0% found this document useful (0 votes)

206 views305 pages

STP402-EB.22497 Testing Techniques For Rock Mechanics

Uploaded by

Francisco

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

206 views305 pages

STP402-EB.22497 Testing Techniques For Rock Mechanics

Uploaded by

Francisco

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 305

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015

Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreement. No furth
TESTING TECHNIQUES
FOR ROCK MECHANICS

A symposium
presented at the
Fifth Pacific Area
National Meeting
AMERICAN SOCIETY FOR
TESTING AND MATERIALS
Seattle, Wash., Oct. 31-Nov. 5,1965

ASTM SPECIAL TECHNICAL PUBLICATION NO. 402

List price $14.75; 30 per cent discount to members

published by the
AMERICAN SOCIETY FOR TESTING AND MATERIALS
1916 Race Street, Philadelphia, Pa, 19103

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreement. No further re
© BY AMERICAN SOCIETY FOR TESTING AND MATERIALS 1966
Library of Congress Catalog Card Number: 66-24783

NOTE

The Society is not responsible, as a body,

for the statements and opinions
advanced in this publication.

Printed In Baltimore, Md.

October, 1966

The papers in this volume were presented at three sessions of the

Fifth Pacific Area National Meeting of the Society on November 1 and
2, 1965.
The first session was conerned with an evaluation of laboratory tests
on rock and description of the apparatus used for such testing. The
second session was devoted to description and analysis of in situ tests
on rock formations, and the final session dealt with applications and case
histories.
All sessions were held under the chairmanship of A. I. Johnson, chief,
Hydrologic Laboratory, Water Research Div., U. S. Geological Survey,
Denver, Colo.

Instruments and Apparatus for Soil and Rock Me-

chanics, STP 3 9 2 (1965), $8.50

Soil Exploration, STP 351 (1964), $8.00

Compilation of Procedures for Soil Testing, D-18

(1964), $7.75

Introduction 1
Equipment for Measuring Pore Pressure in Rock Specimens Under Tri-
axial Load—T. L. NEFF 3
Discussion 17
Plate-Load Testing on Rock for Deformation and Strength Properties—
D. F. COAXES AND M. GYENGE 19
Discussion 36
A Triaxial Pressure Apparatus for Testing of Consolidated or Uncon-
solidated Materials Subjected to Pore Pressure—F. A. DONATH 41
Measurement and Analysis of Rock Physical Properties on the Dez
Project, Iran—R. K. DODDS 52
Techniques Used in Studying the Fracture Mechanics of Rock—B. W.
PAULDING, JR 73
Capabilities of Photoelastic Coatings for the Study of Strain in Rocks—
H. J. PINCUS 87
Discussion 103
Calculation of the Average Ground-Stress Components from Measure-
ments of the Diametral Deformation of a Drill Hole—L. A.
PANEK 106
Deformation Moduli of Rocks—G. B. CLARK 133
Discussion 173
The Role of Microstructure in the Physical Properties of Rock—j. R.
MCWILHAMS 175
Influence of Rock Anisotropy and Time-Dependent Deformation on the
Stress-Relief and High-Modulus Inclusion Techniques of In Situ
Stress Determination—D. S. BERRY AND CHARLES FAIRHURST 190
Laboratory Testing of Rock Strength—B. B. MAZANTI AND G. F .
SOWERS 207
Discussion 228
A Loading System for the Investigation of the Inelastic Properties of
Geologic Materials—H. R. HARDY, JR 232
Discussion 265
Foundation Testing Techniques for Arch Dams and Underground
Powerplants—G. B. WALLACE AND O. J. OLSEN 272
Discussion 290

Introduction

In 1964, the scope of ASTM Committee D-18 was broadened to include

the testing of rocks as well as soils, and its name was changed officially
to Committee on Soils and Rocks for Engineering Purposes. Subcom-
mittee 12 on Rock Mechanics was established at the same time. This
symposium represents the first effort of this subcommittee to present a
formal session to: (1) assess the state of the art of rock-mechanics test-
ing, in the field as well as in the laboratory, and application of such
testing to engineering design problems and (2) evaluate the need for
standard methods and the factors that must be considered in developing
the standards. Because the primary function of ASTM subcommittees is
to develop standard test methods, the symposium was designed to provide
the nucleus around which Subcommittee 12 could build its activities in
test standardization. At the present time there are few or no rock testing
standards, so the field is fertile for development.
If proper consideration is to be given to the most economical design
and the safe performance of a construction project, adequate information
on the properties of the subsurface rocks must be available. Correct
evaluation of rock properties frequently requires laboratory and in situ
tests, supplemented with a high degree of experience and judgment. We
have long recognized the need for the "judgment factor in soils testing;"
the "judgment factor in rock testing" may be even more critical. It is
hoped that this symposium will lead not only to development of stand-
ards but to the philosophical framework within which those standards
can operate with confidence.
The scope of Subcommittee 12 is stated as follows: "It shall be the
responsibility of Subcommittee 12 to develop or to stimulate develop-
ment in other subcommittees of acceptable nomenclature, definitions, and
tests for rocks; for their sampling, identification and classification, and
for the measurement of static and dynamic stress deformation properties
and residual stresses under various environmental conditions of leaching

^ Chief, Hydrologic Laboratory, Water Resources Div., U. S. Geological Survey,

Denver, Colo., and symposium chairman.
^Head, Structural Section, Research Div., U. S. Bureau of Reclamation, Den-
ver, Colo., and chairman of ASTM Subcommittee 12 on Rock Mechanics.

and weathering." Much work is in progress in this subcommittee—^for

example, standard test methods have been drafted for shear strength
and compressive strength of rock materials; a standard engineering
classification of rocks has been distributed to the membership for trial
use and verification; and research is being stimulated on rock reinforce-
ment, in situ stress measurements, and in situ tests for deformation
modulus and shearing resistance. Some of these activities are indicated
in the papers presented in this symposium. Additional papers on new
developments will be presented at the ASTM Annual Meeting, June,
1966, Atlantic City, N.J.
This symposium was the work of many. Acknowledgment is made to
the authors and to those who contributed to the discussions—their names
appear in the body of this volume. Our appreciation also is expressed to
the other members of the Symposium Papers Committee, N. E. Gros-
venor and J. J. Reed. The authors hope that this symposium volume not
only will prove useful to rock mechanics engineers and scientists but
also will stimulate further research and development of testing techniques
and applications of those techniques in the field of rock mechanics.

Equipment for Measuring Pore Pressure

in Rock Specimens Under Triaxial Load

REFERENCE: T. L. Neff, "Equipment for Measuring Pore Pressure in

Rock Specimens Under Triaxial Load," Testing Techniques for Rock
Mechanics, ASTM STP 402, Am. Soc. Testing Mats., 1966, p. 3.

ABSTRACT: This paper deals with equipment development, procedural

details, and pilot tests of larger specimens of intermediate-quality rock,
and rock containing joint systems and weakness planes. The effect of
pore-pressure buildup during loading and its graphical representation, that
is, the vector curve as employed in soil mechanics, is considered. Work to
date indicates that in an undrained triaxial test the pore pressure rises to
a peak value at "yielding" of the specimen and then falls off to some
lower value at failure. Peak values and final values vary widely for dif-
ferent rock types. The vector curve depicts quite clearly the results of
the undrained triaxial test.

KEY WORDS: rock mechanics, triaxial test, pore pressure, undrained

test

Although rock mechanics is a relatively new science, present research

activities encompass a wide range of topics. One that has been the sub-
ject of considerable investigation is the effect of interstitial fluid pressure
on the strength and deformation characteristics of rock. In 1945, Ter-
zaghi [I]2 first discussed the stress conditions for failure in water-satu-
rated concrete and rock. He stressed the need to study pore pressures, or
more accurately, triaxial tests with controlled interstitial fluid pressures.
He believed that, until this type of test could be run, the stress conditions
for failure of rock at depth would remain a matter of speculation. This
challenge was eagerly accepted by many investigators, and some out-
standing work has resulted. Robinson [2] and Handin [3] have evaluated
the effects of pore pressures on sedimentary rocks. Some interesting
work, using pore fluids other than water, has been done by Boozer et al
[4]. They also considered strain rate and temperature as variables.
1
Research engineer, Missouri River Division Laboratory, U. S. Army Corps
of Engrs., Omaha, Nebr.
2
The italic numbers in brackets refer to the list of references appended to this
paper.
3

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
4 TESTING TECHNIQUES FOR ROCK MECHANICS

The work of these and other investigators generally agrees that for
certain rocks, most notably the relatively porous types, the effective
stress theory holds. The effective stress theory states simply that the dif-
ference between the external pressure, a, and the pore pressure, u,
called the "effective stress," c, is the controlling factor influencing fric-
tional strength of the rock, other parameters remaining constant.
Assuming the Mohr-Coulomb failure criterion applies, the shear
strength, r, is:
T = C + a taa<t>
where:
<j> = true angle of internal friction,
C = cohesion or "no-load" shear strength, and
a — (T — U.

For a significant pore pressure increase to occur as the external stress

is increased, the material (such as soil, rock, or concrete) should have
two basic characteristics:
1. A reasonably interconnected system of pores between the crystals.
2. Pores filled with a fluid (liquid, gas, or both) that is less compressi-
ble than the surrounding crystal structure.
These and other factors are well summarized by Lambe [5,6]. Robin-
son and Handin have demonstrated that moderately porous sandstone
and limestone possess these characteristics and develop high pore pres-
sures when saturated.
The presence of pore pressure in very dense rock, obsidian, to choose
an extreme case, is thought to be negligible or nonexistent. It is possible
that fluid pressure in joints may be highly significant, even though the
rock itself may be so dense that little or no pore pressure is developed
within the crystal structure. There is no general agreement as to where
the division lies between rocks that are affected by pore pressure and
rocks that are not.
With few exceptions, the above mentioned work has been performed
on homogeneous unfractured rock. The specimens have been almost uni-
versally small, usually Vi to 1 in. in diameter and 1 to IVi in. in height.
The small size of the test specimens seems to have been dictated by the
loading capacity, o-i, of the testing machine as compared to that required
to cause failure under confining pressures, as, up to 30,000 psi. Also,
the pore pressure was generally held constant throughout the test, and
the resulting data were displayed as Mohr circles and rupture envelopes
using effective stresses.
Initially, in 1963, triaxial rock testing at the Corps of Engineers Mis-
souri River Division Laboratory was limited to undrained specimens
with no provision for measuring pore pressure. By late 1964, certain
findings and observations [7] made apparent the necessity to include pore

pressure measurements in order to provide a comprehensive picture of

the rock strength.
Further, since large masses of rock invariably contain planes of weak-
ness, certain test programs were concentrated on intermediate-strength
materials containing joints and seams. The NX (iVa-m. diameter) size
core was therefore adopted. Equipment which would accomodate cylin-
drical specimens of this size was then developed to induce and measure
pore pressures in a triaxial compression test. It was decided that the

FIG. 1—Major parts of triaxial apparatus. Left to right: top, base with speci-
men in place, and cylindrical chamber.

pore pressure would be allowed to build up during the test, that is, an
undrained test, and an informative method of reporting the data was in-
vestigated. The work raised further interesting questions that suggest
extensive future studies, so the present paper is in the nature of a progress
report, covering development of equipment and techniques, plus early
results.

Equipment
The decision to use NX size specimens involved an increase in loading
area of approximately 1:8 and a corresponding increase in specimen
volume of approximately 1:25 over the size of specimen employed by
earlier investigators. Since ample testing machine capacity was available.

loading posed no real problem; however, complete saturation of the

larger core proved a more formidable task.
Initially, specimens were saturated by simple immersion in water under
pressure. An improved procedure has been developed wherein vacuum
saturation is employed. This provides for the introduction of water into
the previously evacuated specimen within a vacuum chamber. (Water
was the only pore fluid used in this investigation.) This method is fre-
quendy used by current investigators; however, there are no set time re-

\.w^'
FIG. 2—Assembled triaxial apparatus with axial strain measuring device
(linear variable differential transducer).

quirements for application of vacuum or allowing water to permeate the

specimen. For very porous rocks one to two hours will suffice, while for
a more dense specimen a period of several days is necessary. Saturated
specimens are stored in water until time of test.
Photographs of the triaxial apparatus now in use are shown in Figs.
1 and 2. Figure 3 is a schematic sectional drawing of the triaxial com-
pression chamber. This consists of a base, cylindrical triaxial chamber,
top, and accompanying hoses, gages, and accessories. The base, top, and
two bearing platens are all hardened to Rockwell hardness C 58. The
platens have five ports to diffuse pore water into the specimen. The
original apparatus was designed to operate at confining or pore pressures

-To Pore Pressure

Oil Overflow
Source

Packing ; Teflon V-Ring

Packing Gland

'O" Rings

Hydraulic Oil
— Rubber Membrane

Spherical Head

Rock Core
Cylinder

'O' Rings

Packing Gland

Packing; Teflon V-RIng

Hydraulic Oil Under

To Pore Pressure Pressure
Source

FIG. 3—Schematic sectional drawing of tlie iriaxial compression chamber with

pore pressure measuring capability.

up to 3000 psi. A thick-wall chamber has since been constructed in-

creasing the capacity to 10,000 psi. All other parts and appurtenances
were initially capable of high-pressure operation.
To achieve high pore pressures with relatively low pressure sources
(usually a nitrogen bottle), a pressure intensifier similar to that described
by Obert [8] is used. The confining pressure is held constant by hand
manipulation of a positive displacement valve and read on a pressure
gage. The pore pressure can be read on a gage or a pressure transducer
connected to a strain indicator. The transducer increases the accuracy
greatly and to the point where it is limited only by the speed with which
the indicator can be read.
Axial strain during the test is measured by a linear variable differential

transducer (LVTD) which feeds to an autographic recorder coupled to

the testing machine. Corrections are made for apparatus deformation in
order to determine the true strain and strain rate of the rock specimen
during test.

Specimens
The specimens used in this work are right cylinders, the sides of which
are generally smooth and free of abrupt irregularities with all elements
straight to within 0.005 in. over the full length. The ends of the specimen
are cut parallel to each other and at right angles to the longitudinal axis.
They are ground and lapped until flat within 0.0005 in., keeping them
within 0.5 deg of perpendicularity to the core axis. Finished specimens
have a length-to-diameter ratio (L/D) of 2.3 ± 0.025.
The diameter of the specimens is determined to the nearest 0.01 in.
by averaging two diameters measured at right angles to each other at
about midheight. This average diameter is used for calculating the cross-
sectional area. The height of the specimen is determined to the nearest
0.01 in. by averaging each of three measurements made at one-third
points around the circumference of the specimen. Small cavities that might
occur in the sides of the specimen which could cause puncture of the
membrane are filled with gypsum cement or a similar material.

Procedure
The base, with lower platen in position, is set on the table of the
testing machine, and hydraulic lines are connected for confining and
pore fluids. The pore water system is then filled to the top of the platen,
and the specimen (dripping wet) is put in position. The top platen and
spherical seat are carefully aligned before the assembly (specimen and
end caps) is enclosed in an impermeable rubber membrane. The mem-
brane is secured at the top and bottom with rubber 0-rings. Assembly
is finally completed by installation of the cylindrical chamber and top.
The apparatus is centered on the testing machine table, and final
hydraulic connections are made. The pore water path from the top of the
specimen, through the top platen and spherical seat, to the outside of
the chamber is first evacuated and then filled with water. This provides
a continuous system from the entrance in the base, through the specimen,
and out to the top pressure measuring device. There is no pressure ap-
plied to the system at this point.
The head of the testing machine is then lowered and a slight (50 to
100 lb) load applied to the specimen to ensure that the top is not lifted
during introduction of hydraulic fluid to the chamber. The hydraulic
system is then purged of air.
The slight load that is on the specimen is maintained, in addition to
balancing the upward force caused by the confining pressure. As the

confining pressure is raised to tlie working value, sufficient load is ap-

plied through the testing machine to prevent deforming the specimen.
Thus, when the working confining pressure is reached, all apparatus parts
are still in contact, and the specimen has little or no strain other than that
imposed by the confining pressure.
Pore pressure is then induced from both ends of the specimen. It is
permitted to back-pressure momentarily, and then the system is closed
off from the pressure source. Back-pressure, which is the application of
the pressure source to the specimen, here is similar to the technique in-
troduced by Lowe [9] for the purpose of driving any remaining air into
solution, thus increasing saturation. If the pressure remains constant at
both gages, the specimen is considered to be saturated. The condition of
saturation is confirmed by slightly increasing the pore pressure at the
bottom and noting if an equal response (increase) is shown at the top of
the specimen.
The specimen is loaded at a constant rate of strain until a definite
failure occurs, that is, to the peak of the stress-strain curve. In the case
of specimens which bulge or deform considerably or slide progressively
along a shear plane, failure is considered as some predetermined amount
of strain. Since an autographic plot of load versus total deformation is
recorded, it is necessary only to observe pore-pressure values at the top
and bottom gage locations for specific load increments throughout the
test.

Results
As stated earlier, the intent of this paper is to show the type of equip-
ment developed, outline the test procedure, present certain selected re-
sults and tendencies, and discuss certain problems encountered in order
to provide guidelines for future work directed toward ultimately estab-
lishing standards for testing in rock mechanics. In the initial stage of the
program, specimens were tested using only a pore pressure measuring
device at the bottom of the specimen. Although these behaved in a rela-
tively normal manner, it was believed that they were not completely
saturated, and therefore the data were considered somewhat questionable.
The specimens consisted of dolomite breccia, sugary dolomite, sandstone,
and porous limestone, all containing random shale seams, stylolites, or
healed fractures at various angles. Saturation was assumed incomplete
for those specimens that showed no increase or a decrease in pore pres-
sure during the test.
When the double measuring system was installed, it was found that
the pressure rise was unequal at the top and bottom. This was also at-
tributed to incomplete saturation and to impermeable layers within the
specimen. These difficulties were eliminated with a combination of initial
vacuum saturation, by subsequent back-pressure saturation, and by the

use of top and bottom gages. Testing continued on specimens of the rock
types described above with better results, that is, pore pressure increases
were registered at both top and bottom locations, although occasionally
these readings differed. This difference was generally slight and is at-
tributed to variation in response time at the two ends of the specimen.
Possibly, more attention should be given to the determination of in situ
moisture conditions and their preservation or accurate laboratory re-
production, and rate of loading versus expected pore-pressure dissipa-
tion rate.
All of these tests were run at varying levels of confining and pore pres-

16
i
1

^- 4^
14
1
tfl
•J 12 -
t>" 10 • •Inducec Por<i Pre ssure 3°
o = Dsviator Str ess 3
b"
^ 8 --4-,--
(O CO
a>
W3 6 .—l , Q_
a>

1 4 1.0 1
> —^ - — 1 j=-i3
^ - O.b 1
-
Q 2
i 0
X)

a
0 l_ t ^
C) 2-0.5
0 2 0 4 0 6 0 8 10 1
Axi a) St rain, e, pe rcent

FIG. 4—Plot of deviator stress versus induced pore pressure and axial strain.
Test performed on an undrained saturated specimen of St. Peter sandstone.

sure up to 3000 psi, although higher pressures are planned as a con-

tinuation of this program. The pore pressure was always kept below the
confining pressure so that the increase during loading left the effective
confining pressure, as , a positive value. A major factor determining
whether pore pressure increases or decreases during loading is whether
the specimen compresses or expands. An increase of pore pressure is
indicative of a compression of the crystal structure; a decrease is indica-
tive that the crystal structure is dilating or expanding. The latter condi-
tion is most likely to occur during the advanced stages of failure. In a
normal test, the pore pressure rises to some peak value and then starts to
drop off, as the specimen begins to "yield," to a lower value at failure.
This relation can be readily seen on a plot of deviator stress versus strain
and induced pore pressure versus strain as shown in Fig. 4. The pore

!^ 4 —

4 6 8
Normal Stress ( a), ksi.

FIG. 5—Mohr envelope for drained saturated specimens of St. Peter sandstone.
No pore pressures induced or recorded.
10

0 = Failu re Points
40 1 = 3000
\ / u = 700

L
/ ^
.- 6 / - a , = 30C)0 1
/ u = 1500 i

^^-\^ \ " " ' • ^

0,= 2000
u=JOI5
-CT = 3000
U = 2000
m 2-
a = 2000
./ i U = 1000 \

\ \
0 2 4 6 8 10 12 14
Effective Normal Stress (5) , ksi.
F I G . 6—Mohr envelope for undrained saturated specimens of St. Peter sand-
stone. Chamber pressure, 0-3, shown was held constant throughout the test. Pore
pressure, u, shown is value at failure.

pressure at failure can be above or below the initial value depending on

the type of failure mechanism in the specimen.
For tests using top and bottom gages (approximately 35 tests), the
induced pore pressure (average of top and bottom gage readings) from

the initial value to the peak value, varied from 25 to nearly 750 psi. The
pore-pressure readings at failure had a range of 480 psi above the
original value to 600 psi below the original value. While the pore-pressure
values at failure were somewhat inconsistent, the tendency was to rise to
some peak value and then descend to a lower value as failure was
reached.
The Mohr diagram can be used to plot the resulting data from the un-
drained triaxial test. The circles are drawn with effective stresses at failure

Confining Press.= O,- 3 0 0 0 psi

Pore Press.= \i„ = 1000 psi.

en Failure Point
10

,69<
gl-« 3
V^
a _- Induced Pore Pressure
Q. S

«> ^
tf) o n
gj X ^
^ e 69"'{0bserved Failure Z )
p
1, ' I •

0 1 2 3 4 5 6
Effective Normal Stress On Plane Inclined 69°
From H o r i z o n t a l , ffgg , ksi.

FIG. 7—Vector curve for an undrained saturated specimen of St. Peter sand-
stone.

which is normally taken as the peak in the stress-strain curve. A failure

envelope is drawn using several Mohr circles. It should be noted that,
although tests of this kind are now being conducted using a constant
strain rate, all of the work discussed in this paper was performed using
a constant load rate. The constant load rate causes very rapid strains
near failure with resultant rapid pore-pressure changes. This makes it
difficult to accurately pick the pore-pressure value at failure, and an
error here will show up in the Mohr diagram.
Figure 5 is a Mohr envelope for four drained triaxial tests on saturated
specimens of St. Peter sandstone. In a drained test, the pore pressure is
held at zero, and any tendency toward induced pore pressure is per-
mitted to dissipate by drainage through the top and bottom of the speci-

men. Figure 6 is a similar diagram for five undrained triaxial tests on the
same material. In an undrained test the system is closed, that is, no
drainage is permitted, and induced pore pressures are measured. The
pore pressures shown are the pore pressures at failure. Despite some
scatter, the two failure lines agree in slope. This condition must hold if
a material follows the effective stress theory. The points indicated on the
Mohr circles as failure points are used only when a definite failure plane
occurs. The shear and normal stresses on this plane can then be cal-
culated and the actual failure point located in the proper position on the

— —
Failu re Po nt —
u -

j
y r -4--
Failure Point
I '/ ^/\ 61°
3 -
;
(Observed Failure Z )
-/—L / y
40° /

/
^ y^"—^
'
// 1

i'/
//
/ 7 fw p T

67°
\

(
0 1 2 3 4 5 6
Effective Normal Stress{a), ksi.

F I G . 8—Failure line derived from two vector curves. Tests performed on un-
drained saturated specimens of St. Peter sandstone.

Mohr circle. These points are then used to draw a failure or rupture
line, as shown by Lane and Heck [10]. If there is no failure plane pro-
duced by the test, the rupture line is drawn tangent to the Mohr circles.
A vector curve representation can also be used to show the actual pore-
pressure buildup during loading. As introduced by Casagrande and
Hirschfeld [77], "the vector curve is the locus of points on the Mohr
diagram whose coordinates are the shear stress and the effective normal
stress, on the plane on which failure ultimately takes place, plotted for
the entire axial load increase in a triaxial test." If a definite failure plane
occurs, it is possible to show graphically the actual pore-pressure buildup
throughout the test as depicted in Fig. 7. To do this a straight line, in-
clined at an angle equal to the observed failure angle in the specimen, is
drawn starting from the initial point of effective stress, o-, and extending

upward. The deviation of the vector curve from the straight line is the
induced pore pressure—negative if it is to the right of the line and posi-
tive if it is to the left of the line. A failure plane can be assumed, if none
is evident, to get some feel for the pore pressure rise and fall. When
two or more vector curves are plotted on the same sheet, as in Fig. 8,
it is possible to determine the angle of internal friction and estimate the
cohesion intercept. The curves shown are typical for the St. Peter
sandstone tested.

Discussion of Results
Work to date seems to indicate that pore-pressure buildup can be
reliably measured in undrained triaxial compression tests of saturated
rock core specimens under controlled rates of strain. The use of suitable
pressure transducers, rather than pressure gages, has improved con-
siderably the accuracy of pore-pressure readings. The question still
remains, however, whether or not the pore pressure observed is actually
the pore pressure on the failure plane. This may be so for the porous
rocks, but there is some question for dense specimens and specimens con-
taining impermeable layers or seams. Future investigations are planned
that will study this problem more closely.
The use of a constant rate of strain rather than a constant rate of load-
ing appears to be a more reliable method of conducting these tests. This
procedure provides more reproducible results, especially near failure. It
is generally considered that a rather broad range of strain rates can be
used before significant strength changes occur in the specimen, provided
the test time is at least 2 min [72]. In these tests, the strain rate used was
such that failure occurred within 15 to 20 min.
The primary reason specimens are tested in the undrained state is to
study the pore-pressure buildup and thus obtain additional information
concerning the characteristics of the rock. If the pore and confining pres-
sures are both held constant, only the strength of the specimen at a
particular effective confining pressure is determined. If the rock is homo-
geneous, it would seem that the pore-pressure buildup, assuming the test
was started with a fully saturated specimen, depends on the porosity,
permeability, ratio of the compressibility of the pore water and rock
structure, and time rate of loading. For a laminated specimen, that is, one
containing shale seams, etc., a porosity number has little meaning, and,
in tests of these materials, the pore pressures measured at the ends could
be entirely misleading.
Should the specimen not be 100 per cent saturated, the problem of
analyzing the various parameters affecting pore-pressure buildup be-
comes even more involved. An analysis using the pore-pressure coeffi-
cients A and B, introduced by Skempton [13], could be informative. If
the voids are sufficiently large and there is a high degree of interconnec-

tibility, a reasonable load rate should not affect the pore-pressure buildup,
that is, instant response. Should there not be instant response, then the
load rate becomes important and another variable in an already com-
plicated problem. However, assuming all factors regarding the test are
held constant, the change in pore pressure can be assumed to be a func-
tion of the two coefficients, A and B.
It has been found convenient to express the pore-pressure change, Afx,
which occurs under changes in the principal stresses, A<TI , and Ao-a, by
Skempton's equation:
AM = B[A<T3 + AiAai - Aa-i)]
The B coefficient, assuming 100 per cent saturation, is very nearly one
(1.0), but only if the ratio of the compressibility of the pore fluid to the
rock structure approaches zero, that is, the fluid is far less compressible
than the crystal structure. There is some question as to whether this is a
valid assumption for the harder rocks, as water is actually slightly com-
pressible and in very strong rock could be relatively close to the com-
pressibility of the crystal structure. The A coefficient is dependent on the
relative deviation of the rock behavior from the elastic theory, which for
some specimens is considerable. Studies into relative compressibility of
rock structure and pore fluid plus an evaluation of these two coefficients
appear desirable.
It is considered that the vector curve presents the data from an un-
drained triaxial compression test in an informative manner. It is possible
to see the following parameters on a single plot: cohesion intercept, angle
of internal friction, shear strength and effective normal stress throughout
the test, and in addition, the pore-pressure buildup at all stress levels. As
used in soil mechanics, the vector curve furnishes pore pressure char-
acteristics of a particular soil specimen. The slope of the vector curve,
only one parameter, indicates the relationship of the change in pore
pressure versus the change in deviator stress. Whether similar types of
curve analysis can be applied to the plots from rock tests remains to be
seen. It is believed that if additional investigative tests are made with all
pertinent variables controlled, some significance can be assigned to the
shape of the vector curves.

Conclusions
1. The use of constant effective confining pressure in a drained triaxial
compression test does not fully reveal the significant parameters affect-
ing the strength of a rock specimen. It would seem that actual pore-
pressure characteristics can be better determined from undrained tests in
which the pore-pressure buildup is recorded.
2. In future tests of the type discussed, it may be desirable to pre-load
the specimen with the estimated overburden pressure before making

observations of the pore-pressure buildup. This procedure is an attempt

to more closely simulate in situ conditions, although it is realized that
once a core is stress relieved, that is, drilled and removed from its in situ
stress conditions, it probably can never be restored to its exact original
state.
3. The vector curve can be used to compare pore-pressure character-
istics of rock in addition to providing all the information shown by the
Mohr diagram. Continued work is indicated to determine how valuable
this comparison can be and also what additional interpretations, or char-
acteristics, may be derived from the individual curves.

Acknowledgment
The tests described in this paper were performed at the Corps of
Engineers Missouri River Division Laboratory, Omaha, Nebr., under the
supervision of E. J. Deklotz, W. J. Heck, and the author. L. A. Brown
was director of the laboratory during this period. The author also ex-
presses his appreciation to K. S. Lane, L. B. Underwood, and O. A.
Stemler of the Missouri River Division, and J. L. Rosenblad of the Uni-
versity of Illinois, all of whom provided helpful advice and comments
during the investigation.

