PREFACE

PREFACIO

Raúl Marsal is an engineer. Researcher, author, teacher, RaúlMarsal es un ingeniero.Investigador,autor, maestro,

inventor, imovator, tireless worker -but above all, an inventor, innovador, incansable trabajador -pero sobre
engineer-.His written work testifies his technical capacity, todo, ingeniero-. Su obra escrita da testimonio de su
his inventiveness, and his ability for solving problems. capacidad tCcnica, de su inventiva y de su habilidad para
His work attests his capacity and his achievements remain resolver problemas. Sus trabajos dan fe de su capacidad y
there, he participated during 45 years in the design and ahí está su obra, participó durante 45 años en el proyecto
construction of Mexico's hydraulic infraestructure,more y construcción de la infraestructurahidráulica de MCxico,
than 50 medium or large dams for imgation or for más de 50 presas grandes y medianas tanto para riego
generatingelectncity. Results from hisexhaustive research como para generación de electricidad. Y sus
on the properries of Mexico City clay were the basis for investigaciones exhaustivas sobre las propiedades de las
the geotechnicalzoning and the building codes, necessary arcillas del Valle de MCxico, han permitido la
for designing safe foundations in Mexico City. regionalizacióny reglamentación necesarias para diseñar
con seguridad las cimentacionesde las edificacionesen la
Those who read his work and above all, those who use his ciudad de MCxico.
teachings, will appreciate the wealth of his contributions
to civil engineering and geotechnics. Quienes lean su obra y sobre todo, quienes utilicen sus
enseñanzas, apreciarán la riqueza de sus contribucionesa
This book, which presents one of his unpublished papers, la ingenieria civil y a la geotecnia.
brings together contributions o f various authors,
researchers of great value and merit, that have forwarded Este libro, en el que se presenta uno de sus trabajos
theirpapers as a scientific tribute to Marsal's personality, inkditos, está acompañado por contribucionesde diversos
always recognized by his colleagues. autores,investigadoresdegranvalory mCrito, que aportan
sus contribuciones a manera de homenaje científico a la
1 want to mention a few more things about Marsal's other personalidad de Marsal que sus colegas siempre
qualities: as a field engineer, as a practical professional, as reconocieron.
a promotor of technical endeavors, as a solver of crises
-aspects of his personality that are less well known-. Sobre las otras cualidades de Marsal: como ingeniero de
campo, como profesional práctico, como promotor de
1,
, 1had a working relationship with Marsai for almost haif actividades, como solucionadorde crisis -aspectosmenos
a century. 1witnessed his dedication to work, his variety conocidos de su personalidad- quiero decir un poco más.
1 of interests, 1 witnessed how, a researcher capable of
designing a triaxial cell to measure the mechanicai Tuve una relación profesional con Raúl Marsai durante
properties of rockfills, was also able to solve many other casi medio siglo. Me tocó ser testigo de su dedicación al
complex and diverse problems like the handling of coai irabajo, & su plurivalencia, & cómo un investigador,
1 with high amounts of ash in the Rio Escondido capaz de desarrollar una cámara triaxiai para medir las
thermoelectric power plant, the closure of the Maipaso propiedades mecánicas de enrocamientos, era también
dam or the design of the first sewage treatment plant in capaz de hacerse cargo de resolver problemas muy
Mexico City. complejos y variados como el inanejo de carbón con
mucha ceniza en la terrnoelectrica de Río Escondido, el
A memorable markin a lifededicated to engineering is his cierre & la presa de Malpaso o el diseíio de la primera
role as an international authority and consultant,as which planta de tratamiento de aguas negras en el Distrito
he won countless proofs of gratitude for contributing to Federal.
the solution of many serious problems in engineering
projects in many counüies. And, as another aspect of his Testigo de su vida dedicada a la ingeniería, es también su
personality, 1must mention that inhis consultancyjobs he papel como autoridad mundial y consultor en presas y
was never interestedina direct personal retribution.If the mecánicade suelos, que levalieron innumerablesmuestras
electricity board (CFE, Mexico) or the Institute of de agradecimientopor haber contribuido a la solución de
Engineering agreedonhispariicipation as aconsultant,he serios problemas en diversas obras de muchos países. Y,
would go ahead, without charging for these services. como otro aspecto de su personalidad, debo decir que en
sus trabajos como consultor, nunca le importó su
Logically, such a personality formed a lot of people, many retribución directa. Si la CFE o el Instituto de Ingeniería
engineers became better because of him. Through his aceptaban que sirviera como consultor, adelante, él no
example, his tireless dedication to work, his in-depth cobraba por estos servicios.
analyses, his accuteness,heearnedthetitleof "Professor".
Como es lógico, una personalidad así, formó a mucha
It is very fortunatefor me to be able to acknowledgein this gente, hizo mejores a muchos ingenieros. Su ejemplo, su
preface the work of my friend, of a dear colleague, incansable dedicación al trabajo, sus profundos análisis y
Professor Marsal. 1hope that among his disciples, some su agudeza, le valieron el titulo de "Profesor".
will be like him.
Es muy afortunado para mí tener la oportunidad de
reconocer en este prefacio la labor de mi amigo, de un
entrañable colega, el Profesor Marsai. Ojalá que entre sus
discípulos, surjan algunos como él.

Fernando Hiriarí B.
 CONTENTS
CONTENIDO

INTRODUCTION
INTRODUCCION
J. Alberro

Estudio de las propiedades mecánicas de arenas finas

(Construcción de las terracenas de la presa Alvaro Obregón, Sonora)
Study of the mechanical properties of fine sands (Construction of highway embankments
in the Alvaro Obregón dam, Sonora)
R.J. Marsal

4 SOFT SOlLS AND THE SUBSOIL OF MEXICO ClTY

1 SUELOS BLANDOS Y EL SUBSUELO DE LA CIUDAD
DE MEXICO

Static behavior of floating piles in soft clay

Comportamiento estático de pilotes de ficción en arcilla blanda
A. Jaime, E. Rojas, H. Legorreta

Natural laws in geomechanics

Leyes naturales en geomecánica
E. Juárez Badillo

Efectos de la extracción de agua en la zona lacustre del valle de México

Effects of water extraction in the lake zone of the vaiiey of Mexico
M. Mazari, M. Mazari, C. Ramírez, J. Alberro

Constant rate of strain consolidation testing of soft clays

Ensayes de consolidación con velocidad de deformación constante en arcillas blandas
G. Mesri, T. W. Feng

La catedral y el sagrario de la ciudad de México

Corrección del comportamiento de sus cimentaciones
México City's cathedral and sagrario church
Control of the behavior of their foundations
E. Tamez, E. Santoyo, A. Cuevas
gg vi
a i o
e:ga
* 0 u.?
"'u
,af- = 0 45g
U
$83,.
e 2 2 o
3825
'u ag.-
Engineenng properties and microstructure of compacted cohesive soils
propiedades ingenienles y microesmictura de suelos cohesivos compactados
M. J. Mendoza, J. Alberro

Algunas divergencias entre prácticas comunes de compactación y lo que la

investigación parece indicar
Some discrepancies between common compaction practice and what research seems to indicate
A. Rico

A EARTH AND ROCKFILL DAMS

'T PRESAS DE TIERRA Y ENROCAMIENTO

Agrietamiento interno en presas de tierra y enrocamiento

Interna1 cracking in earth and rockfill dams
J. Alberro, F. González

Agua Prieta landslide: A case history

El deslizamiento de Agua F'rieta: Un caso historia
S. Herrera, D. Reskndiz

Rockfill for embankment dams

Enrocamiento para presas de tierra
A. D. Penman

Seismic response analyses of El Infiernillo and La Villita dams.

Análisis de la respuesta sísmica de las presas La Villita y el Infiernillo
M. P. Romo, R. Magaña, J.R. Herndndez

LlST OF PUBLICATIONS BY R. J. MARSAL

LISTA DE P.UBLlCAClONES DE R. J. MARSAL
 J. Alberro, instituto de Ingeniería, UNAM

Por su desinteresada entrega a la Because of his unselfish dedication to

investigación, su espíritu creador, su research, his creative spirit, his earnest
evidente empeño en entender los fenómenos desire to understand phenomena in order to be
para ser útil, para actuar, Raúl J. Marsal useful, to take actions, Raúl J. Marsal
deja una -profunda huella como ingeniero. leaves a profound print as an engineer.
Llegó a México en 1945, incitado por He arrived in Mexico in 1945, prompted by
Arthur Casagrande para que junto con Nabor Arthur Casagrande to dilucidate, together
carrillo dilucidara el problema del with Nabor Carrillo, the problem of regional
hundimiento de la Ciudad de México. Según sus subsidence in Mexico City. According to his
propias palabras consideró que "era own words the topic was important "for the
importante este tema para la estabilidad de stability of its constructions and the
sus construcciones y la preservación de sus preservation of its rnonumentsu. Fourteen
monumento^'^. Transcurrieron 14 años de arduas years of hard work went by, firstly in the
labores, primero en la Comision Impulsora y Committee for the Patronage and Coordination
Coordinadora de la Investigación Científica of Scientific Research (CICC), later in the
(CICIC), luego en los laboratorios de laboratories of Ingenieros Civiles Asociados
Ingenieros Civiles Asociados (ICA) en los que (ICA) in which he designed ingenuous equip-
desarrolló ingeniosos equipos para medir las ment for measuring the static and dy.namic
propiedades tanto estáticas como dinámicas de properties of Mexico City Clay and, finally,
las arcillas del Valle de México y, in the Institute of Engineering, National
finalmente, en el Instituto de Ingeniería de University of Mexico. He was always deeply
la Universidad Nacional Autónoma. Siempre interested in this matter and participated
mantuvo un profundo interés en esta materia y actively during the last 40 years in study cf
participó activamente durante los Últimos 40 the mechanical properties of the soft
añc's en el estudio de las propiedades compressible deposits of the city's subsoil,
mecánicas de los depósitos blandos y in the observation of buildings founded on
compresibles del subsuelo de la zona urbana, different types of substructures and in the
en la observación del comportamiento de drafting of building codes.
edificios sustentados por distintos tipos de
subestructuras y en la elaboración de
reglamentos de construcción.
A partir de 1960, y atento a la necesidad In 1960, aware of the need to rationalize
de racionalizar el diseño de presas altas de the design of large earth and rockfill
tierra y enrocamiento retomó una de sus embankments, he retook one of his oldest
añejas inquietudes: el comportamiento de los interests: the behavior of granular soils,
suelos granulares, o sea de los medios that is, of discrete media. Ever since his
discretos. Ya durante su estancia en Harvard stay in Harvard and rooted in conversations
Marsal, a raíz de sus conversaciones con with Terzaghi, Marsal had shown his liking of
Terzaghi había manifestado su interés por this theme. Terzaghi had pointed out in 1920
este tema. Desde 1920,,- Terzaghi había that "the fundamental error was introduced by
señalado que "el error fundamental fue Coulomb who consciously ignored that sand is
introducido por Coulomb quien conscientemente formed by individual grains and treated it as
ignoró que la arena está constituída por if it were an homegeneous mass with certain
granos individuales y la trató como si fuera mechanical propertiesu. Marsal dedicated at
una masa homogénea con ciertas propiedades least 20 years of his life in clarifying the
mecánicasN. Marsal dedicó por lo menos 20 implications of the discrete nature of
años de su vida en esclarecer las granular soils, without being overwhelmed by
implicaciones del carácter discreto de los the magnitude of the problem. He persuaded
suelos granulares, sin arredrarse ante la the Federal Commission of Electricity (CFE,
magnitud del problema. Convence a la Comisión Mexico) on the need to implement a field
Federal de Electricidad de la necesidad de laboratory at the site of the El Infiernillo
implementar un laboratorio de enrocamientos dam to test large scale specimens of these
en el sitio de la presa El Infiernillo que
Permita probar a gran escala estos
materiales. Diseña los primeros equipos en el materials. He designed the first equipment in
mundo capaces de probar muestras de gran the world capable of testing large size
tamaiío bajo presiones altas: triaxial, samples under high pressures: triaxial cell,
odómetro, deformación plana. Ahonda además en oedometer, plane strain apparatus. He
la investigación de los procesos de rotura de deepened further his research on the process
partlculas, de la distribución probabilística of grain breakage, on the probabilistic dis-
de las fuerzas de contacto interparticulares tribution of intergranular contacts and on
y de la estructura discreta de los medios the discrete structure of granular media. The
grandares. Los resultados de este esfuerzo results of this effort immediately influence
influyen de inmediato en la práctica professional practice in Mexico and quite
profe-sional en México y rápidamente en otras rapidly, in other parts of the the world too,
partes del mundo, suministrando bases para giving the bases for the rational selection
seleccionar los materiales más convenientes of the most convenlent materials for the
para la construcción de presas altas de construction of large earth and rockfill
tierra y enrocamiento. embankments.
Además Marsal se dedicó a comprobar los Marsal also took time to verify the
resultados de estas investigaciones results of these investigations promoting the
promoviendo la medición de campo del installation of field instrumentation to
comportamiento mecánico de numerosas presas measure the mechanical behavior of numerous
diseñadas de acuerdo con sus trabajos dams designed according to his analytical and
analíticos y de laboratorio. Tales mediciones experimental work. Such instrumental meaqure-
instrumentales iniciadas en la presa El ments, which started in the El Infiernillo
Infiernillo y proseguidas en las presas dam and then followed in Malpaso, La
Malpaso, La Villita, La Angostura, Chicoasén, Angostura, Chicoasén, Peñitas and El Carayol,
Peñitas y El Caracol, evidenciaron además la also evinced the existence of phenomena such
existencia de fenómenos como el arqueo contra as narrow canyon arching, the interaction
las laderas, la interacción entre zonas de between zones built with different materials,
las presas construídas con materiales the plastification of important portions of
diferentes, la plastificación de porciones the structure, interna1 cracking due to
importantes de la estructura, el extension of the nucleus and to hydraulic
agrietamiento por extensión del núcleo y el fracturing.
proceso de fracturamiento hidráulico.
Atento a la trascendencia económica que Awáre of the economical implications that
el estudio de los suelos compactados puede the study of compacted materials can have in
tener en el diseño de presas, terraplenes y the design of dams, in 1980 Marsal initiated
plataformas, Marsal inicia en 1980 su their experimental analysis. He started this
análisis experimental. Con el patrocinio de investigation under the sponsorship of the
la Secretaría de Asentamientos Humanos y Secretary of Human Settlements and Public
Obras Públicas, emprende esta investigación, Works. Some of his finfings on this line of
cuyos frutos son múltiples: importancia de la research are: the importante of the method of
preparación del material antes de compactar, prepartion prior to compaction of the soil on
en cuanto a la formación de grumos, y su the formation of clots and its incidente on
incidencia en las propiedades índice y index and mechanical properties, the in-
mecánicas, influencia del contenido de arena fluence of water content on dry volumetric
en el peso voXumétrico seco, efecto del weight, the effect of time on the unconfined
tiempo en la re'sistencia a compresión simple compressive strength of samples stored with-
de especimenes conservados sin variar su out changes in compaction humidity, and the
humedad de compactación, y diferencias entre differences between water and air permeabil-
las permeabilidades al aire y al agua de ities of specimens compacted on the dry or
especímenes compactados en las ramas seca o wet sides of the compaction curve.
húmeda de la curva de compactación.
Es claro q i e la ardua y tenaz labor de Clearly, Raúl J. Marsal's work was fruit-
Raúl J. Marsal fue fructífera, y ful and illuminating.
esclarecedora.
Para honrar su memoria se presenta este This commemorative volume, presented to
volumen de homenaje el que, fiel a sus honor his memory with the participation of
inquietudes como ingeniero e investigador, numerous colleagues and pupils, al1 of them
consta de cuatro partes: Suelos blandos y el his friends, is divided into four parts, each
subsuelo de la Ciudad de México, Mecánica de corresponding to his different interests as
los medios granulares, Suelos compactados y an engineer and investigator: Soft soils and
Presas de tierra y enrocamiento, con la the subsoil of Mexico City, Mechanics of
colaboración de numerosos colegas y alumnos, granular media, Compacted soils, and Earth
amigos todos. and rockfill embankments.
 Estudio de las propiedades mecánicas de
arenas finas*
Construcción de las terracerías de
la presa Alvaro Obri?gón,Son.
Study of the mechanical properties of fine sands'
Construction of highway ernbankrnentsin the Alvaro Obregón dam,Sonora

ES bien conocido que los materiales It is a well known fact that granular
granulares cuando son sometidos a esfuerzos materials experience a change in the volume
cortantes experimentan un cambio en el of voids when subjected to shear stresses. In
volumen de vacíos. Desde 1885, 0. Reynolds 1885 0. Reynolds attracted attention to this
llamó la atención sobre este hecho que tiene, fact which is of particular interest when
tratdndose de suelos no cohesivos saturados, dealing with saturated cohesionless soils. A.
particular interés. A. Casagrande inició en Casagrande started in 1937 a comprehensive
1937 un investigación cuidadosa del resea ch on the behavior of fine
comportamiento de las arenas finas, (1) la sandsfl'which lead to conclusions that are
cual condujo a conclusiones que generalmente generally accepted when applied to the
son aceptadas cuando se trata de la construction of embankrnents where such type
construcción de terraplenes en los que se of materials are used. The concept of
usan dichos materiales. El concepto de critica1 void ratio is related to the
relación crítica de vacios está relacionado so-called vdilatancyl* of the granular
con la "dilatancial* de los materiales materials.
granulares. The possibility of a structural collapse to
La posibilidad de que en una arena uniformr happen in a uniform saturated sand
y saturada se presente un colapso constitutes a phenomenon which has been
estructural, es un fenómeno que se ha observed in Nature . in the case of
observado en la Naturaleza en el caso de cohesionless masses subjected to quick
masas no cohesivas sujetas a perturbaciones disturbances and for certain initial
rápidas y para ciertas condiciones iniciales conditions of the material. The basis of this
del material. Los fundamentos de este phenomenon, which Dr. Terzaghi designated
fenómeno que el Dr. Terzaghi ha designado con with the name of spontaneous liquefaction,
el nombre de licuación espontánea. no son are not well known so far.
bien conocidos. Research conducted at the Soil Mechanics
El trabajo realizado en el laboratorio de Laboratory of the Secretary of Hydraulic
Mecánica de Suelos de la Secretaría de Resources on the fine sands likely to be used
Recursos Hidráulicos sobre las arenas finas for the construction of the earthfills at the
que se piensan usar en la construcción de las Alvaro Obregón Dam, is closely related to
terracerías de la Presa Alvaro Obregón, está both phenomena: dilatancy and spontaneous
ligado íntimamente a ambos fenómenos: la liquefaction. Although not much progress has
dilatancia y la licuación espontánea. Aún been achieved in the course of one year of
cuando no se pretende haber avanzado mucho en research, it might be interesting to inform
el estudio de ambos fenómenos después de un those persons interested in this particular
año de trabajo, puede ser interesante dar a topic of coi1 mechanics about our efforts to
conocer a los interesados en este Capítulo deal with the problem.
especial de la Mecánica de Suelos, nuestro
intentos por atacar el problema.
1.- La Presa Alvaro Obregón (actualmente en 1. The Alvaro Obregón Dam, at present under
construcción) en el río Yaqui, Son., tiene construction across the Yaqui River in the
Por objeto la irrigación y el control del state of Sonora, has been planned for
rio. Una sección típica de la cortina y los irrigation and flood control. A typical cross
diques presenta las siguientes section of the embankment and of the dikes
caracteristicas: Paramento aguas arriba shows the following features: upstream slope
2-5:l; talud aguas abajo 2: 1, con 2.5:l; downstream slope 2:l with rockfill at
enrocamiento al pie de 8:l. La máxima altura the toe laid at 8:l. The maximum dam height
de la presa será de 57m. sobre el nivel del will he 57 m above the river water leve1 and

* Trabajo presentado a la Convención de Verano de la * Paper presented during tlle Summer Convention of
American Society of Civil Engineers realizada en la the American S~cietyof Civil Engineers held in
Ciudad de México en Jmio de 1949. México City, June 1949.
**Departamento de Ingenieria Experimental, Secretaria **Department ~f Experimental Engineering, Secretary of
de recurso^ Hidráulicos de México. Hydraulic Resources, México.
rio y la longitud de la estructura principal the length of the main earth embankmenr
de tierra de 1340m. Las terracerias de tierra structure will be 1,340 m. Earthworks will be
alcanzarán la roca sana en la cimentación por founded upon sound rock by means of a cut-off
medio de un dentellón de 30m. de profundidad. wall 30-m deep. The total volume of materials
La obra en total requerirá alrededor de 3.9 used will be of about 3.8 million cubic
millones de $e m3 de tierra compactada y 4.7 meters of earth and 4.7 million cubic meters
millones de m de enrocamiento. of rockfill.
11.- Los materiales que se van a usar en el 11. Materials to be placed for the
corazón de la cortina se pueden clasificar impervious core of the dam can be classified
como arenas finas uniformes, con un diámetro as unifonn fine sands, with an effective
efectivo y un coeficiente de uniformidad de grain size diameter and a coefficient of
0.Olmm. y 7 respectivamente (Fig. 1) Usando . unifonnity equal to 0.01 mm and 7,
respectively (Fig. 1). Through the use of the
el equipo de compactación estandar (rodillo
pata de cabra), se ha encontrado que los standard field compaction equipment
materiales pueden compactarse en las (sheepsfoot roller) it has been found that
terracerias con un densidad relativa de 75%, the materials can be compacted in the
la que corresponde a una porosidad promedio embankment with a relative density of 75%,
de 40%. La relación de vacíos más alta which corresponds to an average porosity of
obtenida en el laboratorio fue de 1.3. 40%: The highest void ratio determined in the
De acuerdo con el cuadro de valores límites laboratory was equal to 1.3.
dado por Terzaghi sobre arenas inestables With reference to the table of l i m i t j p
(2), la muestra representativa de los values given by Terzaghi for unstable sand ,
materiales que se usarán en la presa no cae the representative sample o% the materials t o
dentro de ese grupo. La mayor parte de las be used for the dam does not belong to such a
pruebas fueron realizadas con un suelo group. Moct of the tests were carried out
(AO-A-54) cuyas caracteristicas with a soil of the AO-A-54 type having a
granulométricas son muy próximas a la muestra grain size di.stribution quite similar to the
promedio de los distintos especímenes average specimen of the different samples
extraidos del préstamo. recovered from the borrow pit.
111.- Siguiendo la técnica de laboratorio 111. Following the laboratory technique
establecida por A. Casagrande en el caso de established by A. Casagrande for the case of
los materiales de la Presa Franklin Falls the construction materials for the Franklin

