (110 t 921
1989 Marmot U.S.A., Portland General Electric/ 48 10,300 1% 120 t 180
Oregon Ebasco Services (14.61 (7870) (381 (71 t 1071
178 Chapter Seven

tection in the design of new embankment dams. In many cases,the ba-

sic equipment used to place and compact the embankment material
has also been used for placement of the RCC.

7.2.1 Foundation improvement

At Cochiti Dam, what was then called cement-stabilized fill was used
as foundation support for a portion of a triple-barrel concrete outlet
conduit. A suitable conglomerate sandstone foundation material was
found to underlie about two-thirds of the length of the outlet works.
However, a variable thickness of shaly clay for about 800 ft (244 m)
toward the downstream toe was found to be too weak to support the
conduit which ultimately would be buried beneath 210 ft (64 m) of em-
bankment fill.
The U.S. Army Corps of Engineers, Albuquerque District, decided
to replace the shaly clay with cement-stabilized fill. The specification
for the work, completed in 1968, stated the intent of this stabilized
sand and gravel backfill is to provide foundational material for the
reach of this conduit with physical properties similar to the sandstone
bedrock. The resulting RCC could thus be termed artificial bed-
Laboratory tests on seven of the most representative samples of the
foundation sandstone indicated an average unconfined compressive
strength of 1190 lb/in2 (8.2 MPa). A 2%day compressive strength of
1000 lb/in2 (6.9 MPa) for the RCC was considered adequate by the
Corps. A cement content of 5.2 percent by dry weight of the maximum
3-in (75mm) aggregate (180 lb/yd3 or 107 kg/m3) was determined to
satisfy the minimum compressive strength criteria.
Tests indicated a permeability coefficient of 5 x 10e6 cm/s after
seven days of curing for a cement content of about 7 percent by dry
weight of aggregate. This compared to 3.9 x lo-* cm/s for a com-
pacted, well-graded sandy gravel with no cement, or a seepage de-
crease of nearly two orders of magnitude. A 15-ft (4.6-m) head of water
was trapped behind the stabilized fill for more than a year with no
through seepage observed through the trapezoidal section.
Construction of the RCC foundation support was typical of later ap-
plications using continuous pugmill-type mixing and 12-in (300-mm)
lifts. Compaction to a minimum of 100 percent standard Proctor den-
sity was achieved by eight passes of a heavy, rubber-tired roller when
the moisture content of the material was between the optimum 8 per-
cent water content and 2 percent wet of optimum. Due to poor control
of the water content at the mixing plant, it was necessary, at times, to
add water prior to compaction on the fill.
The mass RCC, totaling 58,500 yd3 (44,700 m3), shows little evi-
RCC Applications in Embankment Dams 179

dence of cracking. Inspection of a 5ft-deep (1.5-m) trench required for

the construction of a seepage collar on the conduit showed no visible
layering of the RCC.

7.2.2 Upstream slope protection

Soil-cement has been used for upstream slope protection for earth
dams for more than 25 years, primarily in the United States. On-site
sands are stabilized with cement to form a continuous erosion-
resistant facing that competes with rock riprap. Most of the time fine
sands are used, with little material greater than l/4 in (6 mm).
The first dam constructed with soil-cement slope protection, Ute
Dam in 1963, had relatively large diameter sand and gravel deposits
available at the site. No rock suitable for riprap was within economi-
cal haul distance, so the contractor decided to use the 3-in (75mm)
minus, well-graded sandy gravel to produce large-aggregate soil-
cement, or RCC.
Based on freeze-thaw durability tests, a cement content of 7 percent
by dry weight of aggregate was selected for construction. Compacted
at an optimum moisture content of 7 percent, the material exceeded
1000 lb/in2 (6.9 MPa) compressive strength at seven days. Cores
drilled from the facing 21 months later averaged more than 1700 lb/
in (11.7 MPa) (see Fig 7.2).
The 22,600 yd3 (17,300 m3) of large-aggregate soil-cement was
placed in g-in-thick (X0-mm) stair-stepped horizontal layers on the
upstream face of the top 41 ft (12.5 m) of the embankment. The design
provides for a 2-ft-thick (0.6-m) facing measured perpendicular to the

Figure 7.2 Core sample from

Ute Dam.
180 Chapter Seven

Construction was similar to the mass placement for Cochiti Dam ex-
cept that a dual windrow spreader was used to place the outer cement-
treated material as well as an adjacent untreated pervious zone.
The face of what now would be termed RCC was designed to protect
the dam from wave action. It was not subjected to reservoir water un-
til 1983, when a concrete labryinth spillway was constructed to raise
the level of the lake to coincide with the original soil-cement slope pro-
tection. Nevertheless, the facing has been subjected to numerous
freeze-thaw cycles since 1963 with little noticeable deterioration. In
many areas, the imprint of the rubber-tired roller remains on the
When the spillway was raised, an additional 16,000 yd3 (12,200 m3)
of coarse aggregate soil-cement (RCC) was required to protect a dike
at the left abutment. Using basically the same material source, the
Bureau of Reclamation decided to use a l-in (2.5mm) minus material
rather than the 3-in (75-mm) material used in the original construc-
tion. This resulted in an increase in cement content to 9 percent by
dry weight in order to achieve equivalent durability and compressive
A concrete approach was used by the U.S. Army Corps of Engi-
neers, Portland District, to design the RCC upstream slope protec-
tion for the Mount St. Helens Sediment Retention Structure. Resis-
tance to abrasion and erosion from high volcanic sediment loads in
the Toutle River was the controlling design criterion rather than
freeze-thaw durability. The RCC was placed in 2-ft-thick (0.6-m)
lifts and required 320 lb of cement and 80 lb of class F fly ash per
cubic yard (190 kg cement and 47 kg fly ash per cubic meter) to
meet the compressive strength specification of 3000 lb/in2 (20.7
MPa) at 28 days (see Sec. 7.3.3).

7.2.3 Central core

The first example of the use of RCC for the central core of an em-
bankment was for the 210-ft-high (64-m) Shihmen cofferdam in Tai-
wan in 1960. The section is shown in Fig. 7.3 as finally constructed.
When the weathered sandstone originally planned for the core be-
came too wet in the stockpile, the designers Tippetts-Abbett-
McCarthy-Stratton (TAMS) decided to switch to what they termed
rollcrete so that construction could continue.
A 3-in (75mm) maximum size aggregate was used. The RCC was
batched and mixed in the concrete plant used for the conventional
concrete required for the project. Most of the material was placed
using one bag of cement plus one bag of flyash per cubic yard of
RCC (94 lb/yd3 of cement plus 89 lb/yd3 of fly ash, or 56 kg/m3 of
El. (rn)

El. 190



-160 River Cobble Gravel

Boulder cobble

ilty river cobble gravel
ter in original design
-140 pervious core in otginal design


Figure 7.3 Cross-section of Shihmen Dam.

182 Chapter Seven

cement plus 53 kg/m3 of fly ash). A mix containing 1.5 bags of ce-
ment and one bag of fly ash per cubic yard was used at the begin-
ning of construction.
Using a 14-in-diameter (360-mm) mold and a compactive effort
equivalent to modified Proctor, the 2.5-bag mix at optimum moisture
of 5 percent by dry weight produced a 2%day compressive strength of
2000 lb/in2 (13.8 MPa) in the laboratory.
The RCC was placed in 12-in (300-mm) lifts. Originally, 50-ton
(45 metric-ton) rubber-tired rollers were planned to be used for
compaction. Instead, the dump trucks used to haul the material to
the cofferdam started the compaction, which was finished by sev-
eral passes of a D-8 crawler tractor used to spread the material. The
final properties of the core are unknown because no permeability
tests were done and no cores were extracted after completion of con-
The use of RCC for the central core of the upstream cofferdam at
Shihmen led designers Campenon Bernard of Paris to design similar
RCC cores for the cofferdams at Karun Dam in Iran in 1970 and AI
Massira Dam in Morocco in 1975. The 164-ft-high (50-m) cofferdam at
Karun was located between the approach channel to the diversion
tunnels and the main dam. At Al Massira, a 66-ft-high (20 ml coffer-
dam was needed to allow construction of three concrete buttresses to
the right of the diversion channel. Lack of space for construction of a
clay-core rock fill led to the selection of RCC for the core of the coffer-
dam, which was subjected to a head of water during the entire period
of construction of the buttresses.

7.2.4 Spillways
The use of RCC for foundation improvement, upstream slope protec-
tion, and the central core of an embankment dam is based on devel-
opmental work done in the 1960s. However, it was not until the initial
work on RCC gravity dams was completed in the early 1980s that the
use of RCC for embankment dam spillways came into use.
Generally, RCC spillways are gravity overflow sections located in
the central portion of a longer earth embankment. Their design is
identical to that for an RCC gravity section, taking into account the
flow over the structure. In each of the three spillway projects de-
scribed here, galleries and foundation drains were not included in the
designs. Hence, the sections had to be designed to accommodate max-
imum uplift pressures.
The scheduling of construction of RCC spillways in relation to the
adjacent sections of embankment is an important consideration. It is
preferable to raise the spillway and embankment together for con-
RCC Applications in Embankment Dams 183

struction access considerations. Care must be taken to fully compact

the earth sections against completed RCC structures in order to min-
imize the potential for seepage at the interface.
RCC spillways have been constructed for the Dolet Hills Dam in
Louisiana; Bucca Weir in Queensland, Australia; and for Stacy Dam
in Texas. The three quite different sections are shown in Fig. 7.4.
Bucca Weir and the Dolet structure are both uncontrolled spillways,
while Stacy Dam incorporates a large gated spillway. In each case, the
decision to use RCC was based on its cost advantage over a spillway
constructed entirely of conventional concrete.

Spillway face RCC

Interior mass RCC


RCC protection
--.. ._.-
El. 16.2
FI 1.52

Laver of beddina 1
Conventional maSs
concrete facing

rO.6 steps

concreteNom: 0.3 T-
,-Reinforced El.
, 20 RCC l i f t s =
c I -&riq drains&/ &, 4&E El. 5.3
- I _-
z Layer of bedding I
concrete over I
foundation surface
s. 23.3 -I
Figure 7.4 RCC spillways. (a) Dolet Hills Dam; (b) Bucca Weir. (1 ft = 0.3m.j
164 Chapter Seven

El. 1564.0

Interior pier -

NWL El. 1551 .O

Crest El. 1528.0

El. 1517.25

El. 1469.0

Figure 7.4 RCC spillways. (c) Stacy Dam. (1 ft = 0.3m.I Khztinued) Dolet Hills Dam. The RCC spillway at Dolet Hills was de-
signed by consultants Freese and Nichols. It has a crest length of
about 200 ft (61 m) and is located near the left abutment of the 2600-
ft-long (790-m) embankment which impounds a makeup water reser-
voir for a coal-fired powerplant near Mansfield, Louisiana. Because of
the gradual slopes of the upstream and downstream flow surfaces, no
forming was necessary to construct the basic 35-ft-high (10.7-m) RCC
The spillway section is zoned so that a richer RCC mixture was used
for the exterior 6 ft (1.8 m) measured perpendicular to the slope. A
plant-mixed mortar of two parts sand and one part cement was placed
as a bonding mix between lifts at the exterior zones of the section. The
dry mix was hand-broomed over the previously placed RCC lift surface
and then sprinkled with water to form a mortar prior to placement of
the next lift.
The earth embankment was raised at the same time as the spillway
in order to provide access to the RCC work area. Once the RCC con-
struction procedure was established, the contractor had to increase
the rate of fill placement on the embankment to keep up with the pace
of concreting.
RCC Applications in Embankment Dams 185

A conventional reinforced-concrete stilling basin and training walls

were integrated into the spillway design (see Fig. 7.4~). The spillway
concrete was battered on an 0.8 H : 1 V slope where it tied into the
earth embankment.
The exposed RCC surfaces on the spillway have been subjected to
frequent flows since construction was completed in 1985. Little ero-
sion of the exposed surfaces has occurred. Bucca Weir. The 39-ft-high (12-m) Bucca Weir is about the
same height as the Dolet Hills spillway, but it presented a different
problem for design consultants Gutteridge Haskins and Davey. Lo-
cated between riverbanks, the 426-f&long (130-m) RCC overflow sec-
tion is more than one-half of the structures total length of 735 ft (224
Earthfill sections abutting the central RCC weir are 8 ft (2.5 m)
higher than the concrete. The earthen sections are designed to be
overtopped by a 1-in-3-year flood. To accommodate this overtopping,
the downstream slope of the earthfill sections and downstream
riverbanks are protected with a lo-ft-wide (3-m) drained zone of RCC.
For floods with an annual exceedance probability of 1 in 10 years or
greater, the entire structure becomes submerged due to the buildup of
backwater. See Sec. for the design of the section.
Stair-stepped conventional concrete placed concurrently with the
RCC is used on the 0.5 H : 1 V downstream slope of the spillway. The
1 H : 1 V upstream face is unformed. A conventional concrete cap on
the crest is designed for the future attachment of a 2-ft-high (610-mm)
inflatable rubber flash board.
A 6.6-ft-wide (2.0-m) zone of bedding concrete was placed between
RCC lifts at the upstream face to reduce seepage. The conventional
concrete facing on the downstream slope is drained by a network of
strip drains made of polypropylene fiber encased in a geotextile.
A 2-ft-wide (610-mm) zone of noncemented RCC aggregate placed
between the RCC spillway and earthfill end sections was included in
the design to provide support for the RCC during construction.
During construction, a flood overtopped the central RCC weir by 3.0
ft (0.9 m). The material was not damaged and the flood was passed
without incident. Stacy Dam. The 568-ft-wide (173-m) spillway for Stacy Dam
(now S.W. Freese Dam) in Texas differs from the uncontrolled Dolet
Hills and Bucca Weir spillways in that it is a gated structure which
combines both overflow and nonoverlow sections. Located in the cen-
tral portion of a 2-mi-long (3.2-km) embankment, the six-bay spillway
has a capacity of 216,000 ft3/s (6120 m3/s).
Both the overflow and nonoverflow sections require different solu-
186 Chapter Seven

tions using RCC. For the overflow ogee section, the RCC as shown in
Fig. 7.4~ is used as interior mass capped with a minimum thickness of
8 ft (2.4 m) of anchored conventional concrete.
Designers Freese and Nichols devised a unique solution for seepage
control for the 103-ft-high (31.4-m) nonoverflow gravity sections. Bor-
rowed in part from the design of the upstream deck of a concrete-faced
rockfill dam, an H-in-thick (460-mm) reinforced and waterstopped
conventional concrete face was specified. See Sec. and Fig. 3.10
for a description of the upstream face.

7.2.5 The composite dam

One variation on the use of an RCC spillway in the central portion of
an embankment dam is a composite dam. In this case, an RCC gravity
section extends from one abutment to meet an earthfill section placed
against the other abutment. The spillway, outlet conduit, and other
waterways are located in the RCC portion, while the embankment
acts only as a water barrier.
The composite design is based on the concept that an earthfill sec-
tion without appurtenant structures would cost less to build than a
solid RCC gravity section. The design also allows placement of the
concrete structure on whichever side of the valley has the better foun-
dation conditions.
From a construction standpoint, the composite dam allows place-
ment of the RCC portion independently of the earthfill section due to
access from one abutment. Arabie Dam (now Mokgoma Matlala Dam)
in Lebowa, South Africa, is an example of a composite dam. A 1493-
ft-long (455-m) RCC section consisting of both overflow and
nonoverflow portions is located at the right abutment, while a 2526-
ft-long (770 m) earthfill section starts from the left abutment. A 656-
ft-long (200 m) fuse plug spillway 13 ft (4 m) higher than the RCC sec-
tion is located to its right.
Erizana Dam in Spain is a different type of composite dam in that
the main structure is a 14%ft-high (45-m) conventional concrete grav-
ity dam 591 ft (180 m) long, while a 50-ft-high (15-m) dike 361 ft (110
m) long at the left abutment is constructed of RCC.

7.2.6 Overtopping protection

The erosion resistance of RCC allows the material to be used in sev-
eral ways to allow embankment dams to be safely overtopped by
floods. RCC can be used to protect the downstream slope of an earth or
rockfill dam, as a cap on the crest of an embankment, or as a down-
stream gravity section.
RCC Applications in Embankment Dams 187

Am of 2nd - stage upstream cofferdam

Welded wire fabric

Figure 7.5 Revelstoke cofferdam crest detail. Revelstoke cofferdam. The rockfill cofferdam built to allow

construction of the 2700-MW Revelstoke hydroelelctric project in Brit-
ish Columbia is a major structure in itself (see Fig. 7.5). It rises 150 ft
(48 m), has a crest length of 420 ft (128 m) and is zoned in much the
same way as a permanent fill dam.
An RCC cap was used for the uppermost 12 ft (3.6 m) of the coffer-
dam. The entire downstream slope was protected by a grid of steel re-
inforcing bars and wire mesh which were anchored into the rock fill
every 4 ft (1.2 m). The cofferdam embankment settled more than its
rigid RCC cap, creating a void averaging about 3/4 in (19 mm) be-
tween the two materials. The void was filled with grout and some con-
No flood of sufficient magnitude to overtop the cofferdam occurred
during construction of the project, so the RCC design was not tested. Austin detention dams. Three dams for the same owner at
Austin, Texas, had basically the same design criteria. Each had to ac-
commodate a four-lane roadway on the crest, retain a l-in-loo-year
flood, and be able to be safely overtopped for any event up to the prob-
able maximum flood.
The design solution by consulting engineers Freese & Nichols for
two of the dams, called the North Loop detention dams, consisted of a
combination earthfill and RCC gravity section as shown in Fig. 7.6.
The upstream zoned earth portion provided the required crest width
for a four-lane roadway while the downstream RCC section provided
the overtopping protection. The 29-ft (8.8-m) height of the dams was
chosen to store the loo-year flood.
Esthetics were important for the dams, which are part of a business
park development. Precast concrete panels were used to form the
188 Chapter Seven

E x i s t i n g ground l i n e

Figure 7.6 North Loop Detention Dam, No. 1. (Note: 1 ft = 0.3 in.)

downstream face of the RCC gravity sections and also to form a three-
level continuous planter across the dams. Vandals have spray-painted
graffiti on the panels, partially ruining the intended effect.
For the third Austin dam, built at Great Hills at the Arboretum,
consulting engineers Camp, Dresser & McKee faced a similar design
problem. In addition to the hydraulic and roadway requirements, the
owner wanted a permanent lake to enhance the development.
An upstream earth embankment would have taken up too much
room in the proposed reservoir at the cramped site. So the designers
decided to compact select fill between two RCC gravity sections, one
having a vertical upstream face and the other a vertical downstream
face. A chimney drain of coarse material was placed behind the up-
stream RCC section to intercept seepage, which was carried to a
drained crushed-rock blanket and perforated drain at the low point in
the select fill (see Fig. 7.7). RCC was also used as a lining for the lake.

.Y Select - -
it= embankment J=Ell
/G P -

24 drainage blanket 6 chimney

8-in drain pipe

Figure 7.7 Great Hills at the Arboretum Dam. (Note: 1 ft = 0.3 m; 1 in = 2.54 cm.)
RCC Applications in Embankment Dams 199

None of the three dams has been overtopped since they were com-
pleted in 1984 and 1985. Concrete-faced rockfilldams. The use of RCC on the down-

stream slope of concrete-faced rockfill dams provides overtopping pro-
tection during construction. Steel reinforcing of the rock fill has been
used to prevent construction period failures, but the reinforcing is
slow and expensive to install.
The first such use of RCC for this purpose was at Xingo Dam, a
463-ft-high (140-m) concrete-faced rockfill dam in Brazil. Four 52-
ft-diameter (16-m) diversion tunnels and a 164-ft-high (50-m) cof-
ferdam were built to pass a 1-in-30-year flood. To ensure that the
dam could pass a l-in-200-year flood without failure from
overtopping, a zone of RCC with drains was placed across the dam
on the downstream toe.

7.3 Rehabilitation

The fast placement of large volumes of RCC for extensive repairs at

Tarbela Dam in Pakistan made a major impact on the dam-building
community. A total of 3.5 million yd3 (2.66 million m3) of RCC was
used for rehabilitation of the worlds largest engineered embankment
after problems arose following its initial filling in 1974. The Tarbela
remedial work falls into two categories: Repair of the fill around tun-
nel 2 and rehabilitation of the service and auxiliary spillway plunge
pools and adjacent areas.
The work at Tarbela was followed by use of RCC to replace a failed
spillway for a debris detention dam built after the Mount St. Helens
volcanic eruption in Washington State. RCC has also been placed on
the downstream slope at a number of existing embankments to allow
for overtopping by floods.

7.3.1 Tarbela tunnel repairs

About 250 ft (76 m) of the upstream section of the outlet tunnel 2 col-
lapsed during the initial filling of the Tarbela reservoir in 1974. The
failure of the 45-ft-diameter (13.7-m) tunnel was attributed to cavita-
tion from uncontrolled flows caused by a malfunction of the tunnels
intake gate. In addition to the fractured concrete tunnel lining to-
gether with some of its steel reinforcement and supporting ribs, nearly
1 million yd3 (765,000 m3) of embankment material collapsed into the
tunnel and was swept down the Indus River.
The three remaining tunnels were used to draw down the 6.9 mil-
lion acre-ft (8.5 billion m3) reservoir a total of 328 ft (100 m) in 26
190 Chapter Seven

days, The next flood season was due in six months. During that period,
it was necessary to rebuild the damaged tunnel and to provide side
support for the two adjacent tunnels, 1 and 3.
The material used for the rehabilitation of the eroded rock fill had
to have the ability to support itself and have the properties of a low-
strength concrete. It also had to be placed very quickly. RCC, or
rollcrete, was selected by Tippetts-Abbett-McCarthy-Stratton
(TAMS), the owners design consultant.
A total of 460,000 yd3 (350,000 m3) of RCC was placed in 42 working
days with a peak daily placement of 24,000 yd3 (18,400 m3). In the re-
construction of the tunnel it was possible to reduce the diameter to 36
ft (11 m) because the larger bore was no longer needed for diversion of
the river during construction. The work was completed in time to
safely pass the spring floods.
There was a wide variation in the compressive strength of the RCC
fill material. A g-in (230-mm) minus pit-run aggregate with a poorly
graded sand portion was used in the mix. The strength variation was
caused by known variations in cement and moisture content, aggre-
gate grading, and probably compactive effort as well.
With a cement content in the two-bag range (188 lb/yd3 or 112 kg/
m3), the IO-day compressive strength varied from a low of 1260 lb/in2
(8.7 MPa) to a high of 4970 lb/in2 (34.2 MPa). Thirteen 6-in (150-mm)
cores were taken and averaged 2530 lb/in2 (17.5 MPa).

7.3.2 Tarbela plunge pool modifications

Tarbela Dams massive service and auxiliary spillways are located on
the left bank of the river immediately downstream of the 470-ft-high
(143-m) embankment. Starting in the summer of 1977, backflow
caused erosion of the service spillway plunge pool and threatened to
undermine its flip bucket.
RCC was used to construct a massive buttress, groin, and strut to
provide both erosion resistance and slope stability in the plunge pool
area. Most of the RCC was faced with 5 ft (1.5 m) of conventional con-
crete, but a buttress or strut in the plunge pool area was left exposed.
The repaired area was subjected to a test flow of 400,000 ft3/s
(11,300 m3/s) for six hours. An inspection and soundings taken imme-
diately following the test indicated no erosion of the protective works,
including the exposed RCC strut.
Similar mass placements of RCC were used to build a cofferdam and
stabilize the plunge pool area for the auxiliary spillway; to backfill the
stilling basin for tunnel 4 and line its new plunge pool; and to build a
large gravity wall for the powerhouse area in two stages.
RCC Applications in Embankment Dams 191

7.3.3 Spillway replacement

Following the eruption of Mount St. Helens in 1980, the U.S. Army
Corps of Engineers constructed a 3%ft-high (12-m) debris retention
dam on the North Fork of the Toutle River in Washington State. The
embankment was designed for a five-year life and was built to keep
the Columbia River tributary free of sediment and debris from the vol-
canic eruption. One month after construction was completed, high
flows together with rocks, trees, and ash had eroded the shotcrete-
coated gabion spillway and caused its failure.
Faced with building a replacement spillway quickly and economi-
cally, the Corps chose steel-mesh-reinforced RCC for the 300-ft-wide
(91-m) spillway. The 4%ft-thick 11.3-m) RCC slab was supported on
fill and consisted of a 120-ft-long (36.6-m) downstream apron leading
to a 4 H : 1 V sloped section that flattened to horizontal at the crest.
Steel sheetpiling upstream and downstream were the only forms
used. A total of 18,000 yd3 (13,800 m31 of RCC was placed in 60 h over
a six-day period. Mass RCC sections with an exterior slope of 0.7 H : 1
V served as the side training walls. Welded wire fabric was placed
close to the top and bottom surfaces of the RCC spillway.
For the first six months, normal river flows passed over the exposed
RCC at velocities of 15 to 20 ft/s (4.6 to 6.1 m/s). The retaining dam
then filled with sediment and rocks up to the spillway crest. The de-
bris, including rocks as large as 1 ft (0.3 m) in diameter, was passed
over the spillway for the next five months.
Another eruption of the volcano in March 1982 brought masses of
mud down the river, overtopping and breaching the embankment on
both sides of the RCC spillway (Fig. 7.8). An inspection of the spillway
at this time showed that the 6-in-thick (150-mm) RCC cover had
eroded down to the mesh in one location. The remaining surfaces had
been ground down to a terrazzo-like finish. A B-in-deep (150-mm)
groove at a cold construction joint at the center of the spillway was
also in evidence.
Following the failure of the embankment, the Corps decided to aban-
don the spillway and mud-filled reservoir. Instead, it built a 184-ft-high
(56-m) gravelfill sediment retention structure between 1987 and 1989.
RCC is used on that dam for the upstream slope protection and to fill in
a depresssion in the foundation rock downstream of the spillway.