References
[/] Karl Terzaghi, "Stress Conditions for the Failure of Saturated Concrete and
Rock," Proceedings, Am. Soc. Testing Mats. Vol 45, 1945, p. 777-801.
[2] H. L. Robinson, "The Effect of Pore and Confining Pressure on the Failure
Process in Sedimentary Rock," Quarterly, Colorado School of Mines, 3rd
Symposium on Rock Mechanics, Vol 54, No. 3, July, 1959, p. 177.
[3] John Handin, "Experimental Deformation of Sedimentary Rocks Under Con-
fining Pressure: Pore Pressure Tests," Bulletin, Am. Association of Petroleum
Geologists, Vol 47, No. 5, May, 1963, p. 717.
[4] G. D. Boozer, K. H. Hiller, and S. Serdengecti, "Effects of Pore Fluids on the
Deformation Behaviour of Rock Subject to Triaxial Compression," Proceed-
ings, 5th Symposium on Rock Mechanics, University of Minnesota, The Mac-
millan Co., New York, May, 1963, p. 579.
[5] W. T. Lambe, "Pore Pressures in a Foundation Clay," Journal, Soil Me-
chanics and Foundations Div., Am. Society Civil Engrs., Vol 88, No. SM2,
Proceedings Paper 3097, April, 1962, p. 19.
[5] W. T. Lambe, "Methods of Estimating Settlement," Journal, Soil Mechanics
and Foundations Div., Am. Society Civil Engrs., Vol 90, No. SM5, Part 1,
September, 1964, p. 43.
[7] U. S. Army Engineer Division Laboratory, Missouri River, "Tests for Strength
Characteristics of a Schistose Gneiss," MRD Laboratory Report No. 64/126,
Omaha, Nebr., May 1965, p. 16.
[S] Leonard Obert, "An Inexpensive Triaxial Apparatus for Testing Mine Rock,"
Bureau of Mines Publication, U.S. Department of the Interior, 1963, p. 9.
[9] John Lowe, III, "Use of Back-Pressure to Increase Degree of Saturation of
Triaxial Test Specimens," Proceedings, Research Conference on Shear
Strength of Cohesive Soils, University of Colorado, Boulder, Colo., June,
1960, p. 819.
[10] K. S. Lane and W. J. Heck, "Triaxial Testing for Strength of Rock Joints,"
Proceedings, 6th Symposium on Rock Mechanics, University of Missouri,
RoUa, Mo., October, 1964, p. 98.

[11] A. Casagrande and R. C. Hirschfeld, "Investigation of Stress-Deformation

and Strength Characteristics of Compacted Clays," Harvard Soil Mechanics
Series No. 61, May, 1960, p. 36.
[12] S. Serdengecti and G. D. Boozer, "The Effects of Strain Rate and Tempera-
ture on the Behavior of Rocks Subjected to Triaxial Compression," Proceed-
ings, 4th Symposium on Rock Mechanics, Penn State University, University
Park, Pa., 1961, p. 83.
[13] A. W. Skempton, "The Pore-Pressure Coefficients A and B," Geotechnique,
December, 1954, p. 143.

DISCUSSION

B. W. Paulding, Jr}—The author's studies of the changes in pore

pressure during the deformation of rock are particularly significant and
timely. Previous work by myself^ and others^- *• ^ suggest a phenomenon
which the author may wish to consider. This is the observation that,
during uniaxial and confined compression tests on several common rock
types, crack growth and an associated increase in volume begins at about
half the strength and continues to fracture. The onset of crack growth
and the associated volumetric strain can be determined by plots of
volumetric strain versus deviatoric stress. As the author mentioned, the
decrease of pore pressure from the maximum value is probably due to
the dilatency of the material. This hypothesis could be tested by plotting
both volumetric strain and pore pressure versus some other variable, say
axial stress, axial strain, or deviatoric stress.
One question: Would the author explain the existence of the negative
pore pressure of some 350 psi at fracture as indicated in Fig. 7 of the
paper?
T. L. Neff (author)—I wish to thank Dr. Paulding for his pertinent
remarks. We are aware of the phenomenon that you suggest, namely, the
onset of crack growth and associated increase in volumetric strain at ap-
proximately 50 per cent of ultimate strength. We have a study in progress

' Associate research engineer, IIT Research Institute, Chicago, 111.

' B. W. Paulding, "Crack Growth During Brittle Fracture in Compression,"
Ph.D. thesis, Massachusetts Institute of Technology, Cambridge, Mass., 1965.
' P. W. Bridgman, "Volume Changes in the Plastic Stages of Simple Com-
pression," Journal Applied Physics, Vol 20, 1949, pp. 1241-1251.
* Shogo Matsushima, "On the Flow and Fracture of Igneous Rocks.," Journal
of Physics of the Earth, Vol 9, 1961, pp. 2-9.
^E. C. Robertson, 1958, "Creep of Solenhofen Limestone Under Moderate
Hydrostatic Pressure," Geological Society America Memoir, Vol 79, pp. 227-245.

using saturated specimens of Berea sandstone with attached axial and

diametric strain gages that will help us better understand this phenom-
enon. The undrained triaxial test, with recorded pore pressures, will
enable us to plot relationships, such as deviator stress versus volumetric
strain, deviator stress versus load-induced pore pressure, and any others
that may pertain to the failure mechanism.
Regarding the question concerning the existence of a negative pore
pressure of 350 psi at failure as shown in Fig. 7 of the paper, I offer
this explanation. This straight line drawn from the initial effective stress
point (on Fig. 7, aw = 2 ksi, reg = 0), inclined at the angle of the ob-
served failure plane, represents the stress on that plane if no pore pres-
sure increase or decrease occurs during loading. The initial conditions
for that particular test were, o-g = 3 ksi, fiy = 1 ksi, and therefore, ag =
2 ksi. The pore pressure at failure then, lying to the right of the inclined
straight line and indicating a drop in pore pressure from the original
value, represents a pore pressure of approximately 750 psi (1000 — 350),
and not a negative value of 350 psi.

Plate-Load Testing on Rock for

Deformation and Strength Properties

REFERENCE: D. F. Coates and M. Gyenge, "Plate-Load Testing on

Rock for Deformation and Strength Properties," Testing Techniques for
Rock Mechanics, ASTM STP 402, Am. Soc. Testing Mats., 1966, p. 19.

ABSTRACT: Plate-load tests were conducted underground on rock to

determine in situ strength and deformation properties. The plate-load
test provides information on the rock mass properties as opposed to the
strength of the rock substance. When a plate is placed on the surface of
the material to be tested and the contact pressure increased, the plate
deflects as the material deforms, and shear failure of the material ulti-
mately occurs.
Tests were conducted on three different materials: iron ore, paint rock,
and ash rock. The results were compared, in some cases, with the results
of laboratory testing and, in other cases, with the results of analyzing fail-
ures of the rock mass. Moderately good agreement was obtained between
these independent methods of determining the material properties. How-
ever, the principal aspect that emerges is the dispersion of strength values
that must be expected in testing geological materials.
Suggested specifications are given for plate-load testing, conventional
uniaxial compression testing, and classification uniaxial compression test-
ing.

KEY WORDS: plate-load tests, rock mechanics, rock (material), strength,

deformation, shear strength, bearing strength, brittle fracture, specifica-
tions

The mechanical properties of a rock mass depend on the nature of the

rock substance, stratigraphy, and structural features. Laboratory tests
are generally conducted on specimens of the rock substance, which
therefore give no information on the effects of stratigraphy or structural
features. Special laboratory studies have been made in the past on these
other factors, but no substantiation has been obtained that the mass
properties can be predicted from such testing. Consequently, the purpose
of the testing described in this paper was to determine the deformation
1
Head, Rock Mechanics Laboratory, Fuels and Mining Practice Div., Mines
Branch, Department of Mines and Technical Surveys, Ottawa, Canada.
2
Scientific officer, Rock Mechanics Laboratory, Fuels and Mining Practice Div.,
Mines Branch, Department of Mines and Technical Surveys, Ottawa, Canada.
19

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
20 TESTING TECHNIQUES FOR ROCK MECHANICS

and strength properties of the rock mass in various formations in which

instability was being experienced.

Theory

Deformability
Uniform Pressure—The basis of the plate-load test is that if a plate
is placed on an extensive surface, equivalent to a semiinfinite half space,
and the contact pressure is increased, the resulting settlement is a func-
tion of the effective modulus of deformation of the subgrade. The settle-
ment of a bearing area on which a uniformly distributed pressure is act-
ing has been solved in the form of the general Eq 1:
de = qBI/E (1)
where:
dc = deflection of corner of loaded area,
q — uniformly distributed pressure,
B = width of loaded area,
/ = influence value varying with ratio of length to breadth of bearing
area as well as Poisson's ratio of the subgrade, and
E = modulus of deformation of ground.
For a square area the settlement of the center point, d„, is twice that of
the corner settlement.
If Poisson's ratio, jx, is 0.3, the settlement of the center point of a
square bearing area is approximately
rf»= qB/E (2)
For values of Poisson's ratio other than 0.3, the adjustment can be made
by knowing that the settlement varies directly as (1 — /x^).
With the above equations, providing all the appropriate conditions are
fulfilled, it is then possible to determine iJ of a rock mass by measuring
the deflection at either the center or corner of a square area loaded with
uniformly distributed pressure on the rock surface.
Rigid Bearing—^For a rigid foundation resting on a semiinfinite elastic
body, the equation for a rigid circular bearing can be used [7]:^

IRE ^'
where:
Q = load on the foundation and
R = radius of the bearing area.

' The italic numbers in brackets refer to the list of references appended to this
paper.

The contact pressure for this rigid, circular bearing area would be

dv
Q/i^R") (4)
2(1 - {r/RY) 1/2 •

where r = distance from center of bearing area.

A similar equation for a long, rigid foundation of width B has been
derived as follows [i]:

''" Tr(0.25 - {x/BYY'^

where x = distance from center of foundation. Figure 1 shows the
pattern of pressure variation.
Using a square bearing area for the prediction of settlement or for

im^' \ \ \ i'^3=0 t**qf

' 1 -^ VVa

(Ty y^ • ^

FIG. I—Wedge analysis of stresses.

the determination of the modulus of deformation of the subgrade, it is

probable that the practical device of replacing R with a B that represents
a square of equal area to the circle would be satisfactory.

Strength
Yielding Ground—An alternate function of the plate-load test is
based on the assumption that, if the contact pressure of a plate on an ex-
tensive surface of a material to be tested is increased, shear failure of the
material will ultimately occur. The results of such tests must then be
interpreted, using a bearing capacity theory to calculate fundamental
strength parameters. With such fundamental strength parameters it is
then assumed that the test data can, in effect, be extrapolated to other
geometry and other loading conditions for the prediction of failure
pressures.
For yielding ground the bearing capacity theory developed and sub-
stantiated in soil mechanics can be used [2]. The bearing pressure at
failure, Qf, is related to the fundamental strength parameters through the
following equation:

q, = Q.SyBN^ + cNc + pNp (6)

where:
7 = density of ground,
B = width of a long footing,
c = cohesion, as in Mohr's strength theory,
p = surcharge pressure on ground surface adjacent to bearing pres-
sure, and
Ny , Nc, Np = bearing capacity factors, functions of internal friction
angle ^ [2].
Equation 6 could be used for a plate-load test which was performed
on a horizontal surface with a long plate. If the test is run on a wall or a
vertical ground surface, then the first and third terms no longer contribute
to the ultimate bearing capacity. In addition, if the plate is square the
coefficient of c must be increased from 1 to 1.3 [2]. Consequentiy, the
equation for a square plate on a vertical surface is as follows:
qf = 1.3 cNc (7)
Brittle Ground—Mohr's strength theory used in the case of yielding
ground might not be applicable to rocks. A theory that may be found to
be a good representation of the strength of brittle rocks is Griffith's
theory [3]. This theory postulates the presence of microscopic cracks
within the material, whose effect is to produce concentrations of stress
around their boundaries.
If the principal stress in the rock is tension and normal to the crack,
then a tensile stress many times this average stress will be created at the
ends of the cracks. If the principal stress is compression, then tension
can be produced in a direction at right angles to the direction of the
compressive strength.
When the stress concentrations are equal to the tensile strength of
the material, cracks will be propagated. As the length of the crack trans-
verse to the field stress increases, the stress concentrations become
greater; consequently, it is visualized that once initiated, the propagation
of the crack will lead to failure of the material.
By assuming that the cracks are elliptical in shape and that they are
randomly oriented, the following criteria for failure have been established
[4]:
C3 = -T, when cri + 3cr3 < 0 (8)

^~ ~ "'^ = 8r, when <TI -^ 3(r3 > 0 (9)

0-1 + 0-3
and

cos2e = ^ / L r ^ (10)
2(ai -h az)

where:
Ts = uniaxial tensile strength of rock substance and
6 = angle between minor principal plane and plane of failure.
By combining Eqs 9 and 10, a failure equation can be obtained for
comparison with Mohr's strength equation:
r/ = 2 iT,a + Ts'^yi' (11)
where:
Tf = shear stress on plane of failure and
<7 = normal stress on plane of failure.
By plotting such an equation, a curve is obtained indicating a much

B
MM

FIG. 2—Stress concentration in brittle rock under rigid bearing.

lower tensile strength than would be deduced from a linear envelope,

such as might be extrapolated from compression tests using Mohr's
strength theory. It also provides an envelope on the compression side
of the origin with a decreasing slope starting at 45 deg at the Y axis.
Both of these aspects are in general agreement with the results of triaxial
testing on rocks.
If a wedge analysis, as shown in Fig. 1, is made using Griffith's strength
theory, an equation can be obtained for the bearing capacity of rock. As
such an equation would only be applicable to brittle and, hence, generally
strong rocks, it has been assumed that the contribution to the resistance
to failure of the force of gravity on the ground itself is negligible. With
this assumption the simple resultant expression for a long load is
qf = 2 4 r . = 3 e « (12)
where Q^ — uniaxial compressive strength of the rock.

The end effects for a square bearing area and the effect of any sur-
charge, p, have not been analyzed.
The above equations are based on the assumption that the strength of
the rock will be mobilized at the same time along the entire failure
surface. When more is known about the failure of brittle materials, it is
probable that we shall find that failure is initiated at a point due to a
concentration of stress and propagates into a progressive failure. Fur-
thermore, although experimental work on glass has substantiated
Griffith's theory very well, insufficient work has been done on rocks to
determine whether this theory would predict compression or bearing
failures.
Rigid Bearing—For tests conducted on brittle rocks it is improbable
that there will be sufficient yielding, so that average stresses along sur-
faces of impending failure can be used in the analysis as opposed to local
stress concentrations which will cause failure of the rock at a point.
For brittle rocks under a rigid foundation the stress concentration, as
shown in Fig. 2, at the edge of the bearing area could be significant.
Using Eq 5, the average bearing pressure at failure would be as follows:
qf = Q/B - ^a.iy/Byi'il - y/BY'^ (13)
where y = distance in from edge of foundation.
From Eq 12 we can postulate that failure will occur when the
stress in the rock, ay, is equal to three times the uniaxial com-
pressive strength. Also, we can assume that towards the edge of the
foundation the lack of confinement together with the high stress level
would cause some plastic reaction in a bearing medium, such as con-
crete, so that the theoretically infinite stress would not occur; con-
sequently, the maximum stresses would be equal to that which would
be calculated at some distance y in from the edge of the foundation.
Using these assumptions Eq 13 is modified:
qf = 37rQ^(y/Byi^{l - y/Byi^ (14)
This equation is based on the concept that failure will be initiated at a
point under the foundation at a distance y in from the edge, and, because
the foundation rock is brittle, a progressive breakdown would then occur.
Conceivably this equation could have the more general form:
qi^KQJB- (15)
where K and n are parameters and possibly constant for certain ranges
of conditions.
The above equations for bearing capacity suggest that the effect of
the width of the bearing area, B, varies with the type of material. For a
yielding rock that approaches the properties of a soil with the bearing
plate on a horizontal surface, Eq 6 shows that the bearing capacity will

increase with the width of the bearing area. If the subgrade material
can be considered as frictionless (for example, a shale with a high posi-
tive pore water pressure might behave this way) or where the bearing
is occurring on a vertical surface, the bearing capacity is independent
of the width of the bearing area. Also, for brittle rocks under a flexible
loading or a uniform bearing pressure, the bearing capacity is likely to
be independent of the width of the foundation. Then for a rigid footing
on a brittle rock the bearing capacity is likely to vary inversely with the
width of the foundation.
Furthermore, there is an additional case that can occur where a rigid

FIG. 3—Plate-load test on the wall of a drift.

footing bears on a hard rock which is overlying a softer stratum [5]. Here
the stress concentrations under the edges of the footing would be even
greater than represented above. In this case, it is probable that the bearing
capacity would vary inversely with the width of the bearing area raised
to some greater power of n as expressed by Eq 15.

The Tests

Method
These plate load tests were conducted on the walls of drifts, as shown
in Fig. 3. A hydraulic prop was used to apply the load on a circular steel
plate. The load was obtained from the calibrated hydraulic gage. Three
%000-in. dial gages were used to measure the deflection of the plate.

The load was applied in increments, and the deflection was read for each
increment of load.
Besides simply applying a load to the plate, certain time limitations
were observed. It is important for static problems to control the rate of
application to ensure that all viscous components of deformation are
obtained for each increment of load. Failure in some materials can be
by continuous plastic flow at a relatively slow rate. In this case, if a fast
rate of load application were used, the test results would overstate the
actual strength of the material. The test specifications that were used are
included in the Appendix.
In addition, one or more load increments were cycled to determine the
TABLE 1—Plate-load lest results on iron ore.
Plate Bearing n^u^^; a T^X^X t..™ Deflection Modulus of
Test No. Diameter, Pressure at '^°^^°^' r „ ° , ' „ y % at Failure, Recovery,'
in. Failure, psi P*' Content, % -^^ p^j

11 6.5 685 7.5 12.2 0.52 20 000

12 5.0 1555 17.1 4.0 0.24 78 200
13 5.0 1580 17.4 3.9 0.40 40 300
14 5.0 1500 16.5 7.9 0.65 33 700
15 4.0 2000 22.0 4.0 0.33 46 400
16 4.0 2400 26.4 5.2 0.12 164 000
17 5.0 1500 16.5 10.2 0.19 58 500

26 5.0 924 10.1 13.7 0.32 38 300

27 5.0 689 7.6 19.7 0.75
28 6.5 851 9.4 12.9 0.60 41800
29 6.5 851 9.4 12.4 0.42 42 700
30 4.0 2155 23.6 5.4 0.35 99 900
31 4.0 1806 19.9 3.9 0.60 43 400
" Cohesion was calculated assuming 0 = 37 deg, obtained from the laboratory
testing of the recompacted ore.
* This modulus was determined from the recovery of deformation during
an unloading cycle using Eq 3.

modulus of deformation from the recovery curve. This modulus is con-

sidered to be a better measure of the rock properties unaffected by any
surface loosening or expansion.
The purpose of testing was to determine the deformation and strength
properties of three sedimentary rock formations of Proterozoic age: an
iron ore, a so-called paint rock, and an ash rock.

Iron Ore Results

Thirteen tests were run in two drifts in iron ore. The structure of the
rock was that of hard grains set in a fine-grained, usually soft matrix. It
could be classified as a weak to strong, plastic, massive, blocky to solid
rock [6]. The maximum particle size varied between % and IVt in.
The results of these tests are shown in Table 1. The mean bearing

pressure at failure was 1423 psi, and the coefficient of variation was 39.0
per cent. Tlie mean modulus of deformation (or recovery) was 3.76 X
10* psi with a coefficient of variation of 63.8 per cent.
The nature of failure was not the same for all the tests. For Test 11
failure was by sudden yielding with circumferential cracks appearing on
the surface of the rock about 1 in. beyond the edge of the plate. For
Test 12 brittle cracking noises preceded the ultimate bearing failure.
These noises started at a bearing pressure of 923 psi, and pieces of rock
started to fly off the surface at pressures greater than 1000 psi with
failure ultimately occurring at 1555 psi. In Tests 13, 14, 16, and 17

'500
y^
fix
f / .E 075

£> i' /
/i
S 050 Bearing Pressure
9Z3-5 psi
«1000 // / / '' 025
Q. / I1 \ 1

Z)
tf)
at
-
/ 11 / \
\
1
10 20
1
30 W 50
Time.min
60 70

// /
c
- 500
01
CD
//
1
/ /
0 20 0-40 060 080 0100 0120 0140 0160
Deformation, inches

FIG. 4—Typical bearing pressure versus deformation and deformation versus

time curves for iron ore.

failure occurred suddenly in a brittle manner without any noise preced-

ing failure. In Test 15 failure was by plastic yielding; however, pieces of
ore started flying off the face at a bearing pressure of 1889 psi. The
other tests had a similar scatter of failure patterns.
Typical bearing pressure versus deformation and deformation versus
time curves for the iron ore are shown in Fig. 4. These show that the rock
exhibited little viscosity but produced considerable plastic, or irrecover-
able, strain.
Field density tests showed that the void ratio of the ore varied between
0.16 and 0.54. Furthermore, it was obvious from experience that the
strength of the ore varied considerably, which, together with the diffi-
culty of sampling a material essentially composed of hard rock in a
relatively soft matrix, militated against laboratory testing which might

have been used for comparing theoretical relations between bearing

pressure at failure and either uniaxial compressive strength or c/^
parameters.
As it was thought that the bearing failures could have been the result
of yielding and shear failures, the results were interpreted, as shown in
Table 1, using Terzaghi's bearing capacity theory. An angle of internal
friction of 37 deg was assumed to be applicable to the ore, as this angle had

03 04 05 0-6 07 08 0-9
Log. Base Diameter, ins.
F I G . 5—Correlation between hearing capacity and diameter of bearing plate.

been obtained from triaxial testing on recompacted specimens at similar

void ratios that were being studied for other problems.
On the other hand, there was evidence that the ground involved in
the bearing failure behaved like a brittle material. By examining the test
results, it can be seen that the bearing capacity has decreased with an
increase in size of plate. This suggests that the material properties should
be evaluated using Eq 15.
By plotting the bearing pressure at failure against the size of plate on
a log-log graph and also, by assuming that the stress concentration
mechanism is valid only when the size of the plate is greater than the

width of two of the component blocks of the ground, then it might be

logical to extrapolate the curve back to, in this case, approximately 2Y2
in., to obtain a measure of the uniaxial compressive strength of the rock
mass.
In Fig. 5 the log-log graph of mean bearing capacity against diameter
of bearing plate is shown for this rock as well as for some tests by others
[5]. The dependence of bearing capacity on diameter suggests that
brittle failure was involved in these tests. Extrapolating back to twice the
average particle size gives a bearing capacity of 5300 psi, which should
represent the strength of material unaffected by stress concentrations.
Using Eq 12, this bearing capacity would be equivalent to a uniaxial
compressive strength of 1770 psi. Owing to the wide variation in test
results on core specimens, no independent check on the validity of these
deductions can be made. This strength seems a little high compared to
calculated pillar stresses of from 450 to 675 psi, which sometimes caused
failure [7].
Also, from Table 1 it can be seen that the modulus of deformation
calculated from the recovery deflection curve varies with the ultimate
bearing capacity. In other words, these tests provide some substantiation
for assuming that the modulus of deformation of rock is an indirect
measure of its strength.

Paint Rock Results

The so-called paint rock is a fine-grained mass of quartz, pyrolusite,

and kaolin with subangular fragments of chert, hematite, and geothite.
It could be classified as a weak, plastic, layered, solid rock [6]. The
formation is generally made up of laminations less than V2 in. thick.
When first examined it was not known whether or not this material
would have strength that was predominantly due to cohesion strength
with little contribution from friction. Consequently, recompacted labora-
tory specimens were used to obtain some measure of the basic strength
parameters. Drained triaxial compression tests were used for this pur-
pose.
Field density tests showed that the void ratio of the undisturbed
rock varied between 0.4 and 0.7. The laboratory specimens were com-
pacted to different void ratios within this range.
The results of one series of tests run on saturated specimens with a
void ratio of 0.56 showed the material to have an effective angle of
internal friction of 36 deg. Another series of tests run on saturated speci-
mens with a void ratio of 0.49 produced an effective angle of internal
friction of 38 deg. Creep tests and tests varying the rate of strain indicated,
contrary to expectations, that the rock properties were not sensitive to
duration of loading.

TABLE l~Plate-load test results on paint rock.

Plate Bearing rnhp«mn » MnkHirp Deflection Modulus of
Test No. Diameter, Pressure at Cohes on, Mo'sture at Failure, Recovery,''
in. Failure, psi P" Content, /„ j^ p^;

18 8.0 560 6.15 10.6 1.06 19 700

19 8.0 93 1.02 14.3 0.86
20 8.0 135 1.48 27.1 0.70
21 8.0 135 1.48 13.1 0.40
22 6.5 175 1.92 25.4 0.20 15 500
23 6.5 108 1.18 13.3 0.22
24 6.5 205 2.25 13.7 0.37 36 500
25 8,0 268 2.94 7.6 0.37 47 500
" Cohesion was calculated assuming ^ = 37 deg, based on laboratory testing
of the recompacted footwall paint roclc.
>
' This modulus was determined from the recovery of deformation during an
unloading cycle using Eq 3.

250 — Z 0 0 •in — j i '

.E
c"
/
.2
O /-I ^ p c
200
Bearing Pressure
s 51-7 psi

S 150
0 6 10 15 20 25 30 35
Ti Tie, minutes

100

010 0-20 030 040 0-50 0-60 0-70 0-90

Deformation, inches

FIG. 6—Typical bearing pressure versus deformation and deformation versus

time curves for paint rock.

In Table 2 the results of eight plate-load tests are shown. Test 19 was
run on an area that contained some cracks in the face. These might have
been influential in producing a low-bearing capacity at this location.
From the moisture content of the material at Tests 20 and 22 and know-
ing the average specific gravity to be 3.00, it can be seen that in these
cases the void ratio could have been greater than 0.8. Relaxation and
expansion of the ground at the sides of the drifts might have occurred,
which would produce higher than normal void ratios and, hence, lower
than normal bearing capacities. The mean bearing pressure at failure

was 210 psi with a coefficient of variation of 65.7 per cent. The mean
modulus of deformation (in recovery) was 1.29 X 10* psi with a coeffi-
cient of variation of 43.3 per cent.
Typical bearing pressure versus deformation and deformation versus
time curves for paint rock are shown in Fig. 6. These show that the rock
exhibited little viscosity but produced considerable plastic, or irrecover-
able, strain.
As there was no visual or audible evidence of brittle failure, it was
assumed that a yielding failure occurred in the paint rock, which would
be most appropriately interpreted using Terzaghi's bearing capacity theory
for soils. Using this theory, Table 2 shows the calculated values of cohe-
sion. These values have an average of about 1.8 psi with a maximum
value of 6.2 psi. For purposes of comparison the results on recompacted

TABLE 3—Plate-load lest results on ash rock.

Plate Bearing r- i. • a HT • * Deflection Modulus of
Test No. Diameter, Pressure at Cohesion, Mo'sture ^t Failure, Recovery,''
in. Failure, psi P^' Content, % j^ p^;

32 4.0 1070 8.3 21.3 0.52

33 5.0 2075 16.0 20.2 0.31 90 300
34 5.0 689 5.3 19.6 0.59
35 5.0 924 7.1 22.1 0.31 67 900
36 4.0 1071 8.3 21.3 0.25 112 000
37 4.0 1254 9.6 23.7 0,34 108 000
" Cohesion was calculated assuming (/> = 40 deg, which was based partially
on judgment and partially on the failure planes obtained in the triaxial tests on
the massive ash rock.
'' This modulus was determined from the recovery of deformation during an
unloading cycle using Eq 3.

laboratory specimens can be used. In these cases the effect of cohesion

was found to vary from 6 to 13 psi. Also, from the results of analyzing
slope failures in this formation [8], an average cohesion of 7.6 psi was
obtained with a coefficient of variation of 22 per cent from nine slides.
From these tests there is a rough correlation between the modulus of
deformation and the bearing pressure at failure for Tests 22, 24, and
25; however, the results of Test 18 are not consistent with this correla-
tion.

Ash Rock Results

Six tests were conducted in a pyroclastic rock of an unusually basic
type, called locally ash rock. Typical specimens of this rock contain dark
green to black, lenticular, aphanatic, serpentinized fragments generally
less than Yz in. in size in a greenish schistose matrix. The rock varies
from a weak, slightly viscous, plastic altered material to a relatively
strong, brittle schistose rock. Tests were conducted in the altered rock.

Table 3 contains the results of the plate-load tests on the ash rock.
Typically, the vertical rock face at failure included several radial cracks
extending out from the plate in the vertical direction, the direction of
schistosity, for a distance of about 5 in. It was difficult to judge whether
failure was by a brittle or yielding mechanism. However, in view of the
high moisture content and, consequently, high void ratios, it is probable
that a yielding failure occurred. The mean bearing pressure at failure was
1181 psi with a coefficient of variation of 36.8 per cent. The mean
modulus of deformation (on recovery) was 3.45 X 10* psi with a co-
efficient of variation of 36.8 per cent.

c 25

1000
^
> o Beoring Preeeure
i. 793-7 DSi
% 10
7 a
5
>>
0 5 10 15
Time, minutes
500

. <
010 0 20 030 0-40 0-50 0-60 0 70 080
Deformation, inches

FIG. 7—Typical bearing pressure versus deformation and deformation versus

time curves for ash rock.