~ i á m e t r oe q u i v a l e n t e d e l a p a r t í c u l a , e n mm

Envolvente derecho
Riqht boundory

Material 'A"
Frankiin Falis Dom

M0teri01 naOor 2 - X t G q
IFranklin Falls Darn

Valores limites paro orenas lnestobles según K.Terzaghi, Soil

Mechanlcs In Enaineerlna Practice. ~ . 1 0 2
Lirnitin values í o r instóble sand h f t e r K. ~ e r z o ~ h i ' Soil
s
~ e c h a n % sin Enqineerinq Practice, p . 102
Coeficiente de uniformidad

Tamaño efectivo
Effective diorneter
}----)<
Coefficient o f uniformity
mm

~ l a s i f i c a c i ó nM I T
MIT Clossification NOTA
Los curvos 5 y 6 se tomoron de "Compactlon Tests and Critico1
Density Investiqatlon". U.S. Army Corps of Engineers.. Fiq. 01-2
NOTE
Curves 5 and 6 were,faken fram "Cornpoctlon Tests and Critical
Density Investiqation, U.S. Army Corps of Enqineers, Fig 0 1 - 2

Fig. 1.- Curvas de distribución

granulom6trica de dos muestras extraídas del Fig. 1. Grain size analysis distribution
pr6stamo A, presa Alvaro Obregón, y curves for two samples recovered fonn borrow
envolventes que limitan la zona en la que pit A for the Alvaro Obregón Dam and boundary
quedan comprendidas el resto de las curvas envelopes of the rest of the grain-size
granulom6tricas. Con fines comparativos se curves. Curves corresponding t o materials A
mbestran las curvas de los materiales a y b and B used for Franklin Falls Dam, Mass. heve
de la presa Franklin Falls, Mass. been plotted for comparison purposes.
 Propiedades mecánicas de arenas finas 5

(l), la primera etapa del trabajo consistió Falls ~am':'the first stage of the project
en una serie de pruebas de compresión involved a series of drained triaxial
triaxial-drenadas; mediante variaciones en la compression tests; by varying the confining
presión lateral y en la relación de vacios pressure and the initial void ratio of the
inicial de los especimenes, el rango de la specimens, the range of the corresponding
relación critica de vacios correspondiente se critical void ratio was determined from
determinó a partir de mediciones en la measurements of the void ratio variation upon
variación del volumen de vacios, al tiempo de reaching the failure state.
producirse la falla. Tests were made in a triaxial compression
Los ensayos se realizaron en un aparato de chamber to test 3.6-cm diameter cylindrical
compresión triaxial para ensayar probetas specimens. Saturation of the material was
cilíndricas de 3.6cm. de diámetro. La achieved with the application of vacuum and
saturación del material fue hecha por medio using deaired water; in spite of the care
de vacío y empleando agua desaereada; a pesar taken during the preparation of the
del cuidado que se tuvo en la preparación de specimens, the degree of saturation did not
los especimenes, el grado de saturación fue, exceed 85% on the average.
aproximadamente, de 85%. The stress-strain curves and the volume
Las gráficas de esfuerzo-deformación y variations in two of the tests made have been
variaciones volumétricas en dos de las plotted in Fig. 2. Figure 3 presents the
pruebas efectuadas, se muestran en la Fig. 2. variations of the critical void ratio in
La Fig. 3 presenta las variaciones de la terms of the confining pressure u3 for two of
relación critica de vacios con la presión the samples recovered from the earth borrow
lateral u,, en dos de las muestras extraidas
pit.
del préstamo de tierra.
ya que el máximo valor u3 en el prototipo Since the peak value of o- was equal to
es del orden de 6kg/cm2, resultaría necesario about 6 kg/cm2, it would be required to
compactar los materiales, por lo menos, con compact the materials to achieve at least a
una relación de vacios de 0.5 para eliminar void ratio of 0.5 in order to dissipate the
excesos de presión hidrostáticos positivos. positive excess hydrostatic pressure. The
Los ensayos de campo realizados con el equipo field tests performed with the standard
estandar de compactación demostraron que compaction equipment proved that it was
economicamente era dificil obtener en los economically difficult to build embankments
terraplenes una relación de vacios inferior a with a void ratio smaller than 0.65.
0.65. Por lo tanto, la condición establecida Therefore, the condition set by A . Casagrande
por A . Casagrande de que las arenas sean to compact the sands with a relative density
compactadas con una densidad relativa mayor higher than the critical value could not be

Deformación unitaria 8 , en 'lo

Strain 6 ,in 'lo

Fig. 2.- Curvas ~~Esfuerzos-deformacionesl~

y Fig. 2. Stress-strain and volumetric change-
"variaciones volumétricas-deformacionesu de strain curves corresponding to specimens
muestras ensayadas segun la técnica de A . tested with A . Casagrande's technique.
Casagrande.
 O .4
O 2 4 8
Presión lateral, o; en kg/cm2
Canfiriing pressure ,a; in kg/crn2

Fig. 3.- "Relación crítica de vacíos vs Fig. 3. Critica1 void ratio versus confining
presión lateralw, materiales A.O.-A-54 y pressure for materials A.0.-A-14.
A.0.-A-14.

que la crítica, no podía ser satisfecha en la complied with at the lower part of the
parte inferior del terraplén. embankment.
IV. - Con el objeto de tener un mejor IV. In order to achieve a better knowledge
conocimiento del comportamiento mecánico de of the mechanical behavior of fine sands, it
las arenas finas, se propuso efectuar pruebas was proposed to carry out tests under a
a volumen constante midiendo los excesos de constant volume and measuring the excess
presión hidrostática inducidos en las hydrostatic pressure induced at the specimens
probetas durante el ensayo. El primer during the test. The first problem to be
problema que tenía que ser resuelto, fue el solved referred t o the saturation of the
de la saturación de los especímenes; es specimens; it is evident that no reliable
evidente que no pueden hacerse mediciones measurements can be made of the excess
correctas del exceso de presión hidrostática hydrostatic pressure unless the voids are
a menos que los vacios estuvieran completely filled with water. After severa1
completamente llenos de agua. Después de unsuccessful trials with the recommended
varias tentativas sin éxito usando técnicas technicpes, the laboratory developed a simple
recomendadas, el laboratorio desarrolló un method t o guarantee a degree of saturation of
método simple que asegura, en el presente 100% for the material under study. The dry
material, un grado de saturación de 100%. La sample is compacted incide a 3.6-cm interna1
muestra seca se compacta dentro de Un tubo diameter perforated brass pipe; the pipe
preforado de latón, de 3.6cm de diámetro holding the specimen is placed in a vacuum
interior; el tubo con el espécimen dentro es chamber. After severa1 minutes have elapsed,
colocado en una cámara de vacío; deaired cold water (at approximately 15'~) is
transcurridos algunos minutos, se admite agua allowed in the chamber; it was found
fría (aprox. 15 C) desaereada dentro de la convenient to vibrate the chamber to
cámara; es conveniente vibrar ésta para dissipate the vaporized bubbles formed inside
facilitar el paso de las burbujas de vapor the sample (Fig. 4).
que se forman dentro de la pmuestra. (Fig. 4). .
El aparato para los ensayos de compresion The device for the triaxial compression
triaxial a volumen constante lleva en la base tests under constant volume carries at the
una aguja especial de 2mm. de diáinetro y 3cm. base a custo~n needle 2 mm in diameter and
de longitud, (ver Fig. 5) ; el drenaje de la 3-cm long (see Fig. 5); drainage of the
probeta se efectúa por su cara superior. La sample is achieved at top. The neutral
presión neutral es medida con el dispositivo pressure is measured with the Rendulic's
de Rendulic, siendo en líneas generales, device which generally speaking is similar to
semejante al diseñado por el Laboratorio de the one designed by the Coi1 Mechanics
Mecánica de Suelos en el Massachusetts Laboratory oC the Massachusetts Institute of
Institute of Technology. Recientemente este Technology. This device has been recently
dispositivo ha sido reemplazado por celdas de substituted by electrically-controlled
presion controladas eléctricamente, las que pressure cells with the advantage of a
tienen sobre la anterior la ventaja de que el simpler automatic recording.
registro automático es más sencillo.
La técnica seguida en las probetas The laboratory procedure followed for the
triaxiales a volumen constante fue similar a triaxial compression tests under constant
la indicada en el caso de las determinaciones volume was similar t o the one referred t o for
de relación de vacios crítica. Los determining the critica1 void ratio. The
especimenes se prepararon con el material specimens were prepared with the typical
7

Propiedades mecánicas de arenas finas 7

Fig. 4.- D i s p o s i t i v o p a r a s a t u r a r muestrds Fig. 4 . Equipment t o s o t 8 ~ r . i t ec y l i n d r i c a l

c i l i n d r i c a s d e 3 . 6 cm de d i á m e t r o y 10 cm d e specimens 3 . 6 cm i n d i a m e t e r a n d 10 c m h i q h .
altura.
1. P e r f o r a t e d b r a s s plpe t o hold t-h,,
1. Tubo p e r f o r a d o , d e l a t o n , con l a muestra specimen.
en e l i n t e r i o r 2 . Vacuum chamber.
2. Cámara d e v a c í o . 3. D e p o s i t f o r d e a i r e d cold water.
3. Depósito d e agua f r i a d e s a e r a d a . 4 . Trap.
4. Trampa.

Fig. 5.- Aparato d e Compresión T r a x i a l Fig. 5. T r i a x i a l compression charhber.

1. Base 1. Base
2. Aguja e s p e c i a l p a r a mediciones d e l exceso 2 . Custom needle to measure excess
de p r e s i ó n h i d r o s t á t i c a . hydraulic pressure
3. Bureta 3. B u r e t t e
4. Cabeza d e l a p a r a t o 4 . Head o £ t h e d e v i c e
5 Y 6 . Empaques 5 & 6 Packing
7. P i s t ó n d e 5 / 1 6 ' d e d i á m e t r o 7 . 5/611 d i a m e t e r p i s t o n
representativo mencionado antes (AO-A-54). material mentioned before (AO-A 54). Some of
Parte de los ensayos se realizaron aplicando the tests were performed by applying the
las cargas en un aparato de esfuerzo loads in a controlled-stress chamber and the
controlado y el resto de ellos, en un remaining tests were made in a
dispositivo del tipo de deformación strain-controlled device. The specimens were
controlada. Los especimenes fueron sometidos subjected to low-frequency cyclic axial
a esfuerzos axiales oscilantes de baja stresses (1 cycle/min) . In Figs. 6, 7 and 8
.
frecuencia (1 ciclo/min.) En las Figs. 6, 7 the c u w e s for stress-strain and excess
y 8, se muestran las relaciones de hydrostatic pressure are shown for some of
esfuerzos-deformación y exceso de presión the tests. It is worth mentioning that
hidrostática de algunas de las pruebas. Es regardless of the applied confining pressure
interesante notar que, independientemente de and of the initial conditions of the
la presi6n lateral aplicada y las condiciones

Fig. 6.- ' Curvas Esfuerzos-deformaciónI1 en t'ig. Stress-strain

(7. curves from
pruebas triaxiales rápidas consolidadas, consolidated quick triaxial tests (controlled
(deformación controlada) material A.0.-A-54. ,Jeformation): material A.0.-A-54.

o
,:
l 99 k g / c m 2
30 05

1 1o 1O0
Tiempo,en minutos
Time, i n minutes

Fig. 7- Consolidacion uniforme. Muestras Fig. 7. Uniform consolidation. Cylindrical

cilíndricas de 3.6cm de diámetro y lOcm de specimens 3.6 cm in diameter and 10 cm high
altura. drenaje en la base superior de la drained at top; pore-water pressures measured
probeta.- Mediciones de la presión neutral en at the base. Material A.0.-A-54.
la base inferior. Material A.0.-A-54.
 Propiedades mecánicas de arenas finas 9

CONDICIONES INICIALES DE L A
MUESTRA :
I N I T I A L CONDlTlONS OF T H E
SPECIMEN

p r e s i ó n de c o n s o l i d o c i ó n , u c = 5 . 7 5 r q / c m 2

CONDICIONES DE L A PROBETA
DESPUES D E L ENSAYO :
CONDlTlONS OF T H E S P E C I M E N
AFTER TESTING
wi = 19.71 ef = O 51
Densidad de s ó l i d o s , S S :2.75
Specific q r a v i t y , S S = 2 . 7 5

i.1OTfiS
Dferencio entre esfuerzos c i i n c i ~ o l e s( 0 , - u 3 ) en hq/cm2
Presión loterol oplicado u3,en k g / c m 7
Excesode presión hidros'rilica ",en hg/cm2
Deforrnoción a i i o i u n i t o i i a 6 , en
NOTES
Principol stress d i f f e r e n c e (ul-03) .in hg/cm2
Appiied confininq pressure u3,i n k g / c m 2

Tiempo, en minutos Excess h y d r o s t o t i c p r e s s u r e u , in kg/cm'

t i m e , in minutes Axial strain 6 ,i n %

Fig. 8.- Relaciones entre Fig. 8. Stress-strain-time relationships

llEsfuerzos-deformaciones-tiempo".
en pruebas from consolidated quick triaxial tests
triaxiales rápidas consolidadas (esfuerzos (controlled stress); material A.0.-A-54.
controlados) material A.0.-A-54

iniciales de las probetas, para una specimens, for a strain of about 1.5% the
deformación unitaria del orden de 1.5%, el excess hydrostatic pressure (u) reaches a
exceso de presion hidrostática (u) pasa por peak value. The value of u wiien the specimen
un máximo. El valor de (u) cuando el approaches failure tends to a minimum
espécimen se aproxima a la falla tiende a un positive or negative value depending on the
mínimo positivo o negativo, según la presión confining pressure u and on the void ratio
lateral u3 y la relación de vacíos de la of the consolidated specimen. These results
probeta consolidada. Estos resultados are in agreement from a qualitative point of
concuerdan cualitativamente con las view with the observations made by A.
observaciones de A. Casagrande en sus Casagrande in the course of his studies on
estudios sobre densidad crítica de arenas. critica1 density of sands.
Las máximas presiones neutrales y sus The maximum neutral pressure3 and the
valores en el momento de la falla de los values reached upon failure of the specimens
especímenes aparecen dibujados en las Figs. 9 have been plotted in Figs. 9 and 10
y 10, tomando como variables independientes considering the stresses a 3 and the initial
los esfuerzos u3 y las r-elaciones iniciales void ratios (e,) as independent variables.
de vacíos (e,). Para e = 0.65 (relación de For e=0.65 (void ratio for the planned
vacíos en el terraplén proyectado), el exceso embankment), the excess hydrostatic pressure
de presión hidrostática en la falla es a lo at failure is no more than 10% of the
Sumo 10% de u3. u -value.
Las relaciones entre esfuerzos y The relationships among stresses and
deformaciones para los diferentes ensayos y strains for the different tests made and for

Exceso de presión hidrostática y presión 3 ~ h eterms excess hydrostatic pressure and

neutral son sinónimos en el texto neutra pressure are used as synonyms in the
pacer
 4

e , :~ e l a & de vac(os inicial

e i ' I n i t ~ a l void r a t i o

n
O 2 4 6 8 10
Presión laterol efectlva u3, e n kg/crn2
E f f c c i i v e confining pressure u3 ,i n kg/crn2

Fig. 9.- Relaciones entre el maximo exceso Fig. 9. Relationship betweeri the maximum
de presión hidrostática y la presion 1,iteral excess hydrostatic pressure and the effective
efectiva. Pruebas triaxiales rapidas confining pressure. Material A.0.-A-54.
consolidadas. Material A.0.-A-54.

.-
X
w w
- Presión loferol efectiva u 3e n kg/cm2
Effective confining pressure u3 i n kg/cm2

NOTA : L o probeta se considera en estado de falla

cuando (ui-u3) alcanzó SU valor m á x i m a
N O T E : Failure o f the specimen was ossumed
when(u;-u,) reached i t s maximun value

Fig. 10.- Relaciones entre el exceso de FIy. 10. Relationship between the excess
presión hidrostática en el momento de la hydrostutic pressure durinq failure and the
falla y la presion lateral efectiva. Pruebas effective confining pressure. Consolidated
triaxiales rápidas consolidadas. Material quick triaxial tests for material A.0.-A-54.
A.0.-A-54.

dos relaciones iniciales de vacíos (e = 0.74 two initial void ratios (0.74 and 0.57) are
y 0.57) se presentan en la Fig. 11, en la depicted in Fig. 11 followinq3, the
forma propuesta por D. W. Taylor (3). La presentation proposed by D.W. Taylor . The
diferencia fundamental entre los dos grupos basic difference between both sets of curves
de curvas reside en la forma de las mismas. resides in the shape of them. For e,=0.75 the
Para e = 0.75 el material alcanza la material reaches the limiting failure
condición límite de fal1.a con deformaciones condition with strains of about 5% and it
unitarias del orden de '5% y describe luego then follows a curve that approximately
una curva que aproximadamente coincide con la coincides with Mohr's envelope until failure
envolvente de Mohr hasta que se produce la is reached: on the other hand, shear stresses
ruptura: en cambio, los esfuerzos cortantes increase in a practically linear trend in
crecen prácticamente en forma lineal con los terms of the effective normal stresses for
esfuerzos normales efectivos, para el mismo the same soil but with a denser structure
suelo con una estructura más densa (e, = (e =0.57). This distinctive behavior observed
0.57). Este distinto comportamiento observado in the consolidated-quick triaxial tests is
en las pruebas triaxiales rápidas- related to the dilatancy of the material. It
consolidadas, está relacionado con la should be also observed that there is a
dilatancia del material. Nótese, además, la difference between Mohr's envelopes shown in
diferencia en las formas de las envolventes
 Propiedades mecánicas de arenas finas 11

ei:0.75 ~ e l o c i ó nd e voc;os inicio1

1 lnitiol void rotio
Envolvente d e M o h r

10

o
O 2 4 6 8 10 12 "O 2 4 6 8 10 12 14 16

Esfuerzos normales e f e c t i v o s 5,
en k g / c m 2 E s f u e r z o s normoles e f e c t i v o s E, en kg/crn2
Effective n o r m o l stress 5 , i n k g / c m 2 Effectiwe norrnol stress 8,in kg/crnz

NOTA : Los números q u e o p o r e c e n junto o los c u r v o s corresponden o l o s deformaciones

o x i o l e s unitarios e n % , experimentados p o r l o s p r o b e t o s
NOTE: F i g u r e s b e s i d e s e o c h c u r v e c o r r e s p o n d t o the o x i o l s t r o i n s i n '/O ,
e x p e r i e n c e d b y t h e specimens

Fiy. 11.- Esfuerzos cortantes en el plano de Fig. 11. Shear stresses at a 60' plane versus
60 vs esfuerzos normales efectivos. Pruebas effective normal stresses. Consolidated quick
triaxiales rápidas consolidadas. Material triaxial tests for material A.0.-A-54.
A.0.-A-54.

de Mohr que aparecen en la Fig. 11. Fig. 11.

La repetición de los ciclos de carga The repetitive load cycles varying from
variando entre O y un esfuerzo axial próximo zero to a value equal to the axial stress
neutral después del primer ciclo tendía a close to failure (Fig. 8) demonstrated that
aumentar siguiendo una ley semejante a la de the neutral pressure after the first cycle
las deformaciones, aún en los casos en que la tried to increase according to a law similar
densidad inicial del espécimen fuera mayor to that corresponding to the stresses even in
que la critica. Esto sugirió que una probeta the cases where the initial relative density
podría fallar con un esfuerzo axial menor que of the specimen was higher than the critica1
el de la correspondiente prueba estática value. This suggested that a specimen was
(rápida-consolidada estándar), si fuera likely to reach failure under an axial stress
sometida a una excitación de tipo oscilante. smaller than the value of a corresponding
static test (standard consolidated-quick
triaxial test), when subjected to a
cyclic-type of excitation.
V.- En vista de los resultados mencionados, V. When confronted with the results
se decidió construir un nuevo dispositivo de mentioned above, it was decided to make a new
carga para ensayar probetas cilíndricas de loading device to test 3.6-cm diameter
3 . 6 ~ .de diámetro y 10cm. de longitud, con cylindrical specimens 10-cm high under cyclic
esfuerzos axiales oscilantes v frecuencias axial stresses and frequencies ranging from
variables entre 25 y 200 ciclos/min. (véase 25 to 200 cycles/min (refer to Fig. 12).
Fig. 12). Through a lever guided by a suitable
Una palanca guiada por un mecanismo mechanism a harmonic movement 1s transmitted
adecuado impone un movimiento armónico a un to a calibrated spring attached to the
resorte calibrado fijo al marco de carga, el loading frame which in turn transmits the
que a su vez transmite la carga a la probeta load to the specimen by means of the piston
a través del pistón de la cámara triaxial. La on the triaxial chamber. The amplitude of the
amplitud del movimiento armónico puede ser harmonic movement can be adjusted in order to
ajustada para inducir en el espécimen la ley induce the desired variation of the axial
de esfuerzos axiales deseada. Por medio de un stresses in the specimen. It is possible to
regulador "speed master", es posible variar vary the frequencies of the movement from 25
las frecuencias del movimiento entre 25 y 200 to 200 cycles/min through a "Speed Master"
Fig. 12.- Aparato para ensayos con cargas Fig. 12. Equipment for cyclic load tests.
oscilantes.
1. Traxial compression chamber
Cámara de Compresión Triaxial 2. Sliding base
Base móvil 3. Pressure gages
Manómetros 4. Loading frame
Marco de carga 5. Compressed-air compensator
Compensador, de aire comprimido 6. Calibrated spring
Resorte de calibrado 7. Exciting mechanism
Mecanismo excitador 8. Speed regulator
Regulador de velocidad 9. 3/4-HP motor
Motor 3/4 H.P. 10. Counter of revolutions
Contador de revoluciones

ciclos/min.; otros rangos de frecuencias se regulator; other frequency ranges are easily
pueden obtener fácilmente intercambiando las attainable by exchanging ,the transmission
poleas de la transmisión. El peso propio del pulleys. The dead weight of the oscillating
sistema oscilante está balanceado mediante un system is balanced by a compressed-air
mecanismo de aire comprimido. Para eliminar operated mechanism. To eliminate the effect
el efecto de las deformaciones de la probeta of the deformations induced by the specimen
sobre el resorte calibrado, una base on the calibrated spring, a base plate
controlada por engranes mueve la cámara de controlled by gears displaces the triaxial
compresión triaxial hacia arriba con la misma compression chamber upwards at the same rate
velocidad con que se presentan aquéllas. of the strain variation.
LaS deformaciones Y tiempos The deformations and the corresponding
correspondientes se registraron con un equipo times were recorded with a movie-making
cinematográfico; la restante información de equipment; the remaining information of the
los ensayos fue tomada a intervalos de tiempo tests was taken at fixed time periods. An
conocidos. Se está preparando un equipo automatic- controlled electronic equipment is
electrónico de control automático para being readied to facilitate the data
facilitar la obtención de los datos. recording.
La calibración del resorte transmisor de The calibration of the spring to transfer
cargas al espécimen fue hecha con un the load to the specimen was accomplished
dinamómetro especial, aplicando cargas with a custom dynamometer by applying cyclic
oscilantes y para distintas frecuencias. La loads at different frequencies. Well-defined
consistencia de las curvas así obtenidas trends of the curves thus obtained showed
demostró que la frecuencia natural del that the natural frequency of the loading
dispositivo de carga cae fuera del rango para device falls outside the range for which it
el que fue diseñado. was designed.
En la Fig. 14 aparecen las curvas de In Fig. 14 stress-strain-time curves are
"esfuerzos, "deformaciones vs. tiempos" de depicted for one of the tests: axial stresses
uno de los ensayos, en el que los esfuerzos varied from zero to values varying linearly
axiales variaron entre O y valores que se with time in general terms. A frequency of
incrementaron en forma lineal, 40 cycles per minute was used. The
aproximadamente, con el tiempo; la frecuencia stress-strain curve .shown in such a figure
usada fue de 40 ciclos/min. Con los valores was defined from the peak values of the axial
más altos de los esfuerzos axiales en cada stresses corresponding to each of the loading
ciclo de carga, se ha construido la curva cycles together with the same type of curve
"esfuerzo-deformaciónt1que se muestra en la determined from a static test. The 'chape of
misma figura junto a la curva correspondiente both curves is practically the same with the
obtenida en prueba estática. Salvo la
 Propiedades mecánicas de arenas finas 13