7.3.4 Overtopping protection

Many embankment dams in the United States and elsewhere are un-
able to pass current design floods safely. RCC has been used on the
downstream slope of a number of these embankments to allow
192 Chapter Seven

Figure 7.8 North Fork of Toutle River Dam after overtopping.

overtopping during high flows. RCC can be placed quickly without

taking the dam out of service. A large number of applications similar
to the ones described will be constructed in the future. Ocoee No. 2 Dam. Ocoee No. 2 Dam, a 30-ft-high (9.1-m)

rockfilled timber crib dam near Benton, Tennessee, was the first dam
to use RCC to provide increased safety during overtopping. When the
Tennessee Valley Authortity (TVA) started to rehabilitate the 67-
year-old timber-crib dam in 1980, it placed a rock riprap berm on the
downstream slope to improve the stability of the structure. After four
washouts of the riprap from flash floods, the TVA was forced to adopt
another repair method.
The new solution was to place RCC in a stair-step manner over the rip-
rap. Since the rehabilitation was completed, the dam has been subjected
to periodic planned overtopping to accommodate white-water rafters
downstream. The well-compacted RCC remains undamaged by water
and weathering. Some uncompacted RCC mix placed at the downstream
toe has been eroded, however, and the compacted layers adjacent to the
eroded areas have been undercut by about 1 ft (0.3 m) in some areas.
RCC Applications In Embankment Dams 193 Brownwood Country Club Dam. This dam, in Brownwood,

Texas, was also modified to ensure safe overtopping during flood
conditions. The 19-ft-high (5.8-m) earth embankment was con-
structed in 1938 with a 65-ft-wide (19.8-m), 2600-ft3/s (74 m3/s) ca-
pacity spillway at its right abutment. The modification was neces-
sitated by an updated probable maximum flood calculation of
11,600 ft3/s (328 m3/s).
The initial spillway modification alternatives considered by consult-
ing engineers Freese & Nichols were estimated to cost about $225,000.
The engineers then devised a scheme to widen the spillway by 300 ft
(91 m) by lowering the embankment by 5 ft (1.5 m) and placing RCC
in stair-steps on the downstream slope. The RCC was keyed into the
foundation shale to prevent backcutting following overtopping and the
earth fill was drained behind the RCC face. The total rehabilitation
cost $72,000 and took two weeks to complete with the 1400 yd3 (1070
m3) of RCC placed in two days.
Following completion in 1984, the new spillway has been tested a
number of times with flows up to 1-ft (0.3-m) deep with no prob-
lems. Spring Creek Dam. With a structural height of 53 ft (16.2 m),

Spring Creek Dam, north of Gunnison, Colorado, is the highest dam to
date to use RCC on its downstream slope to accommodate overtopping.
The updated probable maximum flood of 11,000 ft3/s (311 m3/s) greatly
exceeded the capacity of the existing concrete chute spillway in the
central portion of the dam.
The design by Morrison-Knudsen Engineers included raising a por-
tion of the dam, flattening the downstream slope, and lining the crest
and downstream face with a 3-ft (0.9-m) layer of RCC measured per-
pendicular to the slope. A unique feature of the design was to turn the
8-ft-wide (2.4-m) horizontal layers of RCC downstream as they ap-
proached the left abutment. This formed a training wall or groin to
control the overtopping flow (Fig. 7.9).
Completed in 1986, the modification has not been tested to date.

7.4 Replacement
Due to failure, damage, or aging, a number of embankment dams need
to be replaced each year. The design and construction of an RCC re-
placement dam is no different than for a new dam, with the possible
exception that the old embankment may be used as a cofferdam dur-
ing construction.
194 Chapter Seven

Figure 7.9 Spring Creek Dam RCC overtopping protection.

7.4.1 Kerrville Ponding Dam

The 21-ft-high (6.4-m), 59%ft-long (182-m) Kerrville Ponding Dam in
the channel of the Guadalupe River in Texas was built as a concrete-
capped clay embankment in 1980. A 198-ft-long (60-m) portion at the
left abutment was depressed by 1 ft (0.3 m) to act as a service spillway.
The entire embankment had been designed to be overtopped under
flood conditions.
However, a portion of the dam in the spillway area was washed out
in late 1984 when a flood overtopped the embankment by 10 ft (3 ml.
After evaluating several alternatives, including repair of the existing
structure, consulting engineers Espey-Huston & Associates, working
with geotechnical consultants Rone Engineers, decided to construct a
new RCC dam immediately downstream of the damaged embankment
as shown in Fig. 7.10.
The downstream portion of the dam was removed and the undam-
aged upstream embankment was left in place to act as a cofferdam.
RCC was placed against the remaining downstream section and built
up to the height of the original dam. A 2-ft-thick (0.6-m) conventional
concrete slab extending 20 ft (6.1 m) downstream protects the toe of
the RCC section from erosion during overtopping.
A 3.5-in (89-mm) maximum-size pit-run aggregate with 10 percent
cement by dry weight of aggregate was used at the crest and base of
the new structure with a 5 percent mix in between. The richer mix
RCC Applications in Embankment Dams 195

P -i

6 in


Figure 7.10 Cross section of Kerrville Ponding Dam in service spillway area.

produced an average compressive strength in excess of 2100 lb/in

(14.5 MPa) at 28 days. A sand-cement mortar was brushed on each lift
near the upstream face to aid in bonding the 1-ft-thick (0.3-m) layers.
Thirty days after the RCC section had been completed, it was sub-
jected to a severe test. Heavy rains in October of 1985 caused the dam
to be overtopped by as much as 14.4 ft (4.4 m). The entire dam was
overtopped for more than four days and water flowed over the spillway
for almost three weeks. Except for washing away uncompacted mate-
rial on the surface of the broad-crested weir, there was no noticeable
erosion of the exposed RCC (see Fig. 7.11).
In July 1987, the dam was subjected to an even greater flood, which
sent 162,000 ft3/s (4590 m3/s) over the entire RCC structure. This lOO-
year event overtopped the dam by a maximum of 16.2 ft (4.9 m).
Again, no noticeable erosion or other distress resulted.

7.4.2 Dryden Weir

Dryden Weir on the Wenatchee River in Washington State consisted
of two timber crib dams built around 1900. The upstream dam is 10 ft
(3 m) high and the downstream dam is 14 ft (4.3 m) high. They are
about 230 ft (70 m) apart and are connected by a reinforced concrete
buttress wall founded on a rock ridge down the middle of the river. An
inspection indicated that the old dams were of questionable stability
to continue to provide a dependable irrigation water supply for nearby
apple orchards.
The solution by consulting engineers CH2M Hill called for construc-
tion of two RCC gravity sections immediately downstream of the crib
Chapter Seven

Figure 7.11 Kerrville Ponding Dam after October 1985 flood.

dams. The two structures combined required a total of 4200 yd3 (3210
m3) of RCC.
Working around the clock, the contractor completed the smaller up-
stream RCC section in 24 hours, took a day off, and then built the
downstream dam in 46 hours.

7.4.3 Cedar Falls Dam

Cedar Falls Dam is part of a two-dam multipurpose project built in the
early 1900s 40 mi (64 km) southeast of Seattle, Washington. A 1979
inspection indicated that modifications needed to be made to the
project to be able to pass the new probable maximum flood. In addi-
tion, the existing timber crib dam was deteriorated and leaking.
Feasibility studies completed by R.W. Beck & Associates for reha-
bilitating or replacing the crib dam indicated a new 34-ft-high (10.4-
m) dam constructed of RCC would solve the problem economically. It
was also required that the selected alternative be constructed during a
30-day low-reservoir period and that it require minimal repairs
should the cofferdam be overtopped during construction. A detailed re-
view of constructability of the various alternatives also favored RCC,
and it was therefore selected for final design.
Even with several shutdowns due to rain, placement of the 5500-yd3
(4200 .m3) of RCC capped with conventional concrete was completed in
RCC Applications in Embankment Dams 197

14 days during the fall of 1986. More information on design of the

gravity dam, which is supported on a nonrock foundation, is presented
in Chap. 3. A section of Cedar Falls is shown in Fig. 3.3.

Bush, E. G. W., Rollcrete in Revelstoke Cofferdam, Proceedings of the Construction
Industry Research and Information Association (CIRLA) International Conference on
Rolled Concrete for Dams, London, June 1981.
Cannon. R. W.. Laboratorv Investieations and Full-Scale Trials bv the Tennessee Val-
ley Authority, Proceed&Is of t[e CIRIA International Confeience on Rolled Con-
crete for Dams, London, June 1981.
Catanach, R. B., and T. N. McDaniel, Cement-Stabilized Fill for Conduit Support,
Journal of the Soil Mechanics and Foundation Diuision, AXE, vol. 97, no. SM6, June
1971, pp. 959-963.
Chao, P. C., Tarbela Dam-Problems Solved by Novel Concrete, Civil Engineering,
AXE, December 1980.
Cooke, J. B., RCC Use With the CFRD Dam, ICOLD Sixteenth Congress, San Fran-
cisco, June 1988, vol. V, pp. 465-468.
Forbes, B. A., Roller-Compacted Concrete-Development in Dams in Australia, Con-
crete Institute of Australia, Queensland Branch, Water Retaining Structures Semi-
nar, September, 1986.
Hansen, K. D., and J. W. France, RCC: A Dam Rehab Solution Unearthed, Civil En-
gineering, AXE, September 1986.
Johnson. H. A.. and P. C. Chao. Rollcrete Usaae at Tarbela Dam. Concrete
InteriationakDesign and Constkction, vol. 1, no. ?l, November 1979.
Launay, J., Soil Cement and Roller-Compacted Concrete: Rollcrete, Conference at
LPPE, Casablanca, October 1986.
Leakage No Problem at New RCC Dam, Highway & Heavy Construction, July 1985.
Lemons, R., A Combined RCC and Reinforced Concrete Spillway, Roller-Compacted
Concrete II, AXE, New York, February 1988, pp. 51-60:
Lowe. J.. III. Use of Rolled Concrete in Earth Dams. Discussion to: Utilization of Soil-
Cement as Slope Protection for Earth Dams, ASCE First Water Resources Engineer-
ing Conference, Omaha, May 1962.
- Roller Compacted Concrete Dams-An Overview, Roller Compacted Concrete
II, AXE, New York, February 1988, pp. l-20.
Morsman, D. E., L. E. Lawler, and J. R. Seimears, Construction of Two Spillways Us-
ing Roller-Compacted Concrete, Roller-Compacted Concrete, AXE, New York, May
1985, pp. 62-70.
Portland Cement Association, Soil-Cement Slope Protection at Ute Dam-A Photo-
graphic Construction Report, Chicago, 1964.
RCC Dam Survives Texas Flood, ENR, April 24, 1986.
Reeves, G. N., and L. B. Yates, Jr., Simplified Design and Construction Control for
Roller-Compacted Concrete, Roller-Compacted Concrete, ASCE, New York, May
1985, pp. 48-61.
Rollcrete Repair is Quick, ENR, March 19, 1981.
Verigin, W. M., Benson, S. A. and Carney, M. J.: Cedar Falls Roller-Compacted Con-
crete Dam, Roller-Compacted Concrete II, ASCE, New York, February 1988, pp 39-

RCC as a New
Construction Method

8.1 General Construction Considerations

There are considerable differences between vertical, block-by-block
construction of dams with conventional mass concrete and horizon-
tal, lift-by-lift construction of dams with RCC. The construction
plant for RCC is usually, less costly, and utilized at a higher capac-
ity factor than that required for conventional concrete gravity
dams. But the fast pace of placement required to achieve the econ-
omies possible with RCC imposes a greater discipline on the plan-
ning, scheduling, and coordination in the field between designer
and contractor.
The RCC design approaches described in Chaps. 4, 5, and 6 diverge
in some construction procedures. In all cases, however, the mixing,
transporting, spreading, and compacting of relatively large volumes of
material are concentrated into a short time interval. In the United
States, the time limit from mixing plant to final compaction is about
45 min. It is about twice that in Japan. In addition, the Japanese RCD
method requires that lifts be cured for as long as 36 h before place-
ment of the next layer of RCC. In the United States and elsewhere,
the goal has nearly always been to come as close as possible to contin-
uous placement of RCC. Production rates for ten RCC dams are given
in Fig. 8.1.

8.1 .l Scheduling
Construction scheduling for RCC is less forgiving than for mass con-
crete. Although some mix designs can have greater tolerances than
conventional concrete, RCC is still a sensitive concrete and the critical

Figure 8.1 Time of placement and production rates for RCC dams.

RCC volume, Days per Shifts No. & type Maximum daily
Dam yda (m3) week (no. x h) mixing units Total time rate, yds (m3)
Willow Creek 433,000 6 2x8 2 drums < 5 months 5800
(331,000) (4460)
Copperfield 183,000 I 2 x 10 2 pugmills < 4 months 3400
(140,000) (2600)
Middle Fork 55,000 I 2 x 10 1 drum 45 days 2000
(42,100) (- 10 days planned (1530)
Galesville 210,000 7 2 x 10 1 pugmill 10 weeks 7500
(160,000) (- 10 days planned (5700)
Grindstone Canyon 114,500 7 2 x 10 1 drum 55 days ( - 4 days 4800
(87,500) rain shutdown) (3670)
Monksville 289,000 6 2 x 10 1 pugmill ~5 months 7800
(221,000) (5960)
Upper Stillwater 1,471,OOO 7 2x8 2 drums and 10 months 11,000
(1,127,OOO) 2 pugmills (over 3-year period) (8410)
Lower Chase Creek 26,000 7 2 x 10 1 drum 23 days -
Elk Creek 348,000* 7 2 x 10 4 pugmills >3 months 12,390
(266,000) (in 2 placement (9470)
Stagecoach 44,500* 7 2 x 10 1 drum 37 days 2260
(34,000) (- 2 days shutdown) (1730)
*Plus conventional concrete face or faces.
RCC as a New Construction Method 201

path for placing RCC is focused on completing one lift at a time. There
are no alternate monolithic blocks to form or place while problems are
being analyzed.
To avoid costly delays over disputes on specification compliance, the
authority to resolve engineering problems must be vested in a well-
trained field organization. Lines of communication between the
project engineer and contractor must be clearly established and used
frequently. Clarity, candor, and cooperation are very important in
achieving the speed of placement possible with RCC.
Thorough field and classroom training of construction crews and in-
spectors in the handling and placement of RCC is essential. Lean, dry
RCC mixes resemble damp gravel and are often placed by road-
building or earth-dam crews using earthmoving equipment. However,
all the desired characteristics of the product are those of concrete.
Meeting the specifications requires educating workers to the specific
requirements of RCC as a new construction material requiring spe-
cific construction methods.
Because so much depends on fast, efficient placement of the RCC,
all related activities such as foundation cleanup, access, assembly
of embedded components, and stockpiling of materials must be me-
ticulously planned and scheduled well before construction starts.
Construction of forms and assembly of embedded items should be
planned and scheduled so that as much of the work as possible is
done off the dam, or if necessary, from the top of a lift during shift
Lift scheduling is often complicated by concerns over heat buildup.
Because the production rate of RCC can be very high, controls to limit
thermal cracking may restrict the season or time of day that place-
ment is allowed as well as the rate of placement.
At Upper Stillwater Dam, all of the RCC was placed during a five-
month weather window each year from May to October and only dur-
ing two 8-h shifts starting at 8 p.m. This was necessary in order to
keep placement temperatures below the 50F (10C) specification
limit. At Willow Creek and Galesville dams, placement was limited to
no more than three and four lifts, respectively, each 24 h.
Conversely, at Elk Creek and Stagecoach dams, the contractor was
encouraged to place RCC as fast as possible. At peak placement, six
1-ft (0.3-m) lifts were placed in a day at Stagecoach. The 150-ft-high
(46-m) dam required 44,500 yd3 (34,000 m3) of RCC which was placed
during 37 days in the summer of 1988 (see Fig. 8.2). To keep place-
ment temperatures low at Elk Creek, the specifications required that
the most massive sections be constructed during the late winter and
early spring.
Chapter Eight

8.2 Aggregate Production and Plant Layout

Maintaining adequate supplies of acceptable aggregate is particularly
important to RCC scheduling. More than half of the aggregate re-
quired for an entire construction season may need to be stockpiled
well ahead of the start of construction in order to keep up with ex-
traordinarily high demand during the placement season This can also
provide some easing of cash flow and scheduling requirements during
the placement season. Large stockpiles also allow for more economical
sizing of production facilities and for blending of material that may be
out of specification. If precooling of aggregate is desired to keep place-
ment temperatures low, stockpiling in the winter provides that oppor-
Continuous raw feed to stockpiles is necesssary throughout the con-
struction season to keep up with RCC production. At Upper Stillwater
Dam, the largest in the United States, the contractor had to produce
9000 tons/day (8160 metric tons) of 3/4- and 2-in. (19- to 50-mm) aggre-
gate from soft sandstone to keep up with its RCC placement schedule
of 7200 yd3/day (5500 m3/day) during the final construction season in
RCC as a New Construction Method 203

The location, size, and shape of aggregate piles must be coordinated

with the concrete plant location and method of feed. If conveyors are
not used, several large loaders may be required to feed the mixers on
large RCC projects. To achieve RCC production of 900 yd3/h (690 m3/
hl, for instance, four 12-yd3 (9 m3) front-end loaders could be needed to
allow for a reasonable interval between loadings. The haul distances,
dumping procedures, and turnarounds need to be planned carefully to
operate efficiently and safely.
At Elk Creek Dam, the aggregate was produced from an andesite-
basalt outcrop about 3000 ft (910 m) upstream and to the right of the
dam. Rock was hauled to the primary jaw crusher, located in the
quarry, using 85-ton (77-metric-ton) end dump trucks. Rock was
crushed to g-in (230-mm) maximum sizes. The aggregate was trans-
ported from the quarry down a 12 percent slope by a 56-in-wide, 2085-
ft-long (1.4 x 635 m) conveyor belt rated at 1000 tons/h (907 t/h) and
stored in a surge pile using a radial stacker. A series of cone crushers
and screens then produced 3-in (75-mm), 1%in (3%mm), and %-in (19-
mm) aggregate sizes. The aggregate was transferred from these stock-
piles by three 7-yd3 (5.4 m31 front-end loaders into movable hoppers
discharging onto five conveyors which fed the batch plant surge piles.
From there, the material was loaded from reclaim tunnels onto 36-in-
wide (910-mm) conveyor belts and transported to the RCC and conven-
tional concrete batch plant aggregate bins. The Elk Creek layout is
shown in Fig. 8.3.
Three concrete mixing plants about 300 ft (90 ml upstream of the dam
were used to produce the bedding grout, conventional concrete, and RCC
for Elk Creek. Conveyors were used to move the mix from the plants to
the dam and deposit it on either side of a regulating outlet structure.
The concrete plant layout and location should be selected to mini-
mize energy requirements whether the RCC is transported to the dam
by conveyor or haul vehicles. The intent should be to minimize haul
distances, vertical lift, and exposure of the RCC to sun, wind, or rain.
If trucks or other vehicles are used for hauling, the plant should be
located on a raised, free-draining area. This avoids making mud and
helps to prevent tracking of foreign material onto the dam.
Fueling and field maintenance of equipment should not be done on
the RCC surface because of the likelihood of contamination and con-
sequent problems in establishing a secure bond between lifts at that
point. Fuel spills were a problem at Monksville Dam because the con-
tractor was allowed to refuel on the dam. That potential problem was
addressed at Elk Creek Dam by specifying that the contractor use a
movable refueling pad on the surface of the dam to collect and contain
fuel spillage. The contractor chose instead to refuel equipment and
perform maintenance off the surface of the dam.
204 Chapter Eight

Figure 8.3 All conveyor delivery at Elk Creek Dam. Each conveyor line delivers both
conventional and roller concrete. (Each line has a separate smaller conveyor for con-
ventional concrete:) The conveyor is self-rising on the ventricle pipe stands.

8.3 Mixing RCC

In most cases, the methods used for transporting, spreading, and com-
pacting RCC will not affect production as much as the speed and eff-
ciency of mixing, Thus, the mixing plant capacity should exceed the
laydown capacity.
Achieving the desired product consistency and quality at continu-
ous, high production rates requires good concrete plant design and rig
orous maintenance. RCC mixes are relatively harsh and sticky, and
the lean, dry material has no fluid properties. The design of mixers,
transfers, and hoppers must take these characteristics into consider-
ation in order to avoid caking and loss of capacity.
The mixing method must produce a homogeneous mixture of the in-
gredients. This factor determines the mixing time and, to a large ex-
tent, the production rate. Tests should be done to determine retention
times required for each mix. The variable-speed pugmill mixers used
during 1987 at Elk Creek Dam required 33 s of mixing for 6-yd3 (4.5
m3) batches. At Willow Creek Dam, the retention time in the 9-yd3
(6.8 m3) drum mixers was 75 s.
The trend in design is to use one RCC mix for the body of the dam. For
large projects that require a variety of mixes, the plant chosen for the job
must be able to change mix designs quickly and with little or no mechan-
RCC as a New Construction Method 205

ical or manual manipulation to plant components. On most projects, dif-

ferent plants are used to produce conventional concrete and RCC.
For major projects the weights of all ingredients should be digitally
recorded as a function of time, date, and mix design. Because of the
sensitivity of RCC to excess water, the plant should be equipped with
instrumentation to determine the fine aggregate moisture content. In
Japan, automatic water-batching systems are adjusted according to
results of continuous monitoring of the moisture content of sand in the
sand storage piles. This system has been used on most Japanese RCD
projects and has resulted in precise control of moisture content. In the
United States most contractors feel continuous visual inspection of the
RCC placement or a Vebe test is a more reliable control of water con-
tent than moisture meters in the aggregate piles.
Proper blending or ribboning of the aggregates and cementitious
material on the charging belt as they are fed into the mixer will help
to speed mixing time and avoid buildup of the sticky material. Achiev-
ing the proper timing and angle of introduction of water into batched
mixtures is also important. The proper sandwiching of material to
achieve best results with, RCC often is different from conventional
mass concrete mixes. Each plant and mix combination has its own
quirks, so the exact method of ribboning mix constituents can only be
determined by trial and error.
High-speed batching and mixing can best be achieved with individ-
ual weigh systems for aggregate. Accumulative weigh systems make
it more difficult to fine-tune the ribboning of constituents on the belt.

8.3.1 Mixing plants

Various combinations of batch and continuous concrete plants using
drum and pugmill mixers have been used to produce RCC.
Continuous-mix plants can provide higher output capability than
batch plants, and the most sophisticated continuous mixers can pro-
duce the same degree of control as batch plants. Land use and labor
requirements are generally less for continuous plants.
Drum mixers have the advantage of using less energy than pugmills
and provide good control of batch proportions. Pugmills are faster and
are generally more portable than drum mixers. Pug-mills have been used
on all RCC dams in Japan. Pug-mills need to be carefully designed to avoid
maintenance problems and excessive wear on paddles and plates, partic-
ularly for mixes with large aggregates, In drum mixers, excessive build-
up of the sticky material can occur, reducing capacity and mixing
efficiency. Redesign of fins has helped to avoid this problem in some cases.
Considerable cleanup of the drums on a daily basis is often required.
For high-volume production, larger-than-average-capacity drum
206 Chapter Eight

mixers may be necessary to cope with problems encountered when

mixing no-slump RCC. Blending a zero-slump mixture with a high
proportion of fines often causes long mix and discharge times.
At Willow Creek, a four-bin Noble 600 plant with two 9-yd3 (6.8m3)
Erie Strayer drums was used. The plant had been proven on an earlier
highway job at 600 to 750 yd3/h (460 to 570 m3/h). For RCC, mixing
times were 30 s longer than for the fluid highway paving mix, and the
drums had to be derated to between 7% and 8 yd3 (5.7 and 6 m3) be-
cause the diameter of the discharge end was increased to speed dump-
ing. On average, the plant produced 400 yd3/h (310 m3/h) of RCC and
peaked during one shift at 438 yd3/h (335 m3/h).
F o u r 250-yd3/h ( 1 9 0 m3h) IHI-Hydam pugmills m a d e b y
Ishikawajima Construction Machinery were used by the Japanese
contractor for mixing batched RCC at Elk Creek Dam. The double-
shaft pug-mills were hydraulically driven and allowed for variable ro-
tation speeds during the charging, mixing, and discharging processes.
The coarse aggregate was crushed basalt sized from % to 3 in (19 to 75
mm) maximum. Batching was computerized using Erie-Strayer con-
trols. The maximum output of the four 6-yd3 (4.6-m3) mixers was 1014
yd3/h (775 m3/h). That was achieved on October 29, 1988.
Two different batch-type mixing plants were used to produce RCC
at Upper Stillwater. The first plant built at the site was a Noble 600
with two 8-yd3 (6-m3) tilting drum mixers. A second plant, a Noble
600 modified to accept two 4-yd3 (3 m3) Nikko high-intensity, spiral-
flow pugmill mixers, was added later. Each plant achieved peak pro-
duction rates of about 420 yd3/h (320 m3/h) in 1986. Abrasive quartz-
ite sandstone aggregate with a silica content of about 80 percent
created extraordinary maintenance problems. Problems were greater
with the Nikko bottom-discharge pug-mill mixers, which suffered from
high wear rates on the paddles and breakage of the liner plates.
For the drum mixers, rubberized plastic liners were installed and
these worked well. The high-wear steel liner plates on the Nikko mix-
ers had to be replaced five times during the 1986 season, however.
That was due mainly to breakage caused by the shock of loading and
mixing RCC. Initially, the contractor switched to milder steel liner
plates. They broke at about the same rate as the high-wear plates but
they cost less and were more readily available.
For the 1987 season, a belt-drive system was installed on the
Nikko mixers to allow a 10 percent reduction in paddle speed. Re-
tention times were kept the same so that wear was reduced some-
what. In addition, a G-in-thick (9.5-mm) layer of ll,OOO-lb/in (76
MPa) epoxy-silica grout was troweled onto the backs of the liner
plates before the sixth set of plates was bolted to the mixer wall.
The grout is commonly used as a coating for nuclear reactor con-
RCC as a New Construction Method 207

tainments and has the beneficial characteristics of being durable

but relatively soft. Placing it between the plates and the mixer wall
solved the breakage problem.
RCC production peaked in 1987 during the first week in June at an
average of 548 yd3/h (419 m3/h) during two production shifts per day.
Production from both plants during the entire month averaged 515
yd3/h (394 m3/h) over 59 shifts. The RCC batching and delivery system
produced 266,000 yd3 (203,000 m3) in June 1987. The dam was topped
out on August 8, 1987.
Two ARAN ASR-200 continuous-mix pugmill plants were used suc-
cessfully to produce 183,000 yd3 (140,000 m3) of RCC placed over 17
weeks at Copperfield Dam in Australia. Crushed alluvial gravel with
a maximum size of 2.5 in (63 mm) was used as aggregate. Each plant
produced a different mix, one with cement and fly ash for interior con-
crete and one with only cement for the exterior faces. Slow progress in
jump-forming the 330-ft-long (100-m), central overflow spillway con-
trolled RCC production rates on much of the dam. Hence, neither of
the self-contained, mobile mixing plants was used to capacity al-
though both were needed in order to produce the different mixes si-
multaneously. The average hourly output was 290 yd3 (220 m3) for the
first plant and 234 yd3 (179 m3) for the other.
A contractor-designed continuous-mix plant produced consistent,
high-quality RCC at Monksville Dam. Mixer proficiency test results
are shown in Fig. 8.4. The plant consisted of three B-yd3 (6-m3) feed
bins with adjustable-speed belts straddling a 36-in-wide (914-mm)

allowable Actual
variability variability
Test Average index index*

Cement Content of full mix, 106 70 88

Ib/yd3 (kgim3) (62.9)
Water content of full mix % by 6.1 75 81
Coarse aggregate content of full 49.6 80 88
mix, % by weight retained on
#4 sieve
Unit weight of air-free mortar, 141.2 85 96
lb/ft3 (kgirn? (2262)

*Average of all tests

Figure 8.4 Monksville Dam RCC mixer proficiency test results. [From Snider and
Schrader (1988j.I
208 Chapter Eight

main feed belt; a 75ton (68-t) cement silo with aerators and one vane
feeder; a Davis pugmill with a capacity of 500 yd3/h (382 m3/h); a
stacker belt; and an 13yd3 (14 m3) two-stage surge bin.
The plant could be operated manually or by a computer pro-
grammed to initiate the production sequence starting at the surge
hopper. Sensing devices included a feed-belt scale; tilt gages on each
feed bin; an electronic water flow meter; and a revolution counter for
metering cement through the calibrated vane feeder. Continuous dig-
ital display was available for real-time measurement of aggregate
feed. For aggregate, cement, and water, the system provided cumula-
tive data every 5 min.