In Fig. 7 bearing pressure versus deformation and deformation versus

time curves are shown for ash rock. These curves show that this rock
has some viscosity, or creep characteristics, and also produces consider-
able plastic strain.
Assuming a yielding failure, the cohesion of the rock has been cal-
culated from the bearing pressure at failure for the various tests. The
average cohesion thus obtained was 9 psi with a maximum value of 16
psi.
Although laboratory tests were conducted on core specimens of the
relatively unaltered rock and also on recompacted pulverized specimens,
it was not considered that the results of these tests should bear any
necessary relationship to those of the plate-load tests.
Slope failures in the exposed formation were analyzed. From the

results of seven slides an average cohesion of 5.3 psi was calculated with
a coefficient of variation of 33 per cent. These slides were also in the
altered material and could have been more affected by exposure than the
material underground at the sites of the drift, thus accounting for the
lower average strength. Cohesion values obtained from the two sources
overlapped to a large extent.
It is interesting to compare the modulus of deformation, as obtained
from the recovery curves on these plate-load tests, with a compilation
that has been made in terms of an alteration index which is obtained by
multiplying the void ratio by 100 and dividing by the specific gravity of
the solid [9]. The alteration index of the ash rock, therefore, would be
about 3.7. Previous tests indicate that for this alteration index a range
in modulus of deformation could be expected of from 80,000 to 110,000
psi [9], which is remarkably close to those obtained from these tests.

Conclusions
1. Some useful corroborative information was obtained on the in situ
strength and deformation properties of the three different rock types sub-
jected to plate-load tests. However, the principal aspect that emerges is
the disposition of strength values which occurs in testing geological
materials.
2. The testing of rocks that are brittle is complicated by the impor-
tance of stress concentrations and the importance of recognising the
mechanics of failure, so that the results can be properly interpreted or
extrapolated through strength parameters to prototype geometry and
loadings. In tests on iron ore, the combination of brittle failure and
stress concentrations under the plate seemingly produced a bearing
capacity that varied inversely with the diameter of the plate, contrary to
what one would expect on yielding materials such as soils.
3. Except for soft rock in tunnels or drifts, plate-load testing is an
expensive method for determining rock mass properties. Large forces
are required for loading, and, on the ground surface, equally large reac-
tions must be supplied similar to those used for pile-load testing.
4. It is probable that the most favorable situation for plate-load test-
ing is when the test requires little extrapolation to the prototype case.
For example, for predicting prop penetration into weak floors under-
ground, the plate-load test has demonstrated its usefulness; similarly, it
could be useful in favorable circumstances for predicting settlement of
foundations.

A cknowledgment
It is a pleasure to acknowledge the important contributions to this
work of J. R. Helliwell and K. L. McRorie of Steep Rock Iron Mines
Ltd. and R. C. Parsons of the Rock Mechanics Laboratory in Ottawa.

APPENDIX

Suggested Test Specifications

Plate-Load Testing
1. Three to six tests should be done at each location; individual tests are
to be separated by five plate diameters or approximately 3 ft center-to-center.
2. Apply a seating load of approximately 500 lb; record the magnitude
of the seating load accurately; allow plate to become stationary.
3. Use load increments of one-fifth or less of the estimated ultimate
capacity; record times and deformations immediately before and after load
application, and immediately before and after load release.
4. Apply three load increments (25, 50, and 75 per cent of the estimated
failure load), maintaining each increment constant and taking deformation
readings every minute until the rate of deflection becomes equal to or less
than 0.001 in./min; then noting deformations at 5-min intervals until the
rate of deformation becomes equal to or less than 0.001 in./5-min; then
noting deformations at 15-min intervals until the rate becomes less than 0.001
in./15-min. This specification can be modified for subsequent tests as a result
of experience with a particular material.
5. After the third load increment, release the load to the seating load, and
record the deflection to the same time rate specification as in Item 4. Reapply
the previous load, and read deformations again to the same rate specification.
6. For special testing repeat Item 5.
7. Load to failure if possible.

Conventional Uniaxial Compression Testing

\. A suite of at least ten specimens of the same rock substance should
be tested to obtain a significant mean and a measure of the dispersion of
strength values.
2. Roller lap the specimens, if necessary, so that the maximum difference
in diameter over the length of the specimen is less than 0.001 in. Lap the
ends of the specimens on a wheel so that they are parallel within 0.001 in.
A standard length-diameter ratio is 2:1, but a ratio down to a minimum of
1:1 is acceptable. After lapping, allow the specimens to dry at room tem-
perature for at least 24 hr.
3. Measure the specimen to 0.001 in. at three points for the lengths and
at three points for the diameter. Weigh specimens to the nearest 0.01 g.
Measure strain either with two strain gages cemented at the midheight of the
specimen and on opposite sides or with a compressometer that measures the
change in length over a 1-in. gage length.
4. Apply the load at a rate of approximately 1000 psi/sec until failure
occurs. Record the maximum load and the duration of the test. Describe
qualitatively the type of failure as indicated by the noise produced, for ex-
ample, very violent, violent, and quiet. Describe the orientation of the frac-
tures, for example, top cone, bottom cone, longitudinal, diagonal, irregular,
along with a description of the fragment size, for example, powdered, highly
fragmented, quarter inch with slivers. Where possible determine the fracture
angle.

Classification Uniaxial Compression Testing

1. Unless otherwise stated the specifications for conventional uniaxial
compression testing apply. The rock substance can be classified with respect
to strength as weak for 2« less than 5000 psi, strong for g„ between 5000 and
25,000 psi, and very strong for g„ greater than 25,000 psi [6].
2. Apply the load in increments equal to approximately 0.25 g u , where
Qu is the assumed uniaxial compressive strength of the rock substance at a
rate of loading of approximately 1000 psi/sec.
3. When the load has been established at the increment value, keep it
constant for 30 min, and record strain readings at 0, 0.5, 1, 2, 5, 10, 20, and
30 min.
4. After maintaining the load increment for 30 min, unload the specimen.
Maintain the specimen at zero stress for 10 min, recording intermediate
strain.
5. Apply an increment of load equal to 0.5 g„ to the specimen and main-
tain for 30 min with the same requirements for strain readings as for the
first increment of load. Cycle the load to zero as for the first increment and
then increase to 0.75 g„ and subsequently to 1.0 (2„ .
6. The prefailure deformation characteristics of the rock substance are
then classified as elastic if the strain rate at a stress of 50 per cent of the
conventional uniaxial compressive strength is less than 2 microstrain per hr,
and viscous if it is greater than this rate. The strain rate is determined by
plotting the strain obtained during the 30 min period at a load of 0.5 Q„
against the logarithm of time and extrapolated to obtain the average rate
applicable for the first hour [6].
7. The failure characteristics of the rock substance are classified as brittle
if less than 25 per cent of the total strain before failure is permanent and plastic
if it is more than this quantity [6].

References
[1] R. J. Roark, Formulas for Stress and Strain, McGraw Hill Book Co., New
York, 1943.
[2] K. Terzaghi, Theoretical Soil Mechanics, John Wiley and Sons, Inc., New
York, 1943.
[3] A. A. GrifBth, Royal Soc, London, Philosophical Transaction A., Vol 221,
1920.
[4] D. P. Clausing, "Comparison of Griffith's Theory with Mohr's Failure Criteria,"
Quarterly, Colorado School of Mines, Golden, Colo., Vol 54, No. 3, 1959.
[5] J. D. Jenkins, "Laboratory and Underground Study of the Bearing Capacity
of Mine Floors," International Conference on Strata Control, 1960.
[6] D. F. Coates, "Classification of Rocks for Rock Mechanics," International
Journal Rock Mechanics and Mining Science, Vol 1, 1964.
[7J D. F. Coates and K. L. McRorie, "Earth Pressure on Multiple Tunnels," Pro-
ceedings, 15th Canadian Soil Mechanics Conference, 1961.
[8] D. F. Coates, K. L. McRorie, and J. B. Stubbins, "Analysis of Pit Slides in
Some Incompetent Rocks," Transactions, Am. Institute Mining, Metallurgical,
and Petroleum Engrs., March, 1963.
[9] J. L. Seraflm, "Rock Mechanics Considerations in the Design of Concrete
Dams," International Conference on Slate of Stress in the Earth's Crust, May,
1963.

DISCUSSION

B. Ladanyi^ and D. Nguyen^.—The authors should be congratulated

for their very interesting analysis of the plate-load test as a means for
determining the deformation and strength properties of the rock mass.
The discussers were particularly interested in the part of the paper
dealing with the problem of the bearing capacity of rocks. They agree
completely with the statement of the authors that "The most favorable
situation for plate-load testing is when the test requires little extrapolation
to the prototype case." On the other hand, if there is no model similitude
between the test and the prototype, the results of the test can be used in
design only if the phenomenon produced is well understood, so that
general strength parameters of the tested material can be properly deter-
mined.
The determination of the strength parameters of the rock from a plate-
loading test will require, therefore, a proper understanding of the mecha-
nism of rock failure under a plate load. While the behavior of soils and
yielding materials under a plate load at the surface is actually rather well
known as a result of a number of experimental and theoretical investiga-
tions carried out during the last 40 years, in brittle materials the same
problem up to now has been relatively little investigated. The analysis
and observations presented by the authors on this subject are, therefore,
particularly welcome.
As the mechanism of failure is concerned, the observations made by
the authors in their tests performed on the walls of drifts are most in-
teresting. The mode of failure observed in different tests was described as
follows:
Test 11—"Sudden yielding with circumferential cracks."
Test 12—"Brittle cracking noises started at a bearing pressure af 923
psi, and pieces of rock started to fly off the surface at pressures greater
than 1000 psi, with failure ultimately occurring at 1555 psi."
Tests 13 to 17—"Failure was by plastic yielding (at 2000 psi); how-
ever, pieces of ore startedflyingoff the face at a bearing pressure of 1889
psi."
In the tests on the ash rock, "at failure several radial cracks [were ob-
served] extending out from the plate."
From these observations, which are in agreement with those made by

' Associate professor, Civil Engineering Dept., Laval University, Quebec,

Canada.
^Graduate student. Department of Mining and Metallurgy, Laval University,
Quebec, Canada.

other investigators in similar tests, it can be concluded that, depending

mostly on the type of rock, a plate-load test can produce: (1) an instanta-
neous brittle failure, (2) a brittle failure preceded by internal fracturing,
or (3) a plastic failure.
In the first case it can be postulated that the failure of the rock will
take place as soon as the condition for brittle failure is satisfied at a
single point. The first crack will propagate immediately and lead to an
instantaneous fracturing of a limited zone beneath and around the plate
accompanied with eventual rockbursting.
In the second case the behavior of rock is similar to that observed
during indentation of brittle materials by a shallow pyramidal indenter.^
It seems that in this case, after starting at a point, the failure englobes
quickly a hemispherical region beneath the plate. In this region the rock is
fractured and has lost most of its cohesion; however, its shearing strength
is still high due to internal friction and high compressive stresses. The
fractured hemispherical zone transmits the pressure radially to the
surrounding material, similarly as in the case of the expansion of a
spherical hole under pressure. Under this radial pressure the rock sur-
rounding the plate will be fractured showing radial cracks at the sur-
face. The radial pressure may eventually lead to rockbursts if the rock
is brittle or to a wedge failure if the rock is more plastic.
The third case of failure is typical for yielding and incompressible
materials but can also be observed as an ultimate state of failure in certain
brittle rocks.
It can be seen, therefore, that no single theory will be able to describe
properly the complete phenomenon of failure under a plate for different
types of rocks. In fact, there are actually three different theories available
which may be found useful in interpreting the failure phenomenon
occurring in the rock under a plate load. The theories are: (1) incipient
failure theory, (2) theory of the expansion of a spherical hole under
pressure, and (3) wedge theory.
The first theory has the object of giving the conditions for the begin-
ning of failure at a single point or in a limited region. One approach to
this problem has been shown by the authors for a rigid plate on a Griffith
material (Eq 14). Another approach will be shown hereafter, for a
flexible load both on a Griffith and on a Coulomb material.
The second theory has already been used in the interpretation of hard-
ness test (see footnote 3) and can be useful for studying the second phase
of failure in Case 2, where a fractured hemispherical zone is formed
beneath the plate before a general failure.
The third theory, which is well known for a Coulomb material in soil

' W . F. Brace, "Behavior of Rock Salt, Limestone, and Anhydrite During

Indentation," Journal of Geophysical Research, Vol 65, No. 6, June, 1960, pp.
1773-1788.

mechanics and which has been applied to a Griffith material by the

authors (Eq 12), is able to describe successfully either the complete
phenomenon of failure in a yielding material or the ultimate state of
failure in brittle and intermediate materials.
From the above considerations it can be concluded that the deter-
mination of strength parameters from plate-load test results may present
considerable difficulties. In order to be able to obtain results close to
reality, the investigator should make correct assumptions on the follow-
ing problems: (1) Which failure criterion is applicable to the rocks
tested? (2) To which phenomenon or to which phase of the total phenome-
non of failure under a plate does the load adopted in the test as the failure
load correspond? Only with this knowledge a correct theory can be
chosen and the strength parameters properly determined. Finally, it may
be interesting to show an alternate approach to the problem of incipient
failure under a plate load.
It is known from a similar analysis in soil mechanics* that the condi-
tion for incipient failure under a uniformly loaded strip can be determined
from stress distribution in an elastic half space according to Boussinesq.
In soils, owing to their yielding character, the approach has not been
very successful. However, for strong and brittle rocks, whose elastic
properties remain practically unchanged up to failure, this approach may
be of more interest.
The concept consists simply in finding the minimum load at which the
failure condition is satisfied in the ground, at least in a single point.
As in the case of a plate-load test the influence of gravity forces on the
stress distribution is negligible, the problem becomes very simplified.
If, for example, a uniformly loaded strip is assumed, the principal stresses
are given by

0-1 = - (i^o + sin ^o)

7r
(16)
o'a = - (l^o - sin i/'o)
where q is the uniform pressure applied on the strip, and i/^o is the angle
between two straight lines drawn from the considered point across both
ends of the strip.
Substituting the stresses according to Eq 16 in the original Griffith's
equation (Eq 10) and differentiating with respect to i/'o , it is found that
the lowest value of the load for which the failure condition is satisfied, at
least in one point, is
qf = l.MQu (17)
where (2« is the uniaxial compressive strength.

*0. K. Frohlich. Druckvertellung im Baugrunde, J. Springer, Berlin, 1934.

If, on the other hand, it is assumed that the modified Griffith failure
theory is valid, which in the most part of compression region does not
differ from the Coulomb theory, it is found from a similar analysis that
the value of q; for incipient failure is not a constant but is a function of
the slope angle <^ of the failure envelope, as expected.
The ratio Qf/Qu is found to vary with <^ from 2.30 at <>/ = 30 deg, over
2.74 at (^ = 40 deg, to 3.43 at <^ = 50 deg.
A similar analysis can be made for a uniform circular loading; how-
ever, in this case the analysis was limited to the points located on the
vertical axis only for which the following simple expressions for principal
stresses are valid:

<ri = ,(.-cos-*^)
(18)
O'S

where \pf,, as before, denotes the angle between two straight lines drawn
from the considered point across two diametrically opposite points on the
edge of the loaded circular surface.
Following the same procedure as before it is found that, in the
vertical axis, the failure will be initiated, when

qf = 2.715 e„ (19)
if the original Griffith theory is assumed. For the modified Griffith theory,
on the other hand, it is found that the ratio qf/Q,,, corresponding to
incipient failure, will have values of 3.12 at 0 = 30 deg, 4.09 at <^ = 40
deg, and 6.25 at <^ = 50 deg.
It is interesting to compare the above values of incipient failure loads
with those corresponding to more advanced phases of failure.
For a Griffith material the above incipient load, qf = 2.18 Qu, should
be compared with the load q, = 3(2,, (Eq 12) obtained by the authors for
a wedge failure. (In fact, a slightly greater value of the load may be ex-
pected in the last case when a kinematically admissible solution will be
found.) For a circular loading the greater incipient failure load of 2.715
Qu suggests that a wedge failure may be attained at about 4 Qu.
For a modified Griffith (or Coulomb) material the ultimate failure load
for a strip can be calculated by using Prandtl's theory. The values found
for the ratio qf/Q^ in this case are as follows:

0 = 3 0 deg 40 deg 50 deg

qf/Qu = 8.7 17.6 48.6

The values should be compared with those obtained above for the

incipient failure in the same material. It will be seen that in such a

material there is a large difference between the incipient and the ultimate
failure loads, respectively. The region of loading between the incipient
and ultimate failure is thought to correspond to the intermediate phase of
failure during which the local fractured zone beneath the plate is acting
on the surrounding material as an expanding spherical bulb.
D. F. Coates and M. Gyenge {authors)—The discussion that has been
submitted by B. Ladanyi and D. Nguyen is a very good review of the
possible mechanisms of rock failure under bearing pressure. Even though
their theoretical comparisons are made with the assumption of a uniform
bearing pressure, the differences between the bearing pressures to pro-
duce the start of failure as opposed to a general and complete failure is
instructive. However, for the majority of cases of bearing pressure on
rock, significantly nonuniform loading probably exists; in other words,
the structure applying the load has some rigidity. Consequenffy, it would
be interesting to have similar analyses made taking into account the rela-
tive stiffness or rigidity of the structure applying the pressure with respect
to the foundation material.

A Triaxial Pressure Apparatus for Testing

of Consolidated or Unconsolidated
Materials Subjected to Pore Pressure

REFERENCE: F. A. Donath, "A Triaxial Pressure Apparatus for Testing

of Consolidated or Unconsolidated Materials Subjected to Fore Pres-
sure," Testing Techniques for Rock Mechanics, ASTM STP 402, Am.
Soc. Testing Mats., 1966, p. 41.

ABSTRACT: To determine the effects of different stress states and pore

pressure on the compaction of sediment and on the deformational
behavior of rock, a triaxial apparatus was developed that permits
control of pore pressure during triaxial testing of materials under high
confining pressure. The apparatus is designed for use with 1-in. diameter
by 2-in. length specimens subjected to confining pressures and pore pres-
sures up to 30,000 psi and axial stress up to 130,000 psi. The axial load
can be held constant or varied continuously during testing. The com-
pact nature of the apparatus permits mobility that makes the apparatus
particularly desirable where field testing is necessary.
KEY WORDS: triaxial test, rock (material), rock mechanics, pore
pressure, compaction, creep testing

Triaxial apparatus for testing at pressures above 10,000 psi are not
common in engineering laboratories, and much of the development of
high-pressure triaxial apparatus has occurred in conjunction with studies
in experimental rock deformation—see Griggs and Miller,2 Handin,3 and
Donath.4 In experimental rock deformation, a cylindrical test specimen
is placed between a piston and anvil, jacketed, and inserted in a pressure
1
Professor of geology, Department of Geology, Columbia University, New York,
N. Y.
2
D. T. Griggs and W. B. Miller, "Deformation of Yule Marble: Part I—
Compression and Extension Experiments on Dry Yule Marble at 10,000 Atmos-
pheres Confining Pressure, Room Temperature," Bulletin, Geological Society
Am., Vol 62, 1951, pp. 853-862.
3
J. Handin, "An Application of High Pressure in Geophysics: Experimental
Rock Deformation," Transactions, Am. Society Mechanical Engrs., Vol 75, 1953,
pp. 315-324.
* F. A. Donath, "Strength Variation and Deformational Behavior in Anisotropic
Rock," State of Stress in the Earth's Crust, Am. Elsevier, New York, 1964, pp.
281-297.
41

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
42 TESTING TECHNIQUES FOR ROCK MECHANICS

FIG. 1—Triaxial pressure apparatus with pore-pressure unit screwed into

place in top of pressure vessel.

vessel. The vessel, in turn, is placed between the platens of a double-

yoke or tie-bar type of press. Radial stresses are produced by pumping
fluid into the vessel, and axial stress is produced by loading the piston with
the press. Most apparatus designed for use with consolidated material
is not intended and, therefore, is not satisfactory for use with uncon-
solidated material.
The apparatus described in this paper consists of a pressure vessel
coupled to a 50-ton ram by means of a steel collar. The collar-coupling
type of press is more compact than the tie-bar type and can be installed
in a small cabinet with casters for easy movement from one location to
another. Because the specimen is inserted through the top of the vessel,
the vessel does not have to be separated from the ram before or after a
test. With tie-bar presses the vessel has to be removed from the press

FIG. 2—Schematic drawing of vessel-press assembly: (1) pressure vessel, (2)

upper piston and seal, (i) specimen, (4) anvil, (5) lower piston, (6) lower seal, (7)
lower retaining plug, (8) ram body, (9) ram piston, (10) cabinet top, {11) retaining
ring, (12) pore-pressure unit, (13) top plug and seal, (14) separator piston.

to permit insertion or removal of specimens. This further requires

breaking pressure Hnes and disconnecting measuring devices, procedures
that are unnecessary with the collar-coupling press.
The complete, assembled apparatus" is shown in Fig. 1. A schematic
drawing of the vessel-press assembly and the apparatus layout are given
as Figs. 2 and 3, respectively. Results of a study on unconsolidated
sediment using the apparatus are reported by Fruth et al."

©_
A

I f
B -
x_g):: Kh-LZ

h^a- ^ -<£)- 0 -^I>-

FIG. 3—Schematic drawing of apparatus layout: (/4) pressure vessel, (B) ram,
(C) pore-pressure unit, (D) intensifier, (£) intensifier-regulator, (F) pore-pressure
bleedoff control, (G) gas pressure regulating valve, (H) hydraulic pump, (I) nitrogen
tank.

Test Apparatus

Pressure Vessel
The pressure vessel is a cylindrical piece of tool steel machined to
provide a pressure chamber and to receive an upper piston, lower seal,
and upper and lower retaining plugs. For tests in which pore pressure
is to be controlled, the upper retaining plug is replaced by the pore
pressure unit described below and shown in Fig. 2. The lower seal and
retaining plug are bored to permit movement of the lower piston into the

"The apparatus was developed with funds from Grant-in-Aid 145 from the
American Petroleum Institute and from Grant DA-ARO(D)-31-124-G149 from
the U. S. Army Research Office (Durham).
"L. S. Fruth, G. R. Orme, and F. A. Donath, "Experimental Compaction
Effects in Bahamian Sediments," Journal Sedimentary Petrology, 1966.

vessel. The vessel body is hardened to 42 Re and the internal parts to

60 Rp. All pressure seals are made with neoprene O-rings.
The specimen is placed between the upper piston and an anvil and
jacketed with a suitable material to prevent the confining pressure
medium from entering pore spaces in the specimen. The jacket also
serves to hold the assembly together. A sleeve of thin (0.008-in.) copper
tubing has been found to be very satisfactory for jacketing the specimen;
it is sealed against the piston and anvil by O-rings. Rubber tubing could
also be used and sealed with suitable clamps. The upper piston and anvil
are drilled to permit access to the pore spaces in the specimen; both are
faced with insert disks having fine concentric and radial grooves with holes
drilled at the intersections. The inlet in the anvil is threaded and tapered
to receive a plug for sealing.
The lower piston, lower seal, and lower retaining plug are left in the
vessel, and the piston-specimen assembly is inserted or withdrawn
from the top of the vessel. The upper piston is threaded so that a large
washer and nut can be used to extract it from the vessel. The upper
end of the lower piston is ground to a concave surface to receive the
convex end of the anvil. This serves as an internal force equalizer, should
the ends of the specimen not be perfectly parallel, and aids in centering
the piston-specimen assembly. The lower end of the lower piston has a
convex surface to provide an external force equalizer where it makes
contact with the ram piston.
The maximum confining pressure and total axial stress to which a
specimen can be subjected in the apparatus is determined by the wall
ratio (ratio of outside to inside diameter) and the maximum thrust that
can be sustained by the threads of the upper retaining plug. The total
force that the upper retaining plug must sustain is equal to the thrust of
the ram plus the force of the confining pressure times the horizontal area
of the upper piston flange. The apparatus described here has a wall ratio
of 3:1 and is designed for operation under confining pressures up to
30,000 psi; it will sustain the full thrust of the 50-ton ram at that con-
fining pressure. Thus, a specimen 1 in. in diameter can be subjected
to radial stresses up to 30,000 psi and an axial stress up to 130,000 psi.
Although the vessel is designed for testing of specimens 1 in. in diameter
by 2 in. long, the apparatus could be rescaled to permit use with larger
specimens; smaller specimens can be tested by utilizing special adaptors.

Confining Pressure System

Confining pressure can be generated in the apparatus in two ways: (1)

with the intensifier and (2) with the confining pressure regulator (see
Fig. 3).
The intensifier consists of a 20-ton ram coupled by means of a steel

collar to an intensifier body of hardened steel. A hardened steel piston

with a pressure seal at the end is connected mechanically to the ram and
is forced into the intensifier body by pumping hydraulic fluid into the
ram. The intensifier piston seal consists of two Teflon and two brass
washer-type packings. The diameter of the intensifier piston is approxi-

FIG. 4—Intensifier-regulator: (A) regulator body, (B) gas pressure chamber, (C)
compound piston, (/)) liquid, (E) gas.

mately one third that of the ram piston, and, because for equilibrium
pressure times piston cross-sectional area in the intensifier must equal
pressure times piston cross-sectional area in the ram, an input pressure
in the ram of 4000 psi would, if there were no frictional resistance in the
system, provide an output pressure from the intensifier of 36,000 psi.
Because movement of the lower piston into the pressure vessel raises
the confining pressure, a very small leak must be introduced in the con-
fining pressure system to offset this increase during testing of low-porosity
materials. The leak can be controlled manually or automatically by

means of a regulator such as that described below. In tests on uncon-

solidated material the decrease in porosity is commonly so large that a
leak need not be introduced, and, to the contrary, a regulator is needed
to introduce fluid into the system in order to maintain a constant con-
fining pressure.
The confining pressure regulator designed for this apparatus can also
be used as an intensifier, although it is principally intended to be a
"reverse-intensifier" to permit its use with standard bottle gas pressure.
The regulator contains a compound piston having difl'erent diameters at
opposite ends; the smaller end is inserted in the regulator body which, in
turn, is threaded into a gas pressure chamber (Fig. 4). The regulator body
is filled with liquid, and a constant gas pressure is maintained on the
large end of the piston by means of a gas pressure reducing regulator
drawing from a bottie of compressed gas. As the specimen volume
decreases in the pressure chamber owing to loss of porosity, the regulator
piston advances to force additional liquid into the pressure vessel and
thus compensate for the change. The liquid pressure remains constant
because of the constant gas pressure. A piston diameter ratio of 4:1 will
permit a gas pressure of 2000 psi to regulate the confining pressure to
30,000 psi.
The regulator can serve as the intensifier for the system if the vessel
and lines are completely filled before the pressure is built up and if there
is little change in porosity during the test. If too much stroke is used to
build confining pressure, there will be insufficient stroke remaining in the
regulator to compensate for large volume changes in the specimen.

Ram Pressure System

The ram pressure system is designed to permit operation of the
apparatus as a standard triaxial unit or as a constant load unit. This is
accomplished by employing a gas-liquid separator in the system below
which is a gas pressure reducing regulator and above which is a hydraulic
pump. The ram is rated at 50 tons for an input pressure of 10,000 psi.
Because standard gas bottle pressure is normally about 2000 psi, the
maximum thrust of the ram from that pressure would be 20,000 lb. The
separator must, therefore, also intensify if the full capacity of the ram
is required. An intensifier-regulator like that described in the preceding
section but with a piston diameter ratio of 2.25:1 has been found to be
very satisfactory for this purpose (see Fig. 4).
For operation as a standard triaxial unit, the pneumatic system is
closed off, and oil is pumped to the ram at the desired rate. For constant
load testing, the intensifier-regulator is filled with hydraulic fluid, and a
constant gas pressure is applied to the low-pressure side of the regulator
to produce a constant hydraulic pressure in the ram and, hence, constant
load on the specimen.

Pore Pressure System

In pore pressure tests the upper retaining plug is replaced by a pore
pressure unit that permits independent control of the pore pressure.
Because in certain testing it is desirable to raise the pore pressure and con-
fining pressure simultaneously, the unit is designed to serve also as a
separator so that it can be connected directly to the confining pressure
system (see Figs. 2 and 3). Pore pressure bleedoff is controlled by con-
necting the pore pressure line to a pressure chamber fitted with a piston
that can be withdrawn at a constant rate. The pore pressure bleedoff
control unit can also serve as a pressure generator by moving the piston
into the pressure chamber. This is, in fact, the more satisfactory way of
producing pore pressure if it is desired to build up pore pressure at some
time after the start of a test. Although the confining pressure could be
held constant by the intensifier-regulator while the valve (6, Fig. 3)
connecting the pore pressure and confining pressure systems is opened,
this would create a serious drain on the capacity of the intensifier-
regulator, such that there may remain insufficient liquid to maintain con-
stant pressure if there is further reduction in porosity of the specimen.
Normally, the pore pressure is built up simultaneously with confining
pressure before the start of a test, and it is more convenient to use the
confining pressure system to create the pore pressure.