1o

8
*
<5
01
1 0
c c 6
--
" -
QJ .-

f'p
bI< 4
b"
-4
8
:2
YU> 2
QJ P
2 ?ñ
W

Densidad de solidos : 5 %= 2 7 5
S Specific r e i g h : o f solids

!$eTfeSncia entre a s esfuerzos principales u,-u3,en kg/cm2 Tiempo, en s e g u n d o s

presión lateral oplicoda u 3 , en hg/cm2 T i m e , in seconds
Exceso de presión hidrostótica u , en k g / c m Z
,
Deformación axiai u n i t a r i a S en
NOTES
Principal stress difference vi-U3.in k g / c m 2
Applied confinig pressure u3,en k g / c r n z
NOTE
Excess hydrostotic pressure u , in kg/crn
Axiol strain 6 , in
The stress-strain curve was plotted DefcrmociÓn caial unitaria S ,%
with t h e rnaximun relotive v a l u e s a f Axial strain ,%
(C,"J3) and 6

Fig. 14.- Relaciones Fig. 14. Stress-strain-time relationships for

"Esfuerzos-deformaciones-tiempo". ensayo de consolidated quick triaxial compression tests
compresión triaxial rápido-consolidado con with cyclic loading.
aplicación de cargas axiales oscilantes. Frecuency: 40 cycles/min; applied confining
Frecuenia: 40 ciclos/m-in. Presión lateral pressure: 4.05 kg/cm2.
aplicada, cm= 4.05 kg cm

diferencia en los esfuerzos (o,-o-),

- - la forma exception of the stress difference (o1-03).
de ambos es enteramente similar. A summary of the results obtained from the
Un resumen de los resultados registrados en tests made in presented in Fig. 15 in which
las pruebas realizadas se presenta en la Fig. the principal stress difference (u,-o-)1 3
at
15, en la que la diferéncia de esfuerzos the moment of failure has been plotted versus
principales (0,-03) para la falla de las the frequency of the cyclic axiai load
probetas se dibuja contra las frecuencias de applied. Three of the curves correspond to
?as cargas axialés oscilantes aplicadas. Tres tests made with confining pressures u- equal
las curvas corresponden a ensayos en que to 0.5, andthe
4 kg/cm2 and esual
an initial void
las presiones laterales o3 eran iguales a ratio (e 11 of specimens to 0.74. In
' l .
0.5, 1 y 4kg/cm2 y la relación de vacíos the fourth curve shown in the figure, the
inicial en los especimenes de 0.74. En la test conditions vere e,=0.65 and o 3=l kq/cm2.
cuarta curva aue se incluve *
en la misma
- ~

figura, las condiciones de la prueba fueron: Frequencies ranged from 1 to

e = 0.65 y u3 = lkg/cmz. Las frecuencias 200 cycles/min. The reSUltS presented in Fig.
e:tuvieron comprendidas entre y 200 15 should be regarded as preliminary since
ciclos/min. Los resultados que se muestran en the effect of the cyclic variations on the
la Fis. 15 deben considerarse ~reliminares. confining pressure u3 induced by the
ya que el efecto de las variaciones hydraulic system as a result of the piston
oscilatorias en la presión lateral u3
inducidas en el sistema hidráulico por el

o l l ! ~ ~ l l ! ' I I
o 1O0 220
Frecuencios,en ciclos/min
Frequency , in cycles /min

Flg. 15.- Variacion de los valores de el-o,, F'ig. 15. Variation of the values ul-u,,
contra frecuencia. Pruebas triaxiales rápidas versus frecuency. Consolidated quick-traixial
consolidadas con cargas axiales oscilantes. tests with cyclic loading. Material A.0.-A-54
Material A.0.-A-54.

desplazamiento del pistón de la cámara displacement incide the triaxial chamber were
triaxial, no se tomaron en cuenta. Se están not taken into account. Tests are being
realizando ensayos en los que se ha carried out in which air has been used
sustituido el agua por aire para transmitir instead of water to apply the confining
la presión lateral al espécimen. pressure to the specimen.
VI.- Por último, el laboratorio de la VI. Finally, the laboratory of the
Secretaría de Recursos Hidráulicos está Secretary of Hydraulic Resources is
experimentando en una cámara triaxial de experimenting with an unusually large
dimensiones desusuales;' las muestras triaxial chamber; the cylindrical specimens
cilíndricas son de 15cm. de diámetro y 40cm. are 15 cm in diameter and 40-cm high (Fig.
de longitud (Fig. 13). Debido a las .
13) Because of the dif ficulties encountered
dificultades para saturar muestras de ese for the saturation of such large samples, the
tamaño, las primerqs pruebas se están first series of tests is being performed with
realizando con material seco; las variaciones dry material; variations in the volume of
en el volumen de vacíos se determinan voids are determined indirectly by measuring
indirectamente por medición de las presiones the pressures developed in the air that fills
que se desarrollan en el aire que llena los the voids of the sample. It is not possible
huecos de la probeta. No es posible adelantar yet to advance results of the tests made
datos sobre los ensayos efectuados debido a since this stage of the project is sífill on a
que esta fase del trabajo está en el periodo trial phase.
de prueba.
 Propiedades mecánicas de arenas finas 15

Fig. 13.- Aparato de Compresion Triaxial, Fig. 13. Triaxial compression chamber for
para muestras de 15.2 cm de diámetro y 38 cm samples 15.2 cm in diameter and 38 cm high.
de altura.
1. Triaxial chamber
1.- Cámara Triaxial 2. Meter of volumetric changes in the
2.- Medidor de las variaciones volumétricas specimen
de la muestra 3. Loading frame
3.- Marco de carga 4. Compressed-air tank
4.- Tanque de aire comprimido 5. Water reservoir
5.- Depósito de agua

VI1.- Las conclusiones que derivan de los VII. Conclusions derived from the studies
estudios en su estado actual pueden as of to date can be summarized as follows:
condensarse en cuatro puntos:
a) En vista de que las arenas finas y a) Since the fine and uniform sands to
uniformes por usarse en la construcción de be used for the construction of the earthfill
las terracerias de la presa difícilmente could be hardly compacted with void ratios
podrían ser compactadas con relaciones de smaller than 0.6, Casagrande's assumption for
vacios menores que 0.6, la condición de the compaction of cohesionless soils cannot
Casagrande sobre compactación de materiales be satisfied at the lower part of the earth
no cohesivos no puede satisfacerse en la embankment in whic$ confining pressures range
porción inferior de la estructura de tierra, from 4 to 6 kg/cm .Laboratory test results
showed that for a critical void ratio of 0.6
en la que las presiones laterales varían
entre 4 y 6kg/cm . De acuerdo con los datos the a~plied confining pressure reaches
de laboratorio, para una relación critica de 4 kg/cm.
vacíos de 0.6 la presión lateral aplicada es
de 4kg/cm2.
b) Las pruebas de compresión triaxial a b) Triaxial compression tests under a
Volumen constante han ,mostrado que, para constant volume have shown that for confining
presiones laterales comprendidas entre O y pressures ranging from O to 9 kg/cm2 and with
9kg/cm2 y material compactado con una the material compacted to a relative density
densidad relativa de 75%, el exceso de of 75%, the hydrostatic excess pressure does
Presión hidrostática no ha sido superior al not exceed 10% of the confining pressure
10% de las presiones laterales aplicadas. .
appl ied
! Los ensayos triaxiales
rápidos-consolidados con aplicación de cargas
C) Consolidated-quick triaxial
compression tests with cyclic axial loading,
axiales oscilantes aún en los especimenes even for specimens prepared with void ratios
Preparados con relaciones de vacios exceeding the corresponding critical void
Superiores a la relación critica ratio, have not shown any trend towards
correspondiente, no han presentado síntomas possible liquefaction of the material.
de una licuación peligrosa del material.
 d) Para la verificación de la estabilidad d) For verifying the stability of the
de la cortina se considera aceptable usar los embankment, it has been deemed acceptable to
valores de la resistencia al esfuerzo use values of the shear strength determined
cortante que se obtengan en pruebas de from consolidated-quick triaxial compression
compresión triaxial rápidas- consolidadas. tests.
Este es un breve resumen de los esfuerzos This paper has presented a brief summary of
realizados hasta el presente en el the work carried out to date by the
Laboratorio de la Secretaria de Recursos laboratory of the Secretary of Hydraulic
Hidráulicos, para estudiar la resistencia al Resources (SRH) aimed at studying the shear
esfuerzo cortante de arenas limosas strength of uniform silty sands. It has been
uniformes. Se estima que queda mucho por considered that tliere is still much to be
realizar antes que se tenga un conocimiento done before a comprehensive knowledge of the
adecuado del comportamiento de estos behavior of this type of material can be
materiales. fully understood.
El autor desea hacer hincapié en el The author wishes to acknowledge the
entusiasta apoyo prestado a este estudio por enthusiastic support received from Messrs.
los Sres. Vicente Orozco Assoc. M. ASCE, José Vicente Orozco, M. ASCE, Chief Engineer
Ingeniero en Jefe (S,R.H.);Aurelio Benassini (SRH), Aurelio Benassini, Assoc. M. ASCE,
Assoc. M. ASCE, 2 Ingeniero en Jefe Deputy Chief Engineer (SRH), Oscar Vega
(S.R.H.): Oscar Vega Argüelles, Director Arguelles, Director General of Studies and
General de Estudios y Proyectos (S.R.H.) y Projects (SRH), and Fernando Hiriart, Assoc.
Fernando Hiriart Assoc. M.ASCE, Jefe del M. ASCE, Head of the Department of
Departamento de Ingeniería Experimental Experimental Engineering (SRH). Thanks are
(S.R.H.). Por su valiosa ayuda, se expresa also expressed to Messrs. M. Mazari, H.
agradecimiento a los Ings. M. Mazari, H. Hiriart and F. Skertchly for their valuable
Hiriart y F. Sckertchly. help.

REFERENCIAS REFERENCES
(1) A. Casagrande, "Compaction Tests and (1) A. Casagrande, "Compaction Tests and
Critical Density Investigation of Critical Density Investigations of
Cohesionless Materials for Franklin Cohesionless Materials for Franklin
Falls Damu, Corps Engineers U.S. Army. Falls Damn, U.S. Army Corps of
Engineers.
(2) K. Terzaghi and R.B. Peck, "Soil (2) K . Terzaghi and R.B. Peck, "Soil
Mechanics in Engineering Practice", Pag.
101. - --
Mechanics in Engineering Practice", p.
IUI.

(3) D.W. Taylor, "Reports on Cooperative (3) D.W. Taylor, "Reports on Cooperative
Research on Stress, Deformation and Research on Stress, Deformation and
Strength Characteristics of Soils", U.S. Strength Characteristics of Soils8',U.S.
Waterways Experiment Station. Army Waterways Experiment Station.

Translated by R . Esquivez
Traducido por R. Esquivez
 SOFT SOlLS AND THE SUBSOIL OF MEXICO ClTY
1 SUELOS BLANDOS Y EL SUBSUELO DE LA CIUDAD
DE MEXICO
 Static behavior of floating piles in soft clay.
Comportamiento estático de pilotes de fricción en arcilla blanda.

A. Jaime, instituto de Ingeniería, UNAM.

E. Rojas, Instituto de Ingeniería, UNAM.

H. Legorreta, instituto de Ingeniería, UNAM.

ABSTRACT: Results of penetration and pullout RESUMEN: Los resultados de pruebas de

series of tests carried out on model concrete penetración y extracción en modelos de
piles driven in a reconstituted sample of pilotes de concreto hincados en una muestra
Mexico City clay have shed light on de arcilla reconstituida de la ciudad de
significant features of the behavior of Mexico arrojan luz sobre los aspectos
floating piles under static load conditions. significativos del comportamiento de pilotes
Friction pullout capacity attains a maximum de fricción sometidos a cargas estáticas. La
value slightly less than the friction capacidad friccionante de extracción adquiere
penetration capacity. Tests on model piles un valor ligeramente menor que la de
with different cross section geometries penetración. Ensayando modelos de diferente
(circular, hexagonal, square, triangular and geometría en su sección transversal
plane) but equal lateral areas show that pile (circular, hexagonal, cuadrada, triangular y
section has a strong influence in the bearing plana), pero con la misma área se demuestra
capacity. An almost linear relntionship was que la sección del pilote influye fuertemente
found between the interna1 angle, 153, formed en el valor de su capacidad. Se encontró una
by the sides of the pile and the bearing relación casi lineal entre el ángulo 3,
capacity. Some piles were driven without formado por los lados del pilote y su
preboring and others in prebored holes of capacidad de carga. Algunos pilotes se
various diameters. It is also shown that the hincaron sin perforación previa, otros con
drivinq processaffects the load capacity of perforaciones de diversos diámetros. También
piles. Test results indicate that preboring se demuestra que el procedimiento de hincado
decreases the bearing capacity of piles. afecta la capacidad de carga de los pilotes.
Finally, results of full scale pile tests are Los resultados de los ensayes indican que la
compared with those on model piles. perforación previa disminuye la capacidad de
carga de los pilotes. Finalmente, se comparan
los resultados de ensayes en pilotes de
tamaño natural con los de modelos.

1 INTRODUCTION capacity, severa1 series of tests were

carried out on piles with equal lateral area
During the September, 1985 earthquakes that but different shapes: circular, hexagonal,
shook Mexico City, severa1 buildings founded square, triangwlar and plane.
on friction piles tilted or subsided and one In Mexico City, preboring is mostly used in
structure collapsed. For this reason a construction practice because there are one
research on the static and cyclic behavior of or two layers about 1-m-thick of stiff sandy
floating piles in soft clays was undertaken. clay or cemented silt, at variable depths,
The first stage of the investigation included interbedded in the upper clay deposit of the
static and cyclic load tests of full scale City. These layers cause pile-driving
friction piles driven in Mexico City clay problems, thus standard practice is to dril1
(Jaime et al, 1989, 1990). To overcome budget them. Besides, preboring helps diminish
limitations and cumbersome field test upheaval and lateral displacement of soil
procedures, the second stage of this study which could affect neighboring structures or
continues in the laboratory with model piles. the final layout of the pile group. In order
In this paper the results of static load to study the influence of preboring on the
tests on model friction piles driven in a bearing capacity of friction piles, severa1
reconstituted soft clay sample are discussed. elements were driven in boreholes of various
The equipment used is described as well as diameters and were statically tested.
the method to prepare the clay specimen. The
tests were performed on precast model
concrete piles scaled to 1/25 the usual 2 EQUIPMENT DESCRIPTION
dimensions of these elements employed in
Mexico City. A detailed description of the equipment is
To study the effect of the cross section given elsewhere (Jaime, 1990; Jaime, Rojas
geometry of the piles on their bearing and Legorreta, 1991), here only a brief
20 Jaime; Rojas and Legorreta

summary is presented. An oedometer was head supported the stemand the core of the
designed and built to prepare the clay measuring devices (Fig. 3). The electric
specimen. It consists of two steel cylinders, signals of the DCDT and force transducer were
1.1 m in diameter and 0.9 m high, one placed monitored by a computer and an X-Y analog
on top of the other, so the final height is recorder.
1.8 m. It has watertight base and top cap,
both with drainage outlets (Fig. 1). It can
whitstand the pressure exerted by a
clay-water mix plus an additional pressure of
50 kPa.

Folded membrone

U p p e r cylinder -

Fig. 1 Oedometer. Cross view

Fig. 2 Multipurpose frame
The lower cylinder of the oedometer has 16
supports equally spaced around its upper
edge. On them, a three-leg frame can be
placed in different positions. This frame has
two functions: As a guide during the drilling
and pile driving operations (Fig. 2); and as
a reaction frame while loading the piles.
Three guiding plates mounted at three
elevations on the frame (Fig. 2), guide a 9
mm drill rod. This rod has an interchangeable
auger-type bit screwed at its lower tip. It
is used to drill holes when required, into
the clay specimen. To do this, the drill rod
is rotated manually while pushing it down 10
cm. Then the auger is withdrawn with clay
clinging to it. The soil is removed for
water content determination. The bit is
inserted again into the hole and turned
further, until a 65 cm depth is reached.
These plates also lead the piles during the
jriving process.
The piles were subjected .fo penetration and
pullout tests. Each test was carried out
applying load increments to the pile at
10-15-minute intervals; the piles were loaded
by meanc of a pneumatic jack attached to the
frame at a convenient leve1 (Fig. 3 ) . To
control the load increments the jack was
connected to an air pressure regulator.
Vertical displacements were measured with a
dial gage (0.01-mm resolution) and a DCDT.
The dial gage and DCDT bodies were firmly
attached to two legs of the frame. Two
horizontal steel bars connected to the pile Fig. 3 Loading system
 Behaviour of floaiing piles 21

3 CLAY SPECIMEN The mean water content of the clay specimen

is 274 % with a standard deviation of 20 %
A soft clay specimen was prepared. Theo final (coefficient of variation 0.07). It was
dimensions of the sample were 1.1 m in obtained from 78 determinations nade with the
diameter and 0.9 m high. The process to soil removed from the prebored holes.
prepare the specimen included severa1 One undisturbed Shelby tube sample was
operations: Mixing, de-aeration and pouring taken from the soil specimen to determine its
of the clay-water mix in the oedometer. mechanical properties. Tnree UU strain
The clay specimen was made thoroughly controlled triaxial tests were performed. The
blending water and natural clay chunks by mean undrained strength was found to be 6.5
means of a concrete-mixer. The clay chunks kPa .
were taken from the first clay deposit of
Mexico City, at 5-40 m depth, with 10-cm 4 STATIC LOAD TESTS
Shelby tubes. A total of 15 kN of soil was
required. The water content of the natural Severa1 types of tests were carried out:
samples varied between 200 and 500 %; the Penetration, extraction and special tests.
liquid Limit between 200 and 500 %, and the The latter were aimed to determine the pile
plastic limit between 40 and 100 %. The mean point bearing capacity at different depths in
specific gravity was 2.4. The clay had a high the soil specimen; also, to observe the
content (25 % ) of microfossils and traces of influence of the driving process on the pile
organic matter. response.
Once a 20-liter batch of homogeneous mud In total eighteen isolated piles were
was obtained, with a water content of 650 %, statically tested, seven with square cross
it was subjected to vacuum for two hours in a section, four with circular section, five
special stee3 bottle (Fig. 4), which in turn triangular, one hexagonal and one
wasosubmerged in water at a temperature of rectangular. The 78 cm long piles were made
60 C. Then the de-aerated mud was poured in of microconcrete using as reinforcement a
the oedometer by means of a flexible tube single aluminum rod 9 mm in diameter. Except
connected to the bottom of the bottle, as for the rectangular piJe, al1 had the same
seen in Fig. 4. While pouring the mud, the lateral area of 568 cm , with the following
free end of the, tube was kept under a small cross section dimensions: Circular, 2.54 cm
depth of water (5 cm). This operation was in diameter; hexagonal, 1.33 cm on the side;
repeated until the oedometer was full and square, 2.0 cm wide; triangular, 2.66 cm on
took about one month. the side; and rectangular (identified as
From the beginniog of the pouring operation plane), 4.0 cm x 0.75 cm. The pile layout in
some sedimentation of the mix ocurred. The tne specimen is shown in Fig. 5.
specimen was left to consolidate under its Piles 1, 2, 6, 13, 14, 16, 24 and 25 (Fig.
own weight for 4 weeks. Then the free water 5) were driven without preboring. The rest in
was removed and a fil'ter material was placed 65-cm long prebored holes with nominal
on top of the sample; at that time the clay diameters (Dpp) of 14.3, 18.5 and 21.6 mm
had a tooth-paste consistency. The latex (table 1). Besides, an addition91 triangular
membrane was placed, folded, over the filter pile was driven in a 10.5 cm diameter hole.
(Fig. l), and the oedometer top cap was In al1 cases the effective embeddment length
placed. was 71 cm. The piles were left at rest at
The consolidation process continued least 3 weeks before testing.
applying air pressure increments of 5 kPa,
uritil a total pressure of 35 kPa was reached. 4.1 Penetration tests
Each increment was left until no more water
flowed through the drains. It took seven more Table 1 summarizes the penetration test
weeks to complete the consolidation of the conditions and results. The load versus
clay sample. displacement curves (P versus 6) of the piles
driven without preboring are depicted in Fig.
6. The results of piles driven in preboied
holes of various diameters are also shown in
Fig. 6.
'--
Atmosphere

;
a) Cloy-water m i x
1

Fig. 4 Specimen preparation Fig. 5 Pile layout

22 Jaime, Rojas and Legorreta

Table 1. P e n e t r a t i o n tests.