8.4 Transporting RCC

Transport of RCC can be by scraper, conveyor, bottom- and rear-dump
trucks, large front-end loaders or a combination of these. Continuous,
high-speed conveyors appear to be the most desirable method for large
RCC jobs. Scrapers have worked well on most medium-size projects,
and rear dumps have been used succesfully on a variety of jobs. Seg-
regation of the larger aggregate usually occurs with end dumps, and
attention must be directed to this potential problem.
Spreading and leveling can be done with a spreader box, bull-
dozer, front-end loader, scraper, or end dump. Wheeled equipment
is generally inferior to tracked equipment, which generally moves
faster and does not tear the RCC surface or cause precompaction of
small areas under spinning tires. Dozers are best for high-
production-rate jobs.
A contractor-built spreader box mounted on a large shovel frame
was used with limited success to place 210,000 yd3 (161,000 m3) of
RCC at Galesville Dam. Dump trucks were used to fill the 16-yd3 (12-
m3) spreader box. The box could place RCC in lanes from 13 to 19 ft
wide (4 to 5.75 m) and to uncompacted depths of between 15 and 17 in
(38 to 43 mm).
Smooth-tired 16-yd3 (12 m3) rock trucks modified with job-designed
air-actuated spreader boxes attached to the tailgate were used suc-
cessfully at Upper Stillwater to dump and spread RCC into H-in (460-
mm) layers. Fine grading was done with a laser-controlled D4 dozer.
The end dumps were filled on the dam by a conveyor system.
End dumps have been used exclusively in Japan for transporting
RCC. Segregation problems have been addressed to some extent in the
mix design. What segregation does occur is rectified by hand labor and
the Japanese method of dumping RCC in piles and spreading several
RCC as a New Construction Method 209

layers of the material with dozers before compacting the layers into a
thick lift.
Whatever transport method is used, the intent should be to get the
material in place quickly, as close to its final location in the lift as
possible and with little rehandling or exposure to weather. If haul ve-
hicles are used, bottom-dump trailer trucks or large-capacity, wheeled
tractor-scrapers offer best results. Because of the higher unit weight of
RCC, weight rather than the volumetric capacity of the vehicle will
probably control the amount of material hauled per trip. If batch mix-
ing is used, matching of the capacity of the haul unit to the batch mix-
ers is critical.
Bottom dumps, because of their short discharge drop distance, min-
imize segregation. Scrapers force the mix out of the bowl and place it
in a relatively uniform layer that requires little secondary spreading.
Scrapers have good maneuverability, top load well, and the bowl un-
loads with few problems. But their basic design is for digging and
rough hauling. Consequently, a scraper with the same horsepower as
a truck will not haul material as efficiently. Scrapers are more mobile
than bottom-dump trailer trucks in difficult terrain, however, and
have better distribution of stress under the tires.
Haul roads need to be laid out, constructed, and maintained to limit
damage from turning and to prevent contamination of the RCC lift
surface from mud, old RCC, or cement spillage and any other foreign
material. Tire cleaning may be required as vehicles approach the dam.
In many cases, access roads near the dam need to be surfaced to pre-
vent tracking of the material onto the lift. The use of clean, crushed
rock for the haul roads at Galesville Dam kept the tires of the haul
fleet clean.
At Monksville Dam, the designers originally specified that RCC
hauled to the dam from surge hoppers near the dam or entirely by con-
veyor belt to avoid tracking of contaminants onto the dam. Instead,
the contractor was allowed to use scrapers to haul RCC from the batch
plant to the point of placement provided that:

. All vehicles entered or left the dam surface in a straight line via a
lane parallel to the axis and immediately adjacent to the down-
stream slope.
n Contaminated haul roads were cleaned immediately by a full-time
crew working exclusively on cleanup.
m The contractor repaired all damaged RCC at the entry and exit
points by removing and replacing it.
n All travel on the dam was in one direction without any turns.
210 Chapter Eight

Because of the extra vigilance, the difficulty of getting RCC repair

work done quickly, and the incautious operation of equipment, project
engineers were dissatisfied with the method used for hauling RCC at
Hauling RCC from the mixing plant to the dam can cause other
complications, Raising the access road to keep up with the daily in-
crease in the height of the dam becomes a cost and scheduling consid-
eration. Also, for most RCC dams, gaining access from the upstream
side can be difficult because of space and scheduling conflicts with
work being done on the upstream face to control seepage.
The use of continuous, high-speed conveyors to deliver RCC to the
point of placing is often the most desirable method of delivery for large
projects. They save road-building costs and equipment operator wages,
and they keep the RCC placement area clean.
At Upper Stillwater, RCC plants discharged directly onto a common
conveyor-belt system that transported the material about 1100 ft (335
m) to the dam. The contractor-designed system included three fixed
sections and a final conveyor that was raised at its discharge point in
4-ft (1.2-m) increments. Belts 48 in wide (1.2 m) and traveling at 750
ft/min (229 m/min) were used. Two tremie tubes discharged the RCC
into trucks which remained on the dam.
An operator at the end of the conveyor system controlled the feed
rate from the RCC plants to the conveyor according to cycle times ex-
perienced on the placement. The contractor experimented with a com-
puterized system for controlling the belts but found that the operator
was better able to respond to unexpected delays in batching and mix-
ing cycles.
At Elk Creek, two separate conveyor systems, each with parallel
belts for RCC and conventional concrete, were used to deliver the ma-
terial from the batch plant to dump trucks on the work surface about
500 ft (152 m) away (see Fig. 8.3). Designed by Rotec Industries, the
hydraulically driven system was made up of 36-in (0.9-m) belts to
carry RCC and 18-in (460-mm) belts to carry conventional concrete.
All main line conveyors were covered and had walkways. The parallel
18- and 36-in (0.46 and 0.9-m) conveyors were mounted on either side
of jack-up platforms. The entire system was raised on 8-ft-diameter
(2.4-m) pipe towers in 5-ft (1.5-m) increments as the dam height in-
A 25-yd3 (19-m3) dual-feed hopper was designed by Rotec to allow
either conveyor system to operate at any capacity at any time without
dead storage in the hopper and without segregation. Each of the sys-
tems terminated in two 50-ft (15.2-m) swingers, one for RCC and one
for conventional concrete. The belts were controlled by a single oper-
ator at the discharge control station.
RCC as a New Construction Method 211

At Middle Fork Dam, a 200-yd3/h (152 m3/h) Ross batch plant and
drum mixers for RCC were situated upstream of the dam on fill com-
pacted behind a sheetpile tieback wall. For the lower elevations of the
dam, a gob hopper discharged RCC onto a stacker conveyor and then
into a 36-in-diameter (0.9-m), enclosed rock ladder. It consisted of a
steel pipe with semicircular steps welded on the inside which kept the
aggregate from segregating as the mix was chuted down a maximum
distance of about 50 ft (15.2 m) to the construction surface.
As RCC placement progressed, sections of the rock ladder were re-
moved and the vertical drop decreased until the conveyor was dis-
charging directly onto the dam. For the uppermost elevations, three
sections of conveyor were used in a scissored arrangement. The final
stacker conveyor was placed atop a tower made of scaffolding. Front-
end loaders were used to catch or pick up the RCC from the chute and
then to spread the mix for most of the lifts. Near the crest, the con-
veyor discharged into 10 yd3 (7.6 m3) dump trucks.

8.5 Compaction

One of the important differences between RCC and conventional mass

concrete is in the way the materials become densified. Mass concrete
is densified, or consolidated, mainly by the influence of gravity with
some help from the reduced friction between particles that results
from externally imposed vibration. RCC is force-densified with exter-
nal high-energy compaction or tamping. Because there is no inten-
tionally entrained air and because the large aggregate is compacted
into its optimum orientation, significantly higher densities can be
achieved with RCC than with conventional concrete. As the degree of
density required increases, however, the compaction effort needed to
achieve it grows exponentially.
The optimum moisture content for compaction is generally in the
range of 4.5 to 7 percent by dry weight of aggregate. Determining the
optimum depends on the grading and maximum-size aggregate used
in the mix, temperature, humidity, and the plasticity of the fines. Ce-
ment content has little effect on the moisture needed for best compac-
tion results. Compaction should be done within 10 min of spreading
and within 30 min of mixing, although full compaction can be
achieved at longer intervals with wetter or set-retarded mixes. RCC is
compacted in lanes. An 8- to 20-in (200- to 510-mm) external strip of
each lane should be left undisturbed and compacted after the adjacent
lane is placed to bond the two together.
Compaction efficiency is affected by the grading and shape of aggre-
gates, the RCC mix design, and the type of compaction equipment.
The compaction effect depends on the mechanical behavior of the mix
212 Chapter Eight

during compaction and on the types of forces exerted. Low-frequency,

high-amplitude vibration has been considered to be the most effective
as lift thickness and/or maximum particle size in the mix gradation
increase. Research in Japan has indicated that high-frequency, high-
amplitude compaction may be more suitable for the kinds of mixes
used in RCD projects. Each RCC mixture behaves differently under
compaction, so field tests are the best way to determine the number of
passes and other parameters.
The most influential factors in achieving densification with vibra-
tory rollers are static force, centrifugal force, the rollers vibrating
mass, and its acceleration. Amplitude, frequency, and vibrating mass
are factors in determining centrifugal force.
Because of a high degree of uniformity in design parameters for roll-
ers, the unit weight per unit drum width of the roller encompasses
such factors as static loading, vibrating mass, and roller dimensions.
The unit weight per unit of drum width is expressed as pounds per lin-
ear inch (PLI) (or kg/cm) of drum width. The PLI can be specified
along with amplitude, frequency, and rolling speed. Recommended
guidelines for RCC construction include:

Average PLI 115 min for compacted lifts to 6 in (20.5 kg/cm for lifts to 15
150 min for compacted lifts over 6 in (27 kg/cm for lifts over
15 mm)
Amplitude 0.025 to 0.035 in (0.63 to 0.89 mm)
Frequency Not less than 1700 vibrations per minute (28.33 Hz)
Rolling speed Not greater than 2 mi/h (3.2 km/h)

Four to eight passes of a self-propelled, double-drum, lo-ton (9.1-t)

vibratory roller have generally been required. The Japanese have re-
quired as many as 12 passes with 7-ton (6.4-t) double-drum vibratory
rollers followed by 6 passes with rubber-tired rollers. In all cases, the
trend is toward using heavier vibratory rollers-14- and 15-ton (12.7-
and 13.5-t) machines. Backup rollers are generally required due to
breakdowns caused by the near continual use of compactors on RCC
At Upper Stillwater Dam, 15-ton (13.5-t) Dynapac CC-50A asphalt
compactors were used for the final season of placement (see Fig. 8.5)
after maintenance problems developed with the contractors original
machines. The Dynapac machines were operated at an amplitude of
0.031 in (0.8 mm) and a frequency of 2400 vpm. The dynamic compac-
tion force of the drum was 429 lb/in (76.6 kg/cm) as compared to 400
lb/in (71.6 kg/cm) for the original machines. The Japanese have fa-
vored Bomag B-2000 compactors for their RCD projects.
RCC as a New Construction Method 213

Figure 8.5 Upper Stillwater Dam compaction.

Initially, 10 passes with a lo-ton (9.1-t) , double-drum vibratory

roller were required for each 2-ft (0.6-m) lift at Elk Creek. After place-
ment of 156,000 yd3 (119,000 m31 during April and May of 1987, how-
ever, the Corps of Engineers Portland District discovered that the
dozer action alone produced the required consolidation, Nuclear den-
sity tests showed that the extensive reworking of the wet, retarded
mix (its Vebe time was 10 s -t 2 s) during the placement of the four
layers resulted in consolidation close to the design density of 148 lb/ft3
(2371 kg/m3). Compaction of the completed lifts with vibratory rollers
produced no additional consolidation, according to 3000 nuclear den-
sity tests done by the Corps. In fact, the use of 10 passes was found to
be detrimental, particularly during hot weather. The roller action
bled out moisture from the RCC, causing it to lose density, stiffen pre-
maturely, and crack.
Based on these findings, the Corps proposed to reduce the number of
vibratory roller passes on the rest of the dam to four or five passes per
lift. Rather than compaction, the intent is to provide a smooth surface
for easier cleanup and to seal the surface against rain damage or wa-
ter loss during hot, dry periods.
Lift surfaces have been gently sloped in a number of dams for drain-
age of rainwater. For ease of RCC placement and lift thickness con-
trol, a constant slope is better than a crowned cross section. In gen-
eral, however, having a flat surface ready to roll in the least amount
214 Chapter Eight

of time is more important for achieving a high-quality product than

delaying compaction to achieve the proper grade. At Upper Stillwater,
the specifications required an initial roller pass without vibration to
aid in leveling the lift surface.
When the roller is operated in the vibratory mode, vibration should
cease before the roller stops. Most manufacturers offer automatic con-
trols that stop the vibration below a fixed speed and restart it above
that speed. This device is especially useful when changing the direc-
tion of roller operation.
The first roller pass should be made 12 to 15 in (300 to 380 mm)
inside an unconfined edge to prevent crumbling of the RCC at the up-
stream or downstream side. The edge can then be compacted with a
heavy-plate compactor of not less than 750 lb (340 kg) or a walk-
behind roller before rolling it with the larger drum roller. Rollers
should be operated in the same lane in the forward and reverse direc-
tions. Lane changes should be made on compacted lanes. The length of
lanes should be staggered to allow rolling out of bumps that can occur
at the end of a pass.
The compacted lift thickness of most RCC dams in the United States
has been 12 in (300 mm). More recently, the trend has been to in-
crease the thickness of lifts for wetter-consistency mixes so that there
are fewer lift joints, thus fewer seepage pathways.
Lifts at Tamagawa Dam were as thick as 39 in (1 m). They were
built up in four or five layers that were spread by dozers. The lifts
were then compacted, cured for up to 36 h, green-cut, and bedded with
mortar before the next series of layers were placed. Except for the long
curing process, a similar system was specified by the Corps of Engi-
neers for RCC placement at Elk Creek Dam. Lifts there were spread
by D7 and D8 dozers in four layers and then the layers were com-
pacted into 24-in-thick (610-mm) lifts. Subsequent lifts were placed as
soon as possible during the coolest months of the year.

8.6 Bonding Lifts

Achieving a tight bond at lift interfaces is one of the most critical con-
struction elements in an. RCC dam. The shear strength and water-
tightness of the structure are largely determined by the integrity of
the bond. Designers and field inspectors direct much of their attention
to the execution of work at these interfaces. Hence, an exact under-
standing of the specifications for treatment of lift surfaces as well as
the designers intentions for ensuring the bond between lifts is critical
for successfully bidding and building RCC dams.
Well-bonded joints are achieved by maintaining a clean, flat, moist,
RCC as a New Construction Method 215

plastic surface condition on the lift surface prior to placing the next lift.
Any aspect of the RCC construction process that influences those desired
characteristics must be carefully executed. For insurance, in some cases
a thin layer of cement-rich bedding mix, mortar, or grout is required to
be placed on all or part of a completed lift before the next layer of RCC is
placed. Most RCC designers specify some kind of bedding for part or all of
each lift lane at the upstream face, except for high-paste content mixes.
A %-in (12.7-mm) layer of broomable, sanded mortar was spread
over the entire surface of each of the compacted, 2-ft (0.6 m) lifts at
Elk Creek Dam to fill voids and increase the bond at the interface.
The mortar was spread using a serrated rubber squeegee mounted on
the front of a small farm-type tractor. A test section built in 1985
showed that the biaxial shear strength at lifts treated with mortar av-
eraged from 205 to 270 lb/in2 (1.4 to 1.9 MPa). Surprisingly, the
higher shear strengths resulted when lift surfaces were not cleaned
prior to placing the mortar and the next layer of RCC. Biaxial shear
strengths achieved for test section lifts are shown in Fig. 8.6.
Transporting and placing a bedding mix can be very complex near
the upstream face. Placing the bedding mix is usually combined with
placement of RCC and facing concrete at the same time and in the
same small area. That is particularly true as the dam narrows near
the crest. Detailed planning and scheduling of the various work crews

Average Shear
Confining Pressure Strength
Lift joint No of
treatment lb/in kg/cm tests lb/in kg/cm*

No cleaning 4 0.3 2 23 1.6

No mortar 100 7.0 2 105 7.4
No cleaning 4 0.3 3 90 6.3
With mortar 100 7.0 2 270 19.0
Washed 4 0.3 5 23 1.6
No mortar 100 7.0 1 155 10.9
Washed 4 0.3 9 65 4.6
With mortar 100 7.0 6 205 14.4
RCC/conventional 4 0.3 7 130 9.2
Through-mass 4 0.3 8 130 9.2
RCC 100 7.0 6 360 25.4
Now RCC mix = 118 lb (53.6 kg) portland cement and 56 lb (25.4 kg) fly ash with
3-m (75 mm) MSA.

Figure 6.6 Elk Creek Dam biaxial shear strengths, RCC test section, 1985. [From
Hopman and Chambers (1988)J
216 Chapter Eight

and pieces of equipment used at the upstream face are required to

avoid slowing RCC placement on the rest of the dam.
At Monksville Dam, for example, a complex procedure for placing a
two-lift-high layer of facing concrete and building up bedded RCC lifts
behind it was abandoned after about one month of construction. Spec-
ified placement rates were not being met and the RCC-facing interface
and the RCC lift bonding were not always satisfactory. A simpler pro-
cedure was adopted for placing one lift of facing concrete and placing
RCC behind it. This worked much better. (See also Set 8.9.1.)
Horizontal construction joints can be intentional or can occur as a
result of a delay in the placement of RCC, leaving all or part of a lift
uncovered long enough for it to reach initial set. Determining when a
cold joint has occurred depends on elapsed time and surface tempera-
ture, the quality and amount of cement used in the RCC mix, and the
effectiveness of set-retarding admixtures. Damage or contamination
of a lift surface can also affect the determination of whether or not a
surface requires extraordinary treatment.
For soils approach RCC dams, the degree of maturity of com-
pleted, uncovered lifts can be measured in degree-hours, i.e., 1 h of ex-
posure of a lift at an average surface temperature of 75F equals 75
degree-hours, The definition of what is a cold joint requiring treat-
ment has changed significantly for soils approach RCC dams. Specifi-
cations at Willow Creek required treatment of lifts after 1600 degree-
hours of exposure. It was half that at Middle Fork Dam and 500
degree-hours at Stagecoach Dam.

8.7 Cleanup, Curing, and Weather

In the Japanese RCD method, all lift interfaces are considered cold
joints and get extensive treatment, including green-cutting, cleaning,
and application of a thin layer of high-slump mortar as a bedding mix
over the entire surface area of each lift (see Chap. 6). Lifts are built up
in layers to thicknesses of as much as 39 in. (1 m), but each lift is re-
quired to cure for as long as 36 h before the joint treatment is started
so production is slowed.
At Elk Creek Dam, the Corps of Engineers initially adopted a sim-
ilarly conservative approach for lift treatment except for the long cur-
ing period. However, results from the test fill in 1985 revealed that
the high-pressure washing of lift surfaces to expose aggregate actually
reduced the bond between lifts. In practice, as long as the surfaces
were kept clean and free of excess water, the Corps required no special
preparation other than placement of the bedding mortar. Lift surfaces
were kept moist with hand-held hoses or by slightly rotating the spray
RCC as a New Construction Method 217

bars on two high-pressure spray trucks to shoot a fine mist ahead of

the trucks.
At Upper Stillwater, a high-paste RCC mix and detailed specifica-
tions for placement were relied upon to provide a tight bond between
lifts. Lifts that were damaged in any way or left uncovered for more
than 72 h at any temperature were considered cold joints in the Bu-
reau of Reclamations specifications.
Up to 24 h, no cleanup was required other than vacuuming to pick
up loose material or water prior to spreading of fresh RCC. For lifts
uncovered for 24 to 72 h, the surface was broomed with a Roscoe
Hydra-broom self-powered sweeper and then vacuumed. In some cases
this process was repeated numerous times. If that failed to clean the
surface, it was hosed down from a water truck or blown using a 185
ft3/min (5.2 m3/min) compressor and blow pipe. After 72 h, a high-
pressure wash was required. That was done only during seasonal
startups and on one section of a lift.
As work progressed, it was discovered that the wet-consistency,
high-fly-ash content RCC mix retained its workability for longer than
the 72 h assumed in the specifications. In most cases where delays
caused completed lifts to remain uncovered for more than three days,
work was allowed to continue based on field inspections of the surface.
In many cases, compaction has made RCC lift surfaces durable
enough to resist damage from all but the heaviest rains, provided that
the lift is not disturbed by equipment or other traffic. Uncompacted
lift surfaces may be seriously damaged by heavy rain and require ex-
tensive cleanup and joint treatment.
In general, a light rain during RCC placement can be tolerated pro-
vided that mud and other foreign material are not tracked onto the
dam and that excess moisture is not worked into the lift surface prior
to compaction. Work should be halted as soon as rollers begin picking
up RCC or other lift damage is evident.
Frequent, heavy rains at Upper Stillwater significantly delayed
production during the 1986 season while crews repaired and cleaned
damaged surfaces. In some cases, the cleanup took two to three days.
The Bureau of Reclamation allowed the contractor to change its
method for handling rain during the final season in 1987. All work on
the dam was halted at the first sign of heavy rain. When it stopped,
instead of repairing the rutted or saturated areas, the most heavily
damaged material was bladed into windrows, picked up with smooth-
mouthed loaders, and discarded. Standing water was vacuumed and
RCC placement resumed. The rain recovery period was reduced from
as long as three days to no more than one shift using the new ap-
Compacted surfaces of RCC must be cured and protected from tem-
218 Chapter Eight

perature extremes in much the same way that conventionally placed

concrete is cured and protected. Lifts should be kept continually moist
with a fine mist spray from trucks or handheld hoses. Too much pres-
sure can loosen particles on the lift surface, triggering additional
cleanup operations. Water trucks should be routed to minimize the po-
tential for damage caused by turning on the completed lift. Unformed,
uncompacted downstream slopes of RCC dams are often considered
sacrificial and may not need to be cured. The final lift of an RCC dam
should be cured for at least 14 days. A layer of damp sand or dirt is
generally more effective than curing compound in covering the final
lift surface and keeping it moist.

8.8 Temperature Control and Contraction

Because the surface area/volume ratio for RCC dams is so much
greater than for conventional mass concrete (as much as five times
greater), it is possible for RCC to gain more heat from the hot sun
than from hydration of cement. It is also possible to lose heat quickly
from only a cool breeze blowing over the warm, large surface area of a
thin lift. Thus, thermal studies for RCC are more exacting than for
conventional concrete. In many cases, construction intensity is greatly
increased in order to meet a compressed schedule dictated by thermal
considerations. To reduce thermal cracking at Elk Creek Dam, for ex-
ample, nearly all of the RCC for the most massive sections was origi-
nally required to be placed between January and April of 1988.
Elk Creek is presently the only RCC dam in the United States with
contraction joints specified for the entire structure. Corps designers
included 10 transverse joints and waterstops spaced a maximum of
300 ft (91 m) apart to prevent uncontrolled cracking. The joints were
formed by vibrating galvanized steel sheets, 24 x 36 in (610 x 910
mm), into the uncompacted RCC. A vibrating blade mounted on the
boom of a Komatsu PC120 backhoe was used to cut the joints. The
sheeting was installed vertically end to end from the upstream face to
the downstream face and for the entire height of the dam. Double
waterstops with B-in-diameter (20-cm) drains were formed in the up-
stream facing concrete which was thickened from 3 to 5 ft (0.9 to 1.5
m) at each waterstop location. (See Fig. 3.11.)
All RCD projects in Japan have been built with transverse contrac-
tion joints cut into each lift using a similar method but at a closer
spacing, usually 49 ft (15 m).
Contraction joints were added to the upper part of Monksville Dam
after delays pushed construction into the warm summer months. Poly-
RCC as a New Construction Method 219

vinyl chloride sheeting was installed as a bond-breaker in the RCC to

create monolithic joints at 120-ft (36.5-m) spacing.

8.9 Special Construction Features

8.9.1 Forms and facing
A variety of methods has been used for construction of the upstream
and downstream faces of RCC dams. Precast concrete panels that
serve as stay-in-place formwork were used for the upstream face at
Willow Creek Dam. Slipformed curbs stacked one atop another were
used for both faces at Upper Stillwater Dam. RCD dams in Japan are
faced on both sides with a thick layer of conventional concrete-as
thick as 10 ft (3 ml-placed against forms and then vibrated and com-
pacted to form a tight bond with the RCC.
Where the intent of the upstream facing is mainly to provide a wa-
tertight barrier, such as at Winchester, Monksville, Elk Creek, and
Japanese RCD dams, construction requirements have been demand-
ing, mainly because of the difficulties of accomplishing a number of
individually trying tasks quickly in a congested space. Where the fac-
ing is intended to provide only freeze-thaw protection and vertical
support for RCC placement during construction, such as at Willow
Creek and Upper Stillwater, the facing methods used have been sim-
At Upper Stillwater Dam, the 12-in (300-mm) lifts of RCC were
compacted between interlocking curbs of 0.5-in-slump (13-mm) con-
ventional concrete on the vertical upstream face and the battered
downstream face. The elements ranged in length from 500 ft at the
base to 2673 ft at the crest (152 to 815 m). The curbs which were 3 ft
high (0.91 m) on the inside, were usually slipformed in a 9-h continuous
pour across the full length of the dam during the daytime mainte-
nance shift. The best run across the dam was 4.5 h.
Two Miller Formless 8100 slip-formers, each equipped with a side-
mounted mold and a 16-ft (4.9-m) finishers platform mounted on the
outside of the element were used (see Fig. 8.7). Eight-cubic-yard (6-
m3) transit mixers equipped with tilting beds were used to transport
the facing mix from the batch plant to the dam. Slump tests were per-
formed on the dam before the concrete was dishcarged down a chute
into the laser-guided slip-form machines. Trucks were placed in neu-
tral and pushed across the dam by the slip-form machines. The com-
pleted vertical elements were required to set for 4 h before RCC could
be placed against the element. A total of 98 mi (158 km) of slip-formed
elements were placed requiring 87,000 yd3 (66,500 m3) of concrete.
This method, which was based on trials done in the United King-
220 Chapter Eight

Figure 8.7 Miller Formless slip-formers with side-mounted mold-placed curbs at Upper
Stlllwater Dam.

dom, worked fairly well at Upper Stillwater. Most of the problems re-
lated to controlling grade. A self-leveling Spectra-Physics laser, model
945L, was used to control grade. For alignment, a Laser Alignment
model 5000 was used. During the longest runs across the dam, con-
struction dust and the heat from the lamps used to illuminate the
work area tended to distort or scatter the laser beams, causing the
slip-former to run up or down.
The facing concrete tended to squat or settle after being slip-formed,
which made it difficult to hold the prescribed grade. A number of so-
lutions were proposed-slowing the machine or using a longer mold.
In the end, however, the tolerances-l in. horizontal or vertical dis-
placement across the dam-were relaxed to 1%. in and these were met
The 16-ft-long and 4-ft-high (1.2 x 4.9 m) tongue-and-groove panels
used for the upstream face at Willow Creek Dam were stacked edge-
to-edge and a cement-enriched RCC mix was placed against them (see
Fig. 8.8). Alignment of the panels was maintained during compaction
of the RCC by use of an exterior strongback system. A 5-ft-long (1.5 ml
threaded coil rod was used to anchor the panels into the compacted
RCC as a New Construction Method

Figure 8.8 Willow Creek facing panels and hand vibrators.