Instrumentation
The important variables to be monitored during a test are confining
pressure, pore pressure, porosity change, axial load, and axial shortening
of the specimen. If the confining pressure is held constant during a test,
this variable need not be recorded. Similarly, in constant-load tests the
load need not be recorded continuously during the test. For standard
triaxial testing, load can be measured conveniently with either a load
cell inserted between the ram piston and lower piston of the vessel, or
with a pressure transducer in the ram pressure line. In the latter technique
the ram pressure is correlated with force exerted by the piston on the
specimen by measuring accurately the changes in pressure produced by
moving the lower piston into the liquid-filled vessel (the specimen and
anvil are omitted from the assembly in this calibration procedure). The
liquid pressure multiplied by the piston cross-sectional area equals the
axial force, which varies linearly with ram pressure. The signal from
the pressure transducer in the ram pressure system is fed into a strip-
chart recorder.
The unit is so designed that during the buildup of confining pressure,
preliminary to actual testing, the confining pressure can act on the
bottom of the anvil as well as on the sides of the specimen. The speci-
men is thus subjected to hydrostatic pressure, and the axial stress is equal
to the radial stress. At the time when the lower piston is moving freely

in the vessel before making contact with the anvil and the specimen is
subjected solely to hydrostatic pressure, whatever value of axial load is
indicated can be considered zero differential load. All subsequent differ-
ential loads can be computed from the recorder chart by knowing the
load interval (pounds of force per division). The differential stress is
equal to the differential load divided by the cross-sectional area of the
specimen at any given time. The cross-sectional area must be determined
from the known original volume, volume change, and axial shortening.
Axial shortening of the specimen is determined from the movement
of the ram piston relative to the vessel body. A linear variable differential
transformer (LVDT) is firmly attached to the vessel body. The trans-
former probe makes contact with an arm extending from the top of the
ram piston through a slot in the coupling collar, and the output from
the transformer is fed into a strip-chart recorder. The piston displace-
ment is used to calculate specimen shortening. This displacement con-
sists of two components: (1) axial shortening of the specimen and (2)
elastic distortion of the apparatus. The latter is determined for different
confining pressures and axial loads by replacing the test specimen with
a steel specimen of identical size and correcting for shortening of the
steel specimen.
Pore pressure can be monitored by a pressure transducer in the pore
pressure system. Porosity change in the specimen is determined from the
volume displacement of liquid from the pore spaces. As the specimen
deforms and pore space decreases, liquid is forced out of the specimen
into the pore pressure unit (Fig. 2). Attached to the separator piston in
the unit is a small rod that extends through a seal in the top plug.
Linear displacement of the separator piston is calibrated in units of
volume displacement, and the signal from an LVDT in contact with the
rod is fed into a strip-chart recorder.

Testing Procedure
For rock or other consolidated material, specimens are prepared with
a 1-in. diameter coring tool. Cores are placed in the collet of a tool and
cutter grinder and the ends ground flat to form perfect right cylinders
1 in. in diameter by 2 in. in length. The exact dimensions of the finished
right cylinders are measured with a micrometer. The cylindrical specimen
is placed between upper piston and anvil and suitably jacketed. The
piston, specimen, anvil, and jacket constitute the piston-specimen
assembly.
The procedure for preparing test specimens of unconsolidated material
is somewhat more involved. The upper piston is inverted, one end of a
thin-walled (0.008-in.) copper jacket is forced over the O-ring seal, and
a disk of filter paper is placed over the piston drain disk. The specimen
is weighed and poured into the copper jacket. The upper surface of the

sediment is smoothed, and a disk of filter paper is placed on it before in-

serting the anvil. The assembly is placed in a vacuum apparatus and filled
with distilled water while under vacuum. The hole in the anvil is sealed, and
the assembly is then placed upright in a special cylinder in which a small
but accurately determined amount of axial force can be exerted on the
assembly, and in which the specimen bulk volume and original length can
be measured accurately. The assembly is again placed in this cylinder
after a test to compare final volume and final length with measurements
made during the test.
For tests in which the pore pressure is initially equal to the confining
pressure, it is desirable to use clamps over the copper jacket at the
position of the O-rings on the upper piston and anvil. The clamps can
also be used with rubber tubing if that material is preferred for jacket-
ing.
The piston-specimen assembly is inserted in the pressure vessel, and
the top retaining plug or pore pressure unit is screwed into place. Hydrau-
lic fluid is pumped into the ram to bring the lower piston just into contact
with the anvil, and the displacement pen of the recorder is moved to some
arbitrary low position on the chart.
The confining pressure system is primed by opening Valves 1,3,4, 5,
and 11 and pumping the confining pressure medium (kerosine in room-
temperature tests) into the system with the hand pump (see Fig. 3).
Valves 1 and 3 are then closed, and Valve 2 is opened. Confining pressure
is built up with the intensifier to slightly below the desired level. Valve
11 is closed. Valve 10 opened, and gas pressure is increased in the in-
tensifier-regulator until the desired confining pressure is reached. If pore
pressure is to be used, the pore pressure unit must first be primed with
water (or other pore solution) so that the separator piston in the unit is
slightly off bottom and the pore spaces, gage, and line in this part of the
system are completely filled. The pore pressure is built up to the desired
value simultaneously with confining pressure by opening Valve 6.
Shortening of the specimen and compressibility of the fluid in the
ram-pressure system allows the confining pressure to force the lower
piston and anvil apart slightly. Hydraulic fluid is pumped into the ram
to again bring the lower piston and anvil into contact, and the axial force
just required to balance the force produced by the confining pressure and
to overcome the frictional resistance in the vessel is determined. During
the seating process the hydraulic pump is run at the same rate as during
a triaxial test, and, in most cases, the pump is not stopped when the anvil
is seated but is allowed to run until the specimen is deformed to whatever
strain desired. At the end of the test the pump is shut off and the ram
pressure dropped until the lower piston moves downward to lose contact
with the anvil. The pump is then used to reseat the anvil and thus obtain
a reseating displacement value. Finally, the confining pressure is re-

moved by opening Valve 3, and the anvil is reseated under atmospheric

pressure. Pore pressure may be decreased simultaneously with confining
pressure by opening Valve 74 or by utilizing the pore pressure bleedoff
control. The axial force (load) is recorded continuously during the seat-
ing, testing, and reseating procedures, and the several displacement
values obtained during seating before and after the test are used to deter-
mine total strain under atmospheric and confining pressure, respectively.
In constant-load tests the procedure described above is the same
except that, after seating the anvil under confining pressure, the desired
axial load is placed on the specimen by closing Valve 17, opening Valves
8 and 12, and building up a gas pressure on the low-pressure side of the
intensifier-regulator in the ram pressure system. Reseating values are
determined at the end of the test as for triaxial testing.
In some testing it is desired to change the pore pressure during the
test. This can easily be done by means of the pore pressure bleedoff con-
trol unit. In other testing it may be desired to hold the axial load con-
stant while the confining pressure is gradually reduced. This, too, can be
accomplished by means of a bleedoff control unit.

Measurement and Analysis of Rock

Physical Properties on the Dez Project, Iran

REFERENCE: R. K. Dodds, "Measurement and Analysis of Rock Physi-

cal Properties on the Dez Project, Iran," Testing Techniques for Rock
Mechanics, ASTM STP 402, Am. Soc. Testing Mats., 1966, p. 52.
ABSTRACT: The foundation rock for the Dez Project is a Pliocene,
limestone cobble conglomerate. The construction of a thin-arch dam on
this young, heterogeneous material required a thorough understanding
of its physical properties. The testing program included: 257 plate bearing
tests, 85 in situ measurements of residual stress using the flat jack
method, 2 large hydrostatic pressure chambers, and laboratory tests on
887 six-in.-diameter rock cores. The application of the principles of
rock mechanics resulted in considerable savings in time and money on the
Dez Project.
KEY WORDS: conglomerate rocks, plate bearing tests, residual stress,
modulus of elasticity, flat jack tests, hydrostatic pressure chambers, rock
mechanics, rock (material), dams, foundations

The Reza Shah Pahlavi Dam, formerly Dez Dam, is a double curva-
ture thin-arch dam 666 ft high, built on the Dez River in Khuzestan
Province of southern Iran, Fig. 1. Besides the dam, the project includes:
a powerhouse 256 ft long, 130 ft high, and 61 ft wide; two spillway
tunnels, one 47 ft in diameter and one 42.5 ft in diameter; and two
34-ft-diameter power tunnels feeding four 13.5-ft-diameter pre-stressed
concrete penstocks. In addition, to gain access to the construction site, it
was necessary to construct 4.65 miles of road tunnel from the top of the
plateau to the bottom of the gorge. All of these structures are built on or
in a (middle to late Pliocene) cobble conglomerate called the Upper
Bakhtiari Formation. The conglomerate is a fluviatile deposit of boulders,
cobbles, and gravel with some intercalated sandstone lenses, and is more
than 2000 ft thick. It was deposited in a subsiding trough during the
middle and late Pliocene time and extends for more than 1000 miles
along the western front of the Zagros Mountains in Iran and Iraq. Dur-
ing the deposition, the velocity of the rivers was high enough to carry
away much of the fine material, and so only subordinate amounts of sand
1
Director, Foundation Sciences Co., Astoria, Ore. Personal member ASTM.
52
Copyright^ 1966 by ASTM International www.astm.org
Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
DODDS ON THE DEZ PROJECT, IRAN 53

FIG. 1—Map showing location of the Dez Project, Iran.

and silt were deposited. The result was a deposit with a high percentage
of initial voids which are partially filled by cementing material. The
mass appearance of the Upper Bakhtiari Formation is that of a heteroge-
neous accumulation of rapidly dumped coarse alluvium.
The conglomerate is composed of subangular to well-rounded gravels
of limestone, dolomite, and 'chert with minor amounts of tuff and mud-
stone. A matrix of sand, silt, and clay partially fills the voids. Calcite
(CaCOg) is the principal cementing agent in the rock, but its deposition
has not been complete, leaving openwork pockets within the rock mass.
The young age, variable composition and cementation, and exacting
foundation demands of the dam combined to make a thorough under-
standing of the physical properties of the rock an absolute necessity.
The testing and exploration program to establish the physical properties
of the Bakhtiari conglomerate consisted of the following:

1. drilling 11,210 lineal ft of NX bore hole and 1432 lineal ft of 6-

in.-diameter bore hole;
2. excavating twelve test adits with a total length of 4000 lineal ft;
3. testing two 20-ft-long hydrostatic pressure chambers (to measure
the large-scale creeping characteristics of the rock);
4. testing 47 flat jacks (to measure in situ residual stress);
5. in situ bearing plate measurements of elastic properties at 257
points using 100-ton hydraulic jacks;
6. measuring rock creep over a period of 16 days using 100-ton
hydraulic jacks in two adits;
7. testing 881 six-in.-diameter rock cores in the project laboratory
for: unconfined and confined crushing strengths, sustained load reaction,
tensile strength, shear strength, effects of saturation on strength, modulus
of compression, and creep; and
8. detailed geologic mapping of the dam abutments and statistical
evaluation of results.
Many of the testing methods gave results for the same physical
property and so acted as checks against one another. As an example,
values for the moduli of elasticity were obtained from five different
methods.
Tests
Hydrostatic Pressure Chambers
Hydrostatic pressure chambers are used to measure the reaction of a
rock to stress over large areas. They give values for modulus of elasticity,
elastic recovery, inelastic deformation, and creep. The results are used
to evaluate the behavior of the dam foundations and related strain dis-
tribution in the structure, and to help estimate the behavior of pressure
tunnel linings.
Hydrostatic chambers cover a much larger surface area than other test
methods used to measure the elastic properties of rock and so, in theory
at least, provide test results that measure the mass behavior of the
material. How close they come to this ideal depends on the spacing, size,
and distribution pattern of the discontinuities in the rock. The test
chambers used at Dez Project were 16.9 ft long and 7.1 ft in diameter
and had a total test surface area of 669 ft^.
Because of their relatively large cost, hydrostatic pressure chamber
tests are used sparingly. This means that there is litde opportunity to
recognize errors statistically; so, to be of value, special care must be taken
in selecting and preparing the site for the test, and the construction of the
chamber must be done properly. In fact, because of the usual primitive
conditions prevailing at the work site at the time these tests are made,
one of the biggst problems with them is to assure that errors do not get
into the tests because of faulty construction of the chamber.

77 ^ 7 :r7 :?-^^
LONGITUDINAL SECTION

FIG. 2—Hydrostatic pressure chamber sections.

FIG. 3—Rough concrete cylinder for hydrostatic pressure cliamber.

Preparation of the pressure chambers—Two hydrostatic pressure

chambers were built and tested on the Dez Project. One chamber was
located in each abutment of the dam. The longitudinal axes of both
chambers were orientated so that deformation was measured in a
direction approximately parallel to the dam thrust. The pressure
chambers were designed by Electroconsult of Milan, Italy, who also sup-
plied the special testing equipment and conducted the tests. For an
additional discussion of the method see Oberti [7].^
Each pressure chamber covers a length of 27 ft in an adit with a
circular reinforced concrete lining. The inside dimensions are 16.9 ft
long by 7.1 ft in diameter. Reinforced concrete plugs close both ends, see
Fig. 2. The outside plug has a steel manhole approximately 2 ft in
diameter to allow access for installing the deformation gages and through
which the water inlet and leads from the gages oass. The lining is made
in four independent sections to reduce stress concentrations during test-
ing. Figure 3 shows the rough chamber before the front plug is placed.
After the concrete lining is in place, low pressure contact grouting is done
to fill voids behind the crown of the lining. Following grouting, any rough
places on the inside concrete surface are ground smooth, and a rubber
bag supported by aluminum braces is placed inside to improve water
tightness.
Eight bolt anchors are now placed around the center of the chamber
for the deformation gages. Four electroaccoustic extensometers are
attached to these bolts as shown on Fig. 2. The extensometers are con-
nected to the bolts by invar steel bars tensioned by springs and are
capable of measuring displacements between 0.0005 and 0.2 in. After
the gages are installed, the chamber is sealed and ready for testing.
Pressure for these series of tests was maintained by a standpipe laid
along the canyon wall. This method permitted more positive control over
the pressure and eliminated the possibility of mechanical breakdown
interfering with a test in progress.
Testing the pressure chambers—Four types of test were made in the
chambers:
1. deformation tests over periods of three and four days at pressures
up to 200 psi,
2. creep tests over periods of 16 days,
3. deformation tests over periods of three and four days at pressures
up to 265 psi (higher pressures could not be used because of cracking
in the concrete lining), and
4. deformation tests over periods of three days at pressures up to
265 psi after grouting the surrounding rock.

" The italic numbers in brackets refer to the list of references appended to this
paper.

TABLE 1—Hydrostatic moduli of elasticity in psi.

Chamber No. 1 Chamber No. 2
Instrument" Low-Pressure Low-Pressure High-Pressure
Grouting Grouting Grouting

1 (426 000) (852 000) 1 063 000

2 1490 000 1775 000 1775 000
3 568 000 1705 000 1490 000
4 1490 000 1561000 1561000
Mean over-all moduli
E 1 138 000 1 561 000
Secant moduli £ „ . . . . 710 000 994 000
( ) Figures not taken into account in working out the mean over-all moduli E.
" Instrument locations:

FIG. 4—Hydraulic jack test equipment at Ciceroz Dam, Turkey.

In performing a test, the pressure was built up in increments of 30 psi,

each increment being held for a few hours to allow most of the deforma-
tion to occur. The same steps were used in reducing pressure at the end
of a test. Table 1 gives the average results of these tests.

Plate-Loading Tests
Because they are relatively inexpensive and easy to set up, plate-
loading tests have been used for some time to measure the elastic
properties of rock. However, they test only a small area and so, for valid
interpretation, a sufficient number of tests are required to permit a
statistical evaluation of the results. At Dez Project, a total of 257 plate-
loading tests were made and the results analyzed by Talobre [2].
Equipment—The equipment used for horizontal plate-loading tests
consists of a heavily reinforced H-beam, supported by an adjustable
cradle and riding on six ball bearings. The test load for these tests was
by a 100-ton hydraulic jack. Ten-inch-square steel bearing plates were
fitted to both the head of the ram and the opposite end of the steel beam.

TABLE 2—Ill-place bearing tests, 1960-1961.

Number of Modulus of Elasticity, average all gages, psi

Location
Tests Min Max

Adit A . . . 6 1.8 X 10«

Adit A . . . . .. . 5 3.6 X 106
Adit B . . 6 1.4
Adit B . . . 4 3.8
Adit C i . . 10 1.2
Adit C i . . . .. . 10 3.8
Adit D i 6 1.0
Adit D i 6 1.5
AditE. . 6 1.4
Adit E 6 3.3
Adit F . 10 1.4
Adit F . . . 10 2.7
Avg. . . . 1.4 X 10« 3.1 X 10"

Figure 4 shows a typical setup of similar but improved equipment used

by the writer at Ciceroz Dam in Turkey.
Deflections are measured each 120 deg around the bearing plate by
dial gages, accurate to 0.0001 in. The dial gages were supported com-
pletely independent of the test beam by a ring and post assembly. The
bearing plates work against a specially prepared rock surface that has a
thin sulfur cap which provides a good contact with the rock test surface
and the plate.
Procedure—When the location for the test has been chosen, all loose
rock is cleaned from the area for the test, and a relatively smooth bearing
surface is prepared using a chipping hammer or chisel and single jack.
After the two bearing surfaces approximately parallel to each other are
prepared, the order of setting up the equipment is:
1. The beam support cradle is set up and leveled by means of the
adjusting screws in the legs.

2. Sections of beam are placed on the cradle and bolted together until
the sum of the length of the beam, ram, and bearing plates is just 2 to 4
in. less than the distance between the rock surfaces.
3. The hydraulic ram and bearing plates are added to the test
assembly, and it is checked to make sure it is absolutely level.
4. The form is built around each bearing plate and the sulfur cap
poured.
5. The gage holding posts and brackets are installed, being sure that
there is no contact between this equipment and the test beam assembly.
6. The dial gages are installed so that they are in contact with the
special gage arms on the bearing plates; by convention. Gages 1, 2, and
3 are at the ram or head end of the test assembly, and Gages 4, 5, and
6 are at the butt end of the assembly; the gages are numbered in a clock-
wise direction starting at the top.
After the sulfur cap has attained a strength in excess of 5000 psi, the
rock test is made.
Testing—Each test usually consists of four cycles of loading and
unloading the bearing surfaces with progressively higher loads and
recording the rock deflections. Ten readings are made during each cycle
of loading and unloading, and the load is held constant for 30 min at
the maximum and minimum load to allow the rock to adjust. When
the 30-min period is not sufficient to complete most movement, a special
sustained load test should be made over a longer period. When rock
strength permits, the cycles consist of loading the bearing surfaces to
50,000, 100,000, 150,000, and 200,000 lb. One completed test usually
takes between 8 and 10 hr.
Results—The results of the last series of plate-loading tests made on
the project are given in Table 2. The minimum value shown is usually
the secant modulus of the first cycle and the maximum value the elastic
modulus of the fourth loading cycle. In using the data from these tests
in design evaluation, they were combined with measurements of rock
elastic properties given by hydrostatic pressure chambers, sustained load
tests, flat jack tests, and compression tests of 6-in.-diameter rock cores
in the laboratory.
The tests in the lower adits gave generally lower results, which is
attributed to the fact that the rock load was much higher near the bottom
of the canyon, resulting in deeper stress relief fractures around the adit
walls.
In his analysis of the test results, Talobre [2] used data from the
plate-loading tests to plot Mohr circles and determine from them
apparent shear strength for the rock. From the shear strength deter-
minations, Talobre then calculated the allowable bearing capacity for the
foundations.

Flat Jack Tests (Internal Stress Measurements)

The internal (or residual) stresses present within a rock mass have an
important effect on the physical properties and behavior of a rock as an
engineering material. The residual stress affects, greatly at times, the
elastic and strength characteristics of the rock in situ and the stress
distribution around an opening made in the rock.
The internal stress in a rock mass is the result of: (1) the weight of
overlying material, (2) hydrostatic pore pressures, (3) remaining tectonic
forces, and (4) chemical stresses due mostly to recrystallization of
minerals.
Measurements of rock internal stress usually rely on the theory that
when an opening in made in a rock the rock will expand into the opening
in direct relationship between its elastic properties and the internal stress
at the point of measurements. The measurements record the strain and
compute stress from laboratory-measured moduli of elasticity, assuming
a seldom realized linear relationship between stress and strain. Flat jack
tests offer the advantage over the above methods of not requiring a pre-
knowledge of the elastic modulus of the rock or the material behaving as
a Hooke solid. They give a direct value for internal stress with only the
acceptance of the assumption that the stress required to return the slot
walls to their original position is equal to the stress that caused the
original deformation. Measurements of residual stress by the use of flat
jacks (thin hydraulic pressure cells) were first made in France by the
Laboratoires du Batiment et des Travaux Publics, and by Tincelin [3].
The method described here is based on Tincelin's work.
Equipment—The flat jacks used for these tests were 70 cm squares
loaded by means of a hydraulic pump. The jacks have a capacity of 2200
psi. Rock deformation was measured by dial gages or a multiposition
strain gage.
Procedures—Two methods of measuring the deflections of the slot
were used in the flat jack tests performed at the Dez Project.
The first method consisted of installing a dial gage reading to 0.0001
in. at right angles to the jack slot and 41 cm from the center of the slot.
The lever arm of this gage moved against a steel plate installed 23 cm
from the center of the slot.
Mechanically this setup was satisfactory, that is, barring mishaps to
the gage or measuring post, the system gave accurate results. How-
ever, since the gage and posts had to be installed and read prior to the
cutting of the slot, they were very vulnerable to jarring while the slot was
being cut and the flat jack installed. These accidents happened often
enough to make an improvement in the measuring system desirable. And
so, a strain gage, of the type used to measure mortar-bar expansion in
the laboratory, was adapted for use by welding an extension of about 15

measuring point
g r o u t e d into \ |0
rock

FIG. 5—Flat jack test measuring point layout.

FIG. 6—Making jack test.

cm in length onto the brass insert measuring studs and installing these
measuring points in the rock on both sides of the jack slot.
This system proved very satisfactory as now the gage was removed
during the cutting of the slot, and only the measuring studs, almost flush
with the rock surface, remained.
The step-by-step procedures for preparing and making a flat jack test
at the Dez Project were as follows.
1. After the location for the test was decided, the rock surface was
smoothed with a chipping hammer, making sure to remove all loose slabs.
2. The inclination of the slot was marked off, and at right angles to it,
the holes for the measuring studs were drilled, see Fig. 5. The measuring

studs were cemented into the holes, and 24 hr later the initial readings
of the distance separating the studs was made to establish the equilibrium
base.
3. A slot approximately 3 in. wide by 30 in. deep was then cut out of
the rock for the flat jack, and the jack was cemented in place. A very dry

TABLE 3—Pahlavi Dam abutments, internal rock stresses measured

by flat jacks.
Right Abutment Left Abutment
Difference,
Location Direction of Stress Num- Stress, Num- Stress, per cent
ber of psi ber of psi
Tests Tests

Adits normal to canyon 334 764

wall
Surface normal 12 552 12 572 4
Adits parallel 10 1385 11 1302 6
Surface parallel 2 580 2 658
Adits horizontal 9 816 5 2222 63

100
T\ "
\ "' " " ' — p- r T r"' ^
- ,__.., . .___.^__._.
..-.- i — ^ - ^ — . — \ —

1
5
^*\
1 ^ ^ ^ "
- -— - . _
. , . —
\
^ X-' r —
^ 50
i\ I '^ '" : i 1 1 i
S J LV.-i ' ; 1
1 ^ ^ 1 \
1 '' ^
L ..
r " ^
— •^

—* ' •

+ ' '
0
\
1000 2000 3000 4 0 00
Con celirif pre sure (psi)

FIG. 7—Flat jack test relationship between canceling pressure and inelastic
deformation.

mix of 50 per cent cement and 50 per cent sand was used, and it was
thoroughly tamped in place. The jacks usually were allowed to set for
seven days before testing to be sure most movement in the slot was
completed and the mortar had developed good strength.
The measuring studs must be protected from bumping during this
work, and the usual practice was to cover them loosely with paper and
then cover the paper with a cement-sand mortar cap.
4. The jack was then tested by removing the protective covering from
the measuring points and measuring the amount of closure the slot had

undergone. The jack was filled with hydraulic fluid and all air expelled
from the pump, jack, and lines (Fig. 6).
The pressure in the jack was raised, usually by increments of 150 psi,
and the deflection of the rock read. This cycle was continued until that
pressure was reached where the rock was forced back to its position
before the slot was cut, that is, the distance between the measuring studs
was equal to the original distance read under Step 2. The loading cycle
was then continued if possible past the zero deformation point about 600
psi. The pressure was returned to zero, again by increments of 150 psi,
with readings taken at each step. This procedure was repeated three times
to constitute a completed test.
Results—Table 3 summarizes the results of the flat jack test measure-
ments of internal stress. The tests listed as "surface" were made on the

FIG. 8—Laboratory triaxial test equipment.

surface of the dam abutments after the excavation of the dam keyway
but before final cleanup. After a number of tests were made, a general
relationship between cancelling pressure and the percentage of inelastic
deformation developed, and thus it seemed could be related, in a rough
way, to the amount of stress relief that had taken place in the rock, see
Fig. 7.
The primary purpose of the flat jack tests was to establish that a cer-
tain amount of internal stress existed throughout in the rock mass. This
they did, with all tests except one showing values in excess of 300 psi,
with a maximum of 4000 psi being recorded.

Laboratory Tests
To develop Mohr's envelopes for the rock so that the in situ measure-
ments could be correctly related to in-place strength, a thorough labora-
tory testing program was required. This program was carried out in the
laboratory on the project site when it became obvious that the remoteness

of the site made protection of the specimens for transportation to other

testing laboratories impracticable.
To supply specimens for these tests, thirty-six 6-in.-diameter drill
holes, with a total length of 1436 lineal ft, were drilled approximately
equally spaced down both abutments.
Equipment—Cores from which the test cylinders were cut were
obtained with a core drill using a 6-in. double-tube core barrel. Drill
bits were internal discharge diamond bits. Specimens were trimmed to
12 or 6-in. lengths using a masonry saw. Specimens from Holes XT-1

FIG. 9—Sustained load test.

through XT-12 were cut using reinforced fiberglass blades. Cutting

time with these blades was approximately 40 min per cut. From Holes
XT-13 through XT-35, a diamond cutting blade was used which reduced
the cutting time to about five min. During cutting, the blade is cooled
with water which means that all specimens tested have undergone at least
two periods of wetting; one during drilling and one during cutting (Fig.
8).
Testing was done using a compression testing machine with a capacity
of 300,000 lb. The pressure gages of this machine were calibrated by the
Ammann and Whitney Laboratory, Teheran, prior to the testing pro-
gram. The results of this calibration show that, within the range of loads
used, the gage readings were 0.0 to 1.5 per cent lower than actual pres-

sures on the 300,000 lb capacity gage, and 0.0 to 3.0 per cent lower than
actual pressure on the 60,000 lb capacity gage.
The confining device for the triaxial tests was fabricated at the con-
tractor's shops at the construction site. Confining pressure was applied
using four 30-ton capacity rams connected to a central pump. The effi-
ciency of this type of confining device depends on the area of contact
between the test specimen and the pressure plates. Obviously, this varies
but visually appeared to be satisfactory.
In an attempt to measure direct shear strength, a three-ring shear
device was made at the contractor's shop. When using this device, about
1-mm space was left between the center ring and the two outside rings.
First tests with this device were not satisfactory because play in the outer
rings allowed the specimen to fail in tension. Later, hard rubber pads
were placed around the specimen, and a confining strap placed horizon-
tally around the base of the device. These changes improved the func-
tion of the device but did not completely eliminate flexure from the
specimens, and so attempts to measure direct shear were abandoned in
favor of Brazilian tests and computed shear strengths from Mohr en-
velopes.
Moduli of elasticity in compression were computed for unconfined
specimens from the compression of the specimen during loading, meas-
ured with a concrete cylinder compressometer.
Most compression tests were made on specimens 6 in. in diameter
and 12 in. long, but 53 were made on specimens 6 in. in diameter and
6 in. long. Brazilian and direct shear tests were made on specimens 6 in.
in diameter and of varying lengths.
A special press was manufactured to be used with a 100-ton jack for
sustained load tests (Fig. 9).
Procedures—When the drill core was delivered to the laboratory, it
was first photographed and then logged. Following logging, all speci-
mens long enough to make a 12-in. test cylinder were trimmed to this
length using the masonry saw. Time lapse between trimming and test-
ing varied between 14 days and a few hours, but generally was about
72 hr. Prior to testing, the area of the top and bottom bearing surfaces
was measured, and the specimen was capped using concrete cylinder
capping compound with a compressive strength of 6000 psi. (Actual
diameter of a 6-in. core was 5.9 in., which gives a surface area of 27.4
for a perfect cylinder.) Only the area of holes appearing in the bearing
surfaces was subtracted to obtain the actual surface area. No adjust-
ment of surface area was made for holes elsewhere in the specimen or
for holes filled with capping compound.
The rate of loading during testing for specimens from Holes XT-1
through XT-12 was about 20 psi/sec. This rate was increased to 50 to
75 psi/sec for the rest of the holes tested. The loading rate is lower than

the 100 psi/sec specified by the U.S. Bureau of Mines (Obert et al [4]),
but the later loading rate of around 50 psi/sec is about standard for con-
crete cylinder tests. However, at the lower loading rates, some creep
could take place in the specimens before failure.
In making the triaxial tests, the specimen was placed in the confining
device, and a load equal to half the test confining load was applied on it.
Then the vertical load was raised to about 500 psi, and, following this,
the confining load was increased to the test load. The vertical load was
then increased to failure in the specimen.
Brazilian (or split-cylinder) tests were run on a number of specimens
with apparently good results. At first the edges of the specimens were

TABLE 4—Test results, 6-in. diameter cores, Pahlavi Dam abutments.