P i l e Shape Dpp Pu 6 Pr Date o f

Pu/Pus d r i v i n g t e s t i n g
NO mm kg mm kg d/m/y d/m/y

6 R O 49.6 1.45 32.3 1.00 120190 060390

8 R 14.3 34.1 1.59 22.7 0.69 150190 070390
10 R 18.5 33.8 1.45 23.0 0.68 150190 080390
12 R 21.6 32.7 2.05 23.0 0.66 160190 090390

24 H O 43.2 1.81 32.4 -- 141090 151190

2 C O 39.2 2.00 27.4 1.00 020390 230390

3 C 14.3 34.4 1.58 22.7 0.88 061289 100190
5 C 18.5 26.9 1.34 18.0 0.69 071289 100190
7 C 21.6 18.1 1.04 12.1 0.46 071289 040189

13 T O 33.2 1.61 21.9 1.00 081289 080190

9 T 10.5 28.2 1.32 19.5 0.85 081289 050190
11 T 14.3 25.0 0.95 15.0 0.75 111289 080190
15 T 18.5 26.0 1.27 18.2 0.78 111289 090190
4 T 21.6 18.0 1.48 11.9 0.54 280290 220390

25 F O 25.5 1.50 20.9 -- 151090 141190

500

z 400
c
,300
O
3 200
O 6 Circular
0 2 Square
1O0 A 13 Triangular

o
O 2 4 6 8 10 12 14 16
Displacernent, i n rnrn Displocement, in mrn

Z
P l L E No CHAPE
c 300
.- O 12 Circuior
u" o 7 Square
,oO 2 0 0 + 4 Triangular

1 o0

o
O 2 4 6 8 10 12 14
Displocement, in rnrn Displacernent,in rnrn

Fig. 6 Penetration t e s t s
 Behaviour of floating piles 23

In Fig. 6, we observe that piles exhibit 4.3 Special tests

peak and residual load capacity (Pu and Pr,
respectively). The ratio Pr/Pu has an average Square pile No 1 was driven in severa1
value of 0.69 in al1 cases. The delayed stages. It was pushed down to a certain
displacements were negligible at low loads depth, then left at rest two weeks, and
and became noticeable, though small, close to tested in penetration; afterwards the element
the maximum. The displacements in the figures was pushed down deeper, left to consolidate,
correspond to the readings taken prior to and so on. The P-6 curves determined at
applying the next load increment. Once the different depths are displayed in Fig. 8. Al1
peak load was reached a sudden displacement curves show a maximum and a residual value,
ocurred, and then the load became stable at a except for the one corresponding to the
residual value. A nearly linear P-6' surface. From a depth of 41 cm to 71 cm, the
relationship can be observed up to a value peak capacity varied slightly. The magnitude
equal to 60 % of the maximum. Beyond this of the peak load at maximum depth was less
load nonlinear effects are noticeable. than half the maximum load exhibited by pile
The displacements at peak load (bF) reached No 2 (which was driven in one stroke). Thus
it can be concluded that the driving process
similar values, independently of the pile affects the load capacity of piles. Similar
cross section geometry and the diameter of effects were found by Kraft et al (1981), in
the prebored hole. Displacement 6F varied tests made with full scale piles driven in
between 0.95 and 2.05 mm (table 1) with a clays.
mean of 1.5 mm. Thus the ratio 6F/B (B, Pile No 16 consisted of a square tube
sleeve (2 cm wide) , through which slipped a
representative cross section dimension) had steel rod. At the lower end of the rod a
the following average values: Circular piles, point like those of the square concrete piles
0.064; hexagonal, 0.07; square, 0.075; and was screwed. Load increments were applied to
triangular, 0.050. the inner rod while holding the sleeve. Once
a test was performed at certain depth, the
casing was pushed down 2-9 mm to touch the
4.2 Pullout tests upper end of the point; then the rod was
fixed to the sleeve, both made to penetrate
Piles 1 to 9 and 13 were subjected to further to the desired deeper level, and so
extraction tests. After being tested in on. A two-week period was allowed between
penetration they were pushed down 2.5 cm into tests. In this way the point bearing capacity
the soil specimen; then they were left at was determined at different depths. The
rest severa1 months before performing the failure loads versus depth are depicted in
pullout tests. Fjg. 9. The P-6 curves showed a continuous
The relevant data and test results are displacement after reaching the failure load
given in table 2. The tension load (Q) versus with no loss in capacity.
displacement curves are shown in Fig. 7. From
the figure, it can be said that pile
responses in tension and penetration are 5 INTERPRETATION OF RESULTS
similar. In both types of tests, piles
exhibit peak and residual loads, and a 5.1 Generali t i es
linear load versus displacement relationship
up to 60 % of the maximum load. In tension In static conditions, similitude. between
the average ratio Qr/Qu is 0.76, which model (m) and prototype (p) requires (Harr,
exceeds the mean ratio in penetration. 1966) that
Table 2. Pullout tests.

testing
in which .a = Volumetric weight of soil; L =
Characteristic length; c = Soil cohesion; and
$ = Angle of interna1 friction of soil.
Equations 1 and 2 mean that to model a
static problem it is necessary to scale not
only the geometry of the problem but also the
soil strength and the volumetric weight.
Moreover, it is advisable to reproduce the
stress-strain behavior of the prototype soil.
In general, it is easy to scale the
geometry of the problem, taking into account
160790 laboratory facilities, budget, etc., so
usually this is the first scale selected. The
second choice is to model either the soil
strength or the volumetric weight of soil.
Displacements at a tension failure varied For soft clays it is possible to scale the
between 0.69 and 2.2 mm, with an average of cohesion by a maximum factor ranging from
1.28 nim. three to five. On the other hand, it is
24 Jaime, Rojas and Legorreta

Displacernent, i n rnrn
NO SHAPE

Squore

10 12 1l
-VI
J
'o

1 O0

o
O 2 4
O
6
,:[
8
Displacernent,in rnrn
,
P l L E NO

10
SHAPE
Triangular
Squore

12 14

Z
300
.-
'o
g 200
1 P l L E No CHAPE
o 7 Square
1 O0

O
O 2 4 6 8 1 0 1 2 1 4
Displa-ernent,in rnrn

Fig. 7 Pullout tests

Point Depth
(cm)
o o
+ 10
o 21
A 31
x 41
v 51

Displocernent, i n rnrn Driven depth/ pile width,(L/B)

Fig. 8 Pile driven in severa1 stages Fig. 9 Point bearing capacity at various
depths
extremely complex to model the volumetric
weight of clays. Thus, with the exception of behavior of real structures, the results
a few special cases, the laws o£ similitude obtained from models are adjusted to make
are difficult to accomplish in a 1-9 generalizations; in this way, implicitly, the
environment (g = acceleration o£ gravity). model is taken as a prototype.
For these reasons, many experiments in soil Pile tests reported in this paper were
mechanics are carried out on distorted models carried out using a distorted model. The
( e . , Terzaghi and Peck, 1967; Whitaker, geometry was scaled to 1/25 of the dimensions
1957; Sowers et al, 1961; Vesic, 1975). Some o £ the piles most frequently used in Mexico
times the results are taken as indicative of City. The clay specimen had a cohesion 5.5
the prototype response. At other times, based times lower than the average value of Mexico
upon full scale experiments and the observed City upper clay deposit; the volumetric
 Behaviour of floaling piles 25

weight was equal to that of the natural clay. wtill known general equation is
The clay sample was prepared to have a
stress-strain response similar to the real Qf=FsAl+FbAp (3)
soil response. in which Fs = Shaft resistance, Al = Lateral
area, Fb = End bearing capacity of pile,tip
and Ap = Pile tip area.
5.2 Full scale pile load tests For determining the shaft resistance in
clays from laboratory strength data, there
We will refer to the results of load tests are two general procedures. The first is
performed on full scale floating piles driven based on an effective stress approach and the
in Mexico City upper clay deposit. A complete other in a total stress one. The latter is
description of the tests is given in Jaime et commonly used in practice.
al (1988, 1989 and 1990). The concrete piles The method most frequently used for
had a 30 cm square cross shape, and an assesment of the end bearing resistance of
effective depth of embeddment of 10 m. piles in clay, using laboratory data, is
The P-6 curves of two tests, including the based almost invariably on a total stress
instantaneous and delayed displacements, approach (Poulos, 1989). Thus in what
carried out in penetration and extraction are follows, a total stress approach is adopted
shown in Fig. 10. These curves were corrected for both shaft and point resistance.
by substracting the structural shortening and The shaft and point capacities are givpn by
elongation of the pile. It can be observed
that in both cases, the piles exhibit peak Fs = a c)avs (4)
and residual capacity. The maximum loads are
attained at vertical displacements of 18 mm and
6001
Penetration Test
where c)avs = average undrained strength of
the soil surrounding the shaft, c)ub =
undrained strength of soil in che vicinity of
pile tip, a = Tomlinson's adhesion factor and
Nc = bearing capacity factor (many authors
suggest a value of 9; i.e. Poulos, 1989).
Tomlinson (1957, 1977) proposed the
adhesion factor a, which depends partly on
the cohesive strength of the soil and partly
on the nature of the, soil above the bearing
stratum of clay into which the piles are
driven. His early studies showed a general
trend towards a reduction in the adhesion
Displacement, in mm factor from unity or higher than unity for
very soft clays, to values as low as 0.2 for
Fig. 10 Full scale pile tests. Penetration clays having a very stiff consistency.
and extraction However, there is a wide scatter in a over
the full range of soil consistency. According
to Tomlinson the scattering seems to be
and 25 mm in penetration and extraction, unrelated to the material forming the pile;
respectively. After reaching the maximum load he does not discuss this point further.
a sudden displacement ocurred, and then the
load became stable at a resjdual value. This
effect was also observed in the model pile 5.4 Cross shape effect on shaft capacity
tests. Below a load of 250 kN the pullout
curve is stiffer than the penetration one; Piles 2, 6 , 13, 24 and 25 were driven without
beyond this load ,the reverse is true. preboring. They had different cross section
The response of the model piles is shapes, as given in Fig 6 and table 3. We
qualitatively similar to that of the full conclude from the figure that the circular
scale piles (Figs. 6 and 10). Both exhibit pile attains the largest peak load, followe?
peak and residual ca~acity;eventhough, the by the hexagonal, square, triangular- and
loss in capacity is less in the real piles plane, in this order. The residual load seems
(Pr/Pu = 0.89) . However ;'~arsal and Mazari to follow the same pattern, at 4 mm of
displacement.
(1959) reported some load tests on piles
driven in Mexico City clay in which, at very Similar results were obtained for piles 3
large displacements (15 cm or more), the loss to 5, 7 to 12 and 15, driven in prebored
in resistance is similar to that of the model holes of various diameters. Table 3
piles. The ratios aF/B of the real and model summarizes the relevant data. The P versus 6
curves are shown in Fig. 6. According to the
piles are almost equal, around 0.07. table and the figure the ratios Pu/Puo (Puo
peak load reached by the circular pile)
corresponding to the triangula-- and square
5.3 Bearing capacity of piles in clay piles are 0.66 and 0.8; except for the piles
driven in holes with diameters close to the
There are different methods to estimate the main cross dimension of the piles. In this
bearing capacity of piles (Poulos, 1989). The case the ratio dimini-hes to 0.55, indicating
26 Jaime, Rojas and Legorreta

Table 3. Influence of the pile cross section geometry on the

bearing capacity. Penetration tests.

Pile Shape Pu 6F Pr Pr/Pu Pu/Puo Pr/Pro aF/B

No k9 ulm k9

WITHOUT PREBORING
6 R 49.6 1.45 32.3 0.65 1.00 1.00 0.06
24 H 43.2 1.81 32.4 0.75 0.87 1.00
2 C 39.2 2.00 27.4 0.70 0.79 0.85 0.10
13 T,, 33.2 2.05 23.0 0.70 0.66 0.71 0.08
25 P 25.5 1.50 20.9 0.82 0.51 0.65 0.02

3 C
8 R 34.1
34.4 1
1.59 22.7 2 6
0.67
6 1.0
1.00 1.0
1.00 0.06
0.08
11 T 25.0 0.95 15.0 0.60 0.73 0.66 0.04

5 C
10 R 26.9
33.8 1 : ~ ~ 123.0
1.45 = 1 ~ ~:
0.68
y~67[ 0.80
1.00 0.78
1.00 0.06
0.07
15 T 26.0 1.27 18.2 0.70 0.77 0.79 0.05
Dpp = 21.6 mm
7 C 18.1 1.04 12.1 0.67 0.55 0.53 0.05
12 R 32.7 2.05 23.0 0.70 1.00 1.00 0.08
4 T 18.0 1.48 11.9 0.66 0.55 0.52 0.06

:. Lateral
Full scale pile penetration test
area corrected

that the preboring effect is more important

1.2 -
I
than the cross shape effect. The ratios
1.0 -
between the residual loads of the triangular
L
and square piles and that of the circular o .
pile exhibit a similar variation.
Chellis (1962) mentions that "the unit
friction value appears to be larger for a
round pile than for a square pile whose side o .
equals the round pile diameter, in the ratio m 0.4 -
L
of about 4 : 3". Based on this, for equal F f = 0.63t0.62 (BETA/180°)
lateral areas, a square pile would have a 0.2 -
shaft capacity of 60 % the capacity of a
round pile. It is pointed out that Chellis O I N 1 1 - 1
made this general statement regardless of the O 0.2 0.4 0.6 0.8 1.0
type of soil and pile material. The authors
did not find onv other reference discussina B E T A / 180"
the effect of the cross section shape on the
friction capacity of piles in clay (we Fig. 11 Cross section shape effect on pile
consulted Little, 1961; Terzaghi and Peck, capacity
1967; Peck et al, 1974; Sowers and Sowers,
1970; Zeevaert, 1973; Tomlinson, 1977; Vesic, in which Pf)m = Pu - Pp (from the tests
1975; Poulos and Davis, 1980; Poulos, 1989). results) and Pf)c is obtained\ from E q . 4
Based on the results of .the piles driven multiplied by the lateral area, considering
without preboring, it is possible to a = 1. Pp is the pile tip capacbity measured
establish a relationship between the shaft with the square pile 16, corrected (when
capacity of the pile and the interior angle needed) to take into account the base area of
formed by the sides of the pile f3, Fig. 11. the corresponding pile. The relationship
As the vertex is more acute the angle f3 tends between the shape factor and the angle f3 is
to zero (plane pile); at the other extreme, shown in Fig. 11. It'can be expressed as:
as the pile has more sides the Sross shape
tends to the circular and f3 to 180 .
To express this relationship in
dimensionless form, a shape factor, Ff, is where f3 is in degrees. Tomlinson (1977)
defined as the ratio of the measured shaft defined factor a in a similar way but did not
capacity and the calculated capacity: give a satisfactory answer to the wide
scattering observed in his results. This
scattering can be attributed to the cross
 Behaviour of floating piles 27

chape effect. The shape factor defined in Eq. diminishes. Something similar is observed for
7 is valid for soft clays. The authors the square and triangular piles. In Fig. 12
believe that a similar trend will be obtained we see that as the diameter of the prebored
for medium to stiff clays, but this is a hole increases the peak load attains values
point to be investigated. as low as 50 % of the maximum (without
The authors have not found a theory to preboring) . The residual load follows a
explain the effect of the cross shape on the similar pattern.
response of piles. However, the circular Results of Fig. 12 can be expressed in
cross shape induces a uniform stress state in dimensionless forrn as shown in Fig. 13. Here
the coi1 as it can be seen from the theory of the ratio between the peak load (Pu) and the
expansion of cavities (Vesic, 1972). On the maximum load reached without preboring (Pus)
other hand, the vertices of the hexagonal, is plotted versus the ratio of the nominal
square and triangular cross shapes are prebored hole area and the pile cross section
singular points (or lines, if the whole shaft area. It is plain that the preboring effect
is considered) on which there are stress is independent of the pile cross shape.
concentrations that cause an irregular stress The more important conclusion to be drawn
state in the coi1 surrounding the pile and from Fig. 13 is that pile bearing capacity is
favor the formation of cracks. The more acute a function of the volcme of soil displaced
the vertex the more pronounced the effect during pile installation. This has a strong
(Fig. 11 and Eq. 7). influence on the size of the zone of remolded
coi1 surrounding the pile, as well as in the
magnitude of the stresses to which the clay
5.5 Preboring effect on shaft capacity is subj ected during the subsequent
reconsolidation. The larger the volume of
The curves P versus 6 of circular piles soii displaced the greater the bearing
driven without preboring and in predrilled capacity.
holes of various diameters are compared in
Fig. 12. Notice that preboring has a strong
effect on the pile response. It causes a 6 CONCLUSIONS
sustantial loss in load capacity, 31 % of the
maximum. Notice that the residual load From the results of the model pile tests the
followinq conclusions can be drawn.

500
CIRCULAR P I L E S
400

O 2 4 6 8 10 12 14
Displacement, in m m Displacement, in mm

Displocement, in m m

Fig. 12 Effect of predrilled hole diameter

on pile capacity
28 Jaime. Rojas and Legorrela

REFERENCES
Chellis, R. D. (1962). "Pile foundations.'
Chap. 7, Foundation Engineering, editor G.
Leonards, McGraw Hill, New York, N. Y.
Harr, M.(1966). "Foundations of Theoretical
Soil Mechanics." McGraw Hill, New York,
N.Y.
Cross section
Square Jaime, A., Romo, M. P. and Reséndiz, D.
(1988). "Comporfamiento de pilotes de

O
O Circular
" "
0.2
'
0.4
~ I
0.6
'
0.8
I
1
I '
1.2
'
1.4
1 fricción en arcilla del valle de México."
Series of the Instituto de Ingenieria,
UNAM, 515, Nov. (in Spanish).
A r e o ratio, A p p / A t p Jaime, et al.(1989)."Static tests on friction
piles in Mexico City ~ l a y .Proc.,
~ 12th
Fig. 13 Preboring effect on pile capacity Int. Conf. on Soil Mech. and Found. Engrg.,
Rio de Janeiro, Brasil, 1141-1146.

Jaime, A., Romo, M. P. and Reséndiz, D.

1. From the P-6 curves in penetration, a peak (1990) l9Behavior of friction piles in
load and residual load can be determined, Mexico City clay." J. Geotech. Engrg.,
the latter being about 67 % of the former. ASCE, 116 (6), 915-931.
Jaime et al (1990) in full scale piles driven
in Mexico City clay reported residual loads
of 85 % of the peak value. However, Marsal Jaime, A. (1990)."Comportamiento de pilotes
and Mazari (1959) determined, from tests in de friccion en arcilla blanda.## Proc.,
woodpiles, residual loads as low as 50 % of Symposium on El subsuelo de la cuenca del
the maximum at large displacements (10 cm). valle de Mexico y su relacion con la
Ingenieria de Cimentaciones, a cinco años
2. The cross section shape has a strong del sismo. Soc. Mexicana de Mec. de Suelos,
influence on the bearing capacity of piles. Mexico City, Mexico, 197-222. (in Spanish) .
The circular cross shape is the more
efficient, followed by the hexagonal, the Jaime, A., Rojas, E. and Legorreta, E.
square, the triangular and the plane. An (1991).1'Pruebas de carga ciclica en modelos
almost linear relationship was found between de pilotes de friccion. Primer informe."
the interna1 angle, (3, formed by the sides of Report of Project No 9526, Instituto de
the pile and the bearing capacity (Fig. 11, Ingenieria, UNAM, Jan. (in Spanish).
Eq. 7).
Kraft, L. M., Focht, J. A . and Amerasinghe,
3. Preboring reduces the ultimate capacity of S. F. (1981) . "Friction capacity of piles
friction piles. This effect is a function of driven into clayn. J. Geotech. Engrg.,
the nominal predrilled hole diameter: the ASCE, 107 (ll), 1521-1541.
larger the diameter the lesser the capacity
(Fig. 13). Little, A. L. (1961). "Foundations." E.
Arnold (Publishers) Ltd. London.
4. The driving process also affects the
bearing capacity of piles. The bearing Marsal, R. J. and Mazari, M. (1959). "El
capacity of a pile driven in one stroke is subsuelo de la ciudad de Mé~ico.~' 2nd ed.,
greater than the capacity of a pile driven in Universidad Nacional Autonoma de Mexico,
severa1 stages, provided that enough time is Mexico City, Mexico. (in Spanish).
allowed for the soil to consolidate in each
stage. A similar result was reported by Kraft Peck, R. B., Hanson, W. E. and Thornburn, T.
et al (1981). Although, this point requires H. (1974). "Foundation engineering." 2nd
further investigation. ed., Wiley, New York, N. Y.

Poulos, H. G. and Davis, E. H. (1980). "Pile

7 ACKNOWLEDGEMENTS foundation analysis and design " John Wiley
and Sons, New York, N. Y.
This pro ject was sponsored by Departamento
del Distrito Federal and Consejo Nacional de Poulos, H. G. (1989). "Pile behaviour-theory
Ciencia y Tecnologia. and application." Geotechnique, 39 (3),
The late Professor Raul J. Marsal advised 365-415.
the first author to continue his research on
the behavior of friction piles in laboratory Sowers, G. B. and Sowers, G. F. (1970).
models, and made very valuable suggestions. 181ntroductory coi1 mechanics and
E. Rosenblueth and J. Alberro critically fo~ndationq.~'3rd ed., Macmillan, New York,
reviewed the manuscript. N. Y.
 Behaviour of floating piles 29

Sowers, G. et al. (1961). "The bearing Vesic, A. S. (1972).'Expansion of cavities in

capacity of friction pile groups in infinite soil mass." J. Soil Mech. and Fdn.
homogeneous clay from model methodsu. Engrg., ASCE., 98.(3), 265-290.
Proc., 5th Int. Conf. on Soil Mech. and
Found. Engrg., 2, 155-159.
Vesic, A. S. (1975). "Principies of pile
~erzaghi, K. and Peck,. R. (1967) llSoil foundation design." Soil Mechanics Series
mechanics in engineering practice." 2nd No 38, School of Engrg. Duke Univ., Durham,
ed., John Wiley and Sons, New York, N.Y. N. C.
~omlinson, M. J. (1957). "The adhesion of
piles driven in clay ~oils.*~ Proc., 5th Whitaker, T. (1957)."Experiments with model
Int. Conf. on Soil Mech. and Found. Engrg., piles in groups." Geotechnique, 7, 147-167.
2, London, 66-71.
Tomlinson, M. J. (1977). "Pile design and Zeevaert, L. (1973). "Foundation engineering
construction practice. A Viewpoint for difficult subsoil conditions. Van
Publication, Great Britain. Nostrand Reinhold, Co., New York, N.Y.
 Natural laws in geomechanics
Leyes naturales en geomecánica

E. Juárez Badillo, Graduate school,of

Engineering, UNAM

ABSTRACT: A single general equation based on RESUMEN: Una ecuación general única basada en
the principle of natural proportionality is el principio de proporcionalidad natural es
used to obtain general equations to describe usada para obtener ecuaciones generales que
the mechanical behaviour of geomaterials. describan el comportamiento mecánico de los
This paper contains a different presentation geomateriales. Este artículo contiene una
of stress-strain-time-temperature general presentación diferente de ecuaciones
equations for geomaterials already presented generales esfuerzo-deformación-tiempo-tempe-
elsewhere. ratura para geomateriales ya presentadas en
otros artlculos.

1. INTRODUCTION similar to those of Eqs. (3) and (4) can be

used to describe the behaviour of
As it is well known, the law of Boyle and geomaterials for al1 values of stresses,
Mariotte for an ideal gas is strains, temperature T and time t (which is
also considered) with a small number of
constant parameters, as long as the material
remains on the same mechanical phase.
where V = volume, o = pressure, T = absolute Graphs, experimental support and the
temperature, and C = a constant. simplest way for obtaining the constant
parameters involved in the general equations
In differential form, Eq. (1) is that will be presented, are contained in the
references included in this paper.