RCC. The rods screwed into inserts in the facing panels and were se-
cured at the other end by a 4-in U02-mm) square washer.
No support was provided for the 0.8 V : 1 H downstream face of Wil-
low Creek. The uncompacted RCC had a tendency to stand on a l-to-l
slope, however, so the contractor placed the downstream side of each
lift with a motor-grader and compacted the material by running the
wheels along the outer edge. Although effective, this method posed
added safety risks and produced a corrugated look rather than a more
desirable smooth surface.
For the downstream face of Middle Fork Dam, conventional con-
crete was placed using 12-in (300 mm) curb forms and stacked in stair-
step fashion. The esthetically pleasing steps were not anchored. In-
stead they were were made monolithic with the RCC by vibrating the
two mixes together.
Scheduling and technical problems resulted at Middle Fork and Ce-
dar Falls dams because the conventionally formed downstream face
was not placed as quickly as the RCC lifts. The lag was two to three
lifts. That caused the RCC to dry out, making it difficult to create a
good bond between the conventional and RCC mixtures.
One solution is to place the conventional concrete ahead of the RCC.
Another is to place the downstream face concrete after the RCC work
is completed. For the Grindstone Canyon Dam spillway, rebar anchors
222 Chapter Eight

Figure 8.9 Grindstone Canyon Dam nearing completion of conventional concrete face
for uncontrolled-overflow spillway.

were set in the RCC before the conventional concrete was placed (see
Fig. 8.9).
At Winchester Dam, a relatively small RCC dam, precast panels
with 65mil-thick (1.65-mm) sheets of PVC liners cast into the back of
them were used to form a permanent watertight barrier on the up-
stream face, Every membrane joint between panels was spliced in the
field and every tieback support or anchor extending through the liner
had to be sealed. The PVC was tied into the foundation by wrapping
the liner under a lift near the base of the dam and tying the plastic
membrane into an RCC cutoff trench. The trained crew heat-welding
the lapped splices between the panels and over the anchor bars was
able to keep up with production. The composite panels have proved to
be an effective water barrier at Winchester.
Schrader has proposed a way of using precast panels to form the up-
stream face, placing RCC behind them and then backing the panels
about 15 in (38 cm) away from the dam for reuse as forms for a con-
ventional concrete cutoff wall. A strongback system with a plate and
double nut arrangement on the threaded anchor rod would position
each panel during RCC placement. The shiplapped panels would be
moved after the RCC had set by pulling them forward on the anchor
rods. All the panels would eventually be repositioned upstream and
RCC as a New Construction Method 223

the space between the completed RCC lifts and panels would be filled
with conventional concrete placed in a continuous operation. This
method has not been used to date.
The upstream faces of Monksville, Copperfield, and other RCC dams
were built with conventional concrete placed against forms and inte-
grated into the fresh RCC by roller compaction at the interface. For
Copperfield Dam, conventional panel formwork placed between sol-
diers anchored with tiebars back into the RCC were used for the
nonoverflow sections on either side of the service spillway. A thin
layer of conventional concrete was placed by transit mixers against
the vertical face at the same time that RCC was being dumped and
spread. Small vibratory rollers and tampers were used to compact the
The method was not successful at Copperfield, in part because of ex-
treme time limitations imposed by the private owner of the remote
dam and a decision made to sacrifice finishing work for speed. The
facing concrete did not provide sufficient cover for the RCC, and the
upstream face has an exposed, segregated RCC strip at the top of each
lift as shown in Fig. 8.10. Over time, wave action is expected to expose
the fully compacted RCC 4 to 6 in (10 to 15 cm) behind the facing con-
crete. Watertightness is being achieved by the use of a bedding con-
crete between each lift at the upstream sacred zone.

Figure 8.10 Upstream face of Copperfield Dam shows exposed RCC lifts.
224 Chapter Eight

The spillway facing method used at Copperfield controlled the pace

of RCC placement. Twenty sliding steel forms were jacked up on
greased pipes embedded in the X-in-thick (3%cm) facing concrete for
each lift of the 330-ft-long (100-m) service spillway. To ease placement
and compaction of the no-slump, lean mix, a super-plasticizer and
water-reducing admixture with an active life of about 30 min was
added just prior to placement by transit trucks. The facing concrete
was placed first and roughly shaped by hand to form a wedge under
the RCC, which was then spread up to and over the wedge. Internal
vibrators were used at the interface, and, after the facing concrete had
lost its fluidity, vibratory rollers compacted the RCC into the stiff but
still workable facing mix. The facing mix design and timing of the
various activities were the critical factors in this method.
The upstream face at Monksville Dam in northern New Jersey was
formed using a more detailed and demanding construction procedure
than was employed at Copperfield. The upstream face varied in thick-
ness from 15 to 27 in (38 to 69 cm) and was designed as the primary
seepage barrier for the water-supply dam. It was placed using jump
forms starting from the top of a small RCC foundation mat. Vertical
crack inducers with epoxy joints and waterstops were installed every
20 ft (6.1 m).
The high-slump mix used for the face was heavily dosed with
superplasticizers and some retarders to provide a durable concrete
with low shrinkage and high creep. It contained 22 gal of free-mix wa-
ter per cubic yard (109 kg/m3). The original sequence for placing a 2-
ft-high (0.6-m) lift of facing concrete and building up RCC behind it
was abandoned after one month due to its complexity and quality con-
trol problems. A simpler method of placing l-ft-high (0.3-m) facing
lifts was adopted and used for the rest of the dam. Diagrams showing
the two placement sequences for the upstream face are shown in Figs.
8.11~ and llb.
For added seepage control, a thin layer of bedding concrete was
spread over a 6-ft-wide (2-m) zone behind the facing concrete. The bed-
ding was originally to be used only on every other RCC lift joint. How-
ever, because the contractor frequently did not meet the 5-h time limit
for placing the next lift of RCC, bedding concrete was placed on about
80 percent of the lift joints.
The difficulty of controlling and synchronizing all the various oper-
ations has led Schrader and others involved with the project to caution
against use of this type of conventional concrete upstream face. In-
stead, they prefer the system used at Winchester Dam-precast pan-
els lined with a synthetic membrane-because it is simpler to con-
struct, is watertight, and has little impact on cracking potential. It is
also more expensive, however.
RCC as a New Construction Method 225

0.02 min

I. Place bedding mix

2. Sped 0.3-m Iiflol RCC 1. Place p-x&Me vertical form
3. Compact 0.3-m Iifi of RCC 2. Place beddmg mix
4. Place coil rod used for form support 3. Spread and compact 0.3-m lift of RCC
5. Place facing concrete (0.6-m lift) 4. Place coil rod used for lorm support
6. Spread 0.3-m lifl of RCC 5. Place facing concrete (0.6-m lift)
7. CompaclO.3-m lift of RCC. Intemafly vibrate facing 6. Place swell seat
COncrete as required while compacting RCC. 7. Placa and compact 0.3-m lift of XC. Internally vibrate
facing concrete as required while compacting RCC
Figure 8.11 Monksville Dam facing concrete placement. (a) Original sequence; (b) mod-
ified sequence. (Note: 1 ft = 0.3 m.)

For watertightness and freeze-thaw protection at Elk Creek Dam, a

3-ft-wide (0.91-m) upstream zone of conventional concrete was placed
against the RCC. An air-entrained mix containing 325 lb of cement
and 126 lb of fly ash per cubic yard (193 kg cement and 75 kg of fly ash
per cubic meter) was used.
For the initial two months of RCC production in early 1987, the con-
tractor chose to place the full 2-ft (0.6-m) lift of RCC to within 3 ft (1
m) of the upstream forms. The facing concrete was then placed and the
two mixes were consolidated at the interface using handheld vibrators
before the RCC was compacted with vibratory rollers.
Voids in the area of the interface resulted in significant seepage
when tested later in the year, however, and the area was grouted. Be-
fore RCC placement resumed in October 1987, the Corps of Engineers
requested that the contractor switch from handheld immersion vibra-
tors to gang vibrators mounted on a backhoe in order to get more uni-
form consolidation of the interface zone. The contractor was unwilling
to add another piece of equipment late in 1987, however, mainly be-
cause of delays resulting from a successful court challenge by environ-
mentalists to the continued construction of the dam beyond a height
safe enough to pass a loo-year flood. For the RCC placed from October
to December 1987, crews did begin immersion vibration of the two
mixes immediately after the facing mix was placed, which helped to
reduce voids.
The Japanese have successfully employed a similar method for the
placement of facing concrete and its consolidation with the RCC lifts
226 Chapter Eight

on both the upstream and downstream faces of Tamagawa and other

RCD dams. The main difference is that a much wider and thicker zone
of conventional concrete is placed-as wide as 10 ft (3 m) placed in
lifts as thick as 39 in (1 m). (That was the lift thickness of RCC used in
the upper elevations of Tamagawa.) After consolidation by banks of
immersion vibrators mounted on a backhoe, the interface between the
RCC and facing concrete is compacted initially with six passes of a vi-
bratory roller followed later by 12 more passes.

8.9.2 Galleries and drainage

Construction of galleries has been accomplished in a number of ways.
All of them have been time-consuming and all have impeded method-
ical, continuous RCC placement. Because of the problems encoun-
tered, RCC dam designers are seeking to eliminate galleries and,
more generally, to simplify appurtenant features.
An efficient gallery-forming method was developed by the contractor
on Upper Stillwater Dam. The same slip-form machine used to place the
upstream facing elements was used to form the vertical walls of the up-
per and lower drainage galleries. The continuous pouring of the 3-ft-high

Figure 8.12 Galvanized corrugated metal pipe sections used for arched roof of galleries
at Upper Stillwater Dam. (Courtesy of U.S. Department of the Interior, Bureau of Rec-
lamation, UINTA Basin Construction Office.)
RCC as a New Construction Method 227

(0.91-m) curbs for the galleries allowed RCC placement to continue un-
interrupted in the confined areas on either side of the galleries. Galva-
nized corrugated metal pipe sections were used for the arched roof of the
galleries. Results of this method are shown in Fig. 8.12.
Conventional formwork was used to form the inspection and drain-
age gallery within the dam section at Middle Fork Dam. All other con-
struction obstacles were eliminated in the design, however. The pri-
mary spillway and outlet conduits, for example, were combined in a
double-chambered tower placed against the face of the dam and con-
nected to a trench under the dam that led to an outlet structure down-
stream. The outlet works for Elk Creek Dam were incorporated into a
similar intake structure built on the upstream face so as not to inter-
fere with the placement of RCC.
Placement of gravel or sand fill in the gallery cavity as RCC rises
around it, then mining the loose material later is another approach
that has been taken. At Willow Creek Dam, the resulting gallery
walls were left in a rough condition and some of the fill material ad-
hered to the RCC. If desired, wood separators can be placed between
the fill and RCC lifts in order to obtain a smooth finish. At Galesville
Dam, a sledge hammer was used to dislodge the gravel which was
then removed by an industrial-grade vacuum cleaner.

American Concrete Institute, Roller-Compacted Mass Concrete, Report No. 207.5R-
Bureau of Reclamation, Guidelines for Designing and Constructing Roller-Compacted
Concrete Dams, June 1987.
Bureau of Reclamation, Roller-Compacted Concrete: Interagency Forum, Divison of
Research and Laboratory Services, Lakewood, Colo., April 4, 1985.
Campbell, D. B., G. C. Elias, and E.K. Schrader, Monksville Dam: A Roller-Compacted
Concrete Water Supply Structure, ICOLD Congr., 15th, Lausanne, 1985.
Carruth, B., A Study of Construction Related Problems Encountered During the Pro-
duction and Placement of Roller-Compacted Concrete, Masters Research Report,
Colorado State University, Fort Collins, Fall 1984.
Dunstan, G., Continuous Volumetric Metering of Concrete Ingredients: The ARAN
Equipment Approach to Accuracy-How and Why It Works, Brisbane, Australia,
October 1986.
Forbes, B. A., Roller-Compacted Concrete: Design, Construction and Performance of
Copperfield Dam, Institution of Engineers Australia (Queensland Division),
Brisbane, September/October 1985.
Hopman, D. R., and D. R. Chambers, Construction of Elk Creek Dam, Ro&F-
Compacted Concrete II, AXE, New York, February 1988, pp. 251-266.
Hopman, D. R., 0. Keifer, Jr., and F. Anderson, Current Corps of Engineers Concepts
for Roller-Compacted Concrete in Dams, Roller-Compacted Concrete, ASCE, New
York, May 1985, pp. l-10.
International Commission on Large Dams (ICOLD), Roller-Compacted Concrete for
Gravity Dams-State-of-the-Art, Bulletin 75, 1989.
Jackson, H. E., Roller-Compacted Concrete for Dams, Electric Power Research Insti-
tute report AP-4715, Palo Alto, Calif., September 1986.
228 Chapter Eight

McKinnon, R., Roller-Compacted Concrete, World of Concrete, Session l-06, Las Ve-
gas, February 1983.
McTavish, R. F., Construction of Upper Stillwater Dam, Roller-Compacted Concrete
II, ASCE, New York, February 1988, pp. 267-276.
Nagataki, S., T. Yanagida, and T. Okumura, Construction of Recent RCD-Concrete
Dam Projects in Japan, Roller-Compacted Concrete, ASCE. New York. _ May 1985,
pp. 90-101. -
Parent, W. F., W. A. Moler, and R.W. Southard, Construction of Middle Fork Dam,
Roller-Comwacted Concrete. ASCE. New York. Mav 1985. DD. 71-89.
Schexnayder, C. J., and R. F.Stewart, ConstructionTechn&es for Roller-Compacted
Concrete, Transportation Research Record 1062, Transportation Research Board,
National Research Council, Washington, D.C., 1986.
Schrader, E. K., Watertightness and Seepage Control in Roller-Compacted Concrete
Dams, Roller-Compacted Concrete, AXE, New York, May 1985, pp. 11-30.
Snider, S.H., and E.K. Schrader, Design Changes and Constructi% at Monksville
Dam. Roller-Comaacted Concrete. ASCE. New York. Februarv 1988. ~a. 220-235.
Spreader Box Aids RCC Placement; Engineering News-Record,July ii,-1985.
Tucker, G., Elk Creek Dam: Roller-Compacted Concrete Mixing and Placing Study,
Report to Corps of Engineers, Portland District, 1984.
Withrow, H., Compaction Parameters for Roller-Compacted Concrete, Roller-
Compacted Concrete II, ASCE, New York, February 1988, pp. 123-135.

Construction Control
and Specifications

9.1 General Considerations

The design of RCC dams is based on properties derived from tests per-
formed on precise laboratory mixtures. Some margin of safety is built
into the design to account for variability in the properties of in-place
materials. This margin should not be compromised by poor construc-
tion control. The properties of the completed dam must be consistent
with those factored into the design.
The rapid pace of RCC dam construction demands that construction
control procedures be focused on all the variables that affect quality
during the mixing and placement of the RCC. Test cylinders can pro-
vide an interesting historical record, but RCC cylinders may not ac-
curately represent the material in the dam. In any case, they are bro-
ken long after anything can be done to rectify a presumed
construction quality problem.
The project specifications are the foundation for the construction
control program. The specifications should provide requirements, test
procedures, and acceptance criteria for materials, mixture propor-
tions, equipment, and construction methods. Items that will not be en-
forced should not be included in the specifications.
To help make sure that the structure will be built according to the
owners specifications, the entire construction control program should
be planned prior to the start of work. The planning should include per-
sonnel and equipment requirements based on the types and frequen-
cies of tests required and on the construction schedule. Methods for
performing each test and procedures for recording and evaluating the
results also should be included in the control plan.

230 Chapter Nine

9.1 .l Test section

Until more standardized design and construction procedures evolve
for RCC dams, a test section should be built for all but the smallest
projects. Potential problems need to be discovered and addressed be-
fore the first, critical lifts are placed on the foundation. In most cases,
the money and time saved in preventing problems more than offsets
the cost of correcting deficiencies after the full construction effort is
Construction of the test section should be well planned in order to
maximize the benefits for the owner and contractor:

m The test section should provide the contractor with an opportunity

for a trial of the equipment and construction method proposed for
the dam.
n The test section should provide the designer with information on the
suitability of the laboratory-designed mix under full-scale RCC con-
struction conditions.
n The test section should give the members of the quality-control
team an opportunity to familiarize themselves with the required
tests so that they can give the contractor a better understanding of
what is an acceptable end product.
. The test section should be located in an area where it need not be
removed after construction and possibly in an area where it can
serve some useful purpose such as for bank protection or a hard
stand for equipment.

In order to evaluate the bond between lifts of RCC, the test section
lifts should be placed at the same rate and using the same procedures
as those planned for the actual dam. Tests should be performed on the
completed section to assure that its properties are reasonably close to
those on which the design is based. If not, changes should be made in
the material or construction procedure. To determine the effect of a
proposed change, construction of a smaller, single-factor test section
may be required prior to the start of dam construction.
On a small project, the test section may be used strictly for deter-
mining the contractors construction capability and to establish test
procedures. In this situation, tests for material properties are not
done. The amount of time the mixing plant needs to be on site also is

9.1.2 Preconstruction training

Preconstruction training of the owners field inspectors and the con-
tractors supervisory staff should be part of the overall construction
Construction Control and Specifications 231

control program. The philosophy behind the design, key specification

requirements, and what was learned from any test section should be
discussed fully. Potential items of dispute such as lift cleanup require-
ments and the basis for acceptance or rejection of RCC mixtures with
inconsistent workability should be explained in detail.

9.2 Materials Control

The quality of aggregates, cement, and pozzolan are generally in-
spected and approved at the source. The construction control program
should focus on maintaining the quality of materials and on methods
for assuring the material design quantities are properly introduced to
the mixer.
RCC can be produced either by weigh-batching followed by mixing
or by continuous proportioning and mixing. RCC mixes that conform
to the concrete approach invariably use weigh-batching, while soils
approach mixes have been produced by either -method. Small soils ap-
proach projects usually tend toward continuous mix methods.
Many modern concrete plants are now computer-controlled, and
batch weights of each material are printed on tape. This degree of au-
tomation does not relieve the inspector from checking to determine if
the plant is properly proportioning and discharging the materials.
Beam balances or dial scales may be used for weighing batches. In all
cases, scales should be inspected and calibrated prior to the start of
production and checked frequently thereafter.

9.2.1 Aggregate
In order to produce a uniform RCC mixture, the grading and moisture
content of the aggregates must be consistent and within specification
limits. The two types of mixes respond differently to variations in
grading and moisture content. Concrete approach RCC mixes are
more sensitive to any change in aggregate grading that increases
voids. If changes occur, paste volumes need to be adjusted so that all
voids are filled. Soils approach RCC mixes achieve their desired
strength properties by maintaining an optimum moisture content, and
therefore are more sensitive to moisture changes in the aggregate.
Minimizing segregation is an important factor for all RCC construc-
Sieve analysis and wash tests to determine aggregate grading
ahould be an on-site requirement for all projects. Proper sampling of
aggregate per ASTM D 75 is important whether the sample is ob-
tained from large stockpiles, a bin, or a conveyor belt. Of particular
concern in the grading are the percentages of sand and minus #200
(0.075-mm) sieve fines. Changes in these percentages can affect the
232 Chapter Nine

consistency or compactibility of the RCC mixture. A marked increase

in fines can increase water demand, resulting in a decrease in
The moisture content of the aggregate, particularly the sand, should
be consistently monitored. The moisture content of aggregate stock-
piles should be constant and as close to a saturated surface dry @SD)
condition as is practical. Changes in moisture content, due to a heavy
rain, for example, must be compensated for at the mixing plant.
Aggregates should be stockpiled and withdrawn from the pile in a
manner that minimizes segregation. In general, allowing aggregate to
roll down the sides of the pile causes segregation because the larger
particles tend to fall to the base more than others. Stockpiles should,
therefore, be built up in layers of uniform thickness. Inspection for
segregated aggregate can either be visual or by a sieve analysis test.
RCC produced from coarse material at the bottom of a poorly placed
stockpile will be difficult to compact into a solid mass, thereby in-
creasing permeability and decreasing strength.

9.2.2 Cementitious materials

Shipments of portland cement and pozzolan (fly ash) delivered to the
job site should be checked to determine that they are of the specified
cement type or fly ash class. Cement and fly-ash can be similar in ap-
pearance, so care should be taken that they are not intermixed in stor-

9.3 Control of Construction Operations

Assuring that time limits for certain operations are met, minimizing
segregation of the RCC mix, and achieving adequate density are the
most critical elements of construction control for RCC dams.

9.3.1 Plant and equipment

The mixing plant must be capable of producing consistently high qual-
ity RCC at production rates in excess of the minimum required by the
specifications. Mixer performance tests may be specified to determine
the acceptability of the mixer or to establish minimum mixing times
for batch-type mixers or minimum retention times for continuous mix-
Although vibratory rollers must meet minimum weight and physi-
cal dimension requirements, the most important test is whether the
roller is capable of producing required density within a reasonable
time. Backup rollers may be required on-site in order to maintain
high production rates.
Construction Control and Specifications 233

Any hauling unit or other piece of equipment moving from off the
dam to the RCC surface must be checked to ensure that it is not track-
ing mud, clay, or other undesireable substances onto the construction
area. A means of removing the material prior to tracking it onto the
dam is preferred to cleanup of the RCC surface.
Conveyor-belt systems should be checked and modified as necessary
to minimize segregation or drying out of the material. For conveyor
belts carrying different RCC mixes or both RCC and conventional con-
crete, markers in the form of colored tags should be placed at the start
of a new material so it may be visually identified and placed in the
desired location.

9.3.2 Control of mixture proportions

The procedures for accurately controlling RCC mixture proportions
that are weigh-batched are identical to those for conventional con-
crete. These procedures have been well-defined for many years and
will not be recited here.
There is a trend toward use of continuous proportioning and mixing
plants for RCC dams, especially for small projects. Most of these con-
tinuous plants are belt-fed, twin-shaft pugmills. The control of mix-
ture proportions for these plants will be covered in this section.
The control of pugmill mixing operations involves: (1) calibrating
the plant, (2) devising a method for checking mixture proportions over
a certain time period, and (3) selecting a means for determining ce-
ment content of freshly mixed RCC. Continuous-mixing plant calibration. For a project planned for

continuous proportioning, as contrasted to weigh-batching, the ce-
ment and fly ash (or other pozzolan) are generally specified as a per-
centage of the total dry weight of aggregate. If the cementitious ma-
terials are specified as batch weights per cubic yard or cubic meter,
they should be converted to percentage of dry weight of aggregate
prior to plant calibration. Calibration consists of setting the cement,
fly ash, and aggregate metering devices to obtain the specified mix
proportions on the conveyor belt that continuously feeds the pugmill
mixing chamber.
Plate feeders or adjustable strike-off gates are used to measure the
volume of aggregate deposited onto the conveyor belt per unit time. If
the aggregate is produced and stored in more than one size, the vary-
ing aggregate sizes must be combined in the correct proportions prior
to deposition on the main aggregate feed belt that supplies the mixing
The first step is to determine an aggregate feed rate for each gate
opening consistent with the planned production schedule. The feed
234 Chapter Nine

rates are determined by running aggregate alone on the belt and

weighing timed loads deposited into a dump truck. This procedure can
be used for each gate opening for various time periods to check the
quantity of aggregate and the uniformity of the feed rate. At the same
time, the moisture content of the aggregate should be determined in
order to calculate the dry aggregate feed rate per unit time for each
potential aggregate feed setting.
In order to minimize the variations in the aggregate feed rate, the
aggregates moisture content must be consistent and the height of the
aggregate source charging the belt must be kept relatively constant.
The next step is to determine the feed rates for cement and fly ash.
Cementitious materials are generally metered by either a vane feeder
or a cleated belt located above the conveyor belt. A surge hopper for
cement fed from the main storage silo is usually set above rotating
vane feeders. This is done to maintain continuity in the height and
thereby the density of the cement stored above the feeder. The height
of the cement source above a cleated belt is less critical because this
metering system slices off a specific volume of cement rather than
completely depending on gravity feed, as is the case with vane feeders.
Therefore, cleated-belt feeders are usually located directly below the
main cement storage hopper.
Because fly ash tends to flow, neither metering device is as accurate
in measuring fly ash as cement. Blending the cement and fly ash in
the specificed proportions before metering is one solution to this prob-
The calibration of cement feed rates is similar to that for aggregate
in that cement is diverted from the cement feeder to a truck, box, or
other suitable container for a certain period of time and then weighed.
The cement-metering devices are run at variable speeds consistent
with project requirements for varying time periods, such as 15,30 and
45 s, to determine uniformity and rate of feed. Cement feed per revo-
lutions per minute is then plotted. Repeated trials should fall within 2
percent of the plotted line.
An Australian ARAN continuous-mix plant with a cleated-belt ce-
ment feeder has only a small box to collect cement so the time to fill
the box is very short, less than 5 s.
With a constant and known rate of aggregate feed, the cement and
fly-ash feeders can be set to deliver the specified percentages based on
weight per unit time.
Water delivery to the pugmill mixing chamber is based on produc-
ing an RCC whose moisture content at the time of compaction is at
optimum or slightly wet of optimum. Therefore, the amount of water
added must account for water contained in the aggregate and also wa-
ter lost to evaporation and hydration prior to the start of compaction.
The required water is usually metered by a positive-displacement
Construction Control and Specifications 235

pump from a constant-head surge tank to spray bars located on top of

the pugmill mixing chamber. Checking mixture proportions for continuous plants. Once the

initial calibration of the continuous proportioning and mixing plant is
made, it should be checked as necessary to determine that the speci-
fied mixture proportions are maintained. This check is similar to ini-
tial plant calibration in that cement, fly ash, and aggregate are me-
tered separately for short periods of time, collected, and weighed. The
percentages are calculated to determine if there have been any
changes from the desired proportions.
In humid or wet weather, the cement feeder should be checked more
often to see if caked or packed cement has restricted the flow. Contin-
uous plants should be equipped with a system that stops the aggregate
feed belt when no cement is being metered. The cement storage silo
can also be equipped with a sensor that actuates a light or horn when
the cement level is low.
There are other methods for checking overall mixture proportions
over a period of time. Modern continuous plants incorporate belt
scales which weigh the amount of aggregate plus cementitious mate-
rials passing a section of the belt. The weight is recorded by a total-
izer. If the plant does not have belt scales, a clock that records the du-
ration of aggregate feed belt operation can be used.
Vane feeders are usually equipped with a revolution counter.
Cleated-belt feeders can determine revolutions using sensors that
count pulses each time a drive pulley tooth passes a certain point.
With known weights per unit time or total weight delivered for each
material, the required percentages can be calculated.
Another method is to check cement and fly-ash delivery tickets
against the total volume of RCC placed since the start of the project. Testing of freshly mixed KC. RCC mixtures can be tested as

they come out of the mixing plant. Cement content alone can be
checked by a titration test (ASTM D 2901). Cement and water content
can be determined by use of a concrete quality monitor (CQM). The
equipment necessary to do the CQM tests is relatively expensive,
making it impractical for some projects. Both tests for cement involve
measuring the calcium content of the RCC.
Another test for determining the cement content in freshly mixed
RCC is the heat of neutralization method, which is based on the fact
that a greater amount of cement produces a higher temperature in the
hydration process. The method was initially developed in Australia
for soil-cement and is undergoing extensive testing by the Bureau of
Reclamation to determine its applicability to RCC mixtures.
236 Chapter Nine

For concrete approach mixes, modified Vebe tests as a measure of

the consistency or the water content of the mixture can be performed
with material sampled from RCC being discharged from the mixer or
just prior to spreading and compacting. At Elk Creek Dam, Vebe tests
were done every half hour near the mixing plant.