Dry Strength, psi Wet Strength, psi
No. Differ-
Test Cubic Cubic ence,
Tests Cylinder Equiva- Cylinder Equiva-
lent lent %

Cylinders, 6 by 12 in.
Unconflned 282 2621 3145
Unconflned 23 2436 2923 7.1
350 psi confinement 75 3533 4240
350 psi confinement 13 3380 4056 4.4
750 psi confinement 83 4697 5636
750 psi confinement 10 4503 5402 4.1
Brazilian (tensile) 269 385
Brazilian (tensile) 6 310
Sustained load u/c 26 2760 3312
Cylinders, 6 by 6 in.
Unconfined 53 4158

trimmed, but, when no effect one way or the other on the test resuhs
could be established, this trimming was stopped. A small piece of ma-
sonite was placed between the test cylinder and the press on all tests.
To compute the tensile strength, the formula: ^3 = —2P/irDL was used.
Specimens from Holes XT-1, XT-3, and XT-4 were tested with the
direct shear device. After Holes XT-4, the Brazilian test was substituted
as being more reliable, and shear strengths were computed from the test
results.
On some of the unconfined compression tests, the compression of the
specimen was measured at increments of 2500-lb load, and these figures
were used to compute the modulus of compression of the specimen.
For sustained load tests, the average unconfined strength for each
specimen was figured from the results of other tests on rock from the
same drill hole. The specimen was then loaded to 80 per cent of this
value, and the load held until the specimen failed or for approximately

three days. If failure did not occur in three to five days the load was in-
creased to failure. Measurement of the deflection of the specimen with a
compressometer was taken about every hour during the test.
Results—The results of the tests conducted in the dam site laboratory
are summarized in Table 4. Four hundred twenty-four lineal meters of
hole were drilled, and 412.7 lineal meters of core were recovered. From
this 412.7 lineal meters of core, 245.3 lineal meters of specimens were
cut and tested, that is, 59.4 per cent of the core recovered was tested.
Because the laboratory test strength given by a specimen is dependent
upon the size and shape of the specimen and because the original design
assumptions for the dam were based upon strength tests on cubic speci-
mens, it was decided to test some specimens with a height-to-diameter
ratio of 1:1 and to compare these results with the normal 2:1 ratio.
Fifty-three specimens 6 in. high by 6 in. in diameter from Holes XT-29
through XT-35 were tested unconfined dry. The crushing strength of
these specimens was 41.5 per cent greater than the average dry uncon-
fined crushing strength of the seventy-two 6 by 12-in. specimens from
the same holes.
Tests of this property on concrete cylinders would lead one to expect
about a 20 per cent increase in strength for the 6 by 6-in. cylinders. The
41.5 per cent increase measured is high and must, in part, be attributed
to the fact that a 6 by 6-in. cylinder has a much lower possibility of
containing a thin weaker seam than a 12-in. high cylinder. The 6-in.
cylinders also contained a smaller percentage of oversize aggregate.
The number of these tests made is large enough to give them some
reliability, so the least one can conclude is that increasing the 6 by 12-in.
cylinder test values by 20 per cent to arrive at the approximate cubic
(height.'diameter = 1:1) strength is conservative.
To evaluate the effects of water on the rock strength, 52 specimens
were tested while saturated. The results of these tests are given on Table
4. It can be seen that slight drops in the average strength for the satu-
rated specimens of from 7.1 to 4.1 per cent were recorded. This phe-
nomenon is also experienced when testing concrete cylinders. Troxell
and Davis [5] list three reasons for this: "(1) the greater density of dry
(and therefore contracted) paste, (2) initial tensile stresses in the paste
due to localized restraint of paste shrinkage by pieces of aggregate, and
(3) possible development of hydrostatic pressure in saturated paste."
After a concrete test cylinder has been cured for about a year. Reasons
1 and 2 are eliminated, and a strength drop of about 10 per cent is all
that occurs in wet concrete cylinders. This is attributed to Reason 3.
Strength drops of this magnitude (7 per cent) are what we experienced
in these unconfined tests. It then seems proper to conclude that, in this
series of tests, no strength loss occurred because of the effects of water
on the cementing agent in the conglomerate.

It can be seen from the results on Table 4 that the presence of even
a moderate amount of confining load is very beneficial to the crushing
strength of the rock. This is partly because confinement tends to over-
come weaknesses in the specimen originating from unfavorable orienta-
tion of flat-sided gravels, large loose boulder fragments, weak zones, etc.
Confinement (P) improves the crushing strength of 6-in.-diameter by
12-in.-high specimens according to the following relationship:
C, = 2621 + 2.68 P
Twenty-six specimens were tested unconfined under a sustained load.
Only three of the sustained-load specimens failed when held at loads
that were 80 per cent of the average breaking strength of the other speci-

TABLE 5—Component classification Bakhtiari conglomerate.

Description of Material Map Symbol"

Well-cemented conglomerate with sandy matrix A + +

Well-cemented conglomerate with sand and red silt
matrix A -|-
Well-cemented openwork gravel A —
Fairly well-cemented conglomerate with sandy matrix. . B 4- +
Fairly well-cemented conglomerate with sand and red
silt matrix B -|-
Sandstone lense B —
Conglomerate with red silt matrix B ^
Fairly well-cemented openwork with sandy matrix C -|-
Fairly well-cemented openwork with sand and red silt
matrix C —
Sand lense D -|-
Poorly cemented openwork D —
Red silt lense D ^
» As used on the geologic maps of the dam abutments.

mens from the same hole. In fact, the specimens used for these tests had
a slightly higher strength than the unconfined compression specimens, so
the load used averaged 76 per cent of their measured breaking strength.
Whether this difference is due to specimen selection or the action of the
rock during testing is not known. However, the specimens were selected
by arbitrarily taking every third specimen from a drill hole.
Geologic mapping—The general appearance of the Upper Bakhtiari
conglomerate is one of a thick heterogeneous accumulation of coarse
alluvium. A more detailed examination of the deposit reveals that it is
composed of diverse, poorly stratified lenses and pods of openwork
gravels, pea-sized gravels, and sand lenses, contained within a boulder
and cobble conglomerate with a coarse sand matrix. To aid in the evalua-
tion of the physical properties of the rock and apply the data collected
during the testing program to the specific problems of foundation design.

detailed knowledge of the extent, location, and distribution of the various

components of the conglomerate was required. This information was sup-
plied by detailed geologic maps of the dam abutments, on a scale of 1
cm equals 1 m (1 in. equals 8.4 ft), delineating all significant variations
in the composition of the conglomerate. For mapping purposes the con-
glomerate was divided into twelve units given in Table 5 in order of
reduced competency.
One of the various lenses or pods occurs approximately every meter
in elevation; and in 11,979 m^ (128,862 ft^) 513 lenses were mapped.
The average thickness of a lense is 14.2 in., and the average length is 20.2
ft.

Application of Test Results to Foundation Analysis

The results of all foundation tests for the Dez Project were evaluated
by Talobre [2,6], and details of the analyses can be found in these re-
ports, which will be summarized here.
The geologic surveys of the dam abutments established that the rock
is free from major structural defects (faults or joints); that lenses and
pods of below average strength rock are of local extent and can be ade-
quately and safely treated during construction by: (1) removal and back-
filling with concrete, (2) special grouting, and (3) additional reinforcing
of abutment to span the dam thrust across the weaker zone. Therefore,
sliding failure in the dam abutments can occur only following the devel-
opment of new failure surfaces by crushing of the rock. It is prudent
then to consider as the critical value, when evaluating the ultimate bear-
ing capacity of the Dez abutments, the crushing strength of the rock.
The crushing strength of the Dez Project foundation rock is affected
by the following factors:
1. confining load (internal stress) that exists within the rock,
2. friction at the contact between the abutment rock and concrete,
3. areas of weak rock in the abutments (sand lenses and lenses of
poorly cemented openwork gravel),
4. effects of reservoir water on the strength of the rock, and
5. action of pore pressure.
Confining load—Twenty-eight determinations of the internal stress
present in the rock at the abutment surface were made using flat jacks.
The average internal stress measured in these tests was 560 psi. The
internal stress measured was more than 300 psi in all but one test which
recorded a stress of 210 psi in rock weakened by blasting. More than 50
per cent of the measurements were over 470 psi. The maximum stress
on the abutment surface recorded by a flat jack was 1260 psi.
For his analysis of the in-place crushing strength, Talobre [6] has
used the very conservative figure of 300 psi for the internal compression
present in the rock at the abutment surface. Correcting this value for the

direction of dam tlirust gives an actual minimum confining load of 339

psi on the downstream edge of the abutment. The upstream edge of the
dam shoulder is deeper within the rock in all cases, so a value of 450
psi for the internal compression at this point is considered conservative.
Using the relationship between confining pressure and crushing
strength developed in the laboratory tests, the mean crushing strength

TABLE 6—Ultimate bearing capacity of dam foundation, psi.

Elevation, m" 354 320 280 240 200

Right bank:
Upstream edge of saddle
All loads, except seismic 284 244 358 324 246
All loads 453 354 444 369 311
Rock crushing strength 4032 3980 3860 3765 3670
Minimum safety factor 8,9 11.2 8.7 10.2 11.8
Downstream edge of saddle
All loads, but seismic 469 480 649 613 460
Allloads 453 580 735 683 491
Rock crushing strength 3735 3735 3735 3735 3735
Minimum safety factor 8.0 6.5 5.1 5.5 7.6
Left bank:
Upstream edge of saddle
All loads, but seismic 392 208 405 377 395
Allloads 545 341 494 436 483
Rock crushing strength 4032 3980 3860 3765 3670
Minimum safety factor 7.4 11.7 7.8 8.6 7.6
Downstream edge of saddle
All loads, but seismic 613 485 672 695 680
Allloads 545 587 758 774 735
Rock crushing strength 3735 3735 3735 3735 3735
Minimum safety factor 6.1 6,4 4.9 4.8 5.1

" Stress normal to foundation in psi.

of the abutment rock at the downstream edge of the saddle block be-
comes:
C, = 2621 + (339 X 2.68) = 3530 psi
And at the upstream edge of the saddle block:
C, = 2621 + (450 X 2.68) = 3810 psi
Friction—The friction along the surface of the concrete and rock
has the effect of providing additional confinement to the rock and thereby
increasing its strength. Talobre [6] computed the value of this additional
confinement equivalent to an increase in crushing strength of 20 per cent.
Sand lenses and poorly cemented conglomerate—Detailed mapping
of the abutment surfaces has established that the total area of sand
lenses and poorly cemented conglomerate does not exceed 1.4 per cent.
Where these rocks did occur at the abutment surface they were removed

and replaced with concrete, specially grouted, or the saddle block adja-
cent to the questioned zones was reinforced to bridge the thrust of the
dam around the lenses should they not be able to carry the full load. Al-
though this rock was removed or otherwise treated, the area of sand
lenses and poorly cemented openwork was subtracted from the bearing
surface when computing the load on the abutment rock. That is, although
it was made satisfactory rock, such repaired rock was not included in the
computation of the effective bearing surface.
Effect of reservoir water on the rock—It has been established that the
Dez River water is completely saturated with CaCOs—the composition
of the cementing agent in the Bakhtiari conglomerate—and is not able
to dissolve an additional amount. It has also been established that
saturation does not have a deleterious effect on the crushing strength
of the rock. However, Talobre [6] applied a 10 per cent reduction of the
mean crushing strength of the rock when computing safety factors to
compensate for any still unknown weakness, present or future, that pos-
sibly could exist in the abutment rock.
Pore pressure—The effect of pore pressure is to lower the effective
confining load in the rock, and also to lower the effective load on the
abutments, Robinson [7]. At elevation 200, the maximum loss of crush-
ing strength for the rock will be 2.68 x 198 =531 psi. The reduction of
the load at this elevation is 198 psi and total loss of strength is 531 —
198 = 333 psi. The average crushing strength of the dam foundation at
the upstream edge of the saddle at elevation 200 is (3810 X 1.20 x 0.98
X 0.90) - 333 = 3700 psi, and at the downstream toe, 3530 x 1.20 x
0.98 X 0.90 = 3735 psi.
Ultimate bearing capacity—Table 6 lists the maximum stress acting
normal to the dam foundations as computed by Talobre [6], and the
minimum factors of safety with these loads.

A cknowledgments
The Dez Project is owned and operated by the Khuzestan Water and
Power Authority, an agency of the Imperial Government of Iran. De-
velopment and Resources Corp. provided design engineering and con-
struction supervision for KWPA on the project. The general construction
contractor was Impresit-Girola-Lodigiani.
In a work of the magnitude of the Dez Project, many people made
important contributions to the collection of the data and the formulation
of the concepts recorded here. To name a few: A. A. Meyer, chairman
of the dam consulting board from its inception; P. P. Fox, who prepared
the first reports on the geology of the site and was a member of the
permanent consulting board for the project; J. A. Talobre, a member
of the dam consulting board and who was responsible for the final analy-
sis of interpretations of test data on the mechanical properties of the

rock; B. E. Warner, Jr., who was project geologist from November, 1958,
until November, 1960, and who conducted the early in situ tests on the
mechanical properties of the rock; and Astorre Rammuccioni, the staff
member of Electroconsult, Milan, Italy, who made the hydrostatic pres-
sure chamber tests. The writer would also like to acknowledge valuable
on-the-site discussion of the foundation problems with L. Miiller and
the late E. Burwell and the very necessary encouragement and support
for the work given by W. L. Voorduin, Development and Resources
Corp. vice president and director of engineering. Special mention
should be made of the contribution of N. T. Mirafuente who, as the
writer's assistant from December, 1960, to April, 1963, was responsible
for the execution of many of the in situ and laboratory tests.
The contributions of these men were essential to the successful solu-
tion of the foundation problems at Dez Project. However, they are not
to be held responsible for the presentation or interpretation of data con-
tained herein; this responsibility rests with the writer.

References
[1] G. Oberti, "Experimentelle Untersuchungen iiber die Charakteristika der
Verformbarkeit der Felsen," Geologic und Bauwesen, Vol 2-3, 1960, pp.
95-113.
[2] J. A. Talobre, "Dez Dam Foundation Tests Analysis," Development and
Resources Corp, New York, N. Y., 1961.
[3] M. E. Tincelin, "Mesure des pressions de terrains dans les mines de fer de
Test," Annates d I'lnstitute Technique du Batiment et des Travaux Publics,
serie: Sols et foundations, No. 58, 1952, pp. 972-990. Translation by S. H.
Britt, U. S. Geological Survey, Open File report No. 28927, Washington, D. C ,
1953.
[4] L. Obert, S. L. Windes, and W. I. Duvall, "Standardized Tests for Determining
the Physical Properties of Mine Rock," Report No. Rl 3891, Bureau of Mines,
1946.
[5] G. E. Troxell and H. E. Davis, Composition and Properties of Concrete,
McGraw-Hill Book Co., New York, 1956.
[6] J. A. Talobre, "Tests," General Report, Development and Resources Corp.,
New York, N. Y., 1962.
[7] L. H. Robinson, "The Mechanics of Rock Failure," Quarterly, Colorado School
of Mines, Golden, Colo., Vol 54, No. 3, 1959, pp. 136-177.

Techniques Used in Studying the Fracture

Mechanics of Rock

REFERENCE: B. W. Paulding, Jr., "Techniques Used in Studying the

Fracture Mechanics of Rock," Testing Techniques for Rock Mechanics,
ASTM STP 402, Am. Soc. Testing Mats., 1966, p. 73.
ABSTRACT: Partially broken material was obtained to study the growth
of cracks during brittle fracture. In uniaxial compression this was ac-
complished by placing a simply supported beam in parallel with the
rock specimen to limit the advance of the ram when the load-carrying
ability of the specimen decreased. Uncertainties associated with straight
cylindrical specimens were avoided by using cylinders with a reduced
central section. Volumetric strain was computed from two perpendicular
strain gages. The onset of crack growth and the work expended against
confining pressure were found with the aid of plots of volumetric strain
versus stress. Measurements of linear compressibility indicated the pre-
dominant direction of crack growth and the increase in porosity.
KEY WORDS: rock (material), rock mechanics, fracture mechanics,
crack propagation, Griffith theory, compression tests

Up to now, study of the mechanics of the brittle fracture of rocks has

been primarily oriented toward determining the applicability of the
Griffith theory [1, 2]. 2 Griffith postulated that materials contain flaws
or cracks and that, in compression, large tensile stresses exist near the
ends of certain critically oriented cracks when the material is stressed.
Griffith proposed that fracture occurred when the most severely stressed
crack propagated.
Until recently, there was a substantial amount of evidence that the
Griffith model of fracture was valid. This included agreement of ex-
perimental data of several investigators with the McClintock-Walsh
modification of the Griffith theory [3], and the observation that, if grain
boundaries are considered to be "Griffith cracks," the compressive
strength of certain rocks was related to crack length as Griffith had
proposed [4],
1
Research engineer, IIT Research Inst., Chicago, 111.
• The italic numbers in brackets refer to the list of references appended to this
paper.
73

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
74 TESTING TECHNIQUES FOR ROCK MECHANICS

FIG. 1—Rock specimen.

However, a recent study of the mechanics of crack growth in rocks

showed that fracture criteria based on the Griffith model could not be
expected to predict the compressive strength of common rocks [5]. This
study necessitated the use of several types of experiments and techniques
of data interpretation which were very beneficial in understanding certain
characteristics of crack growth. Some of these techniques are presented
here in hopes that they may be useful to others studying the deformational
behavior of materials.
The experimental results were obtained from tests on Westerly granite,
a typical, siliceous, crystalline rock which is known to be brittle [6].
The mineralogy of Westerly has been described elsewhere [7].

Description of Techniques

New Specimen Shape

A study of crack growth in homogeneous materials [8] and photo-
elastic analysis of arrays of cracks [5] were completed prior to the tests
on rock. In order for a meaningful comparison to be made between crack
growth in the rock specimens and the predictions of the previously

Lateral Strain Gage

FIG. 2—Specimen position during uniaxial compression.

mentioned studies, the state of stress in the specimens had to be known

exactly.
Straight circular cylinders were first considered. However, the stress
in such specimens is not uniform because of the lateral constraint im-
posed on the ends of the specimen due to the elastic mismatch between
the platens of the testing machine and the specimen [9, 70]. The lateral
constraint imposed on the specimen results in a nonuniform strain field
which commonly manifests itself in barrelling of the specimen [9, 11].
Fractures are commonly observed to begin at the ends of cylindrical
specimens, and the strength is found to increase when the development of
these premature fractures is suppressed [11]. To avoid the uncertainties
associated with these phenomena, specimens of the shape shown in Fig.
1 were used. The total length was 2 in. The length and diameter of the

reduced central section were 1 and 0.5 in., respectively. The specimens
were ground from cores in a cylindrical grinder by shaping the edges of
the grinding wheel to the radius of the fillet and sweeping back and forth,
taking cuts of about 0.003 in. About 15 min additional preparation time
is required for this specimen over that of a straight cylinder of similar
size.
Of obvious concern with specimens of this shape is the effect of the
rather sharp fillet. Theoretically, there is a stress concentration factor of
1.25 at the fillets [72], that is, the compressive stress at the fillets is 25
per cent higher than in the throat region. In uniaxial compression the
specimens failed prematurely along a vertical fracture which went

2.5

2.0
J Lateral / -//Axial

1.5 -

/
m 1.0 _

- /

1 1 1 1 1
0 500 XIO-* 1000 1500 2000
2500 3000 3500 4000 4500 5000
Stroin
FIG. 3—Stress versus strain curves obtained during a uniaxial compression test.

through the fillet. This difficulty was overcome by tightening a steel ring
around the heads of the specimen. This precaution was not necessary
in the confined compression tests, the lowest pressure of which was 0.5
kilobars (kb).^ No other difficulties were encountered, and the agree-
ment with values of Young's moduli and compressive strengths obtained
by Brace [6], on shaped cylinders with a stress concentration factor of
only 1.02 [12], indicated that the stress state in the throat region was
uniform and known. In addition, visual observation of the specimen
during uniaxial compression tests indicated that the fractures started in
the throat region, well removed from the fillets.

' 1 kilobar = 14,500 psi.

Stiff Loading System

It was required that partially broken material be obtained so that
the growth and coalesence of cracks leading to fracture could be studied.
Since it was felt that a large amount of crack growth occurs during the
final stress increments, the specimens had to be loaded to just below the

^ Porosity

Volumetric Strain
FIG. 4—Compressibility of rock containing narrow cracks.

fracture strength and yet be recovered, intact. This required that the
press-rock system remain stable even though the load-carrying ability
of the specimen had decreased. In uniaxial compression this stability
was achieved by incorporating a stiffening element into a conventional
hydraulic press. The stiffening element, which was a simply supported
steel beam, was placed in parallel with the rock specimen as shown in
Fig. 2.
In general, it was possible to load a specimen until the magnitude of
the strains and visible surface cracks indicated that fracture was impend-

ing and yet recover the intact specimen. Typical stress-strain curves
obtained during tliese stiff uniaxial compression tests are shown in Fig. 3.
The curves designated by the words "axial" and "lateral" represent,
respectively, the strain parallel with and perpendicular to the direction
of maximum compression. The positions of the axial and lateral strain
gages are indicated in Fig. 2. The offsets in the curves occurred when the
stress was held constant for intervals of about 5 min. Details of the
testing procedure are described elsewhere [5].
The large permanent lateral strain indicated an increase in volume.
This was interpreted as due to an increase in porosity from either the

0.5

-250«io -lOQO -1250 2000

Lineor Sfroin
FIG. 5—Linear compressibility before and after uniaxial compression test.

loosening of grain boundaries or crack growth. In addition, the fact that

there was very little permanent axial strain suggested that the direction
of new cracks was parallel with the direction of applied compression. It
was thought that the effect of narrow cracks on the compressibility of
rock could be used to determine the orientation of the new cracks.

Use of Compressibility Tests in Determining Orientation of Cracks

Figure 4 shows a typical hydrostatic pressure versus volumetric strain
curve obtained during a compressibility test on a jacketed specimen of
rock which contains narrow cracks. At low pressures the compressibility
is high, that is, there is a large volumetric strain for a small increase
in pressure. This is due to the closing of cracks [13]. As the pressure
is raised more cracks close, and the rock becomes noticeably stiffer.

Eventually, all the cracks are closed, and the compressibility curve is
essentially linear. The slope of the linear portion is governed by the
elastic properties of the solid material and can be predicted to within a
few per cent by averaging the compressibility of the individual minerals
[14]. Walsh [13] has shown that the porosity due to narrow cracks can be
found by extending the linear portion back to the strain axis as shown in
Fig. 4, which, in eifect, subtracts the volumetric strain due to the com-
pressibility of the solid material from the total volumetric strain.
It is known from elastic solutions [15] that the deformation of a crack
depends, in part, on its orientation with respect to the direction of applied
compression. In particular, narrow cracks which are parallel with the
direction of compression experience very little deformation, whereas
those perpendicular to the compressive stress will readily close. There-
fore, if there is a preferred orientation of new cracks, it should be
obvious from comparison of the linear compressibilities in the axial
and lateral directions. The linear compressibility should be greater in the
direction perpendicular to the long axis of new cracks.
The linear compressibilities of jacketed specimens were measured
before and after they were subjected to uniaxial compression. Typical
results are shown in Fig. 5. The output of the axial and lateral strain
gages are denoted by A and L, respectively. The subscript o denotes
the original compressibility, and the primes denote the compressibility
subsequent to the uniaxial test. The arrows indicate whether the curve
was obtained during increasing or decreasing stress. In this particular case
the specimen was uniaxially stressed to 2.1 kb, about 90 per cent of the
compressive strength.
Notice that the compressibility in the axial direction is only slightly
greater than it was prior to the uniaxial test. On the other hand, there was a
marked increase in lateral compressibility. Results such as this indicate
that the predominant direction of crack growth was parallel with the
direction of uniaxial compression. This conclusion was verified by micro-
scopic examination of partially broken specimens [5].

Determining the Onset of Crack Growth and PdV Work with Plots of
Volumetric Strain Versus Maximum Stress Difference
The volumetric strain was computed from the output of the two
strain gages by adding the axial strain to twice the lateral strain. This
computation is valid if the original circular cross section remains circular
or if it deforms into an ellipse, and the lateral strain gage extends halfway
around the circumference of the specimen (see Appendix).
Plots of volumetric strain versus stress were used to determine the
onset of crack growth. Earlier attempts at detecting crack growth con-
sisted of sonic techniques [16] and detection of surface cracks by means
of anomalous behavior of strain gages [17]. The disadvantage of the

sonic method is tliat crack growtli at the ends of the specimen, where it is
in contact with the platens, may be detected in addition to the crack
growth which occurs in the central region of the specimen. The limitation
of the second technique is that it detects only the near-surface cracks
which occur directly beneath a strain gage.
A typical plot of volumetric strain versus stress obtained during a
uniaxial compression test is shown in Fig. 6. The rapid decrease in volume

ZSOOxio''

2000 -^.^^

1500 -~

1000 -
c
s
Z 500 -
at
E
2 0 ^
o
>
-500 - ^~"^
^^"^^te^^A p-1
-1000 - "" -

-1500 -
[
-2000 1 ! 1
05 1,0 1.5 2.0 2.5
Stress,kb

FIG. 6—Volumetric strain versus stress curve obtained during a uniaxial com-
pression test.

up to about 0.5 kb is due to the closing of cracks oriented so as to

close under the applied stress. Similar effects due to the closing of cracks
are the initially low values of Young's modulus and high compressibility
of certain rocks [6] and the rapid increase in sonic velocities with con-
fining pressure [18]. Above approximately 0.5 kb, the variation of the
volumetric strain with stress was essentially linear up to a stress of 1.05
kb. This represents the elastic deformation of the rock.*
At slightly greater than 1 kb the plot of volumetric strain versus stress
departs from the linear portion of the curve. This departure from

* J. B. Walsh, private communication, 1964.

linearity indicates the onset of crack growth. As additional stress was

applied, the departure of the volumetric strain from elastic behavior in-
creased. Thus, crack growth continued as the stress was raised. The
vertical offsets represent the volume increase due to crack growth which
occurred during the 5-min intervals when the stress was held nearly
constant.
The interpretation of the departure point as indicating the onset of
crack growth was substantiated by measuring the linear compressibility

, Fracture

4000

1
3000

i
2000 -

i 1000-

Vc
•/
-1000
• 7
-2000 — ^*,^
A ^ •
•^^!!*-S^ • __,%j^

-3000 --

-4000 "
1

-5000 i 1 1 1 1 1 A' 1 i 1
0 I 2 3 4 5 6 7 8 9 10 II 12
Stress, kb
F I G . 7—Volumetric strain versus maximum stress difference curve obtained
during a compression test at a confining pressure of one kitobar.

of specimens cycled to various levels of uniaxial compressive stress.

There was no measureable increase in porosity until a specimen had been
stressed to a level which exceeded the point at which the curve of
volumetric strain versus stress departed from the linear portion. Further-
more, there was a correspondingly greater increase in porosity as the
specimen was cycled to a higher uniaxial stress.
This method of determining the onset of crack growth worked equally
well for confined compression tests even though confining pressure tends
to inhibit an increase in volume. Figure 7 shows a typical plot of
volumetric strain versus maximum stress difference for a specimen loaded
to failure under a confining pressure of 1 kb. In this particular test the
onset of crack growth occurred at about 4 kb.

As shown in Fig. 7, there was a pronounced increase in volume,

relative to simple elastic behavior, associated with the growth of cracks.
This increase in volume, in the presence of a high confining pressure,
indicates that some of the work required to fracture the specimen was
expended against the pressure medium. In order to determine if this
effect is of sufficient magnitude to be considered in a fracture theory, the
work against the pressure medium was found and compared with the
total work required for fracture.
The volumetric strain due to crack growth was found by extending the
linear portion of the volumetric strain versus stress curve to the stress
at fracture (Line A — A' in Fig. 7). This is the deformation which would
occur in the absence of crack growth. The volumetric strain due to
crack growth, (AF/F),,, then, is the difference between the actual
volumetric strain at fracture and the volumetric strain at point A'.
The PdV work per unit volume was found by multiplying this volumetric
strain due to crack growth by the confining pressure.
The magnitude of the total work was found by measuring the area
under the stress versus axial strain curve with a planimeter and multi-
plying the area times the work represented by a unit area. Comparison
of the PdV work per unit volume and the total work required for frac-
ture indicated that about 15 per cent of the total work is expended against
the pressure medium [5].

Evaluation o£ Techniques
For the benefit of investigators who are involved in studying the frac-
ture mechanics of brittle materials it is appropriate that the techniques
presented in this paper be critically evaluated in the event their use may
be considered.

Specimen Shape
The merit of shaped specimens is obvious when one considers the un-
certainties associated with straight cylindrical specimens. Although the
particular shape used in this study fitted the purpose of the investigation,
there are some limitations which should be pointed out. The rather
sharp fillet, with the associated stress concentration factor of 1.25,
precludes using this specimen for measuring uniaxial compressive
strengths unless the premature failure of the heads is prevented. How-
ever, this procedure is not particularly recommended, since the stress
state in part of the throat region may be influenced by the device
used to strengthen the heads (in this case, steel rings). It is suggested,
therefore, that specimens having a fillet of larger radius, such as those
used by Brace [6], be used for uniaxial compression tests. In the con-
fined compression tests, on the other hand, the rather sharp fillet was
apparently of no consequence. To minimize preparation time, then, it

may be advisable to consider using specimens with a fillet of smaller

radius for confined compression tests. An analytic or photoelastic
analysis should be conducted on any proposed specimen shape, so that
the investigator is assured that the stress state is unform and known in
the region of interest.
Another limitation of the specimen used here is the small volume of
material which is subjected to stress. For Westerly granite this was
thought to be unimportant, since the specimen dimensions were much
larger than the discontinuities (for example, grain boundaries, cracks) of
the rock. However, for determining the mechanical properties of materials
which have larger discontinuities a larger specimen is certainly preferable.