2. GENERAL EQUATIONS
Assuming T to be constant Eq. (1) reduces to
2.1 Cornpressibility
vo = C (3)
A first generalization of Eqs. (3) and (4)
and Eq. (2) reduces to iS

In Eqs. (3) and (4) V and o have complete

domains, (from O to m) . Moreover, when o
dV is proportional
increases V decreases and -v
to G, that is, "natural ~roportionality" which describe the compressibility and
expansibility of clays and of water. For
prevails. practical use, Eq. (6) may be written
(Juarez-Badillo, 1975)
Eq. (3) may be written, for practical use,
as

Using Bridgman's experimental data, Grimer

F.J. and Hewitt R.E. (1968) found the
where (ol, VI) is a known point in o-V space. following expression for the relation between
In this paper, it will be shown that pressure and volume for water at 60' C for
general equations with characteristics externa1 pressures up to 12,000 atmos~heres:
 starts. When a generalized crushing of
particles occurs (second mechanical phase)
where p = o + i , o = externa1 pressure, i = the compressibility is described by Eq. (8)
interna1 pressure = 3750 ,atmospheres. Eq. which is a special case of Eq. (16) with Vo =
(9) may be written in the form of Eq. (8) m. .For the materials mentioned above, a 5 1
(Juárez-Badillo, 1985b) as: in Eq. (16). For a sample of porous solid
concrete 7 = 2 (Juárez-Badillo, 1985b). In
the second mechanical phase a < 1. Eq. (8)
also describes the volume changes in drained
triaxial tests of overconsolidated samples of
clay if o is interpreted as "equivalent
and, therefore, for water, the compressi- consolidation" and "stored" pressures
bility coefficient a = 1 (Juárez-Badillo, 1975) and (Juárez-Badillo y
Rico-Rodríguez, 1975).
For Mexico City clay the compressibility 7
Geomaterials like solids and granular soils varies from 0.20 to 0.60 with a typical value
have a finite volume Vo at u = O. For these of 0.50 in Eq. (8). The expansibility is an
cases a second generalization of Eqs. (6) and = 0.087. That is, the expansibility-
(7) (Juárez-Badillo, 1981) is compressibility ratio is p = 0.08 (Juárez-
Badillo, 1975).
Proceeding in a similar way, many other
equations describing physical phenomena have
if we write already been established. Once the proper
variables are found, the general equations
are easily obtained if natural
proportionality is assumed.

then Eq. (11) may be written as 2.2 P e r r n e a b i l i ty

The well known Darcyls law is: v = ki where

v = discharge velocity, i = hydraulic
and Eq. (7) as gradient and k = coefficient of permeability.
Here k is a function of the volume V of the
sample. Both k and V have complete domains.
If they both are assumed to be proper
variables, then the natural law relating them
should be, keeping T constant,
The variable Z given by Eq. (12) is the
simplest function of V with a complete
domain. Z is the "proper function" of the
"proper variable" V. The proper function of
the proper variable u is u itself. A proper where K = permeachange coefficient.
variable to measure a category is the : For practica1 use, Eq. (17) may be written
simplest variable that defines that category. (Juárez-Badillo, 1983)
For volumes it is just the volume V, not
volumetric strain. Eq. (14) expresses the
natural proportionality between the proper
functions Z and u. For practica1 use, Eq.
(11) may be written (Juárez-Badillo, 1981)
where e = void ratio with respect to the
initial volume of solids.
It should be observed that the author has
found that the void ratio e is not a proper
variable to describe compressibility or
Eq. (15) may be written in a simple way changes in the permeability.
choosing as the known point the Eq. (18) is applicable to the vertical
"characteristic pressure" u* for which VI = permeability kv (with K") as well, as to the
vo
- . Then Eq. (15) reduces to horizontal permeability kh (with K ~ )of a
sample of soil (sand or clay) when it is
compressed one-dimensionally. The
relationship K~ > K is always obtained. At a
high void ratio e = e' it is obtained kh =
kV. The anisotropy of the soil increases as
The compressibility of an accumulation of the volume decreases. The equation results
solid coarse particles, like rockfills, to be (Juárez-Badillo, 1991a)
gravels and sands is described by Eq. (16),
for a first mechanical phase, from u = O to a
critica1 o = uC where crushing of particles
 Rockfill for embankment dams.
Enrocamiento para presas de tierra.

A. D. Petlmafl, Geotechnical engineering

consultant, Sladeleye, U.K.

ABSTRACT: Due to a complete misunderstanding of the RESUMEN: Las presas de enrocamiento con pantalla
behaviour of rockfill, upstream membrane rockfill dams impermeable en el talud aguas amba tuvieron una mala
had a bad reputation. ~eflectionson the first reservoir reputación debido a deficiencias en el conocimiento sobre
filling often caused sufficient damage to a concrete el comportamiento de enrocamientos. Las deflexiones de las
membrane to require repair. It was thought that dumped pantallas durante el primer llenado provocaron daños que
rockfill should consist only of large pieces of the hardest, en algunas ocasiones fue necesario reparar. Se pensaba que
unweathered rock, taken from the heart of the quarry, los enrocamientos colocados a volteo deberían consistir
without any smaller fragments which could prevent good solamente de fragmentos grandes de roca dura no
rock-to-rock contact and so cause settlements. Terzaghi intemperizada tomados de los corazones de los bancos de
showed that points of contact of large pieces of even the préstamo, sin fragmentos pequeños que pudieran impedir
hardest rock suffer considerable crushing, which is contactos roca-roca adecuados, y así provocar
increased by wetting. It is better to bed large pieces of rock asentamientos. Terzaghi demostró que los puntos de
in graded finer fragments to minimise contact stresses and contacto aún en las rocas más duras sufren aplastamientos
prevent rotation of large pieces. The use of this type of considerables cuya magnitud aumenta con el
rockfill, compacted at optimum water content by heavy humedecimiento. Es preferible asentar los fragmentos
vibrating rollers has led to the current success of the grandes de roca sobre materiales graduados de menor
'-0ncrete Face Rockfill Dam. tamaño para minimizar los esfuerzos de contacto y prevenir
la rotación de partículas grandes. El uso de este tipo de
enrocamiento compactado a su humedad óptima con rodillos
vibratorios pesados condujo al éxito que ahora tienen las
presas de enrocamiento con pantalla de concreto aguas
amba..

1. INTRODUCTION. The aim of this chapter is to consider rockfill as a constructional

material, to describe some of its history and the developments that
Much of the surface of the earth where man chooses to live is have taken place to enable it to be used satisfactonly to form the
covered with fine-grained soils and these do not álways form a world's highest embankments.
desirable fill. A distinction is made by the terms "eanhfill" and
"rockíill", although many matenals lie in a border area between the
tWO.
Developments in the art and science of soil mechanics have 2. BACKGROUND.
given us design methods enabling almost any fill to be used
successfully and the choice is usually dependent on cost. Wet, Stone is a traditional building material. It has been used for the
clayey fill can be improved with drainage layers but it may be world's most prestigious buildings: cathedrals and churches;
cheaper to impon more granular fdl and not use the clay at all. government buildings and townhalls; palaces and prisons.
Three aspects have to be considered. One concerns the Quarrying techniques became well-established and some of the
problems of winning the fill, another of placing it and the third earliest of modem scientific research was directed at determining
considers the subsequent behaviour of the fill. Settlement of a road the properties of building stone. One of the first dams built by man
embankment may be of little consequence if the bridge approaches used stone in its construction. The oldest dam in the world
allow for it, but in an embankment forming a water-retaining dam, (Kensel 1985, quoting Helms), dating from around 4000 BC, was
settlements and deformations may be of more senous concem. built of earth with a masonry facing. The Saad el-Kafara darn in
A great deal has been written about the problems encountered in Egypt, constructed about 5000 years ago (Smith 1972) consisted
the use of fine grained soils such as silty clays as fdls, especially essentially of two mbble-masonry walls retaining an infill of earth
when their natural water contents are greater than the optimum between them to form the waterproof element. The dam was 1lm
values best suited for compaction, but relatively little has been Iiigh and the walls were 24m wide at the base, separated by 36m
written about rockfill, which tends to be taken for granted. width of earth fill, as indicated by Fig.1.
304 Penman

complete, there was an exceptional rainfall of 382mm during a 24

hour period, 31 December to 1 January 1934. The crest settled
fCourses o f l i m e s t o n e b l o c k s approx 0.3m 1.8m immediately and continued, to reach 4.lm. To minimise
\ post-construction settlement, extensive sluicing was used ( t h e ~
was now plenty of water) over a period of severai months with the
Sand and grave1 fill intention of washing out fines. This had the effect of increasing
the senlement to 5.3m.
This rockfill "modei" satisfied engineers, ind dams continued to
be built of only the most competent rock, screening off fines w h e ~
F i g 1. Sadd e l - K a f a r a : Dam o f t h e Pagans. necessary. Volumes of sluicing water specified varied from 4 to
about 0.2 of rock volume.

In a detailed history of dam developments, Jansen (1980)

describes many ancient dams in the Middle East, India and China. 3.2 Placemenr from railways.
The Romans built dams throughout their empire, and later, the
Spaniards built hundreds of masonry dums in Mexico. In North In the 1920s and 30s, prior to the advent of heavy earthmoving
America, dams were built to provide power for mills and a dam machinery capable of transporting rockfill by road, railways were
was constructed on the Piscataqua nver in 1623 to supply water for used to bring material from the quarries. To avoid steep gradients,
the first sawmill in America. quarries were opened at three or four levels corresponding to
Contemporary with tne development of masonry dams in quarter or third heights of the required embankment.
Mexico, the Jesuits introduced dam building to California. The The lOOm high Salt Spnngs dam remained the world's highest
Old Mission dam on the San Diego river was built with mortared concrete-faced rockfill dam for 26 years, up to 1958. It was
rubble masonry in 1770 to supply water for one of their first constructed from bridges across the valley to carry a railway and
missions. Discovery of gold in California in 1848 led to the the rockfill \vas discharged from 23m3 side-dump and 15m3 end-
construction of many dams to supply water for the mining dump railway trucks. The greatest drop, into the valley bottom,
operation. Dams were created with stone-filled log cribs. These was 52m, and successive lifts were 23.20, and finally 5.5m high.
soon developed into dumped rockfill dams, made waterproof with
two or more layers of wood planking laid on the upstream face.
Rockfill conshuction rose to a position of major irnportance with
the completion in 1924 of the 84m high Dix River dam for water
supply to Danville in Kentucky. This record height was exceeded
in 1931 by the lOOm high Salt Spnngs rockfill dam. A section of
this dam is shown by Fig. 2. 52n Lrr

,o. .c ,,.,.s
~ ~

u-.-u.u L,,TS IS)~ 1097

3. DUMPED ROCKFILL.
PROrLC ON AX15 (LDO6lNC DOWNSTRCAMI - LIFTS ANO CONSTRKTY)*I SCOJCNCE

Timber soon gave way to reinforced concrete as the material for i r LL I Z U ~ . S ~ U DIcIULD
I I.~..
the water-proof upstream membranes of the rockfill dams, but -7- <.A-

deformations of the dumped rockfill became a major problem.

Under the loaa imposed by the reservoir water as its leve1 began to
rise, the membrane deflected and often cracked or joints opened to
allow leakage into the rockfill. This immediately produced
increased deformation, making the leaks worse.
It was thought at the time that the deformations were caused by a
combination of weak rock, allowing crushing at the points of
contact and the presence of fines which might lodge at [he contacts,
holding the pieces of rockfili apart. It was thought that the effect F l g 2 . Sectlon o F 100m h i g h Salt Spr:igs rnckF111 dam.
of leakage water was to wash fines away from the contacts and so
allow extra deformation of the whole body of the rockfill. In a
rnisguided attempt to reduce deformations. specifications called for
the use of only strong, competent rock from the heart of the quarry 3.3. Sluicing volumes.
and the exclusion of al1 fines. Often large quantities of weathered
rock were led to waste as a quarry was being developed and it was The problems of sluicing granite rockfill has been discussed by
not unusual to pass al1 the best rockfill for the dam over a grizzly Steele and Cooke (1960). who criticised the small amount of water
(screen) with bars so spaced that any material of less than 150mm used at Salt Spnngs during the early stages. Dunng the penod
size would pass through to waste. February 1928 to July 1929 when 575,000m3 of rockfill were
dumped, there was only a 7.5 litrelsec. pump, with a maximum
head of 100m, in use. While the dumping rate was ahout
3.1 Sluicing. 60,000m3 per month, the water supplieci was less than 15,000m3
per month and the pipework distribution such that it was not
High pressure water jets were often used on the rockfill as it was available at every dumping point. A 23 litrelsec pump was added
dumped, with the intention of washing out any remaining fines in July and in October 1929, with 84,000m3 of rockfill in place
from the points of contact. During the construction of the 85m further 90 and 115 litrelsec pumps were brought into use. Thus the
high Cogswell dam, however, there was a shortage of water :o that first 1,150,000m3 of rockfill placed before January 1930 could be
sluicing was not used. It was built from sound granite rockfill and considered to have been only nominally sluiced. The last
consauction began in March 1932. When the dam was 80% 1,450,000m3 had substantial volumes of water, but most of it was
 Rockfill for embankment dams 305

added simply by letting hoses pour on to the tops of the lifts - no In a sezond test, sand done was used, but it was only washed in to
high pressure jets were used to wash out any pockets of dirt and a depth of two rocks by the jet whose kinetic energy had mostly
spalls. dissipated in the fust row of rocks. Below the surface, the water
During construction, a survey point 52m above formation level flowed by gravity through the voids, unable to perform any greater
in the valley floor, settled 0.573m. On first filling of the reservoir, mechanical action than heavy rainfall.
the concrete face slab was pushed down a maximum of 1.317m It was the opinion of Terzaghi that the beneficia] effects of
normal to the face, close to the point of maximum water pressure, sluicing on the subsequent settlement of rockfill were due, not to
i.e. one third of the height of the dam. This caused some cracking any removal of fines from the zones of contact of the large rock
of the concrete slabs and damage to joints, but repairs were pieces, but to the reduction of the strength of rock on wetting. He
possible during subsequent reservoir operation when the water quoted values for the strengths of rock when dry and saturated,
level was low. published by Herschwald (1912) and McHenry (1945) given in
Kaw-ase (1960), in describing construction of the 53m high Table 1.
Ishibuchi rockfill dam in Japan completed in 1953, was also
critical of the sluicing volumes used. Water at a pressure of 685 Table 1: Compressive strengths of rock,
kN/m2 was applied through jets at a rate of twice the volume of net and dry Nlmml
placed rockfill, but it was considered to be too little. As at Salt
Springs, rockfill was dumped from railway trucks from a bridge Source
across the valley. Maximum height of drop was reshicted to 29m.
During construction, a survey point on the surface of the fust lift Sound Austrian granite
(29m above formation) settled 0.44m. On first filling, maximum Granites from Sweden and Germany 240 av. 197 to 237
settlement of the concrete face slab was 0.4m at about third height. Crystalline limestone
A dumped rockfill dam 77m high was built only a few miles Schist from Temessee 96 45
from Salt Sprfngs, across the Lower Bear River, in 1951 and 1952.
Granite rockfill was hauled from the quarry by road and end-
dumped in lifts up to 65m high. Selected loads of larger rocks of 2 4. LABORATORY EXPERIMENTS TO STUDY BEHAVIOUR.
to 15 tonnes size were dumped on the upstream slope which
developed a natural angle of repose in the range of 1 on 1.21 to 1 4.1 Fricrion.
on 1.34. As with the other rockfill dams, this slope was covered
with a layer about 5m thick of placed rock. Carefully chosen, In order to measure the coefficient of sliding friction of quartz on
rectangularly-shaped rocks were handled by crawler crane using quartz, Penman (1953) used two large fragments of quartz crystals
slings operated by a team of six men. They wqre built up, like dry which had been cut and polished. They were bedded in plaster in
stone walling, to forma relatively smooth surface to take the cast the two halves of constant-rate-of-strain shear box so that the two
in situ concrete slabs. On average, 29m3 were placed by each surfaces lay on the line of shear of the split box, as indicated by
crane team during an 8 hour shift. In order to let this time- Fig.3.
consuming work get ahead, the upstream part of the dam was No special surface preparation was used: the quartz surfaces
dumped first, to a height of 25m. were washed with soap and water to remove grease, and finally
Sluicing of the main rockfill of this Lower Bear River dam used rinsed with distilled water. The friction tests were made with the
three 320 litreslsec pumps which could produce a pressure of 820 surfaces submerged to ensure they were wet and that no capillary
kN/m2. Distribution was through 0.3m diameter pipes, leading to tensions developed. The area of the upper quartz surface was
0.15 or 0.2m diameter heavy rubber hoses and jets with 63 or about 775 mm2 and its weight, plus that of the upper half of the
75mm diameter nozzles. Each discharged 60 to--420 11s: four of box was 13.17N so that the minimum normal shess that could be
them gave a volume of sluicing water 3 times that of the rockfill used was 17 kN/m2. Fig.4 shows the pieces of quartz mounted in
being placed. Measured súrface settlements during construction the two halves of a shear box. Load was increased in increments
were small and depression normal to the concrete face after first up to a maximum of 673N (stress 870kN/m2) and the plotted
filling had a maximum value of 0.625m at about one third of the results (Fig.5) showed a good straight line relationship giving a
height. constant angle of friction of 33O.

3.4.Effects of sluicing.

Terzaghi (1960) contributed to the discussion on the paper by

Steele and Cooke (1960), emphasising the effects of sluicing.
Observation of rockfill being dumpeo had shown that as a batch
was tipped over the edge of the advancing fill, segregation takes
place whereby the average particle size and degree of uniformity
increases with distance down the slope. The finest material comes
to rest on the upper part of the slope and the mechanical action of
the sluicing water jets was found to be limited to the uppermost
10m of the slope. Within this distance, the impact of the jet pushes
the smaller particles into the interstices between the larger pieces,
but the effectiveness of this process is limited to a thin layer.
To check on the penehation of fines into the rockfill, Terzaghi
carried out a field hial with a rockfill consisting of pieces of up to
0.75m size. A mixture of sand and gravel was dumped on the
rockfill and sluiced with a jet of water at close range. The surface
layer one rock thick soon choked with the gravel and no significant
amount of fines could be jetted into the rockfill behind this layer. F i g 3 . S h e a r box c o n t a i n i n g ernhedded q u a r t z p i e c e s .
306 Penrnan

The test was then repeated with dry surfaces. The two nalves of added to the apparatus until rne polished surface was well below
the box, containing the embedded quartz pieces, were dned at the water leve1 and the pre-wetted upper half of the box was
1050C for 48 hours and the fnctio-n test made while the surfaces lowered to bnng the three sharp points on to the polished surface.
were still w m , to prevent condensation. Again, a straight-line The weight of the upper part of the box was 18.24N and because
relationship resulted. showing a constant angle of friction of 1l o the points had been spaced equidistant frorn the cenue of the box,
over the wfiile range of normal loads used. it can be assumed that the load on each point was 6.08N. At this
minimum normal load, the angle of friction was 290 (see Fig.5).
As the normal load was increased, the angle of fnction decreased
and at the maximum normal load of 645N. i.e. 215N per contact, it
was 190. Crushing could be heard with al1 loads greater than 148N
per contact, and after the tests, the three points were found to be
badly darnaged and the lower surface had been scratched.
These experirnents demonstrated that the angle of fnction of
quartz on quartz under saturated conditions decreased frorn 330
under low normal stress, to lgOwhen the normal stress was great
enough to darnage the contact surfaces. They also showed that the
coefficient of friction was rnuch lower when the surfaces in contact
were cornpletely dry. Under very low normal stress (polished
surface on polished surface) the angle of friction was 1 lo; at larger
normal stresses (grains on polished surface) the value fe11 to 70.
These results are contrary to the popular belief that water acts as a
Fig 4 . Pieces o f q u a r t z mounted ir- t h e t w o h s l v r s o f
lubncant. In fact it can be said that the water behaves as an anti-
t h e sliear b ~ r .
lubricant.

4.2 Reason for large se;tlements.

The observations discussed by Terzaghi (1960) also indicated that

wetting unpolished rock surfaces did not reduce the coefficient of
friction. H e therefore concluded that the observed large
settlements of dumped rockfill on wetting were not due to
lubricating effects but rnust be caused by the reduction of strength
of rock on wetting. The above expenments show that far from
reducing the coefficient of fnction, the effect of saturating the
contact surfaces was to increase it considerahly frorn the values
rneasured with dry surfaces. This gives confirmation to Terzaghi's
proposal that dumped rockfill settlements were caused by
weakening at the points of contact and not due to reduced sliding
resistance between particles.
The above expenments show, however, that the coefficient of
friction is reduced by increasing normal stress, panicularly when
Normal load, N this stress is great enough to cause crushing at the points of
contact.
Fig 5 Plot of friction force v. normal load. Quartz
4.3 Deformarions at point contacts.
Tschebotarioff and Welch (1948) had previously camed out
similar tests in a stress-controlled shear box. They used a piece of Tests to measure deformations occumng at a point contact of rock
quartz with a polished surface mounted in one half of the box, were reported by Rzadkowski and Zurek (1970). The tests were
beanng on a layer of quartz grains, bedded in plaster in the other made on angular fragments of flysch (a sandstone and schist from
half of the box, with the polished surface pressed against them the Carpathians) obtained from a working quany and typical of
while the plaster was still soft, to ensure contact of each grain material used for dumped rockfill. The pieces of rock were placed
against the polished surface. between the parallel, hardened steel platens of a testing machine
After the plaster had hardened, the surfaces were submerged in and deformations measured as load was increased. A total of 36
water. The shear tests gave a constant angle of fnction of 25O over points were tested, dry and wet. Results, showing the crushing of a
the range of normal loads that could be used with their apparatus. point under increasing load, are shown in Fig.6.
This had a maximum value of 111N. Tests were repeated with the Tests that they had made dunng 1961-3 on pulvensed rock put
surface moist, as distinct from submerged, and the same value of dry into an oedometer, caused strains as shown by Fig.7. Saain
25O was measured. When, however, the surfaces were oven-dry, increased dramatically when water was passed through the sample.
the tests revealed an angle of friction of only 70. This dernonstrated clearly the effect of wetting a dry, severely
In order to obtain higher normal stresses, Penman (1953) scaled-down rockfill under load. It was found that the additional
repeated his tests, using the polished surface that was embedded in strain caused by wetting was greatest at low loads, decreasing as
the lower half of his shear box, but replacing the upper piece of the applied normal pressure was increased, as shown by Fig.8.
quartz with three freshly broken chips of quartz. These were This may have been caused by the fact that compression is due not
bedded in plaster in the upper half of the box so that their extreme only to deformations at the rock to rock contacts, but also to
points lay on a horizontal plane and were equidistant from the rotation and repositioning of the rock pieces into a denser packing.
centre of gravity of the upper pan of the box. Distilled water was The effect of wetting the more highly consolidated sample could be
 Rockfill for embankment dams 307

28, mm expected to cause a lower defonnation because the grains would

5.0 have less opportunity for moving to a denser state.
Tests to measure the defonnations occuning at a point contact of
rock have also been made by Clernents (1981). He used samples
4.0
of sandstone rockfill from the site of Scarnmonden dam in central
England. Cylindrical samples, 25-mm diameter, were machine cut
3.0 to form four-sidedpyramids with points at various included angles,
as indicated by Fig.9. These were placed against a flat surface of a
lower cylinder of sandstone and loaded in a testing machine.
2 .O To test an included angle of 1800 was in fact two complete
cylinders on end, and amounted to a compression test on a
cylindrical sample 50-mm high.
1.0 A blunt point with an included angle of 1640 suffered a
deformation of about 0.4mm under a load of 3.8kN before audible
crushing occurred, allowing the load applied by the rigid testing
o 250 500 750 1000 ,250 1500 N machine to fall considerably, as shown by Fig.9.
o
Sharper points, with included angles of 137O, 113O. and 790
behaved in a similar way, defonnations being successively greater
Fig 6 Deformation of point of rock fragment and crushing damage occumng at lower and lower loads. The
deformations which occurred under a load of 1kN on points of
vanous sharpness are given by Table 2.
Wetting the rock contact caused substantial increase of
deformation under a given load. The results of tests on fairly blunt
0.020 points (included angles P = 169.1 and 135.g0) are given by Fig. 10.
A constant load of 1kN was applied and maintained for two hours
to observe the continuing creep at the contacts, prior to saturation,
0.016 which induced additional deformations as large or even larger than
those caused by the original application of the load.