9.3.3 Transporting and spreading

In no case should haul time exceed 30 min. One hour should be the
maximum allowed between introduction of water to the mix and final
compaction. Any exceptions should be approved by the owners repre-
Visual inspection will reveal if any of the operations involved in
transporting and spreading RCC cause segregation of the larger ag-
gregate particles. For example, depositing RCC in piles that are too
high or that are dumped too rapidly can cause the larger aggregate to
roll to the bottom of the lift. Operating dozers or other spreading
equipment too rapidly may produce inertia, causing large particles to
roll to the front of the material being spread. This causes the segre-
gated aggregate to end up at the bottom of a compacted lift, causing
rock pockets or voids. A U-shaped blade on the dozer helps form a roll
of material ahead of the blade, which tends to keep the RCC well
mixed. Changes in mix design to produce a wetter consistency or in-
creasing the proportion of sand in the mix can also help to reduce seg-
If limits are put on the temperature of the RCC when placed, ther-
mometer readings should be taken and recorded just prior to compac-

9.3.4 Compaction
Consistently achieving required compaction is one of the most impor-
tant requirements of an RCC construction control program. Density
tests, as a measure of compaction or consolidation, provide an indica-
tion of the strength, bonding capability, and permeability of the com-
pacted RCC as well as a confirmation of the unit weight used in the
structural design calculations.
Specifications for compaction can be either prescriptive or require a
desired performance. Prescriptive specifications require a minimum
number of passes by a specified roller, based on its performance on a
test section. Performance specifications require that the contractor
achieve a certain percentage of maximum density for soils approach
mixtures. Concrete approach performance specifications may define
required compaction in terms of an optimum or average maximum
Some specifications combine both types of compaction requirements.
Construction Control and Specifications 237

A minimum density is required as well as a minimum number of

passes, usually four to six. Determination of maximum or target density. A number of soils

approach specifications have required compaction to no less than 95
percent of maximum density or to an average of 98 percent of maxi-
mum density, or both. If a certain percentage of maximum density is
required, there must be a method established for determining the
maximum (or 100 percent) density. It can be established in the labo-
ratory by using a modified compaction test (ASTM D 1557) or it can be
discovered empirically when placing the test section. Maximum den-
sity occurs at a point when additional rolling produces no further in-
crease in density. Once the maximum density has been established,
materials used in the mix must be closely controlled so that the target
does not move.
The term optimum compaction density was used at Elk Creek
Dam and average maximum density was specified for Upper
Stillwater Dam. Both employed concrete approach mixes. The tar-
get density was determined by the owners on-site representative
from results of compaction trials in a control section. Optimum
compaction density is achieved with RCC that has seen ideal place-
ment, spreading and compaction at a water content that also pro-
duces the required properties. A minimum number of roller passes
is then specified that will produce the optimum compaction density.
Rolling continues past the minimum number of passes until the
specified density is achieved.
At Elk Creek Dam, the measured Vebe time was 17 to 21 s for con-
struction during the fall of 1987. Ten vibratory roller passes were
specified for each lift. During construction, however, it was discovered
that the optimum compaction density was almost reached during the
spreading and working of the wet mix by the action of the dozers
alone. Therefore, fewer than the ten roller passes were actually
Determination of the average maximum density (Ah4D) for RCC
of a specific consistency involves use of nuclear density gauges to
determine wet density after each pass of a vibratory roller. When
there is no change in density or where the difference between three
consecutive density measurements is less than 0.2 lb/ft3 (3.2 kg/m3),
the control section is considered to have achieved maximum den-
sity. At Upper Stillwater Dam, the contractor was then required to
average not less than 99 percent of the AMD with no more than 1
test in 20 consecutive tests measuring less than 98 percent of the
A theoretical air-free (TAF) density may also be used as the maxi-
238 Chapter Nine

mum density for a concrete approach mix. The TAF density can be
computed by adding the SSD batch weights and dividing by its corre-
sponding absolute volume, or calculated after extended vibration of
RCC of the desired consistency. The required in-place density should
then be not less than 98 percent of the TAF density. Nuclear gauge for moisture and density measurement. Nuclear

gauges are invariably used for the measurement of both moisture and
density of compacted RCC. Moisture determinations of RCC mixes us-
ing oven drying have proved to be inconsistent due to the hydration of
water with cement prior to testing. Nuclear gauges must be properly
calibrated to account for chemical composition errors that may occur
with certain aggregates in the mix, such as highly siliceous or calcar-
eous aggregates.
Nuclear gauge density testing should be accomplished as soon after
compaction as practical. The density tests should be taken at random
locations by use of a probe placed into the compacted RCC. Use of the
gauge at the surface of the RCC (backscatter method) can produce
misleading results, especially for dry soils approach mixes. Measure-
ments may indicate an acceptable average density, but this can con-
sist of a highly compacted surface area and a low-density area with
voids at the bottom of the lifts.
Recent projects have used a dual-probe nuclear density gauge as
shown in Fig. 9.1 with good results. It is used to measure density at
any desired depth within a lift.

Figure 9.1 Dual-probe nuclear

density gauge.
Construction Control and Specifications 239

9.3.5 Lift treatment

Lift surface treatment is usually required to assure adequate bonding
between successive lifts, especially for lean, dry soils approach mixes.
If a lift is placed quickly and compacted over a previously placed lift
that remains alive, good bonding can be expected. However, if the
lower lift has been allowed. to dry out or there has been a delay in plac-
ing the next lift, core samples have indicated poor or no bonding be-
tween lifts. The initial set or drying out of the RCC has, in effect,
caused a cold joint.
Based on this experience, specifications usually require that ex-
posed lifts be kept continuously moist prior to placing the next lift.
Concern over bonding between lifts also has fostered a concept called a
maturity index, which is measured in degrees Fahrenheit multi-
plied by hours as described in The temperature is the average
reading for each hour measured at the surface of the RCC. Thus, 2 h at
50F equals 100 degree-hours. Degree-hour values have no direct met-
ric equivalent except for a fmed period of time.
On RCC dams where a maturity index is used, lift surfaces are
treated as cold joints after a certain number of degree-hours have ac-
cumulated. However, there are other factors that affect the maturity
of a lift, such as mix proportions and the relative humidity at the site.
Also, there is disagreement about the number of degree-hours above
which a bedding mix must be applied to improve the bond between
lifts. Still, the concept of a maturity index as measured in degree-
hours has been used on a number of soils approach dams with reason-
able success.
Willow Creek Dam was built using 1600 degree-hours as the upper
limit for uncovered lifts, and Schrader has since recommended a 2000-
degree-hour limit for other soils approach RCC dams. He writes in
Sec. 4.2.2 of this book that tests have shown that the bond strengths
between lifts placed anywhere from 12 to 24 h after the previous lift
will be about the same. Other designers have reduced the allowable
maturity value to 500 degree-hours, based on the evaluation of cores
of completed RCC dams.
For all concrete approach RCC dams and some soils approach dams,
the limiting delay between lifts is being specified in hours only. Using
a simple time-elapsed specification allows the contractor to better
plan the number of lifts per shift and, thus, reduce the need for ex-
traordinary lift surface treatment.

9.3.6 Curing and weather considerations

Proper curing of RCC surfaces is important for all RCC mixes. Com-
pacted lifts should be kept continuously moist at a temperature above
240 Chapter Nine

35F (2C) until covered. RCC surfaces that will be left exposed should
be cured for no less than 7 days and preferably 28 days. Water curing
is preferred over sprayed-on membranes, especially for surfaces that
will receive another lift of RCC, because the curing compound may act
as a bond breaker and because of the dry nature of RCC mixes.
Most specifications do not permit placing RCC when the ambient
temperature drops below 35F (2C). The owners representative may
permit RCC placing at lower temperatures if the mix and the surface
of the compacted RCC stay above 35F. If the ambient temperature
drops below 32F (OC), the surface of any exposed RCC less than 7
days old should be covered with heavy tarps, blankets, or other tem-
porary protection until after the ambient temperature rises to above
35F. Many contractors elect to shut down operations due to the added
cost of cold weather protection and reduced production rates when
nighttime temperatures start to consistently drop below freezing.
Most specifications prohibit RCC placement during heavy rain,
which is defined as more than 0.3 in/h (7.6 mm/h) or 0.03 in (0.76 mm>
in 6 min as defined by the U.S. Weather Bureau Glossary of Meteorol-
ogy. If heavy rain is anticipated, the amount of RCC left uncompacted
should be minimized, as the greatest problem with rain is altering the
uniformity and consistency of the mix. If the rain causes the moisture
content of the uncompacted mix to exceed specification limits, it
should be removed. This was done with considerable success during
the final season of RCC placement at Upper Stillwater Dam in 1987.
Specifications originally required the contractor to repair rain-
damaged lifts, but this proved to be too time-consuming and more ex-
pensive than removing all of the overly wet RCC.

9.3.7 Grade control and tolerances

RCC can be placed close to lift thickness tolerances by the use of laser-
controlled spreading equipment, but smaller projects may not be able
to support the added cost. Lift thickness tolerances can be as much as
f 15 percent of the lift thickness, or about 1.5 in (38 mm) for a 12-in-
thick (300-mm) lift. The potential problem with a greater than speci-
fied lift thickness is that the required density may not be achieved
throughout the entire lift, especially at the bottom. Thin lifts are more
susceptible to breakage because of reduced section modulus and
should be terminated before the thickness gets below 6 in (150 mm),
which can occur where RCC intersects with a rising foundation. The
remaining area can then be filled with the next lift.
Other required tolerances should be well-thought-out and not be
more restrictive than required for proper function or appearance of the
structure. For unformed RCC such as might be specified for the down-
Construction Control and Specifications 241

stream slope of the dam, the tolerance for the unformed face could be
from 0 to 12 in (300 mm) as long as the variation in the overbuild is
not excessive.

9.4 Frequency of Testing

The frequency of tests for construction control purposes should be de-
termined by the engineer in order to assure that the materials and
methods meet the specification requirements. Tests should be done
more frequently during the start-up period or when it has been deter-
mined a certain specification item is not being met consistently. The
frequency can be reduced when previous results indicate that the task
or material has been consistently exceeding minimum specification
values. A suggested minimum frequency for certain tests is shown in
Fig. 9.2.

9.5 Action Required for Nonconformance

with Specifications
Most specifications provide requirements or acceptance criteria for
materials, mixture proportions, equipment, and construction methods
but do not provide guidance on what action is required for nonconfor-
mance with specifications.
It is assumed that nonspecification material and mixtures are al-

Minimum frequency
Test Test designation and location

1. Aggregate Grading
Sieve analysis ASTM C 136 Once per shift of aggregate pro-
duction and RCC placement for
each stockpile.
75-pm (#200) sieve ASTM C 117 Once per shift for the combined
fines gradation

2. Aggregate moisture ASTM C 566 Once per day of RCC placement

for each stockpile

3. Temperature of mixed Thermometer Once per shift at both mixing

RCC plant and placement location

4. Mosture Content of ASTM D 3017 Once per 2 h at placement loca-

mixed RCC tion after compaction, but not
less than two per lift or two per
5. Density of compacted 10,000 ft of surface area.
Figure 9.2 Frequency of Field Control Tests.
242 Chapter Nine

ways removed and wasted. There may, however, be situations or loca-

tions where nonspecification materials can be used, where a material
may be modified for acceptance, and cases when the owners represen-
tative will exercise engineering judgment in allowing use of
nonspecification material, mixtures, or equipment.
Three cases where nonspeciiication items may accepted are: (11
where tearing out marginally deficient RCC causes more of a cost or
schedule problem than leaving it in; (2) the original specification may
have been too restictive or in error; (3) material that is only margin-
ally below the desired quality is allowed to remain and is accepted at
no cost or a reduced cost to the owner.
The Corps of Engineers in its specification for Elk Creek Dam
placed all of its action-required items together. In many cases, the ac-
tion required is only that the Corps contracting officer be notified so
that he or she can determine the action to be followed. Whatever ac-
tion is taken, it should be done quickly in order to minimize the ripple
effect of the problem. Some of the Corps proposed actions, together
with some other solutions for nonspecification items, are listed below.

9.5.1 Aggregates
When an aggregate grading test is outside of specification limits, a
recheck sample is taken. If the recheck sample fails, the process is con-
sidered out of control and steps are taken to rectify the situation. If
not rectified, aggregate production and RCC placement can be
The owner may test for verification of field moisture determinations
made by the contractor using the oven-drying procedure (ASTM C
566) or other means. If a discrepancy exists, steps are taken immedi-
ately to identify the problem and correct it so that accurate and veri-
fiable field moisture determinations are obtained. Changes in water
entering the mix with the aggregates must be adjusted for at the mix-
ing plant.
When tests for material finer than the No. 200 (0.075 mm) sieve in-
dicate an excessive quantity, corrective action must be taken immedi-
ately, either in aggregate production or stockpiling.
If the aggregate does not meet grading limits during production
start-up, the nonspecification material may be used in developing a
base for a stockpile. After that it may possibly be blended with other
material to produce an acceptable aggregate or used for site improve-
ments such as road surfaces.

9.5.2 Batching and mixing plant

When weighing or volumetric proportioning accuracy does not meet
the specifications, the plant is shut down until necessary adjustments
Construction Control and Specifications 243

or repairs are made. When a mixer fails performance requirements,

either the mixing or chamber retention time can be increased or ad-
justments can be made to the mixing unit.

9.5.3 Compaction equipment

Rollers or tampers not meeting the weight and physical dimensions
specified must not be used. Rollers having improper frequencies must
be corrected before being used for RCC compaction. The operators of
rollers running at speeds in excess of specification limits should be or-
dered to slow down. If the problem persists, they should be replaced
with another operator.

9.5.4 Depositing and spreading

Whenever uncompacted RCC thickness measurements indicate an ex-
cess of material, the layer should immediately be bladed to establish
the proper thickness. Where a shortage of material is indicated, it
should be supplemented prior to compaction.
If segregation of the RCC mixture occurs during depositing or
spreading, the inspector should determine if it can be reblended with
the remaining RCC or wasted. Segregated RCC that has dried out
should be removed. Also, steps should be taken to correct the situation
that caused the segregation.

9.5.5 Density
Low density of compacted RCC usually can be corrected by additional
rolling. If the moisture content of the mixture or another factor is so
far off that required density cannot be achieved no matter how much
rolling is done, it should be removed. Unacceptable material should be
removed quickly before it gains strength or causes a slowdown in RCC
placement. Nonspecification RCC may be allowed to remain at the
owners discretion, but no payment should be made for the material.
Low density can be caused by one or a combination of the following
factors: insufficient rolling, low or high moisture content, an inappro-
priate vibratory frequency or amplitude for the material, a time delay
prior to rolling, poor gradation or segregation of the mixture, or incor-
rect testing.

9.6 Tests of Hardened RCC

Cores taken from a completed dam provide the most accurate repre-
sentation of RCC properties, the effectiveness of compaction and of lift
bonding methods. Construction control.funds should not be spent on
the preparation of traditional 6 x 12 in (152 x 304 mm) concrete cyl-
244 Chapter Nine

inders unless they are desired for historical purposes. Cylinders may
not represent the properties of the hardened RCC in the dam, they do
not give an indication of shear resistance or bonding at the lift lines
and they are evaluated well after any corrective action can reasonably
be taken.
A coring program is recommended for major dams as a way of track-
ing the success of the construction control program. Cores give an ac-
curate representation of compaction throughout lifts and the bonding
between successive lifts and material properties. The core diameter
should be as large as practical-6 in (150 mm) is considered the min-
imum-and cores should taken from representative locations within
the dam. Comprehensive construction records, such as when each Lift
was placed and under what conditions, when correlated with the cores
can provide valuable information for subsequent projects.
Because of the importance of recovering bonded lift joints and the
possibility of breakage at the lift lines during coring, a double-tube
core barrel is recommended. The coring program should be conducted
prior to reservoir filling. The core holes may serve a double purpose if
they can be used for drain holes or for the installation of inclinometers
or other instruments. The results obtained from the coring program
should be compared with the original design criteria to confirm the
structural stability of the dam.

American Concrete Institute, Roller Compacted Mass Concrete, Report 207~5R-89.
Bureau of Reclamation, Procedure for Consistency and Density of No-Slump Concrete
with Vibratory Table, Procedure 4905-86.
- Guidelines for Designing and Constructing Roller-Compacted Concrete Dams,
- Procedure for Determining Cement Content of Soil-Cement (Heat of Neutral-
ization Method)-Preliminary, 1988.
DeGroot, G., Soil-Cement Short Course-Quality Control and Field Testing Lecture
Notes and Example, U.S. Bureau of Reclamation, 1985 (unpublished).
Lawrence, D. J., Operations Guide and Modification Analysis for Use of the CE Con-
crete Quality Monitor on Roller-Compacted Concrete and Soil-Cement, U.S. Army
Corps of Engineers Civil Engineering Research Laboratory Technical Report M-854/
06 (Revised), July 1985.
Portland Cement Association, Concrete Pavement Construction-Inspection at Batch
Plant and Mixer, EB 084.OlP, revised 1980.
U.S. Army Corps of Engineers, Technical Provisions, Elk Creek Dam, Section 03660,
Corps if Engineer; Portl3ShCt.

Preliminary Design
and Cost Estimates

10.1 General Procedure

Now that RCC dams are a viable solution at many sites, planning en-
gineers require a relatively simple and reasonably accurate method
for the preparation of preliminary design and cost estimates. The
method presented in this chapter is generally based on the paper
Planning, Design and Cost Estimates for RCC Dams presented in
1988 by Tarbox and Hansen. It can best be applied to low and
moderate-height dams on rock foundations.
With a site selected and a presumed adequate foundation, the pro-
cess starts with an estimate of the depth of excavation. This is fol-
lowed by design of the gravity section to include consideration of seep-
age reduction methods. With a known canyon shape, the volume of the
structure can be calculated. The cost of the concrete section can then
be estimated based on unit bid prices received for the construction of
actual RCC dams in the United States.

10.2 Foundation Considerations

General requirements and methods of improving rock and nonrock
foundations are described in Sets. 3.2.2 and 3.2.3, respectively.
Nonrock foundations should be considered only for low RCC dams
in the preliminary stage of design. Foundation excavation guide-
lines for rock foundations are presented in Sec. In all cases,
overburden, such as soil, alluvium, or talus, should be removed
prior to excavation of the rock foundation material. The amount of
rock excavation required depends primarily on the quality of the

246 Chapter Ten

A designer can make accurate estimates of rock excavation require-

ments only if a large amount of subsurface exploration data are avail-
able. In the absence of this information, for preliminary estimates the
depth of rock excavation may be assumed to be 15 ft (4.5 ml for each
abutment and 5 ft (1.5 ml for the foundation at the base of the dam.
These assumptions can be used to calculate the volume of material re-
quired for an RCC dam. The result is sufficiently conservative for pre-
liminary design purposes.

10.3 Dam Cross Section

In preliminary layouts, the axis of the dam is generally located straight
across the valley. The dams cross section is basically triangular in shape
with a vertical upstream face and the downstream slope intersecting the
upstream face at the dams crest as shown in Fig. 10.1.
Figure 10.2 can be used for determining a downstream slope for a
dam with 100 percent efficient drains. Figure 10.2 is based on the fol-
lowing assumptions:

1. There is zero cohesion at the lift lines and at the dam foundation
contact. A redefined factor of safety of 1.5 is suggested for this con-
dition, which depends on the residual sliding strength of the RCC
for shear resistance.
2. The unit weight of RCC is 150 lb/ft3 (2403 kg/m31 or greater and
the unit weight of water is 62.4 lb/ft3 (1000 kg/m3).
3. Drains that are 100 percent efficient are located at approximately 10
percent of the cross-sectional base width in from the upstream face.
The drains reduce the uplift hydrostatic pressure to one-third the dif-
ferential head between the headwater and tailwater pressure.

Axis of dam



Figure 10.1 Typical RCC grav-

L _ ity dam cross section.

l-s, -I
Preliminary Design and Cost Estimates 247

Based on uplift pressure distribution vorying

from full reservoir head otthe upstream face,
reduced by 2/3 ot the droins
and no toilwoter

0.1 0.2 0.3 0.4 0 . 5 0 . 6 0 . 7 0.8 0.9 1 . 0

Downstream face slope (horizontal to vertical 1
Figure 10.2Shear friction factors of safety versus down-
stream face slope. [From Tarbox and Hansen (198LV.l

For preliminary design, a downstream slope for dams with

formed downstream faces of 0.7 H : 1 V may be used. If the down-
stream slope is planned to be unformed, a 0.8 H : 1 V slope is sug-
gested for preliminary design to accommodate the construction
method. If no drains are included in the design, Fig. 5.4 can be used
to determine an adequately safe downstream slope, especially for
higher dams.
If a roadway is planned across the dam, the crest width (Z,) should
be wide enough to accommodate the roadway. If no roadway is in-
cluded, a minimum of 10 ft (3 ml is suggested for dams less than 50 ft
(15 m) high. A minimum crest width of 12 ft (3.6 ml should be pro-
vided for dams 50 to 100 ft (15 to 30 m) high and a minimum of 16 ft
(5 m) should be used for higher dams to allow two construction vehi-
cles to pass safely. With the crest width determined, a vertical face is
used until it intersects the downstream slope.
Various seepage control and upstream facing methods are de-
scribed in Sec. 3.5.2. The actual method to be used may not have to
be determined at this level of study. Except for membrane-faced
concrete panels, the other methods of forming and providing seep-
age control at the upstream face have proved to cost approximately
the same (see Sec. 10.5.2). This includes the horizontal slip-formed
facing elements used for Upper Stillwater Dam. The system of
membrane-faced concrete panels together with some conventional
concrete has been bid higher than the other methods. Based on the
latter systems use at Winchester Dam, it has also performed bet-
248 Chapter Ten

10.4 Volume Determination

Once a dam cross section is determined, the volume of concrete in the
dam can be calculated if the cross-canyon surface profile is known and
rock excavation assumptions similar to that noted in Sec. 10.2 are ap-
plied. The total volume of the dam can be calculated either manually
or with the aid of a computer by slicing the dam vertically into a num-
ber of volumes and adding up the individual volumes.
Another method is to apply the equation below, assuming a cross
section as shown in Fig. 10.1. Possible idealized cross-canyon profde
shapes ranging from narrow to wide sites and from V- to U-shaped
canyons are shown in Fig. 10.3. Regardless of the canyon shape, the
volume can be approximately calculated by the equation. The total
volume is the sum of two volumes which are basically the areas of the
two triangles that comprise the dam section multiplied by a length.
The larger volume consists of the area of the larger triangle that ex-
tends to the base multiplied by the base length Lb plus the volume of two
pyramids whose base is the larger triangle which is vertical and whose
total height is L, - LB. The small volume is simply the area of the small
triangle at the crest multiplied by the crest length (L,). Thus the cross-
canyon distance at the base of the dam CL,) should be measured so that it
best fits the assumptions used to derive the equation.

v = W-O*(L, t 2&J + 3T,2L,


where: V = total estimated volume, yd3.