Stiff Loading System

For investigations where it is desirable to obtain partially broken
material, the ability to control the fracture process has obvious merit. The
stiffening element used in this study allowed the specimen to be stressed
to a point at which the strains were equal to those obtained previously at
complete fracture [6] and yet recover the intact specimen for examina-
tion. It is doubtful if this could have been accomplished without the
stiffening element. However, even with a stiff system, it is perhaps overly
optimistic to expect strains much in excess of those obtained at fracture
in a conventonal loading system when testing brittle materials.

Use of Compressibility Tests

Walsh's analysis [13] of the effect of narrow cracks on the com-
pressibility of rocks proved to be very beneficial in determining not
only the existence of new cracks but also their predominant orientation.
Observation of thin and polished sections of the partially fractured speci-
mens supported the results of the compressibility tests which indicated
that the predominant direction of crack growth was parallel with the
direction of maximum compression. In addition, the increase in porosity
due to crack growth as determined from the compressibility tests agreed
quite well with the permanent volumetric strain obtained during the
uniaxial compression tests [5]. Therefore, it is felt that the method is
reliable and well suited for investigations where indications of the
existence and predominant orientation of original or stress-induced
cracks are desirable for a better understanding of the mechanical
properties of the material.

Use of Plots of Volumetric Strain Versus Stress

This technique provides a convenient means of determining the onset
of crack growth without interruption of the test. It appears to avoid the
uncertainties associated with the other crack-detection techniques men-
tioned earlier. In these particular tests it was felt that the stress at the

onset of crack growth could be determined to within about 10 per cent.

This value was judged from how accurately one is able to determine the
point at which the volumetric strain versus stress curve departed from
linearity. If it is not necessary to obtain the linear strains individually,
the accuracy can be improved by incorporating one axial and two lateral
strain gages into a Wheatstone bridge. The output of the bridge will then
be directly proportional to the volumetric strain.

A cknowledgments
This study was conducted while the author was a graduate student in
the Department of Geology and Geophysics of the Massachusetts In-
stitute of Technology. Finanical support was provided by the National
Science Foundation, Project GP 1470. This support is gratefully acknowl-
edged. Special appreciation is extended to W. F. Brace and J. B. Walsh.
Their pertinent suggestions were of valuable assistance throughout the
investigation.

APPENDIX
The volumetric strain experienced by an anisotropic material may be deter-
mined according to

-p- = «1 + 62 + «3 (1)

where «i, 62, and €3 are the principal linear strains. In this study the volumetric
strain was determined by summing the linear strains according to

^ = 6, + 26. (2)

where CA and e. are the strains measured by the axial and lateral strain gages,
respectively.
Use of Eq 2 assumes that
^2 + t3 ,,,
iL = 2 *• -'

If the lateral strain gage extends over half the circumference of the specimen and
if the originally circular cross section of the specimen becomes elliptical, then
Eq 2 gives the true volumetric strain experienced by the specimen. This is shown
in the following analysis.
Imagine that the originally circular cross section of radius /•„ deforms into an
ellipse of major and minor axes, a and b, respectively. If e^ and €3 are the princi-
pal strains in the plane of the cross section then
a = ro (1 -t- 62)\ ,^^
b = roil +€3)1 "• '

The original length, /„, of a strain gage which extends over half the cir-
cumference of the specimen is
/o = ir/-o (5)

The final length, /, is to a small approximation [19],

[^]
^2-11/2

l=\'^^~-\ (6)

The strain experienced by the strain gage is

(L = —.— (7)

Substitution of Eqs 4, 5, and 6 into Eq 7 gives

€t =
1
XI
2 J -""-"
irr„
- ..211/2

[,+.. + .. + «i^J-_, (8)

The strains are of the order of ICr^. Therefore, to a small approximation,

e^ = [1 + ( e 2 + f 3 ) r - 1 (9)
Using the binomial expansion
(1 + x)" = 1 + nx + higher powers of x
ones finds, from Eq 9 that

€2 + es

to within an error of the order of e'.

Therefore, for the conditions specified, Eq 2 gives the true volumetric strain.

References
[1] A. A. Griffith, "The Phenomenon of Rupture and Flow in Solids," Philo-
sophical Transations, Royal See, London, A., Vol 221, 1921, pp. 163-197.
[2] A. A. Griffith, 'Theory of Rupture," Proceedings, First International Con-
gress Applied Mechanics, Delft, 1924, pp. 55-63.
[3] F. A. McClintock and J. B. Walsh, "Friction on Griffith Cracks in Rocks
Under Pressure," Proceedings, Nat. Congress Applied Mechanics, 4th, Berke-
ley, 1962, pp. 1015-1021.
[4] W. F. Brace, "Dependence of Fracture Strength of Rocks on Grain Size,"
Bulletin No. 79, Pennsylvania, State University Mineral Export Station,
1961, pp. 99-103.
[5] B. W. Paulding, Jr., "Crack Growth During Brittle Fracture in Compression,"
Ph.D. thesis, Massachusetts Institute of Technology, Cambridge, Mass., 1965.
[6] W. F. Brace, "Brittle Fracture of Rocks," State of Stress in the Earth's Crust,
W. R. Judd, editor, Am. Elsevier Publishing Co., New York, 1964, pp. 110-
178.
[7] Francis Birch, "The Velocity of Compressional Waves in Rocks to 10

Kilobars," Part I, Journal Geophysical Research, Vol 65, 1960, pp. 1083-
1102.
[8] W. F. Brace and E. G. Bombolakis, "A Note on Brittle Crack Growth in Com-
pression," Journal Geophysical Research, Vol 68, 1963, pp. 3709-3713.
[9] C. W. Richards, Engineering Materials Science, Wadsworth Publishing Co.
San Francisco, 1961.
[10] L. N. G. Filon, "On the Elastic Equilibrium of Circular Cylinders Under
Certain Practical Systems of Loads," Philosophical Transactions A, Vol 198,
1902, pp. 147-233.
[U] T. R. Seldenrath and J. Gramberg, "Stress-Strain Relations and Breakage of
Rocks" Mechanical Properties of Non-Metallic Brittle Solids, W. H. Walton,
editor, Interscience Publishers, Inc., New York, 1958, pp. 79-102.
[12] R. E. Peterson, Stress Concentration Design Factors, John Wiley and Sons,
Inc., New York, 1953.
[13] J. B. Walsh, "The Effect of Cracks on the Compressibility of Rock," Journal
Geophysical Research, Vol 70, 1965, pp. 399-411.
[14] W. F. Brace, "Some New Measurements of Linear Compressibility of Rocks,"
Journal Geophysical Research, Vol 70, 1965, pp. 391-398.
[15] S. Timoschenko and J. N. Goodier, Theory of Elasticity, McGraw-Hill
Book Co., New York, 1951.
[16] L. Obert and W. Duvall, "The Microseismic Method of Predicting Rock
Failure in Underground Mining, Part II-Laboratory Experiments," Report of
Investigations No. 3803, U. S. Bureau Mines, 1945.
[17] F. A. Blakey and F. D. Beresford, "Tensile Strains in Concrete, Part I,"
Report C2.201, CSIR, Division of Building Research.
[18] Francis Birch, "The Velocity of Compressional Waves in Rocks to 10
Kilobars, Part II," Journal Geophysical Research, Vol 66, 1961, pp. 2199-
2224.
[19] Handbook of Chemistry and Physics, The Chemical Rubber Publishing Co.,
Cleveland, 1961, p. 343.

Capabilities of Photoelastic Coatings for

the Study of Strain in Rocks

REFERENCE: "Capabilities of Photoelastic Coatings for the Study of

Strain in Rocks," Testing Techniques for Rock Mechanics, ASTM STP
402, Am. Soc. Testing Mats., 1966, p. 87.
ABSTRACT: Rock strain can be studied utilizing continuously and
peripherally cemented photoelastic sheets and frozen-stress ring photo-
elastic gages in integrating mounts. Standard errors of estimate of strain
differences, as calculated from normal incidence, range up to 20 ^in./in.
Standard errors of estimate of principal strains taken separately range up
to about 30 Min./in. Strain levels as low as several tens of /un./in. can be
measured with confidence. Strain levels calculated from photoelastic ob-
servations and elasticity theory typically differ from each other by 10 per
cent. Results are consistent among continuously bonded and integrating
mounts, rectangular and circular sheets, different loading schemes, and
different lighting and viewing arrangements. Frozen ring photoelastic
gages are relatively easily read and indicate directions of principal strains.
Work with rock specimens gives results in accord with those obtained
from work with aluminum bars. Photoelastic coatings are very sensitive
indicators of fractures and other types of anisotropy. This method is
applicable to many types of investigation.
KEY WORDS: photoelasticity, photoelastic coatings, residual stresses,
strain optical constant, reflection polariscope, rock (material), rock
mechanics, stress analysis

The purpose of this paper is to present some information on the

capabilities and limitations of the photoelastic coating technique for the
measurement of strain in rocks. This evaluation is based on both labora-
tory and field observations. The devices used have included continuously
and peripherally cemented (integrating) photoelastic sheets and frozen-
stress ring photoelastic gages in integrating mounts. Responses of a
variety of rock types and of 2024 T-4 aluminum test bars comprise the
bases for the evaluation.
This paper does not include evaluations of all commercially available
photoelastic gages, photoelastic plugs with self-contained light sources, or
photoelastic models, nor are dynamic responses part of this study. Work
with these items comprises unfinished business.

1
Professor of geology, The Ohio State University, Columbus, Ohio. Personal
member ASTM.
87

Copyright" 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
88 TESTING TECHNIQUES FOR ROCK MECHANICS

Previous Work
Zandman [7]^ and his colleagues [2-5] have published extremely
helpful notes on the principles and applications of the photoelastic coat-
ing technique. Of the many other informative articles available on photo-
elastic coatings, Refs 6-12 are cited to indicate the coverage available.
Roberts, Emery, and their associates have applied photoelastic coat-
ings to the study of strain in rocks in a variety of situations [13-18], and,
in fact, some of their early reports were my first published source of in-
formation in undertaking this project. In addition, I have had helpful
communications on the application of photoelastic coatings from John
McWilliams of the Minneapolis laboratories of the U. S. Bureau of Mines
and Don Preston of the Houston laboratories of Shell Research Corp.
Considerable use has been made of ideas underlying techniques from
the study of strain in other materials, such as metals.

Basic Concepts

The photoelastic coating technique is based on the fundamentals of

classical photoelasticity. In the coating technique, however, instead of
constructing a photoelastic model and analyzing its behavior, a photo-
elastic coating bonded to the actual material to be studied yields a
birefringent signal that is a function of the surface strain in the material. It
is presumed, with considerable experimental verification, that the coat-
ing's strain is a faithful replica of the test material's strain or that the
latter can be calculated from the birefringent signal in the coating.
When the coating and the test material are bonded continuously
throughout their entire area of contact, variations in strain within that
area can be detected. The typical practical limit of areal resolution is
several times the thickness of the photoelastic coating.
When the coating is bonded around its periphery to the test material, it
yields an integrated signal over the area of bonding. Strain gradients over
the bonding area can introduce vitiating complications. However, valid
interpretations can be made for zero and linear strain gradients.
From normal incidence procedures, one can determine the difference
between principal strains and, of course, their directions. Given an addi-
tional optical path, such as can be obtained from oblique incidence, one
can calculate the magnitude of each principal strain.
Calculations of stresses from strains follow classical procedures. The
greatest difficulty here stems from uncertainties about values of elastic
moduli of rocks, effects of anisotropy, and the like.

''The italic numbers in brackets refer to the list of references appended to this
paper.

Instruments and Materials

The photoelastic coating technique requires that plane or circularly
polarized light be transmitted into the photoelastic coating, that it be
reflected at the inner surface of the coating, and that it then be passed
through plane or circularly polarizing elements, respectively, for con-
ventional photoelastic analysis [/].
A reflection polariscope replaces the transmission polariscope of classi-
cal photoelasticity. A scaled-down polariscope for making observations
in boreholes has proved to be a very useful tool in this work. Reflection is
accomplished by using a reflective cement as the bonding agent or by
using a reflective backing on the photoelastic sheet.
For some simple operations, the polariscope may be replaced by a
single sheet of circularly polarizing material placed over the outer surface
of the photoelastic coating but not bonded to it. This is the basic con-
figuration for commercial photoelastic strain gages, requiring only a
source of ordinary white light.
Although techniques are available for placing coatings on nonplanar
surfaces, the results in this paper are based upon the application of planar
sheets to planar surfaces. However, these surfaces need not remain planar
during deformation, as in bending beams.

Photoelastic Coatings
The photoelastic coatings used in these studies are approximately
0.04, 0.08, and 0.12 in. thick. Thicknesses are furnished by manu-
facturers^' * to ±0.001 or 0.002 in. Thicknesses were verified by
mechanical micrometry and optical microscopy, that is, from travel of the
microscope tube between focus positions for the upper and lower sur-
faces of the coating.
The strain optical constant (strain sensitivity) of the coatings used
ranged from 0.12 to 0.16 (0.14 typical), where

Strain optical constant

_ retardation
twice coating thickness X principal strain diflference
For a strain optical constant of 0.14, the strain required to produce a
retardation of one fringe would be about 675 /xin./in. for a coating 0.12
in. thick and about 1015 ;nin./in. for a coating 0.08 in. thick. Consistent
readings to 0.02 fringe are not unusual.
The greater the thickness of the coating, the greater is its gross optical

' The Budd Co., Instruments Div., Phoenixville, Pa.

* Photolastic, Inc., Malvern, Pa.

sensitivity. However, increasing the coating's thickness results in the fol-

lowing: (a) decrease in horizontal resolution of signals from small struc-
tural units, such as crystals; (b) increase in width of border zone of
unusable readings, resulting from edge effects in coatings not entirely
covering the test surface; (c) increase in the reinforcing effect of the
coating on the test material if the coating's thickness becomes an ap-
preciable fraction of the total thickness of specimen and coating; and
(d) in oblique incidence, increase in the horizontal component of the ray
path and, hence, decrease in horizontal resolution. In connection with c,
reinforcing and bending corrections are available [2]. In connection with
d, the size of the horizontal component is easily calculated; thus, for each
0.01 in. thickness of coating, and at 60 deg incidence, there will be 0.013-
in. spacing between the entering and emerging rays on the surface of the
coating.
The temperature stability of the photoelastic materials used so far has
been quite satisfactory. That is, the range of temperatures encountered
has been within the safe range specified by the manufacturer (below 320
F) for maintaining an essentially stable strain sensitivity. However, tem-
perature changes can enter the picture as a disturbing factor by causing
strain in rock specimens. This effect has been noted when rocks that
had been coated and studied underground have been taken up to ground
level for further study and when specimens have been heated in the
laboratory.
Machining photoelastic coatings and drilling through them into speci-
mens can introduce parasitic birefringence, but techniques are available
for suppressing this effect. Further, where these effects are objectionable,
they usually can be removed in an annealing cycle. Greenwald'^ has
recentiy reported on procedures for drilling clean holes through coatings
and metal specimens.
Moisture can cause some problems, particularly in connection with
bonding under the humid conditions prevailing in some mines. There is
no best method for dealing with this problem other than working out
drying procedures that are effective in each situation.
The bonding agents that have seemed to work out the best are pre-
pack 12-hr epoxy cements. Drying time is longer than one might desire,
but the results obtained have generally been good. A 4-hr cement now
commercially available has not been evaluated by us in the field; an
uncontrolled laboratory test with an early, pilot issue has resulted in
spurious photoelastic patterns. Other products, including high-humidity
epoxy cements developed after our initial success with 12-hr cements and
quick-curing contact cements, have not been evaluated by us.

^' W. E. Greenwald, "Investigation of Residual Stresses Using Photoelastic Coat-

ings," unpublished manuscript of paper presented at Second International Congress
on Experimental Mechanics, Sept. 28-Oct. 1, 1965, Washington D. C.

Standard Photoelastic Gages

Photoelastic strain gages are available in a variety of shapes and types
[6]. Their chief virtue is that they provide a rapid means for obtaining
strain data without requiring a polariscope.
Some of the gages that are commercially available utilize concentric
frozen first and second-order stress rings in 114-in.-diameter circular
patches for indicating direction and magnitude of principal strains [19].
Other commercial circular gages" use the patterns developed around a
central hole in unstressed material to determine direction and relative
magnitudes of principal strains.
A circular gage developed in this project utilizes a frozen zero-stress
ring (Fig. 1) as the basic indicator. The gage is manufactured by first
placing an aluminum pipe cap, about % in. diameter, with its open end
down and centrally located, on a 2V8-in.-diameter photoelastic coating
that is 0.08 in. thick. This assembly is then run through a slow annealing
cycle. On cooling, the zero-order ring appears beneath the cap, with
slight residual compression and tension inside and outside of the ring,
respectively. Attempts to make smaller gages have met with only spotty
success.
All three gage types mentioned in the foregoing are mounted by
peripheral bonding, to achieve integrated strain over the area of the gage.
The photoelastic patterns must be bilaterally symmetrical about mutually
perpendicular axes (principal strains) in order to be useful. This require-
ment precludes effective use of these gages in nonuniform or nonlinear
strain fields, but there are many situations in which they can be very
useful.
Other types of commercial photoelastic gages are available, as for
example rectangular gages that can be used for monitoring strain on
the columns of loading frames.

Some Problems Involved in Studying Strain in Rocl(s

The valid and effective application of any technique to the study of
strain in a material requires consideration of the relevant properties of
that material.
Rock is, in fact, many materials, in terms of both the properties of the
constituent minerals and the properties of the aggregates. An effective
strain-measuring technique for one rock might not work at all for some
other rock. There is often the need to make field measurements under
adverse conditions, effectively beyond the control of the investigator.
This situation is further complicated by the almost complete absence of
widely accepted standards for rock properties.

° Strainline Photoelastic Strain Compass after Oppel [6]: Baldwin-Lima-

Hamilton, Electronics and Instrumentation Div., Waltham, Mass.

These points have much bearing on where and how photoelastic coat-
ings can be used to study strain in roclcs, on the design of polariscopes,
the selection of coatings and cements, and the use of accessory photo-
elastic devices.
Anisotropy of rocks presents many challenges—whether one is work-
ing with anisotropy of the rock mass or with anisotropy across grain or
crystal boundaries or within individual crystals. In some cases, groups or
domains of crystals behave as structural units.
Bedding, foliation, and banding introduce a kind of systematic

Frozen ring integrating gage mounted on aluminum bar (same layout as in Fig.
3). Scribed sheet of circularly polarizing material covers gage. Spacing between
scribed circles in vicinity of frozen ring is 0.1 in. .Sf-axis parallel to long axis of
bar. Polaroid Type 57, f l l , 1 min, white light.
FIG. 1—No load on cantilever.

anisotropy. Incipient, hidden, or open fractures, from the microscopic

scale up through extremely large discontinuities, are very common in
rocks and account for anisotropy of more than one type at different
scales, stress levels, or magnitudes of strain. Stresses locked up in the
rocks as the result of superincumbent loading, tectonic history, cooling,
and the like, can contribute to anisotropy in ways that defy accurate
evaluation at this time.
Given anisotropy in rocks, whatever its cause or basis, the measure-
ment problem can be quite vexing, and interpretation—especially in
terms of stress—can be extremely difficult if possible at all. Classical
elasticity theory does not have solutions to some of the critical problems.
One of the goals of those working with photoelastic coatings should be
to attempt to determine the relation between grain-by-grain ansotropy
and aggregate anisotropy.

Another problem involved in studying strain in rocks arises from their

low tensile strength, typically about one-seventeenth the compressive
strengths This means that experimental studies, in the field and in the
laboratory, are limited by the tensile strength of the rocks if tensile
stresses are part of the stress field. For example, experiments with
flexed rock beams and with uniaxially compressed slabs or blocks into
which holes have been drilled are seriously limited by low tensile strength,
even though compression and shear might be of primary interest in the
study.

Strain is approximately 70 iuin/in. Tension is parallel to A'-axis. Distortion of

frozen ring is continuous for loads below 3 lb. Effects of 1-lb increments (strain
less than 24 iuin/in) between 0 and 3 lb can be detected without difficulty.
Photography same as in Fig. 1.
FIG. 2—Three-lb load on same cantilever as shown in Fig. 1.

Here, the limitation on tensile stress level that can be achieved without
producing fracture may result in a very low level of birefringence in the
photoelastic coating. Effective operations may require optical enhance-
ment of the signal by such methods as increasing the thickness of the
coating, which decreases resolution, or by introducing into the optical
path a full-wave optical plate, which can result in some aberration in
color values and loss of precision in retardation measurements.
Another problem in working with rocks stems from the need for in situ
measurements, for example, in mines and quarries.
Some studies require work in boreholes that penetrate the rock beyond
the influence of the free face and working cavity. This means, then, that
the photoelastic coating, which is placed on the face at the far end of the

' Unpublished report from D. E. Fogelson, Minneapolis Mining Research

Center, U. S. Bureau of Mines, 1963.

hole, must be placed, cemented, and observed at a working distance

requiring special equipment. Overcoring the rock to which the coating is
bonded also requires special care, and the coating must be observed
promptly after coring if time-dependent sequences are to be observed.
In working with mine rocks, it is usually necessary to make photo-
elastic measurements underground even on detached specimens, in order
to avoid spurious effects introduced by changes in the rock when it is
removed to ground level conditions.
Photoelastic patches may be bonded direcdy to pillar surfaces, roof
rock, and the like, to provide monitoring during and after excavation.
However, this technique might be of questionable value because strains
measured on such surfaces need have little systematic relation to the
stress field of real interest to the investigator. It is well known that frac-
tures generated by excavating underground openings can produce signifi-
cant aberrations in the stress field. Further, the mounting of a plane
strain indicator on a curved or irregular surface can lead to complica-
tions in interpreting results. Finally, coatings mounted on exposed sur-
faces are more prone to damage than are those mounted in holes.
It must be remembered that, in studying strain in rocks, thorough analy-
sis usually requires consideration of data on petrofabrics, deformation his-
tory of the area, rockburst and seismic history of the area, excavation
history, and hydrologic data bearing on rock behavior.

Laboratory Experiments

Aluminum Bars
Calibration studies of photoelastic coatings have been undertaken
utilizing 2024 T-4 aluminum bars as the working pieces. The bars have
been loaded as cantilevers and in third-point configurations. Photoelastic
coatings of several types have been used, as have some standard
photoelastic gages. In some cases, normal and oblique incidence measure-
ments have been made, in order to calculate magnitudes of principal
strains.
Using blank photoelastic patches annealed to remove parasitic
birefringence, straight-line plots of calculated strain versus deflection of
the aluminum bars show very little scatter. Standard errors of estimate of
the strain differences, as calculated from normal incidence, range up to
about 20 juin./in. For photoelastic coatings of similar thickness and strain
sensitivity and studied in both oblique and normal incidence, standard
errors of estimate for each of the principal strains taken separately range
up to about 30 ^uin./in.
From the data obtained so far, it appears that strain levels as low as
several tens of /xin./in. can be measured with confidence.

Comparison of strain levels calculated from photoelastic observations

and those calculated from elasticity theory for aluminum bars show
discrepancies that group around 10 per cent, skewed toward lower
percentages. For the range of strains to be expected in our laboratory
work with rocks, this is also equivalent to no more than several tens of
/iin./in. More often than not, strains calculated from photoelastic obser-
vations exceed those calculated from elastic theory, but there is not
enough evidence available to establish these as systematic discrepancies.
Data from continuously and peripherally bonded coatings yield con-
sistent results. Circular and rectangular coatings also give consistent
results. Third-point loading, which gives uniform strain between the two
loading lines, and cantilever loading, which gives a linear strain gradient
along the axis of the cantilever, give consistent results for integrating
mounts. Work with white light and monochromatic light yields com-
patible results, as does work with and without the use of a full-wave
plate. In short, a general exploration of standard techniques and ex-
pectable variations in applications yields results that hang together.
The behavior of frozen stress ring gages bonded to aluminum bars
indicates that these can be very helpful in some studies of rock deforma-
tion.
The manufacturer (see footnote 3) of the circular frozen ring gages
(first and second order; VA in. diameter; 0.125 in. thick) claims that
magnitudes of the separate values of principal strains can be read to 40
/tin./in., that principal strain directions may be determined for strain
levels as low as 50 /nin./in., and that principal strain directions can be
read to within 5 deg.
A study of strain increments in a progressively loaded aluminum
cantilever shows that the values obtained from this type of gage and
from elasticity theory differ, in general, by several tens of juin./in.,
which is also the magnitude of the standard deviations of strain incre-
ments for each of several groups of observations. At low-strain levels,
readings of magnitude and direction are quite erratic; not until strains
of 100 to 200 /An./va. are reached do the data begin to show some order.
These findings, which are preliminary, are actually not inconsistent
with the manufacturer's statements. More data and more experience
using this gage are needed. Its performance in the field, as noted later,
has been encouraging.
The frozen zero-stress ring gage (Figs. 1 and 2) developed as part of
this project gives a clear signal of strains as small as 24 juin./in. For small
strains, deformation of the rings' axes conform quite closely to second-
degree relations, although even linear equations give a fairly good
least-squares fit. Tardy compensation can be used to null small distor-
tions of the ring; second-degree equations give a good fit to compensa-
tions versus load, and linear equations give an acceptable fit.

LOAD
TOP

SIDE

Arrangement for aluminum bars the same, except that bars are Vi in. thick
FIG. 3—Cantilever arrangement for work with rock slab, using circular photo-
elastic coating 0.08 in. thick.

END TOP

Arrangement for aluminum bars the same, except that bars are Vt in. thick.
FIG. 4—Third-part loading arrangement for work with rock slab, using circular
photoelasiic coating 0.08 in. thick.

With the photoelastic materials used in these preliminary studies, the

zero-order ring degenerates at about 100 /tin./in. in white light. However,
this gage has a much larger range, for at significantly higher strain levels,
the zero-order ring becomes circular first, second, and higher order
fringes. These transitions occur at exactly the strain levels required for
blank patches of the same material to show these fringes; this has been
verified in monochromatic light. Thus, the distortion of the ring is the
fine scale, and its change of color as it becomes circular again is the
coarse scale. Variable monochromatic light sources should make it
possible to develop coverage not now available with white light or a single
monochromatic source.

Rock Specimens
A series of experiments was undertaken with slabs of four types of
rock loaded as cantilevers (Fig. 3) and in third-point configurations
(Fig. 4). A slab of Elberton granite, in third-point loading and with a
continuously bonded rectangular photoelastic coating, gave a response
with no more scatter than that observed for aluminum bars in the same

SR4 STRAIN GAGES

ROSETTE
SINGLE ELEMENT

FRONT

Photoelastic instrumentation on duplicate blocks was placed inside and around

the mouth of the drill hole, and in addition SR-4 gages were mounted elsewhere
to provide comparisons with blocks entirely instrumented with SR-4 gages.
FIG. 5—Rectangular blocks that were compressed axially parallel to the long
dimension.

experiment. The calculation of E from these results gave a reasonable

value for this rock type. Further, at surface strains in excess of 100
juin./in., a mottled pattern that is a good replica of the rock's surface
texture appeared as a low-level modulation of the gross birefringence.
The mechanics of generating this textural replica are unknown.
Another series of experiments was undertaken with rectangular blocks
of rocks of three types, measuring 12 by 12 by 18 in., and deformed
uniaxially by compression normal to the 12 by 12 in. faces (Fig. 5). Of
special interest is the strain around NX holes drilled 6 in. deep through
the center of one 12 by 18-in. face in each block. These specimens were

instrumented with photoelastic coatings, frozen-ring gages, and SR-4

gages, both on the face at the far edge of the hole and on the block's faces.
It should be kept in mind that these specimens are demonstrably
anisotropic. This work is the laboratory counterpart of field investiga-
tions of strain relief patterns observed in NX boreholes in mine rock.
Galle's analysis [20] is of particular relevance here.
Results achieved so far indicate sensitivities of photoelastic coatings
and frozen-ring gages that are generally consistent with other laboratory
results. A zero-order frozen ring scaled down to % in. diameter and
cemented to the far face of the hole has proved to be less sensitive than

In later experiments square patches were replaced by circular patches covering

the upper disk face, and some holes were drilled through the coating and specimen
with control holes drilled only through the coating.
FIG. 6—NX rock disk into which relief hole has been drilled after coating had
been mounted {Ref 21).

expected. However, it does show principal strain directions at strains

in excess of 100 /jn./in.
Work with one of these blocks has indicated the high sensitivity of
the photoelastic coatings to fractures. One fracture developed where
theory indicated that tension should be a maximum; yet, the fracture
could not be detected when the block was returned to zero load. Only
when the load was taken back up to a level beyond that at which the
fracture appeared could its hairUne trace be detected with a lOx hand
lens.
Strain patterns of low order birefringence (0.2 fringe) were generated
and studied in coatings cemented to the plane surfaces of NX size rock
disks, 1/2 in. thick, into which test relief holes were drilled (Fig. 6). An
appealing thought is that the signal can be accounted for as the result of

In region around %6-in. hole that penetrates the rock about a half of a
diameter, the very rough cross overrides the fine-textured pattern. This specimen
is a quartz-epidote rock, probably altered granodiorite, from the Blue Ridge.
Polaroid Type 57, fl6, 9 min, monochromatic green (mercury vapor) light, plane
polarization horizontal and vertical. Isochromatics of this specimen are less defini-
tive of fine texture than are isoclinics.
FIG. 7—Fine-textured isoclinic pattern around edges and in corners of
photoelastic coating resembles texture visible in adjoining rock surface.