0.012

0.008

1 Load

0.004

o
O 1000 2000 3000 4000 5000
a, k . ~ / m ~

Fig 7 Compressibility of pulverised rock sample

o 1 2 3
Deformation, mm

Fig 9 Point contact load/deflection curves-sandstone

Table 2: Deformation caused by a load of 1kN

on points of included angle p

Po 11 Deformation mm

Fig 8 Variation of additional compression on wetting

with increasing applied normal pressuve
 Time, min 5. APPLICATION TO ROCKFLLL.
O 30 60 90 120 150
If rock fragments were spheres, and a particular rockfill was made
up of single size material, then the load carried by each rock
contact under agiven applied total normal stress would be directly
proportional to d2, where d represents the diameter of the sphere,
as indicated by Fig. 12.
The load carried by each contact

where o represents the total stress in the rockfill.

If o = yh, the overburden pressure, then

The large difference in the contact loads which have to be

carried by the fragments of a dumped single size rockfill of say,
0.5m average diameter, and a sand with particles of about 0.5mm
diameter, can be seen from the following example.
Assume a depth of 50m in a fill of average bulk density
20kN/m3. In the dumped rockfill

In the sand

It can be seen from Fig. l l that if the durnped rockfill was a

Fig 10 Creep displacernent and additional deforrnations saturated sandstone, the contact load of 250kN would cause
on wetting considerable crushing to even the most rounded fragments,
whereas in sand, the 0.25N loads would cause no darnage even to
points with an included angle as sharp as 70°.

The load at which crushing damage first occurs is of particular

interest in dumped rockfill, because it amounts to failure at the 5.1. Parricle shape and size.
contacts of adjacent rock fragments, permitting deformation not
position of denser packing. Clements made a study of the load A comparison of particle shapes was made by Charles (1973) in his
required to cause damage not only to his machine-shaped points, study of sandstone rockfill. Rock fragments in the size range 76 to
but also to various points on irregular samples of rack fragment.
The results, shown by Fig.I,l, demonstrate the relatively low load
bearing capacity of the sharper points. If a load of 0.5kN is to be
carried by the saturated rock, the included angle of the rock point
should be greater than 150°. --

Section
I I
1

Plan

Fig 12 Contact forces in an idealised rockfill

F i g 11. Damage l o a d v. c o n t a c t angle. Saturated sandstone. consisting of single size spheres
 Rockfill ior embankment dams 309

38- were compared by photography with pieces 4.8 to 2.4mm permeability of the rockfill is reduced to 1x10-5mJs. If the
size and sand from the same quarry of size 0.3 to 0.15mm. permeability is much lower than this, there is a danger that
Reduction and enlargement was adjusted to bring al1 the particles constmction pore pressures may develop and the material should
to a common size, as shown by Fig. 13. There is close similarity of be considered as an earthfill rather than a rockfill.
shape and similar included angles ofAhe points of all the sizes.
This comparison, combined with knowledge of the contact loads
in fill composed of different sizes of particle, shows that while 5.2. Acceptance of the concept of excessfines.
considerable crushing and deformations can be expected in large,
single size dumped rockfill, sand-sized particles are unlikely to
suffer any crushing damage under typical overburden stresses. Although Terzaghi (1960) had made it clear that the cause of large
It is clear that if the large pieces of rockfill are separated by a deformations in dumped rockfill was not due to the presence of
bed of sand, contact loads can be reduced to acceptable arnounts. fines and that, on the contrary, a well graded sand and gravel
It is analagous with the use of mortar to separate bricks and formed an ideal rockfill, it has taken a long time for the profession
masonry blocks. Without mortar, quite apart from problems of to accept that "dirty" rockfill, even containing weathered rock
laying leve1 courses, etc., the bricks would break if the height of excavated during the opening 0f a quarry, can be perfectly
the wall increased sufficiently and the resulting constmction would satisfactory and superior to rockfill formed from the best,
have limited stability. unweathered, hard rock from the heart of a quarry that has been
Separation of the large pieces can best be achieved in rockfill by passed over a 150mm spaced grizzly to eliminate fines.
slightly overfilling the voids with a graded finer fraction to ensure Even in 1973, Bertram continued to recommend rockfill without
that the large pieces can never actually come into contact. In fines. He welcomed the introduction of heavy vibrating rollers that
addition to reducing contact loads, the fines-filled voids prevent could compact rockfill in relatively thin layers instead of dumping
rotation and reorientation of the larger pieces. In this way, it in layers 10 to 12m thick and claimed that Congar dam, 136m
deformations are reduced both by minimising contact distortions high, completed in USA in 1964, was the first major dam to use
and avoiding movements towards denser packing. vibrating rollers. But he warned that pounding by a heavy
Penman and Charles (1976) have suggested that a limit to the vibrating roller can produce what he described as "undesirable
smallest size and amount of fines that can be used, is when the bulk amounts of fines".
He recommended removal of fines by processing and said that
the separation of spalls (usually defined as rock pieces smaller than
150mm size) and sand size matenal from the coarser rock by a
grizzly operation had the advantage that, if matenal smaller than
6mm was not removed, it would remain on the surface of the layer
being compacted, thus reducing the efficiency of the roller
operation and interfering with the bond between layers This has
Fraction been shown, as we shall discuss later, to be completely wrong:
38 to 76mm fines help transmit roller energy into the rockfill and with correct
placing techniques, the large rock sizes punch through the surface
of an underlying layer to such an extent that the boundary between
layers can not be seen in an excavated tnal pit unless the lower
layer surface had previously been marked by a dusting of coloured
sand. The surface is then revealed to be completely irregular and
indented, passing under the larger pieces from the upper layer, the
whole being so integrated that it could be said that no surface
between the layers existed.
Marsal (1973) was unable to shake himself free from this
universal teaching that fines should be kept out of good rockfill. In
his outstanding conmbution to the subject of rockfill (described
Fraction fully by other chapters in this volume) he studied in detail contact
2.4 to 4.8rnm
forces between particles of rockfill and measured the crushing
strength of particles, yet womed about "idle" particles, lying in the
voids between the larger pieces of rock without canying any load.
He did not advocate overfilling the voids with fines to ensure that
they did cany load and drastically reduce contact forces. This may
in part be due to the fact that the rock quamed for El Infiernillo
dam, in which he was primarily concerned, did not have any
surplus of fines. In companng the behaviour of this rockfill with a
natural sand and gravel from Pinzandaran, he said that the El
Infiernillo rockfill, although composed of hard rock grains, was
more compressible, with lower shear strength and underwent more
breakage than the gravel, mainly due to its narrower gradation.
Later, after the El Infiernillo large scale testing equipment had
been moved to the National Autonomous University of Mexico, he
canied out tests on mixtures of rock pieces and sand, i.e. a rockfill
with the voids grossly overfilled with sand (Marsal and Fuentes de
la Rosa 1976). Least compressibility of the rockfill was measured
when it was mixed with 30 to 40% of sand and its shear saength
was found to be improved when the addition of sand lay in the
range 30 to 50%. From the results the authors concluded that,
Fig 13. Sandstone rockfill: shapes of different sizes. from the practica1 point of view, rock-fines, gravels and sands
blended with a rather uniform rockfill, in proportions that range The sandstone had a very pronounced panem of orthegonal
frorn 20 to 50% by weight, improve its behaviour and allows a fissures and bedding planes. At fist, a ripper was hied, attached to
better com~actionby vibrating rollers. a 290kW Caterpillar D9G tractor. This loosened the rock by
opening up the fissures and joints, producing sorne -rery large
pieces of rock. An atternpt was made to fragrnent the rock before
6. WINNING ROCKFILL. ripping with a pattem of 1.5rn deep holes at 1.5m centres, charged
with 0.45kg of gelignite per hole and f m d simultaneously. After
It is clear from the above discussion that the old idea of smpping ripping, however, the range of rock sizes was found to be similar to
weathered rock and carting to waste to obtain hard, sound rock those frorn plain ripping.
frorn a quarry was the result of a commonly held misconception
about the behaviour of rockfiii.
It can be argued that advances rnade in geotechnical engineering 6.2. TriaI blasts.
have provided design techniques which enable airnost any fill from
a quarry to be used. Clearly care is necessary in the use rnade of A vertical face about 6.4m high was prepared to start the quarry
the fmest rnateriais, particularly if the rock strata contain thick beds and for the first mal seven rows of 75mm diarneter vertical holes
of clay. were drilled behind the face at centres of 2.7rn over a length of
Often design can be adjusted to allow acceptance of the whole about 3 0 a . The holes were filled to within 2.7m of the surface
output from a quarry. The slopes of ernbankments rnay be with 70rnrn diameter sticks of gelignite "Q" and stemrned, giving
flattened to cater for lower strength fill, or low grade material can an overall explosive ratio of 0.42kg/rn3.
be ussd to provide toe waiting to irnprove stability. The explosive w?s fired using shon delay detonators, wired in a
Granite rockfill for Salt Springs dam was won rnainly from three senes of V forrnations in plan to give ínitial detonation at the
quarries with different floor levels to suit the railways used to centre of the free face, the explosions occumng in succession
cansport and dump the rockfill frorn three levels. A volume of towards the rear comers of the quarry. The first four delays had
1,600,000m3 of rockfill was won frorn the quarries whicii were incrernents of 25ms. 5 to 12 had 50ms and the remained had 70ms
worked with faces 18 to 55rn high. delay time.
Explosives were packed into 150mm diameter holes drilled This fust blast (trial A) produced quite a lot of large sized rock:
down to the leve1 of the quarry floor at 5.5 to 6.7rn centres, 9 to 15% consisted of pieces larger than 7.5m3 in volume and 45% was
12m back from the face. In addition, 50mm diarneter holes at 2.5m made up of pieces larger than 2.5m3. It was plancrd to place the
centres were used for lifterb at the quarry floor. The largest blast, rockfill in layers about 0.9m thick so the largest piece of rock
made on the 5th November 1929, used 44 main holes 48rn deep acceptable could not be bigger than 0.9m and could be expected to
and 99 lifter holes loaded with 60 tonnes of gunpowder. It have a volurne no more than about lrn3. Thus it was clear that the
fragmented 177,000m3 of solid granite with an average powder explosive force was neither strong enough nor sufficiently well-
factor of 0.36kg/m3. This quarrying method, using large diameter distributed to produce the desired fragmentation.
heavily-loaded holes, produced many large rocks but also a Accordingly, for the second blast (trial B), the holes were drilled
substantial arnount of smaller sizes. Secondary drilling and at the closer centres of 1.8m and filled with the more powerful
blasting was used to break down the larger pieces. The fragrnented Polar Ammon Gelatine Dynamite to within 1.5m of the surface.
rock was loaded by Bucyrus 3rn3 shovels which could handle This produced an overall ratio of explosive to rock of 0.89kg/m3.
pieces of rock up to 25 tonne size. One shovel working two 8.5 A pattem of delayed firing was used as with the first test. The
hour shifts for six days a week could load 38,250rn3 a rnonth. The resulting rockfill had a rnuch more acceptable grading, b u t still,
compressive strength of the granite, rneasured on cubes and about 25% of the lotal was what was regarded as oversize.
cylindricai specimens, lay in the range 100 to 130N:mrn2.
6.3. Oversizepieces.
6.1. TriaI quarry.
A total of eight blasting trials were made and details are
sumrnarised in Table 3. It was alrnost impossible to avoid
When opening up a new site, particularly in an unfamiliar rock producing pieces of rock longer than l m and, during construction
strata, it may be advantageous to start a trial quarry. This can of Scammonden dam, cranes carrying steel drop-bails were used
enable both excavation and blasting rnethods, as well as both in the quarry and in the embankment to break any long pieces
placenents rnethods, to be evaiuated. of rock.
F'rior to construction of a rnotonvay across the Pennine mountain The drop-balls, weighing about 0.75 tonnes, were attached to the
range in central England, this type of mal was made to assess the crane's hooks through an old tyre to act as a shock absorber and
problerns to be expected. The route involved construction of a there was an amusing incident on the ernbankment. The crane
70m high embankment which was aiso to be used to retain water at drivers were instructed to drive around, looking for oversize pieces
Scammonden. At the same time, Scarnrnonden was to be the to break. They would position their crane so that the ball would
highest ernbankment dam in Britain. Adjacent to it, on the line of drop on the chosen piece and the an was to operate the clutch so
the motorway at Deanhead, there was to be a cutting, 46m deep, that as soon as the ball struck the rock, the crane would begin re-
and it was decided to carry out mais in the zone for the cutting. winding to prevent the cable frorn running lwse and jurnping off
These mais have been described in detail by Williams and Stothard the top pulley. One driver, with rnuddy boots, had the misfortune
(1967). to have his foot slip off the clutch pedal as the ball was falling at
The vailey of the Black Brook, which was to be crossed, lay in full speed, but before it hit the rock. The crane began to re-wind
the Millstone Grit senes of Carboniferous sedimentary rocks, and the old tyre stretched as through it would break, but it didn't:
consisting of aitemating, almost horizontal layers of sandstone and neither did the ball quite reach the rock.
shaie: there were aiso beds of mudstone. The energy stored in the tyre, plus the steady upward movernent
At the site of the hiai for the quarry, the rock was Midgeley Grit of the hook, brought the ball back up with an increasing
sandstone. Up to 5m of highly weathered, soft sandstone was acceleration and flipped it completely over the crane, smashing the
stripped frorn an area of about 12,599rn2 and an access road was jib backwards over the cab. Fortunately the strength of the cab w a ~
built to the site for the hial ernbankment. sufficient to prevent it frorn being badly crushed and the driver
 Rockfill for embankment dams 311

Table 3: Explosives and rockfill produced by trial blasís

Trial Hole ' Hole Explosive Explosive Stemming Rockfiü produced

Dia. Spacing Type Ratio Oversize Shape % passing sizes (rnrn)
mm m kgIm3 m % of total % of oversize classi-
Pieces larger than L D 3 W í b 3 fication 914 610 300 150 75 38 10 2.4
1.53 0.76 0.25

A 75 2.7 OGQ 0.41 2.70 O 15 45 45 40 poor 94 82 58 41 31 - - -

B 75 1.8 PAGD 0.89 1.50 O 15 25 45 5 fair 96 83 62 44 29 25 17 14
C 75 1.8 OGQ 1.30 1.00 O 5 10 25 5 good 99 92 73 53 38 31 20 18
D 75 2.3 OGQ 0.89 0.89 O 5 15 25 5 good 97 87 64 46 32 24 15 11
E 75 2.3 OGQ 0.83 1.00 5 20 30 30 10 fair 96 83 60 46 36 27 18 15
H 75 2.3 OGQ 0.77 0.90 O 5 15 60 35 poor 95 83 56 36 25 20 12 10
J 75 2.3 OGQ 0.83 0.90 O 5 12 15 5 good 95 87 75 52 36 27 18 15
K 75 2.3 N 0.95 0.90 O O 15 35 10 fair 100 91 72 49 34 25 17 15
L = leneth. W = width. T = thickness

Details of explosive
Explosive TYP~ Dia. Detonation StrengtNblasting
rnin velocity ( 4 s ) gelignite (%)'

OGQ Opencast GeligniteQ Gelatinous 70 4000 84

PAGD Polar Arnmon Gelignite Dynarnite Gelatinous with arnmonium 70 & 50 2700 90
nitrate and nitroglycerine
N Nobelite Ammonium nitrate based 250 3000 90

escaped unharmed. diameter and each charged with 8.6kg of Polar Ammon gelignite
The experience of quarrying the sandstone demonstrated the fact instead of the 27kg of Nobelite used in each main hole. This
that the shape and size of rock pieces is largely controlled by the resulted in little overbreak and it was felt that this sytem could be
spacing of the fissures and bedding planes in the rock mass. The developed so that the cost of the extra holes would be offset by the
early trial blasts were found to have moved the rock on its bedding reduction ot ~xplosiveand resulting reduction in handling and
planes for some distance around the perimeter of the blast. This treatment of oversize rock pieces.
moved rock was difficult to dril1 through so that further
fragmentation could not be achieved, causing large pieces to fa11
with the produce of the next blast. 6.5. Grading.

A particle size distnbution following a Fuller curve could be

6.4. PTe-split blasting. regarded as an indication of an ideal grading that would enable al1
voids between particles of a given size to be filled with particles of
In an attempt to overcome this difficulty, pre-split blasting was a lesser size, so providing the densest possible fill. In practice,
tried along lines defining one edge of each of the trial blasts D and particularly when working with irregularly shaped pieces resulting
E. A row of 75mm holes was drilled to a depth of 7.6m at 0.6m from blasting, conditions are somewhat removed from the
centres prior to blast D. Each hole waScharged with lOOg theoretical ideal. To embrace Fuller conditions, a band was
cartridges of Polar Ammon gelignite, spaced with 175mm proposed, as shown in Fig.14, to enclose what would be accepted
cardboard tubes to be at 250mm centres over the whole depth. as satisfactory gradings.
Cordtex fuse was passed through the cartridges and the line of To determine the grading of the rockfill resulting from each tnal
holes was fired prior to the main drilling for trial D. blast, samples of 30 to 40 tonnes were taken from each mal. These
Hole spacing along the line adjacent to mal E was increased to were separated by hand, and passed through a complete range of
0.9m and the cartridges in each hole were placed at 460mm grids, from 1.2m square, down to 0.15m square, then through a
centres. The charges were blown simultaneously in an attempt to 75mm sieve. A quartered down sample of the material passing the
create a vertical fissure which would prevent energy from the main 75mm sieve was then wet sieved through a complete range of
blast passing into the adjacent rock mass. It was concluded by Bntish Standard sieves from 38mm size down to a 200 sieve
Williams and Stothard that while the results of these two pre- (aperture size of 0.075mm).
splitting mals were partially successful, the end hardly justified the The results of these gradings were expressed as weight of
means. material passing each of the aperture sizes in terms of percentage
A better way of avoiding damage to the rock mass beyond the of the total weight. The grading curves that complied most closely
limits of the blast area was to increase the number of holes and to the proposed grading envelope are shown in Fig.15. Those
reduce the charge in them, around the periphery of the blast area. considered unsatisfactory are shown in Fig.16. The trial blasts
This approach was tried along one side of trial blasts J and K. The showed that when sufficient explosive was used to fragment the
75mm holes were drilled at half the spacing and loaded with only rock sufficiently so that 90% would pass the 0.6m square grid, then
60% of the charge of the main holes at trial J. This was fairly there was an excess of fines. In general, below a size of about
successful, but it was felt that even less explosive could have been 20mm, there was usually an excess of fines, as compared with the
used, so at m a l K, the line of holes were drilled only 50mm theoretical Fuller curve for the maximum rock size being
 312 Penman

considered. It was felt to be desirable, however, to accept slight another dam began on the River Don. only 11 miles away. The site
excess of fines, rather than have to reject or separately break was in the same Carboniferous series, but sorne of the sandstone,
oversize pieces which could not be put into the 0.9m thick layer called Huddersfield white rock, was considered to be slightly
envisaged for embankment construction. harder than the sandstone at Scammonden, with a fissure and joint
Shortly after Scammonden was completed, construction of pattern which divided the rock into pieces less elongated. In o d e r
to accornrnodate larger sizes of rock, the embankment was built
with layer thicknesses of 1.7m.

W t . passing 6.6. Quarrying slatey argillaceous rock.

The control of size and shape effected by fissure spacing in a rock

stratum was clearly illustrated during excavation of slatey
argillaceous rocks of lower Palaeozoic age in central Wales for the
rockfill Llyn Bnanne dam. The fissure structure of the rock mass
might have caused instability in a high, vertical face, so it was
decided to work the quarry from four levels, at vertical height
differences of 12m. For daily blasts, holes of 125mm diameter
were drilled to the full depth of the face at 4.5m centres along the
line of the cut. Some holes were found to be water-beanng, so a
water-resistant. slurry-based charge was required, although in
general, the effective yet cheap ammonium nitrate explosive was
used.
After some mals, a pattem developed of charging each hole with
2.45 9.5 19 38 75 150 M7 600mm. a 25kg base charge of the.more powerful slurry, to cut the vertical
bedding planes at the toe and leave a leve1 floor, followed by an
F i g 1 4 . P r o p o s e d g r a d i n g e n v e l o p e and F u l l e r C u r v e .
ammonium nitrate - diese1 oil mixture to within 4.5m of the top,
this upper part being correctly stemmed. At normal production,
each blast used 1.8 tonnes of explosive ánd to cater for this
Wt. passing % demand, a special lorry was built with pumps and hoses so that it
could de-water the holes and pipe the explosive directly into them.
Individually, both the ammonium nitrate and the diese1 oil were
not classed as explosives and were carried in separate
compartments in the lorry. They were only mixed together as the
explosive mixture was being pumped into the holes. The
consumption by the quarry was about 15 tonnes of explosive a
week so it was necessary to have mechanical handling for the
ammonium nitrate.
The 125mm holes were drilled by a Gardner-Denver HDC and
Reich 355 rotary drills at arate of about 275m a day.
The rockfill won by blasting consisted of platey rock fragments
and there was some discussion at an early stage that preferential
onentation resulting, perhaps, from tipping from lorries, could
leave inclined planes of weakness in the fill. A popular,
journalistic article at the time spoke of building a dam from broken,
2.4 5 9.5 19 38 75 150 3CO 600 mm. dinner plates! Tnaxial tests on 0.3m diameter samples showed that
values of the angle of shearing resistance of the rockfill, in terrns of
Fic] 1 5 . A c c e p t a b l e r o c k f i l l : f r u r t r i a l b l a s t s C.D,J,K and R.
effective stresses, fe11 from 50° under a minor principal stress d'
70kN/m2 to 400 under a stress of 689kN/m2. Tests with the platey
fragments arranged with their surfaces parallel to the failure plane,
Wt. p a s s i n g N gave results about 4O lower than for random orientation.
Construction of the mal embankment, however, soon showed that
the action of spreading the rockfill in layers and compacting it with
a smooth, vibrating roller (Fig.20), ensured random onentation and
removed any fears of any weakness caused by parallel stacking.
Following these trials of excavation and placing rockfill, a
specification was drawn up designed to utilise produce from the
quarry without processing. The rockfill was specified to have a
maximum size of 0.3m, with not more than 25% below 4mm.
During dam construction, the restriction on maximum size was
relaxed because it was found that about 20% of the quarry run was
larger than 0.3m. The oversize rock was confined to the outer part
of each shoulder and was spread and compacted in l m thick layers.