S = slope of the downstream face (horizontal to 1.0 rise)
H = height of maximum section in ft including an assumed 5
feet, of excavation below top of rock
L, = cross-sectional length of dam at its crest, including 15 ft of

Wide V-shape

_ LB _1

Figure 10.3 Range of typical cross-canyon profile shapes. [From Tczrbox

and Hansen (1988).1
Preliminary Design and Cost Estimates 249

excavation at each abutment measured normal to canyon

Lb = same as L,, measured at the base of the dam
S = slope of downstream face (horizontal to 1.0 vertical)
T, = thickness of crest in feet (see Sec. 10.3)
In metric units, the volume equation is:

v= WV& -+ 2Ld + 3T:Lt

where the measurements are in meters and the volume is in cubic

10.5 Estimating Cost

10.5.1 cost of RCC
RCC prices as submitted by the low bidder for 14 dam projects in the
United States are listed in Fig. 10.4. The table provides the cost of ag-
gregate and all other construction operations related to processing as
a single item separate from the cost of cement and fly ash. The total
cost per unit volume is then the sum of the aggregate and processing
cost plus the cost of the cementitious materials.
The cost per unit volume is a function of the volume required, with
lower costs associated with larger volumes. In order to be able to pro-
vide some consistency in plotting total cost versus volume, several ad-
justments were made to the data. All prices were escalated to 1989
cost levels using U.S. Bureau of Reclamation construction cost trend
factors, and a constant amount of cementitious material was used in
the RCC mix. The cost of RCC per cubic yard as shown in Fig. 10.5
represents the cost of aggregates, mixing, transporting, spreading,
compacting, and curing an RCC mixture containing 150 lb/yd3 (89 kg/
m3) of cement and 50 lb/yd3 (30 kg/m31 of fly ash or 175 lb/yd3 (104
kg/m31 of cement if no fly ash is planned for the project.
The cost curve is based on the assumption that suitable aggregates
are available at or near the dam site. If aggregates need to be im-
ported or a higher-cementitious-content RCC is desired, such as for a
high-paste-content RCC, the cost as determined from the curve should
be increased to account for the addtional cost of these items. Similarly,
if less cement and fly ash are desired in the RCC mix, the cost may be
decreased accordingly. In developing estimated costs for preliminary
designs, the user may want to add a contingency factor to the unit cost
of RCC derived from the curve. A 20 percent contingency factor was
selected for the example problem in Sec. 10.6.
Figure 10.4 RCC prices by low bidder.
RCC Cost per yd (U.S. $)
construction %XWPk
Prospect Bid date date As bid, yd and process cfnnent Fly ash Total
1. Willow Creek Dam, 10-23-81 1982 401,000 Average 11.56 6.23 (117X average @ 1.21 (39X @ 61.60/tori) 19.06
Oregon of four mixes 106.20Aon)
2. Austin Detention 6-28-83 1984 20,670 18.00 7.00 (200# @ 70.00/tori) 1.20 (80# @ 30.00hn) 26.20
Dams, Texas
3. Upper Stillwater 10-17-83 1983-87 1,357,OOO Average 10.78 5.37 (132t @ 81.50/tori) 6.59 (293# @ 45.00/tori) 23.81
Dam, Utah of two mixes 1.07
4. Winchester Dam, 12-13-83 1984 32,000 Bid at lump sum item - 32.50
Kentucky (175#)
5 Dolet Hills Plant 12-20-83 1984 26,123 Average of 27.00 5.60 (160X @ 70.00Aon) 0.91 (64X @ 28.50Aon) 33.51
Spillway, Louisiana two mixes
6. Galesville Dam, 3-14-84 1983 210,500 Average 15.56 3.95 (91X average @ 1.91 (87X average @ 21.42
Oregon of two mixes 87.00hn) 44.00hn)
7. Monksville Dam, 4-10-84 1986 289,000 Average 13.72 3.70 (108X average @ - 17.42*
New Jersey of two mixes 68.58hn)
8. Middle Fork Dam, 3-21-84 1984 35,000 - less than
Colorado 25.00
9. Grindstone Canyon 6-20-85 1986 114,500 20.20 4.38 (125R @ 70.00) 1.00 (50# @ 40.00)9 25.58
Dam, New Mexico
10. Elk Creek Dam, 1-16-86 1987-88 999,000 t 41,860 14.00 4.13 (118X @ 70.00/tori) 1.14 (56% @ 2.90/f& 19.56
Oregon t 0.29 admix
11. Lower Chase Creek 2-3-87 1987 26,830 Bid as lump sum (70X) 33.80
Dam, Arizona contract (105X)
12. Stacy Dam Spillway, 3-5-87 1988-89 103,800 18.05 4.62 (200# @ 46.17Aon) 1.20 (100X @ 23.83/tori) 23.87
13. Stagecoach Dam, 3-5-87 1988 43,500 23.00 4.80 (120X @ 80.00Aor-r) 1.54 (88# @ 35.00/tori) 29.34
14. Cuchillo Negro Dam, 9-19-89 1990 103,700 15.00 5.20 (130X @ BO/ton) 2.00 (lOO# @ IO/ton) 22.88
New Mexico t 0.68 Admix
NOTE : 1 yd3 = 0.765 m; 1 lb/yd3 = 0.593 kg/m3.
Cost of government-furnished sand for RCC mix.
+6-in. ( - 1 aggregate furnished from previous road contract--cost $6.75/yd3 (not included in bid).
*Includes 5.5 percent increase from actual bid due to one-year delay in award.
Actual mix contained an average of 135 lb of cement and no fly ash.
Average of three lowest bidders-low bidder at $19.41/yd3.
Preliminary Design and Cost Estimates 251

35.00 -

= 3 0 . 0 0 -

9 25 OO-
g cost = 1221
ro A A
Ox A
i$ 20.00- \
15 oo-

Volume of RCC, yd3

Figure 10.5 Costs of RCC per cubic yard. (Note: 1 yd3 = 0.76 ma.)

Discretion should be used in trying to apply the costs outside of the

United States, since all of the cost information was for construction of
projects in the United States, with most of the dams in the mountain-
ous western portion of the country. Also, the costs should be adjusted
for any base year other than 1989.

10.52 Cost of upstream face

Figure 10.6 represents the prices submitted by the low bidder for var-
ious upstream face designs. The costs as presented in the table are cal-
culated as either an added cost per cubic yard of RCC or as an added
cost per square foot of face. The cost of conventional facing concrete,
including cementitious materials, includes the cost of forming the ver-
tical upstream face in most cases. The costs listed are actual bid prices
and have not been escalated to 1989 or another base year. However,
all of the dams were bid between 1984 and 1987.
It can be seen from Fig. 10.6 that for dams requiring more than
100,000 yd3 (76,500 m3) of RCC, the added cost of facing per cubic yard of
RCC generally ranges from $4 to $6. In preliminary designs where no
specific upstream facing design has been selected, it is suggested a cost of
$6 per cubic yard be added to the cost of the RCC derived from Fig. 10.5.
Figure 10.6 Upstream face designs and costs. [From Tarbox and Hansen (1988)./

Facing Added
RCC volume concrete cost
Height, as bid, yd3 Added cost cost per per fts
Dam Location ft (m) (kg/m3) Type of face per yd3 RCC yd3* face
1. Winchester Kentucky 74 32,000 Precast concrete panels with $10.04 $ 8.50
(23) (24,500) 65-mil PVC membrane plus +2.34 $ so.00 t 1.98
18-in concrete backing
2. Galesville Oregon 162.5 210,500 Average 18-in-thick concrete 5.88 117.85+ 6.54
(50) (160,800) plus sprayed on membrane t 1.50
3. Grindstone Canyon New Mexico 139 114,500 Average 24-in-thick concrete 5.46 107.a1+ 7.99
(42) (87,500)
4. Monksville New Jersey 157 289,000 Average 24-in-thick concrete 5.38 95.69+ 7.09
(48) (221,000)
5. Upper Stillwater Utah 294 1,357,ooo Horizontal slip-formed 3.88* 58.47+ 4.03:
(91) (1,037,500) elements---al-average (both faces)
6. Elk Creek Oregon 249 999,000 Average 36-in-thick concrete 4.91 71.64+ 7.96
(76) (763,800) plus waterstopped joints t 0.58 jts.
7. Stacy spillway Texas 103 103,800 18-in concrete-reinforced 4.23 110.29+ 8.04
(31) (79,400) plus waterstopped joints
*Includes cement and pozzolan (if used).
Includes cost of forming the face.
Not considered as water barrier-added cost of high-paste RCC must be considered in cost
NOTE: Costs presented were submitted by the low bidder and have not been escalated to 1987 or later
cost levels. See Fig. 10.4 for bid date.
Preliminary Design and Cost Estimates 253

The cost of the concrete dam section can then be determined by mul-
tiplying the volume determined from the equations in Sec. 10.4 by the
unit RCC cost from Fig. 10.5 plus $6 per cubic yard to account for the
cost of the upstream facing method. In this preliminary design and
cost estimate, the volume of the RCC in the section displaced by the
facing has not been reduced and the cost of applying a bedding con-
crete between successive RCC lifts has not been added. These and
other items can be considered once the design is refined at the feasi-
bility level of study.
For dams with less than 100,000 yd3 (76,500 m3) of RCC, the cost of
the upstream face can be considerably more expensive when expressed
as an added cost per cubic yard of RCC. As noted in Fig. 10.6, this
added cost for the membrance-faced concrete panel system at Win-
chester Dam amounted to an increase of $12.38 per cubic yard in the
basic RCC cost.
At the 145-ft-high (44-m) Stagecoach Dam, which was built in a
narrow canyon, alternate bids were taken in 1987 for a formed con-
ventional concrete face averaging 24 in (0.61 ml thick and for a
membrane-faced concrete panel backed by an average of 18 in (0.45
m) of conventional concrete. In this case the cost of the formed con-
ventional concrete reflected an added $16.68 per cubic yard for the
43,500 yd3 of RCC or a 57 percent increase from the basic cost of
$29.34 per cubic yard for the RCC. For the now-patented panel sys-
tem, the added cost was $21.98 per cubic yard of RCC, or a 75 per-
cent surcharge.

10.5.3 Mobilization cost

Mobilization costs submitted on 10 dams in the United States have
ranged from an unbalanced low of 0.03 percent to a high of 10.43 per-
cent of the total low bid. Neglecting these extremes, mobilization costs
have been within a relatively narrow range from 4.70 to 7.97 percent
with an averaage of 5.83 percent of the total bid. In some cases, an
upper limit to the cost of mobilization that can be bid is specified in
the bid documents. If there is no mobilization bid item listed in a unit
price contract, it may be expected that the unit cost of RCC and other
items requiring plant or special equipment to construct must be in-
creased accordingly.

10.5.4 Total cost of dam project

From the previously presented equations and cost data, a simplified
design and estimated cost can be determined for an RCC dam that in-
cludes an upstream facing system.
254 Chapter Ten

It is more difficult to estimate the cost of a completed dam project

due to the varied requirements and many items involved. Some of the
items not considered in thesimplified estimating procedure presented
here include mobilization, reservoir clearing, diversion and water con-
trol, foundation excavation, grouting, drains, galleries, downstream
face, spillway, intake structure, outlet conduit, energy dissipators,
and instrumentation.
For the six U.S. projects in Figure 10.7, the cost of the RCC dam as
compared to the total project bid price has ranged from 40 percent for
Elk Creek Dam to 64 percent for Willow Creek Dam, with an average
of 51 percent.

10.6 Example Problem*

Given a dam site described by the topography as shown in Fig. 10.8a,
develop a preliminary design cross section and estimate the construc-
tion costs of an RCC gravity dam for the site. (The example calcula-
tions are made using U.S. customary units and dollars.)

Crest elevation of dam = 245 ft

Height of dam H = 200 ft
Crest width of dam T, = 16 ft
Drains will be included in dam and foundation
Angle of internal friction, 4 = 45
Cohesion in foundation and on lift lines C = 0
Shear friction factor of safety Q = 1.5
Factor of safety (SF) for compressive and tensile strength of
RCC = 3.0
Compressive strength of RCC, f, = 3000 lb/in2 at one year

Step 1. Draw an axis in plan on the dam site topography, Fig.

Step 2. Develop a profile of the canyon (original ground line) along
the axis looking downstream as shown in Fig. 10.8b.
Step 3. Sketch in an assumed line of excavation as shown in Fig.
10.8b, remembering to add 5 ft of excavation in the foundation and 15
ft of excavation at the abutments measured normal to the rock surface.

<Adapted from Tarbox and Hansen (1988).

Figure 10.7 Gross costs of RCC dams. [From Tarbox and Hansen (1988LJ

Volume, yd (ms) (as bid)

Total cost of RCC Project cost of
Upstream face and upstream face bid concrete/project
Dam RCC concrete* Total concrete price bid price, %
1. Willow Creek, 403,100 t 2,100 = 405,200 $9,040,000 $14,095,000 64
Oregon (308,200) (1,600) (309,800)
2 . Upper Stillwater 1,357,ooo t 35,200 = 1,392,200 34,788,OOO 60,604,OOO 57
Utah (1,037,500) (26,900) (1,064,400)
3 . Monksville 289,000 t 19,500 = 308,500 6,504,OOO 14,678,OOO 44
New Jersey (221,000) (14,900) (235,900)
4 . GaIesville 210,500 t 10,500 = 221,000 5,819,OOO 12,759,ooo 46
Oregon (160,800) (8,000) (168,800)
5 . Grindstone Canyon 114,500 t 5,800 = 120,300 4,237,OOO 7,477,ooo 57
New Mexico (87,500) (4,400) (91,900)
6 . Elk Creek Dam 1,040,800 t 69,500 = 1,110,300 25,366,OOO 62,783,OOO 40
Oregon (770,700) (53,200) (848,900)
*Includes bedding concrete where applicable.
256 Chapter Ten

Figure 10.6 Layout of a RCC gravity dam. (a) Plan; (b) profile
developed along axis looking downstream. (Note: 1 ft = 0.3
m.) (from Tarbox and Hansen)

Step 4. Measure length of dimensions L, and Lb at the top and bot-

tom of the dam, respectively.
Step 5. Entering the chart shown in Fig. 10.4 with a value of
tangent + equal to 1.0, select a value of 0.7 for the downstream
slope, S.
Step 6. Calculate the volume of RCC in the dam using the equa-
tion in Sec. 10.4 and the following parameters: H = 200 ft; T, = 16
ft; S = 0.7 ft; L, = 650 ft; Lb = 130 ft.

v = (0.7 x 200)2(650 t 2 x 130) t 3 x 16 x 650

162 x 0.7
= 161,700 yd3
Step 7. Enter the cost curve in Fig. 10.5 with the estimated volume
to find the estimated unit cost for the RCC dam in 1989 dollars, and
add $6 per cubic yard for the upstream face. A 20 percent contin-
gency factor was used in the example.

($6.00 t $26.20)/yd3 x 161,700 x 1.2 = $6,248,000

Preliminary Design and Cost Estimates 257

Step 8. To estimate a gross direct construction cost for a total

project, providing the project is fairly typical of the six cited in Fig.
10.7, divide the cost from Step 7 by 0.51.

Estimated total project construction cost = $12,250,000

It is emphasized that the above estimated total cost is highly depen-
dent on the complexity of the project. For example, a project with a
significant hydropower component would probably have a direct con-
struction cost larger than the amount predicted using the 51 percent

Bureau of Reclamation, Construction Cost Trends, Denver, Colo. October 1989.
Tarbox, G. S., and Hansen, K. D., Planning, Design, and Cost Estimates for RCC
Dams, Roller-Compacted Concrete II, AXE, New York, February 1988, pp. 21-38.

Data and Performance
of Completed RCC Dams

11 .l Data on Completed RCC Dams

The economic advantage of RCC combined with the long-term safety
record of concrete dams have led to rapid acceptance of RCC dams
throughout the world. By the end of 1989, data were available on 45
RCC dams greater than 50 ft (15 m) high or containing more than
13,000 yd3 (10,000 m3) of RCC that had been completed in 12 coun-
tries. They are listed in Fig. 11.1 with dimensions and some design
details. RCC dams have been completed on all six continents: North
America has 13; Asia, 13; Africa, 7; Europe, 6; Australia, 4; and South
America, 2.
In addition, seven major cofferdams have been constructed by the
RCC method and are listed in Fig. 11.2. They are listed separately
with less design information due to the relatively short design life for
these structures. Nevertheless, they are important dams and can con-
tain large volumes of RCC. In fact, the largest volumes of RCC in
dams at the planning stage are for the cofferdams for the proposed
Three Gorges project in China. If built, the 312-ft-high (95-m) longi-
tudinal cofferdam requires an estimated 2,220,OOO yd3 (1,720,OOO m3),
while the 279-ft-high (85-m) third-stage transverse cofferdam will
contain 1,860,OOO yd3 (1,420,OOO m3) of RCC.

11 .l .I Firsts and records for RCC dams

The following is a historical review of RCC dams, including a list of
the first dams built in each continent and country, the first RCC use
for certain aspects of a dam or dam project, first design and construc-
tion applications in an RCC dam, first performance test, and miscel-
laneous records for RCC dams.

Figure 11.1 Completed roller-compacted concrete dams.

Face of dam
RCC RCC Total Upstream Downstream Spillway
ClIl- State or Ht. ft Length, volume, volume,
Dam/river oleted orovince Owner/Designer h-d ft Cm) yds (ms) yds (ms) Slope Typo Slops Typo Slope Type
Shimaiicrawa/ 1980 Yamarmchi Jaean Ministry of 292 181 216,000 415.000 v/ (1) 0.80 (11 0.80 (11
ShimLjy Construction~tJMC) (89) (2401 (165,000) (317,000) 0.30
Tamagawa/Tama 1986 Akita JMC 328 1448 944,000 1,509,000 VI (1) 0.81 (11 0.81 (11
(1001 (441.51 (757,000) (1,154,000) 0.60
Mano/Mano 1987 Fukushima Fukushima Prefecture 226 104 135,000 286,000 V (1) 0.80 (1) 0.80 (11
(69) (239) (103,0001 (219,000)
Pirika/ 1987 Hokkaido JMC + Hokkaido Develop- 131 2966 212,000 471,000 VI (11 0.80 (11 0.80 (11
Shiribeshitoshibetsu ment Bureau (401 (9101 (162,000) (360,000) 0.80
Shiromizugawai 1988 Yamagata Yamagata Prefecture 179 1204 186,000 412,000 V (11 0.80 (11 0.80 (1)
Shiromizu (54.51 (3671 (142,OOOl t315,0001
Asahiogawa/O 1988 Toyama Toyama Prefecture 276 863 217,000 472,000 Vi (11 0.80 (11 0.80 (1)
(841 (2601 (160,000) (361,000) 0.80
NunomelNunome 1989 Nara Water Resources Develop- 236 1056 144,000 492,000 Vi (1) 0.76 (1) 0.76 (11
ment Corp. (721 (3221 (110,000) (370,0001 0.40
KengkouEingshan 1 9 8 6 Fujian Fujian Provincial Dept. 186 402 56,000 78,500 v (61 0.75 (9) 0.76 (1)
of Water Conservancy & (571 (1231 (43,000) (60,000)
Hydropower/Fujian Water
Conservancy &
Hvdrooower Investiaation
&De&n Institute -
Tianshengqiao 2/ 1988 Guizhou & Guishou & Guangxi 193 1542 171,000 421,000 V (1) 0.75 (9) 0.75 (11
Nanpan Guangxi ProvincesiGuiyang (59) (4701 (131,000) (322,000)
Hydroelectric Power
Investigation & Design
P Institute
Mahui/ Jialing 1989 Sichuan Sichuan Province/Sichuan 75 463 196,000 327,000 V (11 - 1.18 (11
Water Conservancy 84 (231 (1411 (160,0001 (250,000)
Hydropower Investigation
L IbrAn Tnat.it.nts
Longmentatiuxi 1989 Fujian Dehua County, Fujian 189 515 95,000 132,000 V/ (81 0.75 (9) 0.75 (11
Province/Fujian Water (581 (1571 (73,000) (101,0001 0.30
Conservancy &
Hydropower Investigation
Xc Design Institute
Tonljiezi/Dadu 1989 Sichuan Sichuan Province/Chendu 285 3310 589,000 3,544,ooo v (11 0.75 (9) 0.75 (11
Hydroelectric Power (87) (10091 ( 4 5 0 , 0 0 0 ) (2,710,OOOl
Investigation & Design
PajiakoulLuanhe 1989 Hebei Bureau of Panjiakou 80 801 27,000 39,000 v (11 0.60 (9) -
Control Worksflianjin (24.5) (2441 (21,OOOl (30,000)
Prospecting & Design


United States
Willow Creek/Willow 1982 Oregon U.S. Army chps of 169 1780 433,000 433,000 (21 0.80 (41 0.80 (41
Creek EngineersWaIla WaIla, (521 (543) (331,0001 (331,000)
Washington District
Winchester (now Carroll 1984 Kentucky Winchester Municipal 74 1192 32,000 35,000 (31 1.00 (41 1.00 (11
E. E&ml/ Upper Utilities/Palmer (231 (3631 (24,500) (27,000)
Howards Creek Engineering & Parrett,
Ely. & Hurt
Middle Fork/Middle 1984 Colorado Exxon Co. U.S.A.1 124 410 55,000 60,000 (11 0.80 (11 Conduit
Fork of Parachute Morrison-Knudsen (38) (1251 (42,000 (45,OOOl Stepped
Creek Engineers
GaIesville/Cow Creek 1985 Oregon Douglas County/ 162.5 950 210,000 223,000 (1) 0.80 (41 0.80 (11
Morrison-Knudsen (501 (290) (161,OOOl (170,0001
Grindstone Canyon/ 1986 New Village of Ruidoso/ 139 1416 115,000 126,000 (1) 0.75 (11 0.75 (11
Grindstone Creek Mexico Boyle Engineering (421 (4321 038,000) (96,000) Stepped
Monksville/Wanaque 1986 New Jersey No. Jersey Dist. Water 157 2200 287,000 304,000 (11 0.78 (41 0.78 (11
Supply Comm. & (481 (6701 (219,000) (232,OOOl Stepped
Hackensack Water Co./
OBrien & Gere
Lower Chase Creek/ 1987 Arizona Phelps Dodge Morenci, Inc. 64 400 18,000 29,000 (11 0.70 (11 0.70 (11
Lower Chase Creek + Sumitomo Metal Min- (201 (122) (13,800) (22,200) Stepped Stepped
ing Arizona/Dames Xc
Figure 11.1 Completed roller-compacted concrete dams. IContinued)

Face of dam
RCC RCC Toted upstream Downstream Spillway
COTI- State or Ht, R Length, volume, volume,
Dam/river &ted province Owner/Designer (ml ft (ml yd cm31 yds (msl Slope Type Slope Type Slope Type

Upper Stillwater/Rock 1987 Utah U.S. Bureau of 294 2673 1,471,oop 1,675,OOO V (7) 0.32/ (7) 0.321 (7)
Creek Reclamation (90) (815) (1,125,OOOl (1,281,OOOI 0.60 0.60
Elk Creek/ Elk Creek 1988 Oregon U.S. Army Corps of 83 1197 348,000 455,000 V (1) 0.80 (71 0.80 (1)
Engineers (as halted) (25) (365) (266,000) (348,000)
Portland District (as 249 2580 1,041,000 1,100,100
planned1 (76) (7861 (796,000) (841,000)
Stagecoachfllampa 1988 Colorado Upper Yampa Water 150 380 44,000 50,500 (1) 0.80 (1) 0.80 (1)
Conservancy District/ (461 (115) (34,000) (38,500) Stepped Stepped
Woodward Clyde
Stacy (now SW. Freesel 1989 Texas Colorado River Municipal 103 568 117,000 209,000 (11 0.831 (1) 0.831 (1)
Spillway/Colorado Water District/Freese & (31) (1731 (89,000) (160,000)
Marmot (replacement)/ 1989 Oregon Portland General Electric/ 50 194 10,300 13,000 (11 0.80 (11 0.80 (1)
Sandy Ebasco Services (15) (59) (8,000) (10,000)
L a Manzanilla/Ibarilla 1987 Leon Secretaria de Agriculture 118 492 65,400 V/ (1) 0.80 Ill 0.80 (11
y Recursos Hidraulicos (361 (1501 (50,000) 0.24
Copperfield/ Copperfield 1984 Queensland Kidston Gold Mine, Ltd./ 131 1115 183,000 205,000 V (1) & 0.901 (4) 0.80 (1)
Gutteridge, Haskins & (401 (3401 (140,000) (157,000) (51 0.80
CraigbourneiCoal 1986 Tasmania Tasmanian Rivers & 82 810 29,000 29,000 v (21 1.00 (41 Separate
Water Commissionl (251 (2471 (225001 (22,500)
Gutteridge, Haskins, &
Bucca Weir/K&n 1987 Queensland Queensland Water 39 420 31,400 39,300 1.00 (4) Spillway 0.60 (1)
Resources Commission/ (12) (124) (24,000) (30,000) Stepped
Gut&ridge, Haskins, &
Wrights Basin/Point 1989 Australian Australian Capitol 59 282 11,900 11,900 v (2) 1.0 (4) Conduit
Hut Creek Capital Territory Government/ (18) (86) (9,100) (9,100)
Territory Willing & Partners



Castilblanco de 10s 1985 Seville Junta de Andalucia/Hydro- 82 404 18,000 26,000 V (1) 0.75 (1) 0.75 (1)
ArroyosiRivera de Cala graphic de Guadalqvirir (25) (123) (14,000) (20,000)
Erizana (dike) 1987 Galicia Confederaci6n Hidraulica 50* 361 15,700 85,000 -
de1 Norte de Espana (15) (110) (12,000) (65,000)
Los Morales/Morales 1987 New Confederacibn Hidro- 92 656 29,000 V (1) 0.75 (5) 0.75 (1)
Castile grafica de1 TajoiOCISA (28) (200) (22,000) stepped
Santa EugeniaiXallas 1988 La Corwia Sociedad EspaAola de 281 935 286,000 306,000 0.05 (5) 0.75 (5) 0.75 (1)
Carburos Metzilicos S.A./ (85.5) (285) (219,000) (234,000) Stepped


Les Olivettesi 1987 Bas Rhone Conseil General 118 837 105,000 111,000 v (1) 0.75 (10) 0.75 (1)
Tributary of Herault delHerault/ Compagnie (36) (255) (80,000) (85,000) Stepped
Nationale du Bas Rhone
et du Langueduc


IashkumiriNarin 1988 Kirgizia Gidroproject 246 1050 111,000 - V (1) 0.75 (1) 0.75 (1)
(75) (320) (85,000)


South Africa

De Mist Kraal 1986 Cape Department of Water 98 984 46,000 85,000 V (1) 0.60 (1) 0.60 (1)
Diversion/Little Fish Affairs (30 (300) (35,000) (65,000) SkPPed Stepped
Arabie (now 1 9 8 6 Lebowa Department of Coopera- 118 1493 132,000 185,000 V (1) 0.751 (1) 0.75 (1)
Mokgoma Matlala/ tive DevelopmenVTheron, (36) (455) (101,000) (141,500) 0.50
Oliphants Prinsloo, & Van Tonder Stepped

Figure 11 .l Completed roller-compacted concrete dams. (Continued)

Face of dam
RCC RCC T&d upstream Downstream Spillway
com- State or Ht, ft Length, volume, volume,
Dam/river pleted province Owner/Designer (ml ft (m) yd (m3) yds (ms) Slops Type Slope Typo Slops Type
A F R I C A (Continued)
South Africa (Continued)
ZaaihoeWSlang 1987 Natal Department of Water 154 1729 127,000 175,000 v (1) 0 . 6 2 (1) 0.62 (1)
Affairs (47) (527) (97,000) (134,000) Stepped Stepped
Knellpoort (gravity 1988 South Department of Water 164 656 59,000 77,000 v (1) 0 . 6 0 (1) 0.60 (1)
arch)/Rietspruit Eastern Affairs (50) (200) (45,000) (59,000) Stepped Stepped
Wohvedans (gravity 1989 Cape Department of Water 230 819 (196,000) (221,000) v (1) 0 . 5 0 (1) 0.50 (1)
arch)/Great Brak Atfairs (70) (268) (150,000) (169,000) Stepped StePPad
Ain Al Koreima/ 1988 Rabat Direction des 85 407 34,700 39,000 0.20 (1) 0.60/ (2) 0.75 (1)
Akruech Amenagements (26) (124) (26,500) (30,000) 0.75 (5) Stepped
RwedatJ- 1988 Rabat Direction des 76 410 33,000 (35,000) 0.40 (11) 0.40 (11) 0 . 4 0 (11)
Amhmgemente (23) (125) (25,000) (27,000) StePpad Stepped Stepped
Saco de Nova Olinda/ 1986 Paraiba State of Paraibailcoplan 184 755 173,000 187,000 V (1) 0.80 (4) Separate
Gravata (56) (230) (132,000) (143,000)
Urugua/Urugua-i 1989 Misiones Electricidad de Misiones 249 2254 772,000 819,000 V (3) 0.80 (4) 0.80 (1)
Sociedad Anonimal (76) (687) (590,000) (626,000)
NOTE : Type of face for dam: (1) conventional concrete; (2) precast concrete panels; (3) precast concrete panels with membrane plus conventional concrete;
(4) unformed RCC; (5) RCC against forms; (6) precast concrete panels plus asphalt mortar; (7) conventional concrete slip-formed or extruded elements; (8)
Conventional concrete-expansive cement; (9) precast concrete blocks; (10) mechnically compacted unformed RCC; (11) RCC with wire netting.
*Main gravity dam is 148 ft (45 m) high and 591 ft (186 m) long.
Figure 11.2 Completed roller-compacted concrete cofferdams.
com- Country/ Ht, Length, volume
pleted Dam/River province Owner/designer (ml ft. (m) yd (ma) Remarks
1980 Tarbela spillway Pakistan/ Pakistan Water & Power 30 850 50,000
repairs/Indus District Development Authority/ (9) (263 (38,000)
Abbottabad Tippetts-Abbett-McCarthy-
Stratton (TAMS)
1987 Sir/- upstream/ Turkey Cukuroya Elektrik S.A. 131 344 52,000
Ceyhan (40) (105) (40,000)
1988 Yantan upstream/ China/ Guangxi Bureau of Electric 172 1220 216,000 Curved
Hongshui Nanning Power IndustrylGuangxi (53) (372) (165,000) upstream
1988 Yantan downstream/ China/ Hydroelectric Investigation 132 144,000
Hongshui Nanning & Design Institute (40) (3201 (110,000)
1988 Geheyan upstream/ China/ Hubei Province/Yangtze 132 558 178,000 CUNd
Qingjiang Hubei Valley Planning Office (40) (170) (136,000) upstream
1988 Shuikou diversion China/ Fujian Electric Department/ 82 656 371,000
walI/Minjiang Fujian East China Hydroelectric (25) (200) (284,000)
Power Investigation &
Design Institute
1989 Dongfeng upstream/ China/ Guizhou Hydropower 295 26,200
Wujiang Guizhou DepartmenUGuiyang (90) (20,000)
Hydropower Investigation
& Design Institute
266 Chapter Eleven First RCC Dams by Location (Year RCC Completed)