.//////,^,/////.////y//////^//^/^^^/^/^/^/((^/^//^/^

21/8 I

NWWWWWWWWVsWWW k«««.«««<^^^<^

Surfaces on which coatings are mounted are ground flat but are not polished.
FIG. 8—(a) Photoelastic coating cemented to far face in drill hole and {b)
overcoring of coating to obtain relief pattern and to reach greater depth for next
mounting of another coating.

the relief of residual stresses. Greenwald (see footnote 5) has recently

reported on similar studies with metal specimens.
Photoelastic strain patterns were also generated and studied on the
end faces of N X cores, deformed along the major part of their length by
uniform biaxial (hydrostatic) pressure [27]. One arrangement here in-
volves the use of a photoelastic coating on one end of the core and an
SR-4 rosette on the other. The cores reveal some interesting aspects of
anisotropy, including differences in strain between adjoining beds and
evidence of shear along their interfaces.

Some of the work with both the disks and cores mentioned above has
already been published [22], and need not be repeated here. Additional
work has confirmed the need to develop reliable techniques for measur-
ing strain directions and magnitudes on the scale of individual grains
(Fig. 7). Isoclinics are often more sensitive indicators than isochromatics
of the presence of apparent grain-by-grain anisotropy. Whether aniso-
tropy at this scale is intergranular or intragranular, and whether or not
such strains are any indication of the fine-scale stress field beneath the
surface, are problems yet to be tackled.

Field Investigations
Field studies, utilizing overcoring techniques (Fig. 8) with photoelastic
and U. S. Bureau of Mines borehole gages, have been conducted at Iron
Mountain, Mo.; Moab, Utah; Colorado Springs, Colo.; Lyon Mountain,
N. Y.; and Barberton, Ohio.
These operations have posed difficulties described earlier in this paper.
The greatest single source of difficulty in terms of measurement has come
from the development of anomalous, asymmetrical patterns in some of
the standard photoelastic gages. In some cases, it has been possible to
estimate only the directions of the principal strains, and even this with
some trepidation.
About 80 per cent of the commercial frozen ring gages have yielded
patterns from which at least the directions and relative magnitudes of
principal strains can be determined. About 50 per cent of the gages
should yield data on strain magnitudes. These patterns have been re-
corded underground for interpretation.
About half of the zero-order frozen ring gages have yielded usable
patterns, that is, those from which at least the directions and relative
magnitudes of principal strains can be determined. The cementing
problem has been more serious with these gages than with the commer-
cial gages.
Some blank patches with continuously cemented mounts have also
been tried out, but none of these has produced signals that could be
used.
Conclusions
Photoelastic coatings can be placed where most other strain-indicat-
ing devices can be placed, except that the photoelastic method requires
access for viewing or photographing.
Photoelastic coatings have the great advantage of being able to pro-
vide either integrated strain data or a panorama of a nonuniform
strain field.
Integrated strain data yield reliable results, sufficiently sensitive for
many investigations in rock mechanics. Methods for working on fine-

scale problems, such as grain-by-grain strain, require refinement in

terms of both application and theory.
Problems posed by temperature and humidity, particularly in mine
situations, are generally not susceptible to solution as of this time, but it
is possible to by-pass most of them.

Possibilities for Future Investigations

Many types of studies utilizing photoelastic coatings can be initiated
or refined with reasonable prospects for useful results. Among these are
investigations of:
(a) Effects of temperature on fine-scale strain.
{b) Effects of water pressure on strain field.
(c) Effects in the rock during drilling while the rock is under load.
(<f) Fracture propagation under controlled conditions.
(e) Reliable monitoring techniques in hazardous areas.
(/) Techniques for studying residual stresses.
(g) Methods for measuring low birefringence, very small changes in
magnitude of birefringence, and fine-scale areal changes in birefringence.
{h) Deformation of curved surfaces in boreholes, and torsional de-
formation of cores and slabs.

A cknowledgment
I wish to thank L. A. Obert of the U. S. Bureau of Mines for calling
my attention in 1962 to the photoelastic coating technique and for his
continuing stimulation, advice, and assistance.
C. E. Norman and Chakhib Khelil, both former graduate students at
The Ohio State University, obtained some of the data used in this
report. In particular, Mr. Norman obtained the bulk of the field photo-
elastic data.

References
[1] Felix Zandman, "Photostress—Principles and Applications," Tatnall Measur-
ing Systems Co. (Budd Co. subsidiary), abstract from Handbook of Society
of Non-Destructive Testing, 1959.
[2] Felix Zandman, S. S. Redner, and E. I. Riegner, "Reinforcing Effect of
Birefringent Coatings," Experimental Mechanics, Vol 2, No. 2, 1962, pp. 2-11.
[3] S. S. Redner, "New Oblique Incidence Method for Direct Photoelastic
Measurement of Principal Strains," Experimental Mechanics, Vol 3, No. 3,
1963, pp. 67-72.
[4] S. S. Redner, "Oblique Incidence Formulae and Data Reductions," Bulletin
PS-5052, Instrument Div., Budd Co., Phoenixville, Pa., Jan. 29, 1962.
[5] Daniel Post and Felix Zandman, "Accuracy of Birefringent-Coating Method
for Coatings of Arbitary Thickness," Experimental Mechanics, Vol 1, No. 1,
1961, pp. 1-12.
[6] G. U. Oppel, "Photoelastic Strain Gages," Experimental Mechanics, Vol 1,
No. 3, 1961, pp. 1-9.
\7] Thomas Slot, "A Study of the Photoelastic Coating Technique," General
Electric Report 60 GL 72, April 1, 1960.

[8] Thomas Slot, "Reflection Polariscope for Photography of Photoelastic Coat-

ings," Experimental Mechanics, Vol 2, No. 2, 1962, pp. 41-47.
[9] T. C. Lee, C. Mylonas, and J. Duffy, "Thickness Effects in Birefringent Coat-
ings with Radial Symmetry," Proceedings, Society for Experimental Stress
Analysis, Vol 18, No. 2, 1961, pp. 134-142.
[10] J. Duffy, "Effects of the Thickness of Birefringent Coatings," Proceedings,
Society for Experimental Stress Analysis, Vol 18, No. 1, 1961, pp. 74-82.
[11] Joseph Der Hovanesian, "Sign Determination in Oblique Incidence for
Photoelastic Coatings," Experimental Mechanics, Vol 5, No. 4, 1965, p. 128.
[12] Richard O'Regan, "New Method for Determining Strain on the Surface of
a Body with Photoelastic Coatings," Experimental Mechanics, Vol 5, No.
8, 1965 pp. 241-246.
[13] A. Roberts, C. L. Emery, and P. K. Chakravarty, "Photoelastic Coating
Technique Applied to Research in Rock Mechanics-Part I, Laboratory In-
vestigations," Bulletin, Institute of Mining and Mellallurgy, London, July,
1962, pp. 581-601.
[14] A. Roberts, C. L. Emery, I. Hawkes, P. K. Chakravarty, and F. T. Williams,
"Photoelastic Coating Technique Applied to Research in Rock Mechanics-
Part II, Field Investigations," Bulletin, Institute of Mining and Metallurgy,
London, July, 1962, pp. 602-617.
[15] A. Roberts, "Progress in the Application of Photoelastic Techniques to Rock
Mechanics," Proceedings, 6th Symposium on Rock Mechanics, RoUa, Mo.,
1964, pp. 606-648.
[16] C. L. Emery, "The Measurement of Strain in Mine Rocks," Proceedings,
International Symposium on Mining Research, RoUa, Mo., 1961, pp. 541-
557.
[17] C. L. Emery, 'The Photoelastic Technique for Studying Rock Strains,"
Transactions, Canadian Institute of Mining and Metallurgy, April, 1962, pp.
546-550.
[18] C. L. Emery, "Strain Energy in Rocks," Proceedings, International Con-
ference on State of Stress in the Earth's Crust, Elsevier Publishing Co.,
New York, 1964, pp. 235-269.
[19] "Photostress Rosette Gage," Bulletin PS-3050, Instrument Div., Budd Com-
pany, Phoenixville, Pa.
[20] E. M. Galle, "Photoelastic Analysis of the Stresses near the Bottom of a
Cylindrical Cavity due to Non-Symmetrical Loading," thesis. Rice Inst.,
Houston, Tex. April, 1959.
[21] John Fitzpatrick, "Biaxial Device for Determining the Modulus of Elasticity
of Stress-Relief Cores," Report of Investigations No. 6128, U. S. Bureau of
Mines, 1962.
[22] H. J. Pincus, "Some Experiments Utilizing Photoelastic Coatings for the
Study of Strain in Rocks," Proceedings, International Conference on State of
Stress in the Earth's Crust, Elsevier Publishing Co., New York, 1964, pp.
269-279.

DISCUSSION

B. W. Paulding, Jr.^—Some of the techniques you employ are similar

to those used by others^ in studying the stress relaxation of rock. In their
Studies, photoelastic material is aifixed to the surface of rock specimens
obtained from highly stressed areas in mines. The photoelastic patterns
are then observed over periods of days or weeks and are interpreted as
due to the relaxation of residual stresses in the rock. In view of com-
ments by Frocht^ and your own experience with regard to the time-edge
effect of atmospheric moisture on photoelastic materials, would you
comment on the validity of this technique in studying stress relaxation
phenomena?
H. J. Pincus (author)—As noted in my paper, our field mounts of
blank, continuously cemented photoelastic coatings have not produced
signals that we could use. This has been the case inside holes in over-
coring experiments and in monitoring observations on the walls of drifts.
Blank coatings have yielded useful signals only in laboratory experiments.
All of the effects studied by us, with both blank and frozen-ring gages,
have been observed within a very short period after overcoring; hence,
aging effects are likely to have been insignificant.
Some of the fine-textured, tweedy patterns reported by others look
much like those that have appeared in some of our laboratory specimens,
after having aged for weeks to months. When these specimens are de-
formed, the tweedy pattern is largely or entirely overridden by a general-
ized response affecting most or all of the coating, or, in some cases, it
gives way to the mottled pattern, described earlier, that is a replica of
the texture seen in the bare rock surface.
As far as the time-edge effect is concerned, I am not aware of any clear
indications of this, per se, in published work dealing with applications
to rocks. We have observed effects in our own work that might be of
the time-edge type, but these have been of minor significance.
Certainly, there are situations in which spurious signals appear in
photoelastic coatings. With experience and reasonable care, one need
not be misled by them.
D. F. Coates*—We have investigated the use of photoelastic strain

' Associate research engineer, IIT Research Institute, Chicago, 111.

^ C. L. Emery, "In Situ Measurements Applied to Mine Design," Proceedings,
6th Symposium on Rock Mechanics, University of Missouri, Rolla, Mo., 1964.
" M. M. Frocht, Photoelasticity, Vol I and II, John Wiley and Sons, Inc., New
York, 1941.
'Head, Mining Research Laboratories, Mines Branch, Department of Mines
and Technical Surveys, Ottawa, Canada.

gages for ground control work. From our experience we would question
their use on the rock in place as this usually is on surface rock which
often is, and we must assume to be without evidence to the contrary,
loose or semidetached. Consequently, any measurements that are ob-
tained on such rock would not be representative of the rock mass.
Also, we have found the thermoelastic effects resulting from taking
rock from underground, where the average temperature is about 45 deg,
to the surface is so great that it inevitably masks any stress effects. This
would probably apply also to any overcoring operation using such strain
gages.
Finally, the effects of volume change in the cement can lead to some
erroneous deductions; J. E. Udd at McGill University has shown that
for commonly used cements a cycle of expansion and contraction, which
produced significant strains in a sheet of photoelastic plastic, occurred
over a period of about a month.^
On the positive side we can say that on one particular study the use of
unidirectional photoelastic strain gages was undoubtedly the simplest
method of obtaining a measure of the loads being sustained by a series
of steel supports underground."
Mr. Pincus—^All of Mr. Coates' points are valid causes for concern.
Each of these points is consistent with or implied by comments made in
the body of my paper. And, part of my response to Dr. Paulding is
relevant here.
In response to Mr. Coates' first point, I agree that the use of photo-
elastic strain gages mounted on surface rock—or for that matter any
strain gage mounted on surface rock—is likely to result in nonrepresenta-
tive data. It is for precisely this reason that we have concentrated our
efforts in borehole observations.
In response to the second point, the paper points out that we had
noted thermal effects when coated specimens were brought up to ground
level. For this reason, our photoelastic measurements on rocks collected
underground have been made underground.
In response to the third point, our procedure has been to overcore
within hours after the cement has set. Observations in the hole and in
cores broken from the hole have been made immediately following over-
coring, and usually for a period of several days thereafter. The major
part of the change in pattern has been observed within about 15 min fol-
lowing overcoring.

' J . E. Udd, unpublished report, 1962.

" D. F. Coates and M. Gyenge, "The Mechanics of Support and Caving in Long-
wall Top-Slicing," Proceedings, International Conference on Strata Control and
Rock Mechanics, Columbia University, New York, N. Y., 1964.

In my opinion, frozen ring photoelastic gages can be used effectively

in the field (and in fact have been) if the proper precautions are taken in
preparing, mounting, and reading the gages; if the gages are mounted at
valid sites, namely, in boreholes; and if the observations are made (as can
be done in overcoring) within a short time after mounting.

Calculation of the Average Ground-Stress

Components from Measurements of the
Diametral Deformation of a Drill Hole

REFERENCE: "Calculation of the Average Ground-Stress Components from

Measurements of the Diametral Deformation of a Drill Hole," Testing Tech-
niques for Rock Mechanics, ASTM STP 402, Am. Soc. Testing Mats., 1966,
p. 106.
ABSTRACT: Determining the change of stress in rock based on measure-
ments of the deformation of a drill hole is a procedure that is increasing in
popularity. A closely related procedure involves the application of this prin-
ciple to determine the existing stresses in a rock medium by cutting free (core
drilling) an annulus containing the drill hole.
In this report, expressions are derived from which one can calculate the
stress ellipsoid—the three-dimensional change of stress or the existing three-
dimensional stress field, as the case may be. Study of these equations shows
that the stress components in three dimensions can be determined by measure-
ments in only two drill holes and yields general principles that are useful in
planning an efficient program of drilling and measurement.
The need for some method of averaging the measurements arises even
in the simplest situation, because one usually has available several sets of
measurements and wishes to combine them into a single set of average stress
components. Procedures, therefore, are given for calculating the least squares
estimates of the stress components and their standard errors, and for evalu-
ating or comparing calculated values by means of standard statistical inferences.
KEY WORDS: rock (material), rock mechanics, drilling, core testing, earth
pressure

As the deformation of a drill hole in an isotropic elastic rock body

can be related to the corresponding change of stress by fairly simple
theoretical equations, involving a minimum of assumptions, one can
readily envision the possibility of determining the change of stress by
measuring the change of diameter of the drill hole [1,2]}
If a small volume of rock containing the drill hole is cut away from
the main rock body, then one may assume that all stresses are relieved
in the cut away portion and, therefore, that the change of stress is nu-
merically equal, although opposite in sense, to the stresses existing in
the parent rock body. This principle is the basis for the following method
1
Supervisory physical research scientist, Bureau of Mines, College Park, Md.
2
The italic numbers in brackets refer to the list of references appended to this paper.
106

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
PANEK ON CALCULATION OF GROUND-STRESS COMPONENTS 107

of determining existing rock stress [3]. A small diameter (1.5 in.) hole
is drilled into the zone where the stresses are to be determined. A gage
is inserted in the hole to measure change of hole diameter. A larger hole
(for example, 6 in. diameter) is core drilled, concentric with the first
hole, thus relieving the stresses within the annulus of rock that is formed.
The resuhing change of diameter of the small hole is measured along
several diflferent diameters. Provided that the diameter changes are
measured by a gage that does not interfere with the deformation (a
"soft," or low modulus gage), then one can make use of the elastic
theory solution for the deformation of a circular hole in an infinite

FIG. I—Coordinate system for any drill hole.

elastic medium to calculate the stresses that act in the medium. Direct
use cannot be made of the simple elastic theory solution for Hast's
method [4], which employs a gage that exerts a substantial force against
the wall of the drill hole.
An investigation employing the overcoring method, as described
previously, commonly yields a number of measurements from several
drill holes, which one would like to combine into a single set of average
three-dimensional stress components. This is a typical situation in which
to seek a solution based on the least squares principle; of the many stress
solutions that are possible, one chooses the particular solution that
results in the minimum sum of squared deviations from the fitted re-
gression line.
At the other end of the scale is the situation in which only enough

measurements are available for a single solution. The efficiency of the

investigative program, in terms of the information obtained per foot of
hole drilled and per measurement made, can be improved by giving
consideration to the optimum distribution of the measurements and the
minimum amount of drilling required to obtain a solution.
Several papers have been published that deal with the relation between:
(1) the change of hole diameter and (2) the change of rock stress [1,2]
or the existing rock stress as determined by "overcoring" [3,5,6]. These
are treatments based on special simplifying assumptions; furthermore,
none deal with the averaging of a set of measurement data.
The present report considers the problem of calculating the least
squares estimates and standard deviations for the three-dimensional
stress components from any number of measurements taken in any
number of drill holes that are oriented at any azimuth or inclination in a
homogeneous stress field. As no simplifying assumptions are made with
respect to the magnitudes or directions of the components of stress or
strain in the rock, this solution is completely general.

Formulation of a General Equation for Diametral Deformation

Consider an infinitely large cube of rock, isotropic, homogeneous, and
linearly elastic, containing a central drill hole, and subjected on its faces
to a homogeneous state of stress, as shown in Fig. 1.
Removal (relief) of the applied stresses from the block causes a change
in the hole diameter, which is equal in magnitude but opposite in sense
(extension versus contraction) to the diameter change caused by the
application of these stresses. After the applied stresses are removed from
the block of rock, all rock elements within the block are stress free.
Hence, no further change of the hole diameter will be caused by cutting
out an annulus, containing the central hole, from the stress-free block;
moreover, the annulus may instead be a square prism, an elliptical prism,
or irregular in cross section and any size (diameter). The axis of the
central hole need not be parallel to the axis of the outer surface of the
annulus, and the center of the hole need not coincide with the center of
the annulus.
Also, because of the superposition principle of the theory of elasticity,
the diameter of the hole in the stress-free annulus is independent of
whether the annulus is cut out from the block before or after the applied
stresses are removed from the block; the essential consideration is simply
that the annulus be stress free. Determining the diameter change in the
drill hole due to overcoring or trepanning of an annulus from the strained
block of Fig. 1 is, therefore, equivalent to determining the diameter
change that occurs on removing the stresses applied to the block. Further-
more, since the diameter change is not influenced by the size or shape of
the annulus, any convenient size of overcoring bit may be used, and no

particular precision is required with respect to the orientation of the

axis of the overcoring hole. For purposes of subsequent testing to de-
termine elastic properties, a well-centered coaxial drill hole may be
desirable in the annulus, but this has no bearing on the validity (with
respect to the stress field calculations) of the determination of diameter
change that accompanies the overcoring operation.
Because of the superposition principle of the theory of elasticity,
application of a specified change of stress (increase or decrease) has the
same eifect on the diametral deformation of the drill hole, whether the
change is superimposed on an already existing state of stress or is applied
to the stress-free block. Throughout this paper, therefore, the reader
may interpret the discussion in terms of an application of stress to a
stress-free block (Fig. 1), or a mining-induced change of stress in rock
(for example, due to an approaching cave line), or an overcoring-induced
relief of stress, whichever is most pertinent to his specific circumstances.

Diametral Deformation in hi, h^, h Coordinates

We proceed now to formulate an equation giving the deformation of
the drill hole diameter as a function of the stresses applied to the block.
Fig. 1. The derivation is made for the case of stresses applied to the
block, so as to maintain a consistent mode of expression and also so
that Fig. 1 can serve as a reference to indicate the positive (tensile) direc-
tions of all stress components or changes therein. If a calculated stress
component (or change) comes out negative (compressive) then that
component acts in the direction opposite to that shown in Fig. I.
Several coordinate systems are required. Each drill hole is assigned a
right-hand, rectangular coordinate system hi, h^, h , as shown in Fig.
1. The positive hi axis is horizontal to the right for an observer facing
in the positive hi direction; the hi axis is parallel to the drill hole axis,
which may have any azimuth and inclination; and the h^ axis is in a
vertical plane (/jg is vertical only when h^ is horizontal). There would be
advantages to designating the axes hn , h^ , and h^ for the first drill hole,
hi2, ha , and h^^ for the second drill hole, etc. However, the writer be-
lieves that the notation is easier to follow if the second subscript is
omitted. In doing so, one must clearly understand that the hi axis for
the first drill hole is separate and distinct from the hi axis for any other
drill hole (in general, they do not indicate the same direction) and simi-
larly for the hi and h axes.
At any point far from the drill hole, that is, in the medium, the three-
dimensional state of stress is defined by specifying the magnitudes of the
three components of normal stress ai, a^, and 0-3 and the three compo-
nents of shearing stress Tn , rn , and m (owing to the "equality of cross
shears," rn ~ m , Tn = 732, and T^ = TU). That is, a total of six inde-
pendent stress components are required in order to define the stress

tensor. Of these six components, we may assume that the stresses TU and
723, which act parallel to the axis of the drill hole, have neghgible in-
fluence on the diametral deformation of the drill hole, because a linear
element experiences no change of length due to a shearing stress that
acts parallel to it or at right angles to it [7].' Consequently, the diameter
change of the drill hole is for practical purposes a function only of the
three stress components (o-i, 0-3, and TU) that act normal to the drill hole
axis and the component ((Xi) that acts parallel to the drill hole axis, which
implies that the stresses and strains are the same in any slice of unit
thickness normal to the drill hole axis. Therefore, the state of stress and
strain along the hole can be represented mathematically by combining
a constant value of the strain «2 with what is known as the plane-strain
condition of the theory of elasticity.
The change U of drill hole diameter d due to <ri, 0-3, and m can be
found by the complex variable method of Muskhelishvili [8] by taking
ri2 = T23 = €2 = 0 (plane-strain condition):
£/(,!) = aid (1 + 2 cos 26»)(1 - v'')/E, due to o-i (1)
£/(„3) = aad ( 1 - 2 cos 20)(1 - v^)/E, due to 0-3 (2)
!/(.„) = Tud (4 sin 2e)(l - v^)/E, due to ri3 (3)
where E is the modulus of elasticity, v is Poisson's ratio, and d is the
angle with respect to hi measured in the hi, hz plane, the positive direc-
tion of d being as shown in Fig. 1.
The derivation of these equations is too lengthy to be reproduced in
this paper. However, Eqs 1-3 can be verified as follows. Sokolnikoff
[8, p. 291] gives the solution for the radial displacement due to a stress
CTi (in Sokolnikoff's expression for Ur one must set r = R, R being the
radius of the hole, /c = 3 — 4c, ju = E/2{\ -\- v), and note further that
2UT = U, 2R = d). Equation 2 can be obtained by substituting 90 deg +
6 in place of ^ in Eq 1, so that the direction of stress is at right angles to
hi, and therefore the stress is 03 rather than ai. Equation 3 can be ob-
tained by combining a stress ai, as given by Eq 1, with a numerically
equal stress —(73 as given by Eq 2. These two stresses together are equiva-
lent to a pure shearing stress, numerically equal to <7i, that acts at 45 deg
to the hi and hi axes. This pure shear is transformed to TH by substi-
tuting 6 — 45 deg for 6, so that the shearing stress acts parallel to the
hi and h^ axes.
To properly take into account the influence of 0-2, it is necessary to
add to the above solution (superpose) the effect of the actual value of
e2, which in general is not zero as assumed, in order to obtain the solu-
tion. When the stresses are applied to the block, the unit strain £2, owing

' The effect of shearing stress is only to change the angle between two hnear elements,
one of which is parallel to, and the other perpendicular to, the direction of the shear.

to the Poisson effect, is accompanied by a unit strain — ve2 in all direc-

tions normal to the hi axis. Therefore, an element of length d, which is
oriented in any direction normal to h^ (at any angle 6), experiences a
change of length equal to d{ — vt2). Consequently,
[/(.2) = -dvti (4)

is the additional increment of diameter change due to the amount by

which the actual ez differs from zero. To express this increment in terms
of stress component increments we substitute for £2 in Eq 4 the Hooke's
law expression
£2 = [(Tl — v{ai + (T-i)]/E (5)

obtaining
f/((2) = <Tidv^/E — aidv/E + (T^dv^/E, due to £2 (6)
Finally, the total change of drill hole diameter, due to application of
the six components of three-dimensional stress is the sum of the four
increments given by Eqs 1, 2, 3, and 6:
U = cTifi + 0-2/2 + 0-3/3 + nifi (7)

where
fi = d{l +2 cos 26){I - v')/E + du'^/E
/2 = - dv/E
(8)
fz = d{\ - 2 cos 2e){\ - v'')/E + du^E
/4 = c?(4sin2^)(l - v-')/E
If the increment due to e^ is not expressed in terms of the stress com-
ponents then the total change of drill hole diameter may be written as
follows:
U' = tri/i' + 03/3' + 731/4 (9)

where
V = U+ dvei 1
/ i ' = d{\ + 2cos2e)(l -v-')/ E\ (10)
/' = d{\ - 2 cos 2(9)(1 - >'2)/£'
Equations 7 and 9 express the diametral deformation, in a specified
direction, as a linear function of the applied stresses. Equation 7 applies
when U alone is measured; Eq 9 applies when measurements of both
U and «2 are obtained. To every distinct orientation of U there corre-
sponds a distinct set of values of t h e / (or/,') which depend only on the
directions of measurement and on the elastic constants. By determining

U for three different directions and substituting the three sets of values
either in Eq 7 or 9, we obtain three independent simultaneous equations,
which, however, because a or €2 is unknown,^ cannot be solved for all of
the three independent stress components a i , 0-3, and Tn that act normal
to the axis of the drill hole, unless some additional information is em-
ployed, such as the following:
1. the value of 0-2 is inserted in Eq 7, which means that ai must be
obtained by some other means (estimated, or iterated from a trial solu-
tion, etc.); or
2. the strain €2 is inserted in Eq 9, which means that the axial strain
as well as the diametral change must be measured in the drill hole; or
3. some additional assumption is made as to the state of stress or strain
[7,5]. For example, Morgan et al [5, Eqs 11-14] solved for 0-3 by making
the additional assumptions that T31 = 0 (principal stress directions
horizontal and vertical) and ei = 0 in the medium, which seemed to be
consistent with the measurements obtained at their particular mine.
An important implication of Eq 7 or 9, which will be referred to sub-
sequently, is that measurements of diametral change in a single drill hole
provide a maximum of three independent conditions (three independent
equations) for the solution of the stress components.
If only Ua and [/90 are measured (change of drill hole diameter at
0 = 0, 90 deg) then/4 = 0 in Eq 7, which means that T31 has no influence
on U. This does not of itself enable one to obtain a solution, for no in-
formation is obtained with respect to 0-2, that is, statements 1, 2, and 3,
mentioned previously, remain true. Leeman [7] considered the applica-
tion of this system of measurement to the determination of change of
ground stress. He obtained a solution for Uw as a function of 0-3 only
by making assumptions as to the constraint in the hi and hi directions.
For example:
1. Assuming that ai and m remain constant when m changes (no
lateral constraint), then
^90 = 'J-iJl = C,d{-i - V)IE (11)
2. Assuming lateral constraint in the h\ and h^ directions (ci = 62 = 0),
then (71 = (72 = <Sivl(\ — v), from which
U,, = <Tl/l + .7 A + <T;/3 = .73^(3 - V - Av-')/E (12)
(Here a and t refer to change of stress and strain). Two other condi-
tions of constraint were also considered (ei = 0 and €2 = 0), the results
faUing between those given by Eqs 11 and 12. Since v = 0.15 for the Rand
rocks, Leeman concluded that the change in 0-3 could be calculated to
within ±3.5 per cent by taking the value of C/90 to be the mean of the

* Solution for <r2 cannot be obtained from Eq 7, even with four determinations of U,
because /2 is constant for all $.

two extremes given by Eqs 11 and 12. This procedure is effective as long
as: (1) the changes of en and c^, when 0-3 changes, actually lie between 0
and asv/il — »<), as assumed, which impUes that the direction of hz should
be approximately that of the major (independent) stress change (that is,
additional prior knowledge is required), and (2) Poisson's ratio is not
large (the error is ± 11 per cent for v = }i). For a situation in which
the above requirements are met, this procedure offers the practical
convenience of determining the major stress change from measurements
in a single drill hole.