6.7. Effecr of rock hardness.

2.4 5 9.5 19 38 75 150 MO 600 mrn.
In general, the harder a rock, the more angular the fragments and
F i g 1 6 . U n a c c e p t a b l e r o c k f i l l : f r u r t r i a l b l a s t s A,B,E,H &Q the less fines result from blasting. Where tunnels and undergronnd
 Rockfill for embankment dams 34-3

chambers have to be excavated, as for hydro-powei schemes, much compaction energy equivalent to that applied by the field
of the spoil comes from the unweathered heart of the rock mass. In machinery is applied to small samples of about 0.001m3 at a range
hard rock, such as basalt. the resulting fragments can be very hard of water contents in a steel cylindrical mould. Plots of density v.
and sharp. water content reveal optimum values: the tests have been descnbed
The Foz do Areia darn (160m high, the world's highest in detail by Bntish Standard Specification 1377: 1990.
concrete-faced rockfill dam) on the Iguacu River in the state of The size of rockfill particles preclude the use of small moulds
Paranfi, Brazil, was built from basalt and bassaltic breccia. for canying out this type of test and there is no standard test for
12.5hm3 of the required 14hm3 of rockíill for the darn was spoil obtaining p t i m u m conditions for ,rockfill. Some idea of
from the tunnels, powerhouse and associated excavations. The conditions may be obtained from the construction of trial
massive basalts were a fine-grained igneous rock, moderately embankments, but often these can not be built until preliminary
jointed, with a subvertical fracture system. On the other hand, the contracts have been let.
basaltic bíeccias were hybnd rocks, made up of fragments of Fortunately, the relatively free-draining properties of rockfill
vesicular basalt in a silty-sandy to clayey-calcitic mamx and were usually permit an excess of water to be added without harm. In
only slightly jointed. some cases it has been found that the fines are of such a
Fill for the main part of the dam, predominantly denved from consistency that they forms balls during working at high water
excavation in massive basalt, though some material from the contents and this prevents adequate filling of the voids between the
breccias was included, was of fairly uniform size and short of larger pieces. In such cases it rnay be advantageous to limit
fines. Grading c w e s for the matenal, given by Pinto et al. (1985) addition of water to the fill.
is shown in Fig.17. It did not entirely meet the specifications, Placement of todayrs rockfills, containing plenty of fine
which required a more extended curve, with a minimum of 20% matenal, is usually in layers of thickness equal at least to the size
below no.4 sieve to provide for better compaction and lower of the largest piece of rock in the fill. Placement mals, such as
perméability. The rockfill was accepted as produced so as not to those descnbed by Williams and Stothard (1967) have shown that
increase costs. Deformations on reservoir filling were larger than the best method of placement is to tip the rockfill on to the layer
in some other comparable dams consuucted of compacted rockfill, being placed, 3 or 4m back from the advancing edge. The lorry
but were aiiow for in the membrane design and caused no trouble. load of tipped fill is then bulldozed over the edge to leave a level
Quoich darn (38.4m high), completed in 1956, was an early surface at the required height above the surface of the previous
example of compacted rockfill. The valley was in a schist and a layer, producing the desired layer thickness. The larger pieces fa11
majonty of the required rockfill came from tunnel spoil. It was on to the lower surface (Fig. 18) and are covered with fines which
built at a time when fines were thought to be deletenous to the form a relatively smooth surface (Fig. 19) that readily accepts
properties of rockfill. Following the old idea of ensunng rock to compaction rollers and the wheels and tracks of transporting and
rock contact, al1 material smaller than 10mm' was removed by haulage machines. The surface of a slatey argillaceous rockfill after
screening and washing. The fill was placed in 0.6m layers, sluiced compaction by a smooth vibrating roller 1s shown by Fig. 20.
and rolled with a 3.5 tonne vibrating roller. Bishop (1959) in
discussing description of the work by Roberts et al. (1958), pointed
out that this compaction would produce fines from the damaged 7.1. Compaction.
points of contact and questioned the need for removing the below
lOmm size matenal. The result was very satisfactory: darn An early example of compaction with vibration was given by
deformations have been small. Hellstrom (1955). The 15m high Nissastrom darn was built from
tunnel spoil, compacted in 0.6m layers by a 10 tonne steam roller
followed by a plate vibrator of 1.6 tonnes static weight. Together
they caused 18% compression of the layer. The embankment was
completed in January 1950 and crest observations showed very
small vertical movements: the maximum was a nse of 7mm.
To determine the compaction achieved by vanous passes of the
various machines at the Scammonden mals, the level of the surface
of the placed fill was measured dunng compaction. T o test
separately rockfill from the various blasts, the placement area was
divided into nine bays, each 60 x 18m. Because the placement area

Fig 17. Foz cb Areia: m i n rockfill(1B) ard t r a s i t i m m t e r i a l ( I I 6 ) .

Graiug curves as recorded i n tk field. (Pinto e t al 1985).

7. PLACING ROCKFILL.

The aim of placing fill is to achieve desired deformation and

strength characteristics at least cost. With earth fill, specification is
often based on a simple concept of placement at an optimum water
content to an optimum density. Optimum values are obtained from
tests such as that devised by Procter (1933) in which an amount of F i g 18. Advancing e d g e o f r o c k f l l l l a y e r .
314 Penman

Fig 21 Layout of area for compaction trials

F i g 19. Surface of placed and compacted r o c k f i l l .

Best results were obtained by tipping the rockfill 3 to 4m back
from the advancing edge of the layer and then pushing it over,
controlling the bulldozer blade to produce a level surface at a
height to produce the required thickness. Laser control is now
available to make this process semi-automatic. Concern was
expressed that layers of fines might be left in the rockfill,
producing surfaces of weakness, but mals pits, particularly those
made dunng subsequent construction of the dam, showed that the
larger pieces had been pressed, dunng compaction, well into the
lower layer and the irregular junction between the layers could
only be detected where coloured tracer layers had been used.
After placement in the trial areas, the rockfill was compacted
using, at vanous times, the vibrating rollers, a gnd roller and a
pneumatic-tyred roller. Details of these rollers are given in Table
4. After a given number of passes, the thickness of the layer was
measured by levelling. Before this could be done, track and tyre
marks were smoothed out by a light tandem smooth roller and
while surface measurements were being made, any change in the
level of formation was obtained from readings on the overflow
tig LU. burtace o f a s l a t e y a r g i l l a c e o u s r o c k f i l l settlement gauges. The achieved density was calculated from the
a f t e r compaction by a smooth v i b r a t i n g r o l l e r . known weight of placed rockfill and measured water content.
Subsequently, mal pits were excavated so that the packing of the
rockfill could be examined. At the same time, pea-grave1
was underlain by a st'iff clay, it was necessary to measure any replacement density tests were made to give further information
settlement that might occur under the placed rockfill so that the about the density achieved and, in particular, variations which
measured levels of the compacted surfaces would give a correct might occur over the depth of the layer.
value for the layer thickness. Accordingly 72 overflow settlement Three layer thicknesses of 0.6, 0.9 and 1.2m were med but best
gauges were installed in the clay at positions indicated by Fig.21. results were obtained with a 0.9m thickness and this was used for
These gauges have been descnbed by Penman71969). the majofity of the tests. The results of the compaction mals, given
A stable level referente was built in the form of a reinforced by Williarns and Stothard (1967) are shown by Table 5.
concrete column founded in solid bedrock and red, bumt colliery Physical properties of the rock were determined from tests made
shale was spread as a marker layer so that dunng subsequent on samples taken frequently dunng the tnals. Most of the tests
excavation to study the condition of the placed rockfill, the conformed to Bntish Standard 812. Compressive strength was
position of the layer base could easily be found. Rockfill from the measured on cylindrical samples 25mm diameter x 25mm high in
various trial blasts was weighed before being placed and spread in an oven-dry condition. It has been found that saturation of the
appropnate bays of the test area. sample reduced the compressive strength to between a third and a
Each of the nine bays was split into two lengths, each 30 x 18m half of the oven-dry strength. There is no laboratory tests that will
and after placement surface levels were measured at the determine the optimum placement water content for rockfill, but
intersections of a square grid of 1.5m side. Only the central 15 x dunng the trials, weather conditions caused placement water
9m area was measured, leaving an outer strip 4.5m wide as content to vary over the range 5 to 14% of dry weight. In general,
support, to minirnise the effect of lateral spread. high densities were obtained when the placement water content
Placing methods tried included (1) tipping on the prepared was 5 to 6%, rather than 8 to 10%: the results of the tests at water
surface, then spreading the heaps to form a continuous layer; (2) contents greater than 10% are not included in Table 5.
the traditional approach of tipping over the edge of the advancing
layer (this minimised the amount of spreading required) and (3)
tipping on to the advancing layer, following by bulldozing over the 7.2. Method of specification.
edge. These methods caused segregation, voids were caused by
large pieces bndging over each other and the layer surface was left These mals enabled a method specification to be prepared for use
too rough for a smooth vibrating roller. in the consmction of the rockfill embankments on the route of the
 Rockfill for ernbankment darns 315

Table 4: Details of rollers used for trial compaction

Roller Manufacturer Working Width of Dia. of Power Frequency of Cenuifugal

weight roiier roiier vibration force
(tonnes) (m) (m) (kW) (Hz) (m)
Gnd Hyster 13.5 1.60 1.68
I1172T Stothert and Pitt 3.75 1.83 1.30 36 40 125
nAfi2T Stothert and Pitt 5 1.83 1.37 40 33 169
SAW ABG 8.5 2.00 1.60 49 24 to 26
Pneumatic tyred Albaret 52 Tyre pressure 620 kNlm2

Table 5: Results of compaction trials on sandstone rockiill

Layer Blast m/c Dry density (tonneIm3) Compacting plant

source (%) Number of passes
O 4 6 8 10 12 16 20

(1) 0.9m layers

1IB B 6 2.07 2.08 2.10 25 ton dumptmck
11C B 6 2.07 2.15 8.5 ton vib. roller
21.4 B 5.5 2.15 2.20 2.21 8.5 ton vib. roller
2D BIC 5 2.08 2.15 2.20 52 ion roller
2.24 8.5 ton vib roller
2/c C 5 2.08 2.18 2 18 8.5 ton vib. roller
31A D 8 1.94 1.99 2.02 3.75 ton vib. roller
2.04 8.5 ton vib. roller
31C D 6 2.08 2.15 2.16 2.18 8.5 ton vib. roller
3/8 D 8 1.99 1.99 13 ton grid roller
2.02 2.02 52 ton roller
2.07 8.5 ton vib. roller
2/F E 8 1.73 1.79 1.81 8.5 ton vib. roller
28 E 9.5 1.91 1.97 2.02 2.02 2.02 8.5 ton vib. roller
4/8 E4H 9.5 1.96 1.99 2.02 52 ton roller
2.04 2.05 8.5 ton vib. roller
WJ 8 1.88
2D
3/F K 8 2.05 2.10 2.12 2.13 2.15 5 ton vib. roller
2.15 8.5 ton vib. roller
4lA J 10 1.86 1.88 1.89 1.89 5 ton vib. roller
1.89 8.5 ion vib. roller
1/H R 9 1.89 1.94 1.96 8.5 ion vib. roller
11J Q 9 1.89 1.94 1.94 1.94 1.94 8.5 ton vib. roller
(ii) 0.6m layers
11A Rip H 6 1.97 2.02 2.04 2.08 2.08 8.5 ton vib. roller
1/F C 5 1.78 1.83 1.84 1.88 8.5 ton vib. roller
lE C 5 1.71 1.79 1.81 1.84 8.5 ton vib. roller
ID1 c 5 1.76 1.83 1.84 3.75 ton vib. roller
1.84 8.5 ton vib. roller
ID2 C 5 1.76 1.83 1.88 52 ton roller
1.88 8.5 ton vib. roller
(iii) 1.2m layer
41C UH 9.5 1.83 1.86 1.88 1.92 8.5 ton vib. roller

trans-Pennine motonvay and particularly for the rockfill placement 7.3. Limesrone as rockfill.
for Scammonden dam. Size of rockfill was restncted to 0.371113
with a maximum dimension of 0.9m, to be placed in layers with a Sedimentary rocks commonly fragment into a wide range of
loose thickness of 0.76 to 0.9m. Compaction was to be effected by particle sizes when quarried with explosives and so produce a
eight passes of an eight tonne vibrating roller, o r equivalent. rockfill that satisfies the requirement of sufficient fines. As with
Dunng construction of the dam, an 11.5 tonne vibrating roller was the sandstone considered above, limestone can produce a very
used and it produced the required density with five passes. T w o satisfactory fill.
five tonne vibrating rollers were also used, in tandem and the The 140m high Ben Chao Nen dam (renamed Snnagarind, when
required compaction was obtained after six passes. officially opened in 1980) was built in a limestone area and is
316 Penman

founded partially on limestone and constructed with limestone main rock fill in the shoulder would be placed in 0.5m layers,
rockfül. Of the 9x106m3 of rockfill required, 3x106m3 came from cómpacted by six passes of either a 15 tonne Dynapac CH61 or a
tunnel and power station excavations. A quarry was opened in an 12 tonne Stothert and Pitt T208 vibrating roller.
outcropping strata of hard durable limestone to obtain the
remaining two thirds of the volume. It was worked in 15m high
benches, drilled with lOOmm diameter and 175mm diameter holes
and fragmented by ANFO, a fertiliser grade ammonium niuate
mixed with diese1 oil. used at arate of 0.4 to 0.5kg/m3 of rock.
Maximum size of rock pieces was 1.5m. The rockfill was
hauled directly from the quarry without any treatment other than
addition of water, and dumped back from the advancing edge of
2m layers. As with the Scammonden rockfill, it was pushed over
the edge, leaving a fairly smooth surface of the finer fraction and
compacted by four passes of a 13.5 tonne vibrating smooth roller.
The grading of the rockfill before compaction is given in Fig.22.
The material did not contain an excess of fines and it was found
possible to determine the volume of voids in the compacted
upstream rockfill by the foliowing unique, large-scaie test.
To avoid sudden changes which might occur when, during
reservoir impoundment, the water leve1 reached the top of the
cofferdam and overfiowed into the rockfill upstream of the dam
core, this zone between core and cofferdam was pre-filled with
riy 23. Evretou dam: yradiny o f reef limestone rockfill.
water pumped slowly into it. Measured volumes of water showed
that the voids in the compacted rockfill amounted to 22% of the
total volume of the rockfill.
An example of the use of a very soft limestone was given by the Table 6: Results of compaction trials
construction of Evretou dam in Cyprus. lt was considered by on reef limestone rockfill
Brown (1986) that, judged by criteria used earlier in this century,
the soft rockfill would have been considered unacceptable for use
in dam construction.
Quarries were opened in the massive Koronia reef limestone.
The compression strength of the rock was measured on 60mm
cubes cut from pieces produced by mal blasts. Values of 24 to
30Nlmmz were obtained.

7.4. Compaction control.

Method specifications, derived from trial placements, are used

more frequently than specification of end product because of the
difficulties involved in attempting to control placement and
compaction by in situ measurements of achieved densities of the
rockfill. This does not preclude use of large mal pits for inspection
of the compacted rockfill and measurement of density and in situ
O
. . . . . *. m . m.
0 0 0 N 07C9n 00E permeability. It is 1,usual to make these measurements and
7 u-, OLMO \DO
7 0
inspections from time to time to check on the end product, but in
practice they are both too difficult to make and too expensive for
use in placement control.
F i g 22. Crinagarind dam: grading o f llmestone
Method specifications usually define grading limits, maximum
rockfill before compaction.
particle size, loose layer thickness, compacting plant and number
of passes and amounts of water to be added. The inspectors can
often judge grading by eye, confirmed from time to time by Spot
The absence of joint and bedding planes in the massive rock checks and quantities of water can be calculated from pump or
caused fragmentation by quarry blasting to produce irregularly bowser sizes. The number of passes of compacting equipment,
shaped pieces and a great deal of fine material. The grading usuaily a smooth, vibrating roller can be checked by ybservation,
curves, reproduced in Fig.23 show 25% of the fill passing the 200 but a compaction meter can be used to measure the degree of
sieve (Le. less than 0.075mm). This large volume of fines compaction achieved.
produced smooth surfaces to the placed layers, but addition of
water, to avoid subsequent problems of collapse settlements, could
cause excessive bogging of the construction plant. Water was 7.5. The compaction meter.
added by spraying it on to the rockfill during tipping and spreading
and, by limiting the rate to 0.1 of the volume of the rockfill, The fundamental idea is the use of the vibrating drum of the roller
excessive rutting by plant was avoided. as a measuring tool. The compaction meter, designed by
Tests to determine the effectiveness of compaction were made Geodynamik in conjunction with Dynapac research in Sweden, has
by measuring depression of the fill surface after passes of the been described by Forssblad (1980) and Thumer and Sandstrom
roller. From the results, summarised in Table 6, it was decided the (1980). Behaviour of the drum is sensed by accrlerometers
 Rockfill for embankment dams 317

attached to the bearing of the dmm. The signal is analysed by two pressure of 2,500kNIm2 which was considered desirable in view of
band-pass filter units set to detect the fundamental and first the proposed height of the dam.
harmonic frequencies of the drum. When working on loose Bedrock consisted of silicified conglomerate which had almost
material, the amplitude of the fust harmonic is low, but as the fill been melted by heat from nearby dioritic intrusion and the diorite
becomes compact, the first harmonic amplitude increases and itself. The rockfill came from tunnels, powerhouse excavations
causes an increase of meter reading. Improvement with repeated etc. and quames. Results of maxial tests on this material given by
passes is shown by Fig.24. Marsal and Ramirez (1964) are shown in Fig.25. The failure
The apparatus has been used with the meter in the driver's cab envelopes for both loose and dense samples show a distinct
so that he can see from the digital display when an agreed amount curvature: angles of sheanng resistance exceeding 4S0 at low
of compaction has been achieved. pressures, but reducing at increasing confining pressures.
The meter will record continuously and has been used on control
rollers,-driven over the compacted fill to provide a printed record
of the degree of compaction achieved.
Thus a permanent record of compaction can be made and
combined with an electronic system for recording roller position on
the fill, the equipment could be used to provide evidence of the
compaction over the whole area of each layer of rockfill.

", "
O 100 200 300 400 500
Normal s t r e s s , kN/m2

Fig 25 Results from triaxial t e s t s on c o m p a c t e d

rockfill

8.1. Curved failure envelopes.

Most particulate materials exhibit curved failure envelopes.

Values of the angle of shearing resistance for a silt at a voids ratio
of 0.6 were found by Penman (1953) to fa11 from 42.3O iinder a
confining pressure of 100kNlm2 to 380 under a pressure of
690kNlmz. De Mello (1977) drew attention to the curved failure
envelope found for rockfills and suggested that the shear strength
could be descnbed by a relationship such as:

where z represents the shear strength of the rockfill

o represents the normal stress on a potential shear
plane in terms of effective stress
A and b are constants.
He quoted, for the diorite rockfill at El Infiernillo tested by
Number o f passes. Marsal, values of A = 1.10 and b = 0.87. Values of other types of
F i g 24. Compaction m e t e r r e a d i n g s . rockfill, calculated by de Mello from test results elsewhere, are
given in Table 7.
Charles and Watts (1980) carried out tests on four rockfills
specifically to study the effect of increasing confining pressure and
8. MEASURING THE PROPERTIESOF ROCKFILL. to obtain values of shear s t r ~ n g t hat low confining pressures
corresponding to those that would apply to a potential shear plane
Strength is seldom a matter for concem with rockfill. Because of just below a slope in rockfill. The maximum size of sample they
its high permeability, construction pore pressures rarely develop could test in triaxial compression was only 0.23m diameter and
and it behaves as a fully drained material. The relatively high 0.5m high, but the rockfill they tested had surplus fines. I t had
densities achieved in a well graded, compacted rockfill ensure been found with this condition that strength was predominantly
adequate interlocking of particles and values of shearing resistance conaolled by the fraction below 38mm size, the maximum size that
are high. could be used in these shearing tests. The values for A and b found
The large sizes of individual pieces of rock prevent use of by Charles and Watts for the four rockfills tested under low
normal laboratory equipment for measunng strength but, prior to confining pressures are also given in Table 7.
constmction of the 148m high El Infiernillo dam, the Comisión de
Elecmcidad, Mexico, commissioned a special maxial apparatus
capable of testing specimens 1.13m diameter and 2.5m high. The 8.2. Stabiliry analysis.
maximum particle size that could be used in samples for shearing
tests was 0.2m, to enable a large fraction of the actual rockfill to be To enable these parameters A and b to be used in stability analysis
tested. The apparatus was designed for a maximum confining of rockfill slopes, Charles and Soares (1984) modified a computer
318 Penman

Tabk 7: Values f w the parameters A and b

for several compacted rockfills

Rock type A b Reference

Diorite 2.0 0.870 Marsal

Silicified conglomerate 2.6 0.846 Marsal
Pizandarau sand and grave1 2.2 0.876 Marsal et al (1976)
Nelzahualcoyotlconglomerate 2.1 0.881 Gamboa & Benassini (1967)
Malpasocenglemerate 3.8 0.808 Marsal (1973)
Carboniferous sandstone 6.8 0.67 Charles and Watts (1980)
Palaeozoic slate 5.3 0.75 Charles and Watts (1980)
Palaeozoie slate (weathered) 3.0 0.77 Charles and Watts (1980)
Basalt 4.4 0.81 Charles and Watts (1980)

programrne for circular a f t stability analysis to accept non-linear F i g 26. Circular arc stability analysis.
failure envelopes. They introduced a dimensionless stability
number so that they could produce charts to help the engineer to F represents factor of safety
make a rapid assessment of rockfill slope stability. H represents height of slope - m
The use of parameters obtained from triaxial tests could b e T represents stability number
criticised because most field situations correspond to plain strain. y represents bulk density of the rockfill.
Marsal (1973) however, has been able to demonstrate that
laboratory tests made under plane strain conditions, always give The value for A is that obtained from a log-log plot of test results
larger values for shear strength than corresponding tests under in which the shear strength and confining pressure are expressed in
triaxial compression. It was therefore considered by Charles and units of kN/m2.
Soares that the use of parameters derived from maxial tests will be As a guide to the values of confining pressures appropnate in the
highly conservative. Any resulting gain, however, could readily be maxial tests used to determine the parameters A and b, Charles and
eroded by small errors due to density or scale effects and should Soares have calculated the maximum normal effective stress (a'?)
not be included in design calculations. that will act on the critical slip surface. This has been presented in
The computer programme enabled circular arc stability analyses dimensionless form by the chart reproduced in Fig.28. A
to be canied out by both the traditional ~ellenius'(l936)method of knowledge of the slope, expressed as horizontal over vertical, and
slices and by the Bishop (1955) semi-rigorous method, in which its height, together with the b u k density of the rockfill, enables a
allowance is made for inter-slice reaction. The situation was value of (a',) to be obtained. This will vary according to the
slightly simplified by the fact that, by definition, rockfill can not value of b but initially can be taken as 0.75 until test results give a
develop consmction pore pressures so the analyses were made for better value.
the condition of full drainage. It was also assumed that failure
would not be caused by weak foundations and the circular arcs
analysed were of the form shown by Fig.26, with the slope over a
strong rock foundation. By making al1 the slices of the same
width, the factor of safety for the Fellenius approach could be
expressed as:

A L@ cos(2b-l)al
F = -
$1-b) Ch sin a

With a given slope height H, this was re-arranged as:

F(y~(l-b) q w ) b cos (2b-1) a

-
A Z(h/H) sin a
Charles and Soares (1984) designated the dimensionless right hand
side of the equation the stability number T.
A similar construction can be used with the Bishop semi-
rigorous approach and Charles and Soares have calculated the
stability number for this method of analysis, shown by Fig.27.
This chart can be used to obtain the factor of safety against the
occurrence of a rotational circular arc slip in rockfill from the
equation:

where A represents a strength parameter

b represents a smngth parameter Fig 27 Stability nurnbers v. slope angle
 Rockfill for embankment dams 319

01
O.5
I
1
1
1.5
Slope l(vert) on X(hor1
I
2
-
Fig 28 Maximum normal effective stress on critica1
slip surface

Fig 29. Post construction settlement o f rockfill darns.