1. Asia:
a. Japan, Shimajigawa (1980)
b. China, Kengkou (1986)
c. Turkey, Sir Cofferdam (1987)
2. North America:
a. United States, Willow Creek (1982)
b. Mexico, La Manzanilla (1987)
3. Australia:
a. Copperfield (1984)
4. Europe:
a. Spain, Castilblanco de 10s Arroyas (1985)
b. France, Les Olivettes (1987)
c. USSR, Tashkumir (1988)
5. Africa:
a. South Africa, DeMist Kraal Diversion (1986)
b. Morocco, Ain Al Koreima (1988)
6. South America:
a. Brazil, Saco de Nova Olinda (1986)
b. Argentina, Uruguai (1989) First specific RCC application in a

dam project
1. Gravity retaining wall: Yale Hydroelectric Project, United States,
2. Central core in embankment cofferdam: Shihmen Dam, Taiwan,
3. Foundation build up: Cochiti Dam, United States, 1968.
4. Emergency repairs: Tarbela Dam, Pakistan, 1974.
5. Overflow structure: Chena Dam, United States, 1978.
6. Cap for embankment cofferdam: Revelstoke Dam, Canada, 1979.
7. Main portion of cofferdam: Tarbela Dam, Pakistan, 1980.
8. Main portion of gravity dam, RCD method: Shimajigawa Dam, Ja-
pan, 1980.
9. Rehabilitate overflow section (timber crib dam): Ocoee No.2 Dam,
United States, 1980.
10. Entire gravity dam, lean RCC dam: Willow Creek Dam, United
States, 1984.
11. Spillway in a new earth dam: Dolet Hills Dam, United States,
Data and Performance of Completed RCC Dams 267

12. High-paste RCC dam: Upper Stillwater Dam, United States, 1985. First design or construction application in an RCC dam project

1. Design concept
a. Similar to conventionally placed concrete dam: Shimajigawa
Dam, Japan, 1980.
b. Exposed RCC downstream face and spillway: Willow Creek
Dam, United States, 1982.
c. Design for zero cohesion in shear: Copperfield Dam, Australia,
2. Upstream face design
a. Waterstopped transverse joints in conventional concrete face
and transverse joints in RCD: Shimajigawa Dam, Japan, 1980.
b. RCC placed against fill: Tarbela Dam cofferdam, Pakistan,
c. Precast concrete forms, no transverse joints in RCC: Willow
Creek Dam, United States, 1984.
d. Crack inducers in conventional concrete face: Middle Fork
Dam, United States, 1984.
e. Plastic-membrane-faced precast concrete panels: Winchester
Dam, United States, 1984.
f. Unformed RCC on sloped face: Bucca Weir, Australia, 1985.
g. Horizontal slip-formed elements: Upper Stillwater Dam, United
States, 1985.
h. Asphalt mortar behind precast concrete panels: - Kengkou,
China, 1986.
i. Expansive cement concrete: Longmentan, China, 1989.
j. Reinforced concrete and waterstopped joints: Stacy Dam Spill-
way, United States, 1989.
3. Foundation
a. Nonrock foundation: Cedar Falls Dam, United States, 1986.
b. On RCC mat foundation: Lower Chase Creek Dam, United
States, 1987.
4. Dam shape
a. Curved in plan (partial): Saco Dam de Nova Olinda, Brazil, 1986.
b. Curved gravity dam: Knellpoort Dam, South Africa, 1988.
5. Construction methods or features
a. Bedding mortar on every lift: Shimajigawa Dam, Japan, 1980.
b. Inclined railway transport of RCC: Tamagawa Dam, Japan,
c. Conveyor transport of RCC and stepped conventional concrete
downstream face: Middle Fork Dam, United States, 1984.
d. Special vibrating device for compacting stepped RCC down-
stream face: Les Olivettes Dam, France, 1987.
266 Chapter Eleven

11 .I .1.4 First performance test for an RCC dam

1. Reservoir filled: Shimajigawa Dam, Japan, 1981.
2. Overtopped after construction: Kerrville Ponding Dam, United
States, 1985.
3. Overtopped during construction: Craigbourne Dam, Australia,

11 .l .1.5 Miscellaneous records for RCC dams

1. Highest: Tamagawa Dam, Japan, 328 ft (100 ml.
2. Longest: Pirica Dam, Japan, 2986 ft (910 ml.
3. Largest RCC volume: Upper Stillwater Dam, United States, 1.471
million yd3 (1.125 million m31.
4. Greatest RCC volume placed in 24-h period, dam repairs: Tarbela
Dam, Pakistan, 23,500 yd3 (18,000 m31, 1975.
5. Greatest RCC volume placed in 24-h period in gravity dam: Elk
Creek Dam, United States, 12,392 yd3 (9474 m31, 1987.
6. First RCC dam constructed in one day: Dryden Weir, United
States, 1986.
7. Largest-capacity reservoir: Urugua-i, Argentina, 950,000 acre-ft
(1175 X lo6 m3).

11.2 Performance of Completed RCC Dams

It is important for designers to obtain accurate records or reports of
performance of dams in service so as to compare actual with predicted
performance and to learn from unanticipated or unsatisfactory perfor-
mance. Doing so moves the state of the art forward.
RCC dam designers are to be complimented on the amount of infor-
mation published on the performance of their dams. While most of the
results have been positive, there have been some negatives, especially
with respect to seepage and seepage related phenomena. Because this
information was published, engineers could quickly apply what had
been learned to more recent designs.

11.2.1 Structural performance

For the dams which have been subjected to full reservoir load, there
have been no failures or unanticipated movements or deformations in
the RCC structures. Following initial filling at Upper Stillwater Dam,
a downstream movement of 0.4 in (10 mm) was measured in a
Data and Performance of Completed RCC Dams 269

multiple-position borehole extensiometer in the foundation rock about

30 ft (10 m) below the upstream face and near the center of the spill-
way. There was no movement measured at the toe of the dam at the
same station, and there has been no further movement at the up-
stream face since early 1988. The movement has not been considered a
safety problem. It has been theorized that the movement may be due
to the closing of vertical fissures in the rock due to the horizontal force
of the reservoir on the dam.
There is no record of an RCC dam having been shaken by a signif-
icant earthquake to date, even though seven RCC dams have been
completed in Japan and others are located in seismically active areas.

11.2.2 Seepage
This section deals with measured overall seepage through RCC
dams and their foundations as contrasted to permeability values of
the RCC material. Permeability of RCC mixtures is discussed in
Sec. 2.5.3 and initial permeability values for certain RCC dams are
plotted in Fig. 5.1. Seepage in this discussion refers to the total
measured volume of water passing through the dam and adjacent
foundation material. Leakage is water passing through joints and
cracks in the structure.
To evaluate overall seepage through or around an RCC dam and its
foundation properly, information on the designed seepage collection
and seepage control system is listed in Fig. 11.3 for the dams discussed
The measured seepage is an indication of the performance of the de-
signed seepage control system. The total measured seepage can con-
sist of the following items: (1) leakage through joints and cracks, (2)
seepage through a conventional concrete face if used and the RCC ma-
terial itself, and (3) seepage through foundation materials. The water
passing through the RCC depends upon the void characteristics of the
mixture in addition to construction-related voids, such as those pro-
duced by segregation of large aggregate at or near the bottom of the
lift and possibly inadequate compaction of a lift.
Most measurements of seepage have been from weirs located in the
gallery, at the downstream gallery entrance tunnel, or at a point in
the waterway downstream of the dam. Some seepage that may proceed
through the dam has been measured by weirs at the intersection of the
dam with the abutment just prior to the water being deposited into the
stilling basin or other waterway. Comparing seepage on an equal basis. To compare seepage

through various dams on an equal basis, the size and shape of the
Figure 11.3 Seepage collection and seepage reduction systems for various RCC dams.

hydraulic Max. wetted
Dam PuIpCLse head, ft (m) area, Its (mc) Seepage control system Seepage collection system

Shimajigawa Flood and river 246 89,300 Upstream facing concrete, 10 ft (3m) Drain pipes downstream of water
control (75) (8290) thick stops in contraction joints
Municipal water Double waterstop in contraction Drainage gallery
storage joints through entire section at 49 ft Foundation drainage curtain
(15 m) on center
Bedding concrete entire lift
RCC interior
Foundation grout curtain
Willow Creek Flood control 141 149,000 Zoned RCC Bedding concrete 1 ft (0.3 Drainage gallery
(43) (13,900) m) wide upper portion of dam Foundation drainage curtain
Copperfield Industrial water 95 62,100 Thin wedge of conventional concrete Drainage gallery
storage (29) (5,770) (not full height of lift) Geofabric drain tubes
Zoned RCC Internal and foundation drainage
Three PVC waterstopped joints at curtain
key locations Geofabric strip drain every fourth
Foundation grout curtain lift near downstream face connected
to voids produced by form jacking
Middle Fork Flood control 94 16,800 Upstream facing concrete +1.5-R- Drainage gallery/tunnel
C29) (1560) thick (0.46 m) Internal and foundation drainage
Partially caulked crack-inducer curtain
grooves 12 ft (3.7 m) on center Geofabric drain tubes near
Bedding concrete 6 ft. (1.8 m) wide downstream face below gallery
RCC elevation
Foundation grout curtain Porous concrete zone hear
downstream face
Winchester Municipal water 61 29,600 PVC lined (65-mil = 1.65-mm) precast Geofabric/crushed rock toe drain
(now Carroll storage (19) (2750) concrete upstream facing panels
E. Ecton) Conventional concrete
f 1.5~R-thick (0.46 m) downstream of
Pl.nrl.tin mm,,+ f.llrtSi
Facing concrete + 2 ft (0.6 ml thick
Bedding concrete 6 ft(1.8 ml wide
Zoned RCC
Foundation grout curtain

Grindstone Municipal water 128 90,400 Upstream facing concrete f 2 ft Drainage gallery
Canyon storage (39) (0400) (0.6 m) Internal and foundation drainage cur-
Partially caulked crack-inducer tain
grooves 16 ft(4.9 ml on center
Bedding concrete
Foundation grout curtain
Monksville Municipal water 141 131,000 Upstream facing concrete + 1.8 ft. (0.55 Partial drainage gallery
storage (431 (12,lOOl ml thick Internal and foundation drainage
PVC waterstops in crack inducers at curtain
20-R (6.1-m) and 40-R (12.2-m) spac-
Contraction joints through dam at
120~ft (3.7-m) spacing upper 40 ft
(12.2 ml
Bedding concrete 8 R (2.4 ml wide
plus hydrophytic rubber strip at end
Foundation grout curtain
Arabic (now Agricultural and 98 Upstream facing concrete 22.5 ft (0.76 hainage sallery
Mokgoma municipal water (301 ml thick Internal and foundation drainage
Matlalal strage PVC waterstops in crack inducers at curtain
3.9 ft(12 ml on center
Bedding concrete H/4 wide
F o u n d a t i o n grout curtain

Upper Municipal, 280 528,000 Slipformed upstream concrete Drainage gallery/tunnel

Stillwater agricultural and (88) (49,000) elements 2 ft (0.6 ml average Foundation dminage tunnel
Industrial storage thickneea
Zoned high-paste RCC
Y Foundation group curtain
272 Chapter Eleven

dams upstream face exposed to reservoir pressure must be taken

into account. A unit seepage factor whose units are in gallons per
minute per 10,000 square feet of wetted surface area per foot of av-
erage head (liters per second per 1000 m2 per meter of average head)
is used. The average head is the vertical distance from the water
surface to the centroid of the wetted surface area and is th4 level of
average pressure acting upon the face. Thus, as the reservoirs wa-
ter surface fluctuates, both the wetted area and average head
The location of the average pressure is different from the location of
the average force acting upon the structure. The location of the aver-
age pressure is generally between one-third to one-half the total hy-
draulic head at any reservoir level. The one-third factor corresponds to
a triangular-shaped upstream face while one-half is for a rectangular
face. For a dam with a rectangular shaped face, the location of the re-
sulting hydraulic force is two-thirds the head measured from the wa-
ter surface. Unit seepages for various RCC dams plotted versus time
after initial filling are shown in Fig. 11.4.
Because there is considerable variation in the accuracy of seepage
measurements, due mainly to variations in what is being measured,
relative unit seepages for each dam and reductions over time are more
significant than the absolute values. Factors which overestimate the
actual amount of water passing through the RCC structure are flows
through foundation and abutment rock that are collected from foun-
dation drains or drainage adits and downstream rainfall that is col-
lected and flows through the measuring device. Water passing
through the dam that evaporates at the downstream face or is not col-
lected in a downstream weir leads to underestimates of actual seep-
Low reservoir levels may produce higher unit seepages due to the
increased effect of water passing through foundation rock when com-
pared to that acting upon and seeping through the dam. In Fig. 11.4,
the degree of seepage reduction with time for a certain dam should be
helpful in predicting the future seepage performance for a recently
completed dam constructed with similar seepage control measures and
RCC mixtures. Seepage history for specific RCC dams. Information on seep-

age measurements with time together with remedial action where re-
quired is available from the following dams in service. The dams
listed were all designed as lean-RCC dams with the exception of
Shimajigawa Dam (RCD method) and the high-paste RCC design for
Upper Stillwater Dam. The RCC mixtures used to construct each of
the dams is listed in Fig. 2.14.
Data and Performance of Completed RCC Dams 273



Middle Fork

- ,- _
I-..,_ hht
..-. Plnitf=Y-l
- - -

mOx u-tit SeeWae = 0.009

(8 yrs)
Winchester-no seepage reported (4yrs) 22.3 1

6 Willow Creek (after grouting)
c Initial unit seepage = 42 h

in- Grindstone

1.2 - /fv Copperfield


).l I I I I I I I I 0.22
0 6 12 18 24 30 36 42 48 54
Months (after statiof initial filling)

re I I A Unit seepage versus time for various RCC dams.

1. Shimujiguwa Dam. Seepage has been continuously measured

i monitored since its initial reservoir filling in 1981. No seepage
; been visible at its downstream face, and the amount of seepage
m foundation drains has never exceeded 8 gal/min (0.5 L/s). The
kage from the contraction joints is as small as that experienced at
ventional concrete dams in Japan.
2. Willow Creek Dam. When the reservoir for this.dam, which is
marily for flood control, was initially filled during the spring of
33, seepage in the drainage gallery and at the downstream face was
#iced within 12 h after filling. The initial total seepage, which in-
des water exiting the gallery and stillway spilling basin, was
n-ly 3000 gal/min (189 L/s) with a water depth of 48 ft (14.7 m). This
lduced an initial unit seepage value of 42 gallmin per 10,000 ft of
tted area per foot of average head (93.5 L/s per cubic meter per
ter). Over a two-month period, the total measured seepage dropped
2380 gal/min (150 L/s).
274 Chapter Eleven

At that time, the U.S. Army Corps of Engineers decided to reduce

seepage by grouting. In preparation for the grouting, the reservoir
level was lowered to 35 ft (10.6 m). An initial chemical grouting pro-
gram from the upstream face did not significantly reduce seepage. A
subsequent cement grouting program from the dam crest proceeded
during the second half of 1983 and early 1984.
Following the completion of grouting, the water level was raised to
the original water depth of 48 ft (14.7 m) and then to 95 ft (28.9 ml. At
the lower level, the total seepage was about 135 gal/min (8.5 L/s) com-
pared to the seepage prior to grouting of 2380 gal/min (150 L/s). At the
higher reservoir level, the total seepage collected was 2030 gal/min
(128 L/s), which takes into account the higher head and a greater wet-
ted upstream surface area which was exposed to seepage water for the
first time. Seepage was noticeable on the downstream face of the dam
as shown in Fig. 11.5. After one year with the reservoir at about the
same 95-ft (28.9-m) level, the seepage had reduced to 465 gal/min
(29.3 L/s). The following two years saw even further reduction of the
seepage to 340 gal/min (21.5 L/s) and then 250 gal/min (15.9 L/s), re-
With unsealed joints between the upstream precast concrete form-
ing panels, the RCC was considered to be the primary water barrier.
The intial seepage was attributed to voids at the lift lines caused by
coarse aggregate segregation, which in turn contributed to decreased

Figure 11.5 Willow Creek Dam (April 1984).

Data and Performance of Completed RCC Dame 275

bond between the untreated successive lifts of RCC A bedding mix

placed between successive lifts above the level of the gallery was not
effective because it was too narrow, about 12 in (300 mm) wide, and
located too close to the upstream facing panels for RCC to be properly
compacted into the bedding mix. The reduction in seepage with time
was determined to be caused by silting, calcification (production of cal-
cium carbonate), additional maturity, as well as the grouting pro-
3. Copperfield Dam. Initial filling of this dam, which was con-
structed to provide water for gold mining operations, started in late
1984. With a water depth of 69 ft (21.0 m), the total measured initial
seepage was 391 gal/min (24.7 L/s). The seepage was measured in V-
notch flume located about 0.6 mile (1 km) downstream of the dam.
Although the reservoir level was raised to 78 ft (23.7 m), the seepage
had reduced to 282 gal/min (17.8 L/s) after one month. Two months
later it had further reduced to 162 gal/min (10.2 L/s). Seven months
following initial filling, a vertical transverse crack developed through
the spillway which increased seepage to 254 gal/min (16.0 L/s).
The downstream face of the dam is shown in Fig. 11.6 shortly after
development of the crack. Some seepage has bypassed the vertical
drain holes and is draining through the spillway facing drainage sys-
tem. Seepage decreased by natural autogenous healing (calcification)
to a value of 89 gal/min (5.6 L/s) in late 1986 and had further reduced
to 34 gamin (2.15 L/s) by mid-1989.

Figure 11.6 Copperfield Dam (July 1985).

276 Chapter Eleven

4. Middle Fork. Middle Fork Dam was constructed in 1984 to

provide flood protection and supplemental water supply for a proposed
shale oil mine and processing plant. The reservoir was filled in the fall
of 1984 to produce a water depth of 44 ft (13.4 m). At that time an
initial seepage of about 150 gal/min (9.5 L/s) was measured, which re-
duced to 110 gal/min (6.9 L/s) after 2% months.
The reservoir was raised in the spring of 1985 to the elevation of the
primary spillway, which produced a water depth of 86 ft (26.2 m). To-
tal seepage then peaked at 475 gal/min (30.0 L/s), but decreased to less
than 48 gal/min (3.0 L/s) 18 months later. This was less than the 50 to
125 gal/min (3.2 to 7.9 L/s) predicted during design. The seepage re-
duction was attributed to calcification of the RCC mass and was most
rapid during the first 3 months after complete reservoir filling.
At the time of the maximum seepage, it was determined that 80 per-
cent of the total seepage was through the dam as collected from roof
and floor drains that exited in the gallery or from the gallery walls.
The remaining 20 percent was collected from drainage tunnels con-
structed in each abutment. After four years of service, the amount
passing through the RCC dam is estimated to be about equal to that
passing through the rock abutments due to the decreased permeabil-
ity by further calcification within the dam.
5. Winchester (now Carroll E. Ecton) Dam. When the reservoir
for this municipal water supply dam was initially filled in 1985, no
seepage was measured or noticeable through the dam. However, leak-
age through the jointed limestone foundation was considered exces-
sive. The water level was lowered and remedial cement grouting of
the foundation was performed. The reservoir was filled again, and af-
ter four years there has been basically no seepage through the dam,
indicating the effectiveness of the membrane-lined precast concrete
panel upstream facing system. The very small amount of water col-
lected downstream of the dam has been determined to be passing
through the foundation rock.
6. Galesville Dam. Initial filling of this multipurpose reservoir,
which includes hydroelectric power generation, started in December
1985. By late March 1986, a hydraulic head of 110 ft (33.5 m) had been
achieved and the maximum total seepage was measured as 713 gal/
min (45 L/s). The total seepage reduced over the following 10 months
to a value of 315 gal/min (20 L/s), which was attributed to siltation
and calcification maturing of the RCC.
In early 1987, the reservoir was raised to about 5 ft (1.5 m) below
the spillway crest, thus producing a head of 134 ft (41 ml. Seepage in-
creased to a maximum of 950 gal/min (60 L/s), not including seepage
exiting the downstream face, which could not be measured. Most of
the leakage was through seven major cracks in the dam. A coal-tar-
Data and Performance of Completed RCC Dams 277

based elastomeric membrane that was sprayed on the upstream face

in two 20-mil-thick (O.&mm) layers did not effectively bridge the
cracks, but was effective in reducing seepage through the area be-
tween the cracks.
A reduction in seepage to 790 gal/min (49.8 L/s) was attained after
pelletized bentonite was dumped from a boat into the reservoir near
the widest crack soon after recording the maximum seepage. This
worked well because there was sufficient velocity of water to draw the
bentonite into the crack. Then, a further decrease in seepage to 510
gal/min (32.2 L/s) occurred after divers caulked the cracks to a depth
of 50 to 60 ft (15 to 18 m) below the water surface using quick-set ce-
ment. Since reaching its maximum water level in early 1987, the res-
ervoir has been steadily lowered due to less inflow. In the spring of
1988, when a head of 110 ft (33.5 m) was reached, a seepage of 200 gal/
min (12.6 L/s) was recorded. This compares to the 713 gal/min (45 L/s)
seepage with the same head during initial filling two years earlier.
7. Craigbourne Dam. Initial seepage measured at the down-
stream toe of this irrigation dam was 140 gal/min (8.81 L/s) when the
hydraulic head was 61 ft (18.5 m) following filling that started in Oc-
tober 1986. Five months later, the seepage had reduced to 43 gaI/min
(2.7 L/s). This seepage reduction was attributed to autogenous calcifi-
cation healing. After two years of reservoir operation the seepage had
further reduced to 24 gal/min (1.5 L/s).
The design for Craigbourne, unlike Copperfield by the same de-
signer, included precast concrete panels rather than conventional
unreinforced concrete for the upstream face. Both designs incorpo-
rated a 6 ft 8 in wide (2 m) partial bedding concrete between each RCC
lift placed near the upstream face.
8. Grindstone Canyon Dam. Although the RCC for Grindstone
Canyon Dam was completed in July 1986, filling of the municipal-
water-supply reservoir did not start until late March 1988. The delay
in initial filling was due to a number of reasons, including a missing
outlet works valve, a dispute over water rights, and the need to seal
some cracks. By the latter part of May, the water level was 93 ft (28
m) above the reservoir floor and total seepage measured at a flume in
the channel downstream of the dam was 1146 gal/min (72.3 L/s). Dur-
ing June, the water level continued to rise and the seepage exceeded
the capacity of the flume and was estimated to have reached a maxi-
mum of 1400 gal/min (88 L/s).
Calcification and siltation together with repair of one hole in the
RCC at the right end of the gallery [estimated to be flowing about 100
gal/min (6 L/s)] brought the total seepage down to nearly 400 gal/min
(25 L/s). During cold weather in early 1989, the RCC mass contracted
and existing cracks or joints in the conventional concrete upstream
278 Chapter Eleven

facing and the leakage increased to 533 gal/min (33.6 L/s). It was then
decided to empty the reservoir to allow for repair of these joints and
cracks during 1989. It was reported that the original joint sealant
used in the dummy joints spaced 16 ft (4.9 m) on center was not of the
type recommended for underwater placement and was not applied to
the depth specified.
9. Monksuille Dam. The initial filling of this municipal-water-
supply reservoir started in July 1987. Full pool producing a hydraulic
head of 141.5 ft (43 m) was reached in April 1988, at which time a
maximum total seepage of about 250 gal/min (15.8 L/s) was measured
in weirs located in the gallery, and at both abutments. Of this total,
the measured seepage from the gallery which extends only 300 ft (91
m) across the dam and corresponds to the spillway width, was 61 gal/
min (3.8 L/s). During the next five months, total seepage had de-
creased to 120 gal/min (7.6 L/s) and that from the gallery to 25 gal/
min (1.6 L/s). With the advent of cold winter weather, the total
seepage increased to 290 gal/min (18.3 L/s) in March 1989. Of that, the
flow collected in the gallery rose to 55 gal/min (3.5 L/s). Three months
later the total seepage had reduced to 150 gallmin (9.5 L/s). Wet spots
were noticeable on the exposed RCC downstream faces of the
nonoverflow sections soon after filling and remain visible two years
The overall low permeability of this dam was attributed to design
response to a refined thermal analysis and improvements in the RCC
mix design to provide reduced permeability. When it was determined
a delay in construction due to warmer weather would produce higher
peak temperatures than predicted in the original design, the design-
ers decided to increase the number of waterstopped joints for the up-
per 40 ft (12 m) of the dam. The basic watertightness of the RCC ma-
terial was aided by a reduction in the maximum size aggregate from
the usual 3 in (75 mm) to 2 in (50 mm) and by using a 40 percent sand
fraction in the mix. Both of the mix design factors tended to reduce the
potential for segregation and minimize voids for an RCC dam that had
a low cement content of 105 lb/yd3 (63 kg/m3).
10. Arabie (now Mokgoma Matlala) Dam. Reservoir filling for this
dam started in early 1987 and measured total seepage has not ex-
ceeded 16 gal/min (1 L/s). The low amount of seepage has been attrib-
uted to design of an improved seepage control system as well as the
dam being located in a warm area. At this location in Lebowa, there
was a low-temperature drop from the peak RCC temperature in the
mass to the annual average ambient temperature, thus minimizing
the potential for thermal-induced cracking.
The seepage control system includes waterstops installed down-
Data and Performance of Completed RCC Dams 279

stream of formed grooves in the upstream conventional face concrete

spaced at 39-ft (12-m) centers. Bedding concrete was placed atop each
RCC lift for a horizontal distance back from the upstream face equal
to one-fourth of the height of the dam at that elevation. Segregation of
coarse aggregate remained a problem, but increasing the sand content
of the RCC mix to 40 percent helped reduce the permeability of the
RCC mass.
11. Upper Stillwater Dam. After three years of RCC construction,
filling of the reservoir started in early December 1987, at an average
rate of 0.3 ft/day (90 mm/day). By early June 1988, the water surface
had risen to an elevation of 8140 ft (2482 m), 165 ft (50 m) above the
tailwater elevation. The streambed is at elevation 7990 ft (2436 m). At
that time, total leakage measured from weirs within the gallery and
at both abutments was slightly greater than 700 gal/min (44 L/s). This
produced an initial unit seepage value of 0.18 gal/min per 10,000 ft2 of
wetted area per foot of average head (0.40 L/s per square meter per
meter). Most of the seepage measured in the gallery, whose invert is
at the same elevation as the streambed, came up from the foundation
drains, indicating the greatest volume of seepage was passing the
grout curtain rather than seeping through the RCC.
A major transverse crack then formed to the right of the 600-f&wide
(183-m) spillway, which is located near the center of the dam. With a
full reservoir [water at el. 8172 ft (2491 m)] the flow through this sin-
gle crack, which reached a maximum width of 1/4 in (6 mm), over-
flowed the gutters in the gallery. Leakage was estimated to be about
1600 gal/min (100 L/s) into the gallery plus another 2200 gal/min (140
L/s) exiting from the crack on the downstream face.
Even though the crack had formed and was producing leakage, it
was determined that it would not affect the structural safety of the
dam. The reservoir was lowered to the inactive storage pool elevation
of 8028 ft (2447 m) in order to cap some fine red sand piles in the res-
ervoir that had discolored the water. With the reservoir lowered, the
crack repair work using polyurethane resin was accomplished during
the winter of 1988-1989 and the reservoir was filled again in the
spring of 1989.
With a full reservoir, the estimated leakage was now 900 gallmin
(57 L/s) from the crack at the downstream face. Leakage into the gal-
lery from the crack had been stopped by the repair program. Some ad-
ditional repairs in the vicinity of this crack are planned for 1990.