9. DEFORMATIONS OF COMPACTED ROCKFLL.
specific numbers of years after construction. Their records
Small deformations are a particularly attractive feature of included measurements made over periods of more than 30 years.
compacted rockfill. The move towards its use was accelerated by The use of this type of information in the prediction of
the unacceptable behaviour of dumped rockfill: an aspect of movements at the design stage is beset with many difficulties.
particular importante in ernbankments designed to support Perhaps greatest among them is the fact that every embankment
uvstream membranes for the retention of reservoirs of fluids. exhibits individual properties both in site and material available.
Design against the unacceptable possibility of rotational slip
failure by use of a factor of safety gives little indication of the
amount of movement to be expected, other than that, in general, the 9.1. Analysis using finite element techniques.
greater the factor of safety the less is likely to be the movement.
In design for acceptable movement there are two questions: (a) What might be considered as a more logical approach to obtaining
what amount of movement is acceptable and (b) how can predictions of movement at the design stage is analysis by use of
magnitude of movernent be predicted? An answer to the first can finite element techniques. By dividing a section of embankment
be sought by a study of field measurements. The settlements of into discrete elements, calculation can be made of the load
embankments have been measured fairly commonly, but transmitted through each one due to gravity and other forces acting
measurements of horizontal movements are more unusual. The on the embankment. Knowledge of the deformation properties of
crest settlements of embankment dams have always been of the fill enables calculation of the distortion of each element caused
concern because of the danger of loss of freeboard. The potential by the load and hence the deformation of the whole embankrnent.
for damage to concrete upstream rnembranes caused by horizontal It is assumed that the elements are interconnectedonly at specified
movements has led to interest in measurements of face deflections. nodal points.
Details of the measurements of ernbankment deformations have The analysis of two-dimensional linear-elastic structures by the
been published and Sowers et al (1965) assembled 14 case records
of post construction vertical settlement. Values were plotted as a finite element method has been fully described (e.g. Clough
(1965), Zienkiewicz (1967)). and its application to the analysis of
percentage of fill height against log time, as shown in Fig.29.
embankments has been discussed in some detail by Clough and
Most of these embankrnent dams had been built with dumped
rockfill: their heights varied from 23 to 100m. The results show Woodward (1967). The key to the whole approach lies in defining
clearly that creep movements continue for a very long time in the stiffness properties of each element in tems of force-deflection
dumped rockfill and can readily reach values of 0.015H (where H relationships at its nodes. Once this has been done, the analysis of
represents the height of rockfill) after construction has been the whole assemblage of elements can be camed out by routine
completed. and well-established stmctural anaiysis techniques with a digitai
~ ~ c o l l e c t i ohas
n been made by Soydemir and Kjaernsli (1979) of computer.
the measured deformations of 23 uvstream mernbrane dams built The simplest shape of element for the two-dimensional analysis
between 1876 and 1970. They sep&ated them into two grocps: 14 is a triangle with a node at each apex. This shape is particularly
built from dumped rockfill,(Type A) and 9 of compacted rockfill useful when dealing wi:h complicated geometries. When
(Type AA). The collected measurements were presented in the formulating stiffness properties of the element, it is necessary to
form of three plots of deflection against dam height, reproduced in assume that interna1 defonnations vary linearly across it so as to
Fig.30. The fust plot (a) shows maximum crest settlements, (b) maintain compatability between the edges of adjacent elements.
shows maximum horizontal movement of the crest and (c) shows This implies that the stress and strain within each element are
maximum membrane deflection in the direction normal to the constant and they are usuaily taken to refer to the centroid of the
membrane surface. These plots show that the dumped rockfill mangle.
embankments suffered much larger deflections than those Three attributes of the finite element method which are of
constructed from the more modern compacted rockfill. particular value in the analysis of embankments are:
Al1 the results relate to post-construction movements and their 1. There is no resmction, other than the number of elements
increase with time was shown by Soydemir and Kjaernsli by use of that can be handled, on the complexity of geometry which
smooth curves to indicate the movements which occurred at can be anaiysed.
320 Penman

The analysis of an eanh or rockfill embankment is complicated

by the fact that it is consmcted in successive layers. Each layer
not only applies loads to the underlying fiU due to its self-weight,
but it also adds stiffness to the structure as a whole. The final
displacements are very dependent on the particular construction
sequence adopted.
Analysis made to predict movements of Scammonden, Llyn
Brianne and Winscar rockfill datps have been described 6y
Penman et al (1971 and 1975) and Penman and Charles (1973,
1975 and 1985).

9.2. Deformation paramerers from a large odeometer.

To obtain deformation pararneters for rockfill, a large odeometer

was constructed. It could accept samples of l m diameter in a
floating nng, shown by the section of the apparatus, Fig.31. The
pistons were perforated and faced with filter plates to permit

1
drainage. They were fitted with inflatable seals so that permability
tests could be made.
Type A dams Maximum particle size of the test specimen was limited to
Type A A dams 125mm and samples were prepared by simply sieving off larger
sizes fmm the rockfill. To achieve densities comparable with field
values, material was placed in thin layers, watered by spray to a
required water content and heavily compacted with a kango
electric harnmer. The floating nng was fitted with circumferential
strain gauges to measure the lateral pressure developed by the

FJfer
plares

Type A dams Pis!ons

Type A A doms

1.0 -

0.5 --

_---
o
O 50 1O0
D = lnitial reservoir filling Darn height ( m )
Fig 30 Deflections .vl. darn height. ( Frorn Soydernir 8
Kjaernsli, 1979 )

2. Self-weight can be incorporated simply by applying the

appropriate forces at each node.
3. The stiffness properties of each element are evaluated
individuaily on the computer, and it is therefore as simple to
deal with inhomogeneity and anisotropy as with
homogeneity and isotropy. Fig 31. Oedometer for t e s t i n g r o c k f i l l .
 Rockfill for embankment dams 321

---
sample. Compression curves obtained from tests on four rockfills 7-
are given by Fig.32.
Defomation parameters of the rockfill required for the linear-
elastic finite element analysis are Young's modulus E and

F i g 33. L a y e r s u s e d f o r c a l c u l a t i u n c f e q u i v a l e n t
cornpressibility.

This is used to calculate the compression of each layer which,

when surnmed, gives the compression at mid-height of the whole
thickness H.
The definition of the 'equivalent compressibility' for thickness
H is the average vertical strain over the half height hc divided by
the increase of vertical stress when the thickness is increased to H.
From this value of m,, the value of E is found from the
expression
1 2v2
E = - 1- -
mv 1-v

Strengh initiai \vater

D m Rock type
ofrock
index
dry content
density m plnced
Porosity

(hlN/rn2) (hlg/m3) (S) (56) Values of Ko can be obtained from the measured lateral pressures
Llyn B r i a ~ e Paiaeozoic 57 2.34 4.5 15 developed against the side of the oedometer ring, and values of
mudstone Poisson's ratio can be calculated by the expressions:
Winscnr Carbonüerous 120 2.06 7 21
snndstone
Lefkara Sheeted 252 1.95 3 3O
diabue
or alternatively
Thnmes 2.06 2 22
grave1
v-
2 - sincp
F i g 32. Cornpression c u r v e s f o r f o u r r o c k f i l l s . where cp represents the angle of shearing resistance of the rockíill.
This approach was used to predict the movements that would
occur during the construction of Scammonden, Llyn Brianne and
Poisson's ratio v. A1:hough rockfill clearly does not exhibit linear- Winscar embankment dams, from the results of oedometer tests on
elastic properties, its behaviour can be analysed by allocating samples of their rockfills. The movements which actually cccurred
separate values to E for each pressure increment. Each layer of were measured with horizontal plate gauges idescribed in the
rcckfill could be given a separate value of E for each successive section on instrumentation) and the compansons of predicted and
layer placed above it, but to make the computer programme more observed movements at Scamrnonden dam are shown by Fig.34.
manageable, an average value is taken tnat will correctly represent
the whole thickness being considered.
Penman, Burland and Charles (1971) have shown in Appendix 2
to this paper that a correct value of m, to represent a given
thickness of rcckfill is that which will give when applied to the
whole thickness, the correct value for maximum settlement, at the
mid-point. T o obtain this value of m,, the lower half of the given
thickness, H, is divided into five or six equal layers, as indicated by
Fig.33. The assumption is made that the vertical stress o, at the
mid-height of each layer, will equal the overburden pressure, 9. A
computation is then made of the stress increase at mid-height of
each layer when the rockfill is brought up from half-height hc to
full height H. i.e.
Layer 1 Initial stress Final stress
1 "I(hc - 0.5hI) yiH - 0.5hl)
2 ?/[hc- (hl + 0.5h2)l $H - (hl + 0.5h2)] -o- [bcerved r m m t
etc. -a- Predicted r m m t
The vertical strain caused by the increases from initial to final
vertical stress is read from the compression curve, e.g. Fig.32. F i g 34. P r e d i c t e d & o b s e r v e d rnovernents. Scanimonden data.
322 Penman

It can be argued that use of defomration parameters obtained

from one-dimensional cornpression is too simple for this type of
analysis. It has been found, however, that predictions based on
parameters derived from triaxial compression tests have not
necessaril y given better agreement. Charles (1976) defended the
use of one-dimensional compression tests and showed that during
embankment construction, most of the stress paths were
characterised by large changes in mean stress with only gradual
changes of principal stress ratio. Conditions are close to the
constant stress ratio of the one-dimensional compression test.
Where stress paths depart from constant stress ratio, either by
stress-ratio increasing qt constant minor principal stress or
decreasing at constant major principal stress, one dimensional
compression tests may still give a reasonable prediction of overall
deformations. It is of interest to see that an analysis of Llyn
Brianne dam by Cathie and Dungar (1978) using both two and
three dimensional finite element meshes and elastic, non-linear
elastic and elasto-plastic constitutive relationships, failed to
improve on the elastic equivalent compressibility method, as
described above, which was used by Penman and Charles (1973).

9.3. safety factor.

It has been argued that design for acceptable deformations does not
provide knowledgeof the factor of safety against failure. This
criticism has been overcome by Tan and Donald (1985) who have Fig 36 Vector movenients as factor of safety reduced
used an analysis of movements, using finite element techniques,
not only to predict movements but also to give a value for factor of
safety. This they ,have done by reducing the deformation development of a rotational slip in the upstream face. Reference
parameters by multiplying them with a h c t o r N until large should be made to Tan and Donald (1985) for fuller details of *he
movements occurred. A plot, Fig.35, of the movement of a node method.
shows when movements become unacceptable. The value of 1/N
at this point gives a factor of safety F, i.e. fmm Fig.35, N critica1 =
:.
0.48 1 F = 2.08. 10. INSTRUMENTATION.
When applied to the 162m high Talbingo dam, the movement
vectors, Fig.36, for N = 0.485, 0.474 and 0.400 show the The most direct measurement that can usefully be made on
rockfill is that of movement.

10.1. Measuring surface movements.

Surface movements can be measured in relation to stable reference

points by precise surveying.
It is convenient to make the stable reference points in the form
of pillars fitted with an instrument base to accept the surveying
instruments and position them accurately at an operating height. A
typical reference pillar is shown by Figs.37 and 38. Use of the
traditional mpods is not sufficiently accurate.
Because surface rocks tend to be disturbed, the observation
station should be founded well into the rockfill as shown by Fig.39.
T a l b i n g o darn. This may be difficult when a slope is finished with large, loose
boulders. It may be worth considering provision of a zone of
small size fill at the surface to accept the base of a substantial
monument.

Sockets can be grouted into concrete monuments, buried

instrument houses, etc, so as to be flush with the surface, to
minimise accidental damage (Cheney (1974). Under severe
conditions, they can be funher protected by bolted-down heavy
metal plates tightened on to xubber gaskets.
Two types of targets are used for observation by theodolite or
distance measuring device:
1 Bull's-eye of concentnc rings, white on black ground on a
flat surface at the centre of rotation of the target or a l a r g e ~
F i g 35. Nodal rnovement v. r e d u c i n g f a c t o r N. cruciform shape of the type illustrated by Fig.39.
 Rockfill for embankment dams 323

2. Cube comer reflector. end of the sight-line at the time of taking each reading. This
Both types of target can be screwed into the sockets and tumed to enabled an accuracy of +lmm to be achieved over a distance of
face the o b s e ~ n ginscniment. Several are required so that they do 300m.
not have to be moved during a set of observations from one
reference ~illar.

For highest accuracy, positioning should be made with a

combination of trilateration and triangulation. The basic three-
dimensional shape containing al1 reference points and targets
should be established using both methods. Subsequent readings
taken to measure changes of target position can usually be made
with only one method. The fastest to use may be either
mangulation or mlateration depending on the layout of the site.
Many electronic measuring devices use the speed of infrared or
other electromagnetic propagation. By timing the return from a
cube comer reflector, the double distance can be determined if the
speed of propagation in air is known. This varies with
temperature, humidity and pressure. In using the Tellurometer
MA100, Penman and Charles (1974) measured al1 three at each

F l g 38. DKM 2 A Theodolite i n u s e .

F i g 37. S t a b l e r e f e r e n c e p i l l a r . F i g 3 9 . T a r g e t on downstream s l o p e of dam.

 324 Penman

The Mekometer, invented by the Standards Division of the 10.6. Measuring infernal movements.
National Physical Laboratory, uses the principie of distance
measurement by means of a modulated light-beam (Pippard 1970). Settlements at specific positions in the rockfill can be measured by
It followed the discovery that the electro-optical effect in a crystal an overflow water level. A simple type, used to measure at several
of ammonium dihydrogen phosphate can function at microwave positions on a given lift of the rockfill is illustrated by Fig.40.
frequencies. This instrument is laid in trench to an outward fa11 during
Distance obtained from the phase difference between the rockfill construction. Its cost is largely in the trenching and
outgoing and returned beam with a modulation wavelength of backfilling operations of installation and in taking subsequent
32mm (frequency 9400 Mcls) gives a sensitivity of about 0.05mm. readings. Without very much increase in the cost of pipes and
Lower frequencies have to be used for distances greater than about fittings, a more useful horizontal plate gauge can be installed. This
100m. but a quoted accuracy was 3ppm f0.lmm. In practice this will measure both settlements and horizontal movements of

m'
allowed field measurements over distances of the order of 300m to discrete points within the rockfill.
be measured to an accuracy of flmm.
To obtain comparable accuracy in triangulation, angles should
be measured to an accuracy better than 1 second of arc. The Kern
DKM2A (a 1 second theodolite) read to an estimated 0.1 second,
was found to give a satisfactory accuracy in the field. Attempts to
Water t.ank 5 t
high l e v e l
,
improve this by using the much heavier DKM3 theodolite proved
to be self-defeanng because of surveyor fatigue.
6mm p l a s t i c
Transparent
standpipe \I 1 ;
I
l

tube Steel Ov/erflow

Convoluted
rubber hose
Accuracy in triangulation relies on sight lines being completely
straight. In practice they are readily bent by passing through air of
:
varying densities caused by temperature gradients (heat shimmer).
This effect may be reduced by arranging for sight lines to be well IlOmm s t e e l
clear of ground surface. Distance measurements by EDM sleeve
instniments are less seriously affected. Benefits may accrue from 100mm g a l v . ~onrrete
working at night, but bending can still be caused by residual s t e e l plpe
ground heat and cool night air.
Angular changes are most sensitive to target movement when F i g 4 0 . Water o v e r f l o w s e t t l - e m e n t g a u g e .
the sight line lies normal to the expected direction of movement.
Distance measurements should be along the direction of movement
for greaiest sensitivity. In practice a compromise usually has to be 10.7. The horizontal plate gauge.
stnick between these relatively conflicting requirements.
The reference pillars should be positioned so that they are: This consists essentially of a line of plastic pipes bedded in pea
1. on a stable foundation far enough away from the new grave1 in a trench across the rockfill at the required position. It
rockfill so as not to be affected by its mass passes through marker plates of metal or fitted with permanent
2. inter-visible and with clear sight lines to the positions magnets. The position of each plate can be accurately detected by
proposed for al1 the targets. This requires detailed a sensing unit carrying an induction coi1 or reed switches. The
consideration of the construction programme to ensure that level of the plates is given by a second sensing unit carrying a
sight lines will not be cut by temporary stock piles etc., water overflow orifice or other liquid device that will show height
particularly during crucial phases of the main works. change. One type uses a sensitive electrical transducer in the
3. at a sufficient spread and height difference to enable the readout unit to measure pressure of the fluid in a flexible tube
target positions to be accuractely established in three- leading to the sensing unit. The tube ends in a small, very flexible
dimensional space. balloon inside a wire cage. In use, the fluid presses the balloon
against the cage until the sensing unit is in position. A few
millilitres of fluid are then withdrawn into the readout unit to
10.5. Results. collapse the balloon slightly and the pressure of fluid measured to
show the vertical head difference between balloon and readout
Variation of distance does not usually provide high accuracy for position.
settlement measurements. Change of vertical angle may give a A problem is that of moving the sensing units into the line of
better measurement. The use of reflected light from a liquid pipes:
surface in the Kern DKM2A theodolite gives a very accurate 1. If the pipes go right through the rockfill, a rope can be
measurement of vertical angle. pulled from the other end.
Three-dimensional coordinates should be established from each 2. If the pipes stop in the fill (e.g. in a dam where they meet
reference pillar, using instruments and targets of matching height. the impewious element) the sensing units can be pulled in
Severa1 repeated sets of readings are advisable initially to obtain by a duct motor, operated by compressed air.
accuracy and check repeat sets of readings should be taken from 3. Sensing units may be provided for each marker plate. They
time to time in case there should be any pillar movement. Land can be connected to an interna1 smng of metal rods which
Rovers have been pulled out of ditches from the anchorage of a can be moved a shon distance in and out of the line of pipes
reference pillar! to detect each plate in turn. This greatly simplifies the
A simple computer programme is required to reduce instrument reading process and improves accuracy. A horizontal plate
readings to give three-dimensional coordinates of the target gauge using this method has been described by Penman and
positions. These should be written to accept actual instrument Charles (1982) and is shown by Fig.41.
readings. Two lines of plastic pipes were laid side by side in the Saine
 Rockfill for embankment dams 325

n h a r i a i cable

U
% in. BS pipe (class C) Rigid PVC pipe
(89 mm 0.d.. 77 mm i.dJ
Ovedow orifice \

i Nylon 1 1 tube I/ I
(8 inm o.d., 5.25 mm 1.d.)
F i g 41. H o r i z o n t a l p l a t e g a u g c .

trench, Fig.42, one to take the water overfiow sensing units and the 10.8. Total pressures.
other, fitted with marker plates, the induction coi1 sensing units.
Once installed, there was no need to move the overflow units. Where rockfill is used to support another structure, such as a
Each was connected to a standpipe by 6mm I.D. tubing. The concrete dam or a retaining wall, there may be interest in
induction coils were connected by coaxial sheathed cable to a attempting to measure the pressure of the fill against the smicrure.
switching unit so that they could be connected in turn to an Ths size of pieces in a rockfill preclude the use of a normal
inductance bridge. The smng of rods was moved in and out of the commercial earth pressure cell unless a local zone of small size fill
pipes with a rack and pinion which had a travel of 0.5m, although is included.
only about O.lm movement was needed to define the position of Some success has been achieved with oil-filled cells both
each plate. As the sensing unit moved through the plate, the mounted on the concrete structures and placed in finer material in
indicating meter of the bridge moved as shown by Fig.43. Length the rockfill
readings were taken on an invar steel scale attached to the rack at The important aspect is to compact the poqket of finer fill so that
the 0.5 full deflection of the meter rising and falling. The mid- it is no more compressible than the main body of the rockfill.
position was taken as that plate position. The cells consist essentially of two steel plates partially cut
through and welded together around their edges as indicated by
Fig.44. The space between them is completely filled with oil and
pie-stressed by a small pressure. Oil pressure is measssured by an
eioctncal, pneumatic or hydraulic transducer and total pressure is

-
060

Fig a s sensing u n i t passes through p l a t e .

O i l f i l l e d space

~ l e x i b l éa n n u l u s Tube t o t r a n s d u c e r

F i g 42. H o r i z o n t a l p l a t e g a u g e d u r i n g i n s t a l l a t i o n . F i g 44. O i l - f i l l e d e a r t h p r e s s u r e c e l l .