11.2.3 Chemical effects Production of calcium hydroxide and calcium carbonate. The
appearance of calcium carbonate (CaCOs) in the gallery, on the down-
280 Chapter Eleven

stream face, or on the bottom of the waterway downstream of the dam

has been reported on most completed RCC dams. The calcium carbon-
ate (also called calcite) is formed when calcium hydroxide that is re-
leased from the cement hydration process is carried by seepage water
to a surface where it combines with carbon dioxide from the air as in-
dicated by the following formula:

Ca(OH), t COB + CaCO, t Hz0

In conventional concrete and concrete masonry construction, this
phenomena is termed effluorescence. The formation of this white
salt tends to clog voids adjacent to the surface and thus reduce seepage
through the RCC mass. Still, the formation of calcium hydroxide or
ultimately calcium carbonate has produced some negative effects on
several RCC projects in service. The negative effects include clogging
of drain holes, increasing the pH*of the seepage water, making gal-
lery floors slippery, and possibly creating undesirable visual effects
downstream of the dam.
Calcium carbonate as great as 2 in (50 mm) thick has been noticed
on the face of the gallery walls at both Willow Creek and Middle Fork
dams. If this amount of calcite can be produced, it could clog internal
drain holes, which are generally 3 in (75 mm) in diameter. At Middle
Fork Dam, the drain holes have required periodic cleaning or
redrilling. The drain holes at Copperfield were reamed out in 1989. It
was apparent that the holes exiting the roof of the gallery were clear
of calcium carbonate beyond approximately 3 to 6 ft (1 to 2 m) and the
drain holes from the foundation which were full of water were clear of
calcium carbonate buildup. This confirms that carbon dioxide present
in air is required to produce the calcium carbonate precipitate. Clog-
ging of drainage system holes by calcium carbonate also was reported
in the RCC used for the foundation of a spillway chute and ski jump
for a large embankment dam in Iraq.
Increasing the pH of the seepage water has its greatest negative ef-
fect when releases from the reservoir are low or nonexistent. In this
case, there is not sufficient water to dilute the highly alkaline water
being produced by calcium hydroxide being dissolved and carried
downstream by seepage water. Such was the case at Grindstone Can-
yon Dam where the calcium carbonate precipitated at the bottom of
the stilling basin and stream downstream of the dam soon after initial
filling in the spring of 1988.
The water exiting the dam was determined to have a pH greater

*pH is a measure of the hydrogen ion concentration of solutions, with the pH of neu-
tral water being 7.0. Values less than 7.0 indicate acidity and those above 7.0 indicate
Data and Performance of Completed RCC Dams 281

n 11. Initially, the volume of seepage from the dam was nearly as
at as the receiving waterway. As a result, rapid dilution of the al-
inity does not occur. Instead, precipitation of calcium and some
gnesium carbonate occurs. This mineral precipitation was aggra-
ed by the fact that the water in the first stream downstream of the
n was already supersaturated with calcium carbonate. Water soft-
ng by pH adjustment is another way of explaining the phenomena.
lile the white-colored bottoms of the streams produced an unnatu-
visual effect, there was no indication of either aquatic or terres-
11 biological damage.
.he immediate solution to the problem was to pump a majority of
seepage water collected in the stilling basin back into the reser-
r. The small amount of water that was proceeding down Grindstone
:ek was then rapidly diluted when it entered larger flows of down-
earn creeks. The mitigation measure effectively stopped the down-
earn carbonate precipitate problems and no long-term detrimental
rironmental effects are expected. Pumping the seepage water back
o the reservoir did increase the pH of the total reservoir storage,
educing a hard water and caused a fine precipitate which gives the
ter a distinctive turquoise color.
although the pump-back scheme produced immediate positive re-
ts, the most effective long-term solution was to reduce the volume
;eepage passing through the dam. As noted in Sec., the res-
roir was emptied in early 1989 for the purpose of repairing cracks
d resealing joints.
deports on the performance of Copperfield, Craigbourne, and Mid-
I Fork dams provide added insight into the calcium hydroxide leach-
; and calcium carbonate buildup phenomena. At Copperfield Dam,
alysis of seepage water varied from a high pH value of 11.6 for a
w leak to 7.45 through a crack, with an average of 9.3 for a flow
wnstream of the dam of 160 gal/min (10.3 L/s). Thus, water passing
ickly through a crack has less opportunity to dissolve the calcium
droxide than water slowly seeping through the RCC mass that
nes into contact with more cemented surfaces.
4 greater amount of calcium carbonate effluorescence appeared on
: exposed downstream face of Craigbourne Dam (see Fig. 11.7) than
Copperfield. This was attributed primarily to a more porous RCC at
s location, due mainly to greater difficulty in compacting the outer
ge to achieve a high density.
4x-1 unusual situation occurred at Middle Fork Dam where initial
emical analysis of seepage water confirmed that calcium hydroxide
ts being leached from the dam and precipitating to calcium carbon-
3 within the gallery and downstream of the dam. Within a few hun-
ed yards of the dam, sufficient dilution had taken place and the
282 Chapter Eleven

Figure I I .7 Craigbourne Dam.

creek returned approximately to its original chemistry. After several

months, the reaction had reversed, whereby the seepage exiting the
dam had less calcium hydroxide and other dissolved salts than the
creek inflow, indicating some solids were now being deposited in the
With the relatively dry no-slump mixes used for RCC, there is a pos-
sibility of partially hydrated cement being present in the structure
prior to filling. As water seeps through the mass, the cement hydrates,
producing greater strength but also releasing additional calcium hy-
droxide for dissolution and transport downstream. If the reservoir
level remains constant, the amount of leaching diminishes with time.
This is due both to reduced seepage attributed to the natural siltation
and calcification process but also a reduced amount of available cal-
cium hyroxide. There is a fixed amount of calcium hydroxide within
the structure, and if some has already been carried downstream, the
amount remaining is less. If the reservoir is raised to levels not pre-
viously exposed to water, the chemical phenomena repeats itself.
Basic concrete technology indicates that fly ash will react with the
calcium hydroxide produced from the cement hydration process and
make less calcium hydroxide available for dissolution and leaching.
From actual performance of RCC dams, it is unclear if and to what
degree fly ash in the mix actually improves the situation. At Upper
Stillwater Dam, where the fly ash content in the RCC was more than
twice that of the cement, there was still some calcite buildup at cracks
in the gallery walls.
Data and Performance of Completed RCC Dams 283

.2.3.2 Reservoir-induced gases. The potential problem of reservoir-

duced gases is not directly related to RCC dams, but its occurrence
Willow Creek Dam generated some concern in newspapers about
e long-term stability of the dam. Willow Creek dam is located in a
rming area where runoff fills the reservoir with water high in or-
nit nutrients, mainly from fertilizers. The nutrients, especially dur-
g warm weather, use up oxygen in the water in a biological process
educing hydrogen sulfide (H&S) and methane (CH,) gases. Seepage
tter passing through the dam carried some gas into the gallery,
lere the distinctive smell of hydrogen sulfide has been noticed at
Hydrogen sulfide by itself will not directly harm concrete, but re-
lting aerobic bacteria could produce a dilute sulfuric acid (HzSOJ
nilar to that experienced in sewer pipes. The entire process does not
em capable of producing a concentration of acid that would start to
jintegrate concrete to any degree. If a problem with reservoir-
duced gases presents itself, it can be minimized by the introduction
air into the reservoir or the gallery. Following improvement of ven-
ation in the Willow Creek gallery, tests performed on cores removed
lrn the gallery walls in 1989 confirm that no deterioration or retro-
ession of RCC strength due to the effects of the gases combined with
epage water. Still, the main solution to this potential problem is to
sign and construct an RCC dam with a high degree of impermeabil-

.2.4 Lessons learned from seepage and

epage-related phenomena
!epage and seepage-related phenomena from early RCC dams have
ught designers a number of lessons:

. . Initial seepage volumes from early lean RCC-dams were in some

cases more than anticipated.
!. The amount of seepage through the RCC has been reduced consid-
erably with time due to external effects such as siltation into
voids and natural internal autogenous healing due to production
of additional concrete gel and calcification after contact with seep-
age water.
I. Where measured seepage has increased significantly, it is usually
due to leakage passing through a newly formed crack.
1. Side effects of seepage such as visual wet spots, calcium hydroxide
leaching, calcium carbonate buildup, or passing reservoir-induced
gases through the dam have generally created more public con-
cern than the total seepage through the dam and foundation.
284 Chapter Eleven

5. As expected, seepage is greater with increased head, with in-

creased wetted surface area, and during cold weather when the
RCC mass shrinks, thus creating greater crack widths.
6. RCC dams can be designed to have equal seepage to dams con-
structed of conventionally placed concrete.
7. Designs incorporating conventional concrete faces with water-
stopped joints (including the RCD method) and membrane-faced
precast concrete panels have proved to provide a high level of wa-
8. Leakage through cracks can be repaired more efficiently than
seepage through the entire RCC mass. Leakage through cracks
also has less opportunity to leach out calcium hydroxide from the
dam in the initial stages due to higher seepage flow velocities.
9. The permeability of the RCC material can be improved with
higher cementitious contents, higher sand contents, and construc-
tion methods to minimize voids in the compacted mass concrete.
10. Construction control is extremely important in ensuring that the
planned design is executed in a manner to minimize seepage. Spe-
cial attention should be placed on installing waterstops and apply-
ing sealants at planned joints or grooves.
11. The contact of the dam with its base foundation or abutments as
with any dam is a prime potential seepage path and care must be
taken to ensure this intersection has a high degree of watertight-

11.2.5 Cracking
Cracking in any nonreinforced concrete dam can be expected. From a
design standpoint, joints or crack inducers may be installed in the
dam to help control cracking or the dam can be constructed with no
crack relief as discussed in Sec. Cracks that extend below the
waterline and have sufficient width to pass water are of greatest con-
cern to dam designers. Therefore, leakage and cracking performance
are directly related. Cracks in gravity dams are generally vertical and
transverse to the dams axis and pose no threat to the structural sta-
bility of the dam. Cracking case studies. Information on cracking from the fol-

lowing RCC dams provide considerable insight into the formation of
cracks and their repair.

1. Shimigawa. Detailed investigation of the dam which was con-

structed with full-section contraction joints spaced at 49 ft (15 m) re-
Data and Performance of Completed RCC Dams 285

aled no cracking in the upstream face, the downstream faces, or in

e gallery which has caused water leakage. The maximum opening of
e joints was 0.20 in (5 mm) and the mean opening width was 0.11 in
7 mm).
2. Willow Creek Dam. Narrow hairline cracks have been noticed
the exposed RCC at the crest and within the gallery. These cracks
e closely spaced and not of sufficient width to pass any water to the
wnstream face. In the gallery roof, however, there is evidence of
eater water leakage at the location of the cracks.
At the change in geometry where the spillway meets the non-
erflow section, instrumentation detected an internal crack within
e RCC starting near the foundation after the reservoir level was
ised. It did not penetrate through the RCC mass, as no indication of
:rack can be detected at either the upstream or downstream surface
this area.
3. Copperfield Dam, Craigbourne Dam, and Bucca Weir. The ex-
rience gained at these three Australian dams provides an insight
to the temperature drop required to cause cracking, taking into ac-
unt tensile capacity of the RCC mixes used. Crack widths were also
At Copperfield, a single transverse uncontrolled crack developed
rough the spillway as a result of a temperature drop of 20F (11C)
)rn the peak temperature of 95F (35C). Leakage increased follow-
g the formation of this crack and then naturally diminished there-
;er as noted in Sec. In addition to the crack through the
illway, three transverse contraction control joints opened as in-
lded with no significant seepage bypassing the waterstops. Crack
dth measurements made of the three joints indicated a total width
0.18 in (4.5 mm) two months after cracking. During the following
mmer the joints and cracks closed fully, reopening to their initial
dth the next winter.
No cracking has been identified after one year at Craigbourne Dam.
I this location RCC temperatures peaked at 68 to 70F (20 to 21C)
.d were expected to stabilize at 54 to 57F (12 to 14C). Thus the
aximum temperature drop of 16F (9C) was not expected to cause
ermal cracking.
Delays at Bucca Weir pushed construction into the summer months,
Id RCC temperatures peaked at 113F (45C) or about 18F (10C)
3re than anticipated in the design. After the peak temperatures
opped 22F (12C), three transverse, nearly vertical cracks devel-
ed through the section at approximately the mid and third points,
Id the transverse contraction control joints spaced at 381 ft (116 m)
ened. The width of the opened joint was about 0.04 to 0.06 in. (1 to
j mm) through the RCC.
286 Chapter Eleven

4. Galesville Dam. A delay in the start of RCC placement contrib-

uted to formation of seven cracks which are the primary cause of the
leakage at Galesville. The uppermost lifts of RCC were placed during
the hottest part of the year. The cracking occurred when unusually
cold weather hit 60 days later and caused rapid cooling of the concrete.
The delay in construction caused the peak temperature to be more
than 21F (12C) greater than originally predicted due to the higher
temperatures of both the RCC and ambient air. Capping the crest of
the dam with conventional concrete on three sides contributed to the
heat rise in this area.
The cracking occurred when there was a differential of measured
temperature between the downstream face and interior mass of about
28F (16C). The cracks started at the crest and extended vertically
down both faces. The first crack noticed was located near the right
abutment angled from the vertical so that it intercepted the entrance
adit to the gallery, a hole which provided the least resistance to the
thermally induced tensile stresses.
5. Cedar Falls Dam. An error in control of cement feed to the mix-
ing plant produced an RCC mix containing 320 lb/yd3 (190 kg/m31 of
cement plus 155 lb/yd3 (92 kg/m31 of fly ash. This-higher-than planned
cement content was used for the lower two-thirds of the 30-ft-high (9-
m) of the dam until the error was discovered and rectified.
Calculations indicated the richer mix would produce a 20F (11C)
greater temperature rise than originally anticipated, which was con-
firmed by sensors within the dam. A maximum temperature of 120F
(49C) was actually reached compared to the predicted 100F (38C) for
the specified mixture. The nonrock foundation (see Sec. 3.2.3) for this
small dam was forgiving to thermal movement and potential cracking
as the lack of restraint at the soil abutments reduced tensile contrac-
tion stresses. An inspection of the completed structure indicated the
higher internal temperatures in the RCC produced no more cracking
than originally anticipated in the conventional concrete faces. Narrow
transverse shrinkage cracks spaced at approximately 20-ft (6-m) spac-
ing were observed.
6. Upper Stillwater Dam. As the dam was being constructed with
horizontal slip-formed elements of conventional nonreinforced con-
crete on both the upstream and downstream faces, narrow vertical
cracks spaced 20 to 30 ft (6 to 9 m) apart were noticed in the concrete
surface within two days after placement. These shrinkage-related
cracks were limited to the facing elements and did not extend into the
RCC mass.
The dam was topped out in August 1987. By December, 13 cracks
had developed in the dam and spillway crest. All started at the joints
Data and Performance of Completed RCC Dams 287

the parapet or spillway crest where the steel reinforcement had

en stopped. Parapet contraction joints were produced by hand-
)ling a 2-in-deep (50-mm) groove at 40-ft (12.1-m) centers on both
les of the 4.5-ft-high (1.4-m) slip-formed parapet wall. These cracks,
rich started at the crest and continued down the faces, had an aver-
e spacing of 160 ft (49 m). As the weather warmed up, crack-width
strumentation showed the cracks had closed up and little or no leak-
e passed through them.
In the original design, cracking was predicted. However, higher-
an-anticipated peak temperatures near the top of the dam produced
scking deeper into the mass than expected. The spacing between
acks was greater than anticipated, which resulted in fewer but
.der cracks in the 2673-ft-long (815-m) structure.
While the reservoir was filling, the single major crack described in
C. occurred. The crack reportedly was noticed first at the
se of dam and then propagated upward to the crest. The external
mperature restraint condition (see Sec. 35.1) best explains the for-
ation of this crack, but it appears to have been triggered by a slight
lwnstream movement in the foundation. Upper Stillwater Dam is
.own in Fig. 11.3. Note that all cracks are vertical and transverse to
e dams axis and have already started to heal naturally.

gure 11.8 Upper Stillwater Dam (July 1988).

288 Chapter Eleven

7. Elk Creek Dam. Actual peak temperatures during construction

of the dam in the spring of 1987 were significantly higher than pre-
dicted in the Corps of Engineers computer program. The higher tem-
peratures were due to both higher ambient air and RCC placing tem-
peratures than had been put into the program. These factors, plus a
belief that finite-element analysis does not accurately predict the ef-
fect of radiant solar heat into the RCC surface, produced a peak tem-
perature of 101.6F (337C) on the left side of outlet works compared
to a computer-projected peak of 75F (24C). The difference on the
right side of the outlet works was not as pronounced, with an actual
88.3F (31.3C) peak compared to the same maximum computer
project temperature of 75F (24C). The lower peak was due to cooler
air temperature at the time of this RCC placement. Six months after
placement, the high RCC temperature had caused no noticeable prob-
lems. The contraction joints installed in the upstream face at a spac-
ing up to 300 ft (91 m) had opened up approximately 0.06 in. (1.5 mm).
Three vertical transverse cracks located about halfway between the
joints and in the middle of the 120-ft-wide (37-m) spillway have also
been identified and ultimately will be repaired. Summary of cracking performance. From the performance of

the RCC dams in service, the following conclusions can be made, most
of which are self-evident

1. There is less potential for cracking in RCC than in conventional

concrete due to less contraction of the RCC mass combined with
generally lower elastic modulus and higher creep rates for RCC.
The lesser amount of shrinkage is due to lower water and
cementitious contents in the RCC mixtures as compared to con-
ventionally placed concrete.
2. Most cracking in RCC dams can be attributed to thermally in-
duced stresses. Cracking occurs when the thermal stress exceeds
the tensile capacity of the concrete. Cracks can occur with a tem-
perature drop of as little as 20F (11C) from the peak RCC tem-
perature for lean weaker mixes to as much as 36F (2OCO for
stronger mixes. Cracks in conventional concrete faces are also in-
fluenced by drying shrinkage stresses.
3. The spacing between cracks apparently depends upon the tensile
strength of the RCC, with greater spacing noticed m dams con-
structed of high-strength RCC such as Upper Stillwater Dam than
those of lean lower-strength mixes (i.e., Willow Creek). With in-
creased crack spacing comes greater individual crack width as the
total volume reduction must be accommadated in fewer cracks or
joints. Obviously, wider cracks have a greater potential for leakage.
Data and Performance of Completed RCC Dams 299

1. Full-section transverse contraction joints with upstream water-

stops and drain holes are an effective means of controlling crack-
ing through the entire RCC dam. When the spacing between
joints is too great, cracking will occur between the joints. Crack
width of uncontrolled cracks between joints is less than the width
of the ind/uced
i cracks in the transverse joints.
i. Cracks invariably will occur in intentionally planned transverse
contraction joints or at a point of reduced dam section where the
overall tensile resistance of the section is less. This crack location
can be at a reentrant angle in the foundation rock, producing a
stress concentration, through a central spillway section, at a
transverse entrance adit, or at a planned groove in the conven-
tional concrete face.
i. Cracks in RCC dams are generally vertical, transverse to the dam
axis, and pose no problem to the structure stability of the gravity
section. The preceding conclusions provide an insight into the po-
tential spacing and location of transverse contraction joints
through all or a portion of an RCC dam if desired.
. Initial cracking can usually be attributed to the internal temper-
ature restraint condition where the temperature at the center of
the concrete mass is greater than at the exposed faces of the dam.
These cracks invariably start at the dams crest, where the section
of the dam is at a minimum and there is a greater exposed surface
area to cool. Then the cracks propagate down both upstream and
downstream faces.
. Greater cracking than initially predicted by thermal analysis has
occurred due to delays in in the construction schedule, forcing
placement during warmer weather, thus producing higher peak
temperatures in the concrete than anticipated. Also, the thermal
analysis being used does not appear to properly account for radi-
ant heat effectson exposed RCC surfaces.
. Cracking in conventional concrete faces can be effectively con-
trolled by either joints or crack inducers formed in the concrete.
Horizontal spacing of vertical-face crack-inducer grooves 1.5 to 3
in deep (38 to 75 mm) in a face that averages 1.5 ft (0.46 m) thick
have ranged from 10 to 16 ft (3 to 5 ml.
. Cracks may be sealed or repaired by a variety of materials and
methods. Sealants formulated from polyurethane resins and
polysulfides have been used to the greatest degree to date. Crack
repair with the reservoir lowered is the most efficient situation,
but underwater crack-sealing methods have been used, including
dumped pelletized bentonite and use of quick-set cement by
290 Chapter Eleven

11.2.6 Durability
The durability of RCC dams is directly related to the properties of the
exposed concrete, whether placed conventionally or by the RCC
method. For RCC, greater strength of the mix and aggregates produce
greater durability. The same applies to conventional concrete, except
that the addition of air entrainment to the mix significantly improves
freeze-thaw resistance. Erosion resistance. Erosion resistance of RCC can best be

evaluated following high-velocity, high-volume flow over its surface.
The most noticeable examples of the erosion resistance of RCC are at
Tarbela Dam (see Sec. 7.3.2) and Kerrville Ponding Dam (Sec. 7.4.1).
The 21-ft-high (6.4-m) all-RCC dam at Kerrville, Texas, was
overtopped by as much as 14.4 ft (4.4 m) and 16.2 ft (4.9 m) in 1985
and 1987, with little noticeable erosion of the RCC except for washing
away some uncompacted material at the downstream face.
The 132-ft-high (40-m) upstream cofferdam at Geheyan Dam in
China was overtopped in 1988 by a flow of 12,360 ft3/s (350 m3/s) for 4
h without sustaining any damage. During the construction of
Craigbourne Dam and Bucca Weir in Australia, their cofferdams were
overtopped, producing very little evidence of erosion to the recently
placed RCC on the surface or downstream face (see Sec. 3.6.1). Ex-
posed RCC has proved to have a high degree of erosion resistance due
to the high percentage of aggregate in the mix. The degree of erosion
resistance is directly proportional to the compressive strength of the
RCC, which is dependent upon both the quality of the aggregate, the
mix proportions, and the degree of compaction.
At Copperfield Dam, some RCC was exposed during construction of
the upstream face. After four years of exposure to reservoir water,
there is erosion of the paste at or above the waterline due to wave ac-
tion, exposing some segregated RCC. The quality of the RCC 6 in (150
mm) in is good. It is difficult to achieve a high degree of compaction for
RCC placed directly against forms, resulting in lower-density, lower-
strength RCC at this location. (See Fig. 8.10.) Freeze-thaw resistance. Both Willow Creek and Galesville

dams have exposed RCC downstream slopes for both the spillway and
nonoverflow sections. They have been subjected tofreeze-thaw action
while being wet from seepage through the dam. While there undoubt-
edly has been some freeze-thaw damage to RCC on the exterior that
was poorly compacted, the effects to date have been minimal, result-
ing in negligible loss of surface concrete.
Data and Performance of Completed RCC Dams 291

mson, S. A., Verigen, W. M., and Carney, M. J., Cedar Falls Roller Compacted Con-
crete Dam, Roller Compacted Concrete II, ASCE, New York, February 1988, pp. 39-
unstan, M. R. H., Recent Developments in Roller Compacted Concrete Dam Con-
struction, Water Power & Dam Construction Handbook, 1989.
nbes, B. A., RCC Dams in Australia, Roller Compacted Concrete II, AXE, New
York, February 1988, pp. 323-339.
uzina, B. J., Uzelac, S., Savic, M., Application of Roller Compacted Concrete at Ap-
purtenant Hydraulic Structures of a Large Dam, ICOLD Sixteenth Congress, San
Francisco, June 1988, vol. III, Q62, pp. 515-534.
ansen, K. D., Roller Compacted Concrete Dams Worldwide, Water Power and Dam
Construction Handbook, 1988.
opman, D. R., and Chambers, D. R., Construction of Elk Creek Dam, Roller-
Compacted Concrete II, AXE, New York, February 1988, pp. 251-266.
kckson, H. E., Thermal Cracking in Roller Compacted Concrete at Galesville Dam,
ICOLD Sixteenth Congress, San Francisco, June 1988, vol. 5, pp. 462-465.
cDowel1. T. E.. Grouting of Willow Creek Dam and Foundation, USCOLD Newslet-
ter, November 1985. -
oler, W. A., and Moore, J. F., Design of Seepage Control Systems for RCC Dams,
Roller-Compacted Concrete II, ASCE, New York, February 1988, pp. 61-75.
agayama, I., and Yamanaka, T., Watertightness of the RCD Dam-Monitoring
Shimajigawa Dam, Engineering for Dams, Japan Dam Engineering Center, vol. 4,
no. 1, 1986.
ichardson, A. T., Upper Stillwater Dam-Design, Construction and Performance,
U.S. Army Corps of Engineers Structural Conference, St. Louis, June 1988.
:hrader, E. K., Watertightness and Seepage Control in Roller Compacted Concrete
Dams, Roller-Compacted Concrete, AXE, New York, May 1985, pp 11-30.
:hrader, E. K., and Namikas, D., Performance of Roller Compacted Concrete Dams,
ICOLD Sixteenth Congress, San Francisco, June 1988, vol. 3, 662, pp. 339-363.