Diametral Deformation in x, y, and z Coordinates

When determinations of U are made in more than one drill hole, it
becomes necessary to relate all measurements to a common coordinate
system. Therefore, in addition to the coordinate system for each drill
hole, we assign a set of right-hand rectangular coordinates x, y, and z
to the medium, chosen in any convenient direction. For example, these
might be parallel to the north-south, east-west, and vertical directions,
respectively; or, they might be taken parallel to the strike of the orebody,
parallel to the dip, and normal to the plane of the ore body.
The six independent components of the stress tensor, for the medium,
are designated CTI , o-j,, and a^ (normal stresses), and Txy, T^^ , and TZX
(shearing stresses). According to the rules of transformation from one
rectangular coordinate system to another, the normal component of
stress in the hi direction (any drill hole) is given by [10]
ffi = Ux/i^ + (7ymi^ + a^ni^ + iTxyhmi + 2ry,mini + IT^JIJI (13)
the normal component in the h2 direction is given by

0-2 = (Txl-? + (Tym-i^ + ajt-? + iTxyhmi + iTyjn-ini + Ir^xnik (14)

and the normal component in the hi direction is given by

0-3 = Oxk^ + (Tymi^ + a^nz^ + Itxyhms + Iry^mmi + Ir.xmh- • • (15)

where h , mi, and ni are the direction cosines for the positive hi axis with
respect to the positive x, y, and z axes, respectively, and similarly for the
hi and hs axes. That is, h is the cosine of the angle between the positive
hi and positive x axes, mi is the cosine of the angle between the positive
hi and positive y axes, • • • , and na is the cosine of the angle between the
positive hi and positive z axes.
Similarly the shearing component of stress that acts parallel to the
hi, hi plane is given by

TK = o-x/i/s + (Tymim^ + (T,nin-i + Txyihms + hmi)

+ TyzimiUt + mini) + T^x{nik + mh) (16)

Copyright by ASTM Int'l (all rights reserved);

Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia
114 TESTING TECHNIQUES FOR ROCK MECHANICS

Substituting the Eqs 13-16 into Eq 7 and collecting terms:

U = 6iA + M a + 63/3 + 64/4 + 65/5 + b,J,. .(17)
where
U = change of drill hole diameter
bi = o-x

hi = (Ty

bs = <Tz
.(18)
hi = Tiy

bi = Tyz

hi = T^i^

^..(19)

/4 = 2/i/imi + 2/2/2W2 + 2/3/3^3 + /4(/im3 + kmi)

/ s = 2/imini + 2/2m2«2 + 2/3W3«3 + /4(/«l«3 + W3«l)

^6 = 2/ini/i + 2/2n2/2 + 2/3«3/3 + fi{nik + K3/1)
Equation 17 expresses the diametral deformation as a linear function of
the hi, the applied stresses. To every distinct orientation of U there
corresponds a distinct set of values of Ji that depend only on the direc-
tions of measurement and on the elastic constants E and v. That is, the
Ji may be regarded as "known" coefficients as soon as the orientation
(direction of the hole and of the diametral measurements within it) are
specified. By determining U for six independent conditions, and sub-
stituting the six sets of values in Eq 17, we obtain six independent simul-
taneous equations, which can be solved for the "unknowns," the values
of the hi (the six independent components of the stress tensor).
Minimum Requirements for Complete Determination of tlie
Stress Tensor
Equation 17 can be used for any number of U determinations made
along any number of diameter orientations in any number of holes that
are drilled in any number of directions. Assuming that no two drill holes
are parallel, each has a diff"erent set of direction cosines h , irii, iii, k ,
nii, rii, h , ma, and tn for its axes with respect to the x, y, and z axes
of the medium.
In any one drill hole, assuming that all U determinations are made
Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreemen
PANEK ON CALCULATION OF GROUND-STRESS COMPONENTS 115

along different diameters, then each U has a different set of/i and, con-
sequently, a different set of "orientation coefficients" / , . For a single
drill hole only three of the six Ji are independent, however, because each
Ji is a linear function of/i ,fi,fi, and/4 which, as pointed out following
Eq 7, contain only three independent conditions for solution of the stress
components. In fact the relation between Eqs 7 and 17 can be seen by
considering the determination of U for several different values of 9 in a
single hole drilled in the positive ;; direction (positive h^ coincides with
positive_y). Then, for every U, h = m^ = m - I, h = h = rrii = m-i =
«i = «2 = 0, and from Eq 19,

Ji = fi]
Ji = fi

h =/3
J- .(20)
^4 = 0
A =0
/e = / 4 ,

so that Eq 17 for each U determination in the hole parallel to ;; becomes

U = biJi + biJ^ + hJs + ba^ (21)

or, substituting from Eqs 18 and 20,

U = aji + a J, + <7j, + T,J, (22)

which is equivalent to Eq 7.
As a solution for all six components of the stress tensor cannot be
obtained from determinations of f/ in a single drill hole, we consider
next the nature of the solution when there is a second drill hole.
Case 1—First drill hole parallel to the y axis, positive h^ coinciding
with positive y; second drill hole parallel to the x axis, positive h^ coin-
ciding with positive x. For the second drill hole, 4 = — mi = Wa = 1,
h = h = nii = rris = Hi = 112 = 0, and from Eq 19,

Jx = / 2

J2 = / i

Jz =fz
.(23)
Ji = 0
A = -A
Je = 0

SO that Eq 17 for each U in the hole parallel to x becomes

U = biJi + biJi + bzJz + biJi .(24)

U = (Txfi + O-j/l + cji — Tyzfn- .(25)

By determining U for three different values of Q in the first drill hole and
substituting the three sets of values in Eq 21, we obtain three independent
equations for the solution of the bi (the six stress components). Simi-
larly, by determining U for three different values of 6 in the second drill

^ I T
zx
V—h
\ 90-l\ I /

•y(N)

x(E)
FIG. 2—Coordinate systems for an arbitrarily oriented drill hole when x and y are
taken to be East and North; hi = axis of drill hole, I = angle of inclination, B = bearing
angle-

hole and substituting in Eq 24, we obtain three independent equations

for the bi. Solution for only five of the six 6, can be obtained from the
combined six equations, however, because bi = r^y is absent from all
of them.
As the orientation of the holes becomes more complex, one must pay
increasing attention to the quadrants of the coordinate system. For con-
venience of exposition throughout the remainder of this report, the
positive coordinate axes x, y, and z are arbitrarily taken to be east,
north, and vertical, respectively, as shown in Fig. 2.
Case 2—First drill hole horizontal, with positive h^ axis bearing north,
second drill hole horizontal, with positive hi axis bearing N 60 deg E.
For the first hole, the / , are given by Eq 20. For the second hole, 1% =

^3 = «i = «2 = 0 (the other direction cosines are nonzero), which when

substituted in Eq 19 yield the following expressions for the Ji :

Ji = f\m^ + f-m-S-

,(26)
Ji = Ifihmi + 2/2/0^2

Ji = fimiHs

Je = fitsh
Three determinations of U for three different d orientations in the first
drill hole yield three independent equations of the form 21. Three deter-
minations of U for three different d orientations in the second drill hole
(most of the Ji are nonzero) yield three independent equations of the
form 17. Solution for the six 6, can be obtained from the six combined
equations, because there are six independent conditions, although this
does not necessarily follow from the fact that all the •/, are nonzero.
In general, the number of independent conditions cannot be deter-
mined simply by inspection of the sets of Ji in Eq 17, because most of
the Ji are nonzero for a drill hole that is off the coordinate axes. Table 1
gives the coefficients Ji that correspond to eight different directions of
measurement in nine difiTerent drill holes. Among a set of equations, the
number that are independent can be determined by proceeding as if
with the actual solution, employing the method of successive elimination.
The U values need not be known, as they do not enter into the question
of independence. If a solution can be obtained for only five of the six
unknowns, then only five independent conditions exist among the set of
equations. That is, only five sets of Jt are independent; in mathematical
terms, the coefficient matrix is only of Rank 5. By applying this method
to appropriately selected pairs of drill holes one can easily verify that a
solution for all six independent components of the stress tensor can be
obtained from measurements of U in two drill holes only if the holes are
not mutually perpendicular.
If the two drill holes are mutually perpendicular, solution can be made
for only five components of the stress tensor. To obtain an explicit solu-
tion one must choose a pair of coordinate axes parallel to the two drill
holes, as in Case 1, mentioned previously. If at least one drill hole is off
the coordinate axes, then all six components of the stress tensor appear
in the equations. In this case one can solve for any five stress compo-
nents, but only as a function of the sixth component, which is unknown.
Influence of Axial Stress and Strain
If the diametral deformation due to the applied stresses were inde-
pendent of the axial stress and strain, the six stress components could

TABLE 1—Orientation coefficients Ji showing the influence of drill hole direction

and measurement direction (u = 14, d = 1.5).

e, deg EJi EJi EJ, EJ, EJi EJe

Drill hole bearing N, inclination 0 deg

0 4.167 -0.500 -1.167 0.000 0.000 0.000

30 2.833 -0.500 0.167 0.000 0.000 4.619
45 1.500 -0.500 1.500 0.000 0.000 5.333
60 0.167 -0.500 2.833 0.000 0.000 4.619
90 -1.167 -0.500 4.167 0.000 0.000 0.000
120 0.167 -0.500 2.833 0.000 0.000 -4.619
135 1.500 -0.500 1.500 0.000 0.000 -5.333
150 2.833 -0.500 0.167 0.000 0.000 -4.169

Drill hole bearing N, inclination 30 deg

0 4.167 0.667 -1.000 0.000 0.577 0.000

30 2.833 0.333 0.000 -2.309 -0.577 4.000
45 1.500 0.000 1.000 -2.667 -1.732 4.619
60 0.167 0.333 2.000 -2.309 -2.887 4.000
90 -1.167 0.667 3.000 0.000 -4.041 0.000
120 0.167 0.333 2.000 2.309 -2.887 -4.000
135 1.500 0.000 1.000 2.667 -1.732 -4.619
150 2.833 -0.333 0.000 2.309 -0.577 -4.000

Drill hole bearing" N, inclination 90 deg

O" 4.167 1.167 -0.500 0.000 0.000 0.000

30 2.833 0.167 -0.500 -4.619 0.000 0.000
45 1.500 1.500 -0.500 -5.333 0,000 0.000
60 0.167 2.833 -0.500 -4.619 0.000 0.000
90 -1.167 4.167 -0.500 0.000 0.000 0.000
120 0.167 2.833 -0.500 4.619 0.000 0.000
135 1.500 1.500 -0.500 5.333 0.000 0.000
150 2.833 0.167 -0.500 4.619 0.000 0.000

Drill hole bearing N 45 E, inclination 0 deg

0 1.833 1.833 -1.167 -4.667 0.000 0.000

30 1.167 1.167 0.167 -3.333 -3.265 3.265
45 0.500 0.500 1.500 -2.000 -3.771 3.771
60 -0.167 -0.167 2.833 -0.667 -3.265 3.265
90 -0.833 -0.833 4.167 0.667 0.000 0.000
120 -0.167 -0.167 2.833 -0.667 3.265 -3.265
135 0.500 0.500 1.500 -2.000 3.771 -3.771
150 1.167 1.167 0.167 -3.333 3.265 -3.265

Drill hole bearing N 45 E, inclination 30 deg

0 1.750 1.750 -1.000 -4.833 0.408 0.408

30 0.095 2.405 0.000 -3.167 -3.236 2.420
45 -0.583 2.083 1.000 -1.500 -4.491 2.041
60 -0.905 1.405 2.000 0.167 -4.869 0.787
90 -0.250 -0.250 3,000 1.833 -2.858 -2.858

" In a North-bearing vertical hole the positive hi axis bears East (see Fig. 2).

TABLE \-~Concluded
e, deg £/i £/j EJi EJi EJi EJt

Drill hole bearing N 45 E, inclination 30 deg—-Continued

120 1.405 -0.905 2.000 0.167 0.787 -4.869
135 2.083 -0.583 1.000 -1.500 2.041 -4.491
150 2.405 0.095 0.000 -3.167 2.420 -3.236

Drill hole bearing N 4 5 W, inclination 45 deg

0 1.667 1.667 -0.833 5.000 0.471 -0.471

30 2.966 -0.300 -0.167 3.000 1.838 2.781
45 2.885 -0.885 0.500 1.000 1.253 4.081
60 2.300 -0.966 1.167 -1.000 -0.048 4.666
90 0.333 0.333 1.833 -3.000 -3.299 3.299
120 .. -0.966 2.300 1.167 -1.000 -4.666 0.048
135 .. -0.855 2.885 0.500 1.000 -4.081 -1.253
150 -0.300 2.966 -0.167 3.000 -2.781 -1.838

Drill hole bearing N 6 0 E, inclination 0 deg

0 0.667 3.000 -1.167 -4.042 0.000 0.000

30 0.333 2.000 0.167 -2.887 -4.000 2.309
45 0.000 1.000 1.500 -1.732 -4.619 2 Ml
60 .. -0.333 0.000 2.833 -0.577 -4.000 2.309
90 .. -0.667 -1.000 4.167 0.577 0.000 0.000
120 .. -0.333 0.000 2.833 -0.577 4.000 -2.309
135 0.000 1.000 1.500 -1.732 4.619 -2.667
150 0.333 2.000 0.167 -2.887 4.000 -2.309

Drill hole bearing N 60 E, inclination 30 deg

0 0.542 2.958 -1.000 -4.186 0.289 0.500

30 .. -0.542 3.042 0.000 -1.588 -3.753 1.500
45 .. -0.780 2.280 1.000 0.034 -4.866 0.809
60 .. -0.708 1.208 2.000 1.299 -4.907 -0.500
90 0.208 -0.708 3.000 1.588 -2.021 -3.500
120 1.292 -0.792 2.000 -1.010 2.021 -4.500
135 1.530 -0.030 1.000 -2.632 3.134 -3.809
150 1.458 1.042 0.000 -3.897 3.175 -2.500

Drill hole bearing N 60 E, iinclination 45 deg

0 0.417 2.917 -0.833 -4.330 0.333 0.577

30 .. -0.831 3.498 -0.167 -0.965 -3.161 1.055
45 .. -0.883 2.883 0.500 1.019 -4.266 0.153
60 .. -0.498 1.831 1.167 2.499 -4.495 -1.254
90 1.083 -0.417 1.833 2.598 -2.333 -4.041
120 2.331 -0.998 1.167 -0.767 1.161 -4.519
135 2.383 -0.383 0.500 -2.751 2.266 -3.617
150 1.998 0.669 -0.167 -4.231 2.495 -2.210

not be determined from measurements of U in two drill holes. This can

be demonstrated by taking six sets of /, for (2 = 0, corresponding to the
six directions of measurement, and finding the number of independent
observation Eq 17, which is at most five.
Properly taking account of the axial stress 0-2 causes the term in fi
to be present in Eq 7. Owing to this term, one cannot solve for the stress
components in the plane normal to the drill hole axis solely from U
measurements in a single hole, because the effect of a^ is the same for
all U and, therefore, cannot be determined from a set of observation
Eq 7. But when there are two nonperpendicular^ drill holes, the two sets

TABLE 2—Orientation coefficients /, calculated on the assumption that a = 0.

e, deg EJi EJi EI, EJi E/s EJi

Drill hole bearing N 60 E, inclination 0 deg

0 1.000 3.000 -1.333 -3.464 0.000 0.000

30 0.667 2.000 0.000 -2.309 -4.000 2.309
45 0.333 1.000 1.333 -1.155 -4.619 2.667
60 0.000 0.000 2.667 0.000 -4.000 2.309
90 . -0.333 -1.000 4.000 1.155 0.000 0.000
120 0.000 0.000 2.667 0.000 4.000 -2.309
135 0.333 1.000 1.333 -1.155 4.619 -2.667
150 0.667 2.000 0.000 -2.309 4.000 -2.309

Drill hoi e bearing N 60 E, inclination 45 deg

0 0.500 2.833 -0.667 -4.041 0.667 1.155

30 . -0.748 3.414 0.000 -0.677 -2.828 1.633
45 . -0.800 2.800 0.667 1.308 3.932 0.731
60 . -0.414 1.748 1.333 2.787 -4.161 -0.677
90 1.167 -0.500 2.000 2.887 -2.000 -3.464
120 2.414 -1.081 1.333 -0.478 1.450 -3.942
135 2.466 -0.466 0.667 -2.463 2.599 -3.040
150 2.081 0.586 0.000 -3.942 2.828 -1.633

of U measurements involve two diiferent axial stress components. This

dependence of the observation equations on the axial stresses in the two
holes provides the additional condition needed to solve for the six stress
components.
If the two drill holes are mutually perpendicular, however, the two
axial stress components contribute no additional information because of
* Since the same homogeneous, three-dimensional state of stress must exist for all
diametral deformation measurements that are to be combined into a single stress sys-
tem, any drill hole may be shifted to the origin of the x, y, and z coordinate system with-
out affecting the stress solution in any way, provided the bearing and inclination of the
drill hole are unchanged. For purposes of discussion, therefore, any set of drill holes
that are to be used for a single stress solution may be treated as if they intersect at the
origin of x, y, and z. In this sense, any pair of drill holes, irrespective of their bearing,
inclination, or location are considered to define a plane.

double determination of one of the stress components. For example, in

a pair of mutually perpendicular horizontal drill holes the axial stress
for one hole is identical to the horizontal stress ai for the other hole.
In addition to making possible a two-hole stress solution from diam-
etral deformation measurements alone, the axial strain has a significant
influence on the magnitudes of the calculated stresses.
Table 2 shows, for two drill holes, the sets of Ji that were calculated
for the condition 62 = 0. Comparison with the corresponding "correct"
values of Ji in Table 1, for the same drill holes, shows that the influence
of the axial stress on the Ji, and, therefore, on the calculated stresses, is
neither negligible nor systematic. Deletion of £2 does not alter all the
stress components by the same percentage, nor does it always affect the
same components.

Choosing Directions of Measurement

In choosing the orientation of a drill hole and of the U measurements
within it, one may wish to consider also the relative contribution of each
of the six stress components to the diametral deformation. For example,
Eq 21 shows that U in no way reflects the intensity of the two compo-
nents bi and b^, if the drill hole is parallel to the y axis. From the sam-
pling viewpoint, it is desirable that the information obtained from any one
drill hole be influenced approximately to the same degree by each of the
six stress components, that is, the six 7, should be approximately of the
same magnitude, numerically. Investigation of the relative contributions
from the six stress components is facilitated by calculating the set of y,
corresponding to each of several possible directions of measurement in
each proposed drill hole, as shown in Table 1.
In order to approach a reasonably good balance with respect to the
determination of each of the six stress components, one must seek an
appropriate combination of: (1) directions d of measurement within the
holes and (2) hole direction with respect to the x, y, and z axes (which
should be chosen to coincide with what are believed to be the directions
of principal stress or directions that are otherwise important). For sim-
plicity, one would prefer to choose
1. for a 3-element gage,^ three directions at 60 deg to each other, for
example: (a) 9 = 0, 60, 120 deg, or {b) 6 = 30, 90, 150 deg, or
2. for a 2-element gage, two directions at 90 deg to each other, for
example: (a) 9 = 0, 90 deg, or (b) 6 = 45, 135 deg .
Table 1, however, shows that the components are difficult to balance
if one considers only a single orientation of a 2- or 3-element gage in the
drill hole; four or more equally spaced measurement directions in each

« By "3-element gage" is meant a device that can measure change of diameter of the
drill hole in three different directions 6, simultaneously or in quick succession, without
being repositioned in the hole, and similarly for a "2-element gage" that can measure
in two different directions.

drill hole provide significantly better balance than only two or three
directions. Furthermore, in the case of the 2-element gage, a solution for
the six Stress components cannot be obtained if there is but one gage
orientation in each of two drill holes, as only four conditions are pro-
vided by the measurements. A much better procedure for overcoring is
to alternate the gage positions for successive stress reliefs in each drill
hole, that is, to employ the directions given under la and b for successive
reliefs with a 3-element gage and the directions given under 2a and b
for successive reliefs with a 2-element gage. For a single-element gage,
one might make successive reliefs at the orientations ^ = 0, 45, 90, 135,
0, 45 deg, etc.
As an example of the appropriate orientation for two drill holes. Table
2 shows that good balance can be achieved by making measurements in
holes bearing N 45 E and N 45 W, respectively, inclined at 30 deg. The
angle between these two holes is 75.5 deg, the angle between their re-
spective hi axes,' which is given by
cos (hi, hi') = W + mmi + «2«2' = (0.612)(-0.612)
+ (0.612)(0.612) + (0.500)(0.500) = 0.250 = cos 75.5 deg. .(27)
We have shown that the complete solution for the six components of
the stress tensor can be obtained from measurements of diametral defor-
mation in only two drill holes, if the holes are not mutually perpendicular
and if the measurements are performed in six independent directions. A
solution for five stress components can be obtained from diametral
deformations measured in two holes drilled along a pair of coordinate
axes; the shearing stress parallel to the plane of the holes is not deter-
mined. The relative contribution by each of the stress components to the
diametral deformation provides a basis for choosing directions of meas-
urement that are appropriate for the type of gage to be used.
Although more than two drill holes are not needed for complete solu-
tion of the stress tensor; nevertheless, one may wish to make measure-
ments in three or four drill holes in order to obtain a better sampling
or to investigate effects due to nonisotropic rock properties. The choice
of an appropriate orientation for a set of three or more holes will depend
on the particular objectives to be achieved, over and above those previ-
ously discussed in this paper.

Least Squares Solution for the Stress Tensor

As shown in the preceding section, solution for the six components
of the stress tensor can be obtained by determining U three times (each
in a difi'erent direction d) in each of two drill holes. Six independent
determinations of V provide the minimum information required to

' The two drill holes are treated as coplanar, in the sense defined by footnote 5.

calculate the six stress components, but provide no information as to the

precision of the stress values. Each determination involves a measured
value, which is subject to random eifects; measurement along a given
diameter orientation will not ordinarily yield the identical result when
repeated one foot away in the same rock. Six measurements of U, each
in a different direction, at a single point in a rock body yield information
equivalent to the estimate of mineral composition of an ore body based
on the analysis of a single rock sample. Obviously one cannot place a
high degree of confidence in such an estimate. It is, therefore, desirable
to make a number of sample measurements of U and calculate from
them an average value for each of the six stress components, as well as
the precision of these estimates.
The degree to which the sample is representative, and hence the ac-
curacy of the stress components (degree of closeness to the "true"
values), can be increased by measuring U at four or more different
orientations in each drill hole or making measurements in three or more
drill holes or both. The precision of the average stress values can be
increased by making two or more measurements at each orientation.
To capitalize on these two principles, one must be able to pool all the
U measurements and to calculate the average values of the stress com-
ponents from the combined data. The method of calculation given here
yields the least squares estimates of the stress components and makes
use of all available U measurements, irrespective of the number of
measurements in any drill hole, the number of orientations in any drill
hole, the number of drill holes, or the directions of the holes.
Mathematically, the least squares solution for the stress components
can be treated in the same way as the problem of multilinear regression
analysis in mathematical statistics. The treatment presented in this paper
may be interpreted as follows [77]. Given the Eq 17 for the measured U
as a function of the parameters / , , one wishes to calculate average
sample values of the coefficients fc, such that the values of U predicted
by Eq 17 will have minimum deviation from the actual measured values,
according to a criterion stated below. It is assumed that a sample set of
U values is randomly selected from a population in which each measured
value of U can be represented by
C/ = |3i/i + /32^2 + • • • + /Je/e + e (28)
U = biJi + hJ2+ ••• + beJi + e (29)
where the 0i are the population parameters and the bi are their respec-
tive estimates; e is the true error of each C/measurement and e the resid-
ual about the regression line. The least squares solution consists of
finding a set of estimates bi which minimizes the sum of squares of the
residuals e. From Eq 29, the residual for each U measurement is given by
e = U - {bJi + b2J2+ ••• + b,J,) (30)

Copyright by ASTM Int'l (all rights

Downloaded/printed by
Pontificia Universidad Catolica de Chile
1 24 TESTING TECHNIQUES FOR ROCK MECHANICS

Then
Se^ = 2[u - (bJi + ^2^2+ • • • + bMf
= 2[f/2 - lUbiJi - 2C/62/2 - ••• - lUbiJ,
+ biUi^ + Ibib^JiJi + • • • + IbibiJiJi
+ biUi' + 26263/2/3 + • • • + Ib^bJiJ,
+ ••• + 6 6 W ] (31)
where the summation is taken over all the U measurements.
The value of 61 that minimizes Se^ is found by taking the partial deriva-
tive with respect to 61 and equating it to zero:
2
ase = 0 = S [ - C / / i + 61/1' + 62/1/2 + 63/1/3 + + 66/1/6].. (32)
dbi

Repeating this operation for 62, 63, • • • 65, the following equations
are obtained, which can be solved for the bi :
6i2/i2 + 62S/1/2 + 63S/1/3 + • • • + 66S/1/6 = S f / / i

612/2/1 + 622/2^ + 632/2/3 + • • • + 66S/2/6 = 2 [ / / 2

.(33)

6 i 2 / 6 / i + 62S/6/2 + 632/6/3 + • • • + 662/6^ = 2 t / / 6

This set of equations is applicable to any number of stress components.
For example, if there is only one drill hole and it is parallel to the ;;
direction, Eq 20 show that J^ = J^ = 0, which imphes that the fourth
and fifth rows and columns must be deleted from the above array.
If only the bi are wanted, any of the conventional procedures for
solving a set of simultaneous equations can be employed. If the precision
of each bi is to be calculated also, then certain additional steps are re-
quired, which can be conveniently incorporated into a systematic proce-
dure that yields both the 6, and their errors. For purposes of discussion,
it is convenient to define

\ .(34)
gi = ^UJi (ij = 1,2,3, 6)
so that Eq 33 can be written
61^11 + 62012 + 63013 + ••• + 65016 = gl

6lfl21 + 62022 + 63023 + ••• + 66026 = g:

.(35)

blU^l + 62062 + 63O63 + ••• + 66066 = gi

where a-a = an, a^ = a^, etc. The major mathematical operation

consists of finding a set of elements c,y such that
bi = Cii^i + Cugi + Ci3g3 + ••• + Ciege

bi = C2lgl + C22g2 + Cngi + • • • + C26g6

(36)

^6 = Ceigi + C62g2 + C(agi + • • • + C(,ig6

where cn = C21, cn = C31, etc. The set of elements c,y is the inverse of
the set of elements Oiy. Once the elements c,y have been found, the bi
may be obtained from Eq 36. The cy are needed in order to calculate
the precision of the b,, the stress components.

Statistical Interpretations
The method of least squares yields unbiased minimum-variance esti-
mates of the (3,, if the true errors e in Eq 28 are independent and have
a normal distribution with zero means and a common variance. The
advantages to be gained from this interpretation are that one can evaluate
the precision, or scatter, of these estimates of the fii and of the measure-
ments of U and formulate confidence statements [12] with respect to
them. The least squares calculation does not depend on nor require
any assumption as to normality of the underlying distribution, but
simple probability statements based on the statistical estimates of pre-
cision do require normality.
For the set of U measurements as a whole, an important quantity is
the error sum of squares, which is given by Eq 31, but which is easier to
calculate from [11, p. 171]
Se2 = s[/2 _ (6igi + h2g2 + ••• + b,g,) (37)
The quantity SeVS U^ is a measure of the relative dispersion of the U;
it is the fractional part of the observed variation of i/that is not ascribable
to the relationship, Eq 17. Alternatively, the square of the multiple
correlation coefficient R,
2
^2 ^ bigi + b,g2 + • • • + ^6g6 ^ 1 _ ^ (33)
SC/2 ' HIP'
is the fractional part of the total variability among the U measurements
that is explainable by Eq 17 [12, p. 215].
The standard deviation about the fitted regression line, Eq 17, is given
by
s = [2eV(« - p)Y" (39)
where n is the total number of U measurements and p is the number of
hi calculated from the data. Confidence limits for the average of all the U

measurements that might be obtained for a specified orientation (for a

specified combination of the J^) are given by
0 ± U (40)
where
D = biJ, + biJ, + ••• + b,J, (41)
and /„ is the tabled value^ of "Student's t" corresponding to the desired
probability level a (for example, 0.05) and n — p "degrees of freedom."
That is, the probability is 1 — a (for example, 95 per cent certainty) that
the interval specified by Eq 40 includes the true average value of U in
the population.
One may wish to know whether the dispersion observed in one part
of the mine differs significantly from that in another part, perhaps because
of a suspected difference in rock properties or the degree of rock frac-
turing. The F statistic may be used to compare ^i with ^2, the error or
dispersion in two sets of U measurements [12, p. 92]. If

-,> F^ (42)

where 5i > .^2, and F„ is the tabled value^ of F for the desired probability
level a, with ni — pi degrees of freedom for s{- and «2 — p^ degrees of
freedom for ^2^ then the probability is less than a that the two sets of
measurements were obtained from populations having the same dis-
persion. That is, statistically si differs significantly from .^2.
The "standard errors" of the bi are given by [12, p. 216]
\l/2
Si^ = i-(cii)
/ \l/2
Sb, = S(C22)
.(43)

where s is given by Eq 39.

Confidence intervals for the 6, are given by
bi - t^Sbt < 0i < bi+ t^st- (44)
where /„ is the tabled value of Student's t for the desired probability a,
with n — p degrees of freedom. For example, the probability is I — a
that the value ^2 ( = o-„) calculated from the measurements is within
± /„S62 of the population value ^-i.
One may test, statistically, hypotheses as to the magnitudes of the
stress components. For example, one may wish to know whether the

' Tables of t and F are given in standard textbooks on statistical analysis.

lateral stress is the same in all directions. Given that bi and 62 are the
two horizontal stress components, if

^ ' ^ ^ > , . (45)

where ^6, — bs = s{cn — 2cn + €32)^'^, s is given by Eq 39, and /„ is the

tabled value of Student's t for the desired probabiHty a with n — p
degrees of freedom, then one may state, with less than a probability of
being incorrect, that bi differs significantly from bi . Of course, if the
difference between the two horizontal stress components is less than 10
per cent, one may consider that the difference is not technically signifi-
cant, even though it is statistically significant.
One may test the hypothesis that a given stress component bi does
not differ significantly from some theoretically expected or assumed value
B(B may be zero). If
\bi - B\
~ -^t^ (46)
Sbi
where ?„ is the tabled value of Student's t for probability a and n — p
degrees of freedom, then there is insufficient evidence from the data
to support the contention that bi differs from B.
If calculations are made as set forth in this paper, using the individual
U measurements, then each U will have an equal influence on the out-
come, and each drill hole will automatically have a weight proportionate
to the number of U measurements obtained in that hole. Depending on
circumstances there may be valid arguments for preserving equal
weight for each U measurement, or for giving equal weight to each
direction of U measurement (irrespective of the number of measurements
made in that direction), or for giving equal weight to each drill hole
(rrespective of the number of measurements made in each hole). If
the bi are calculated from U values that are averages or interpolated
values, then the weighting in general is altered. Altering the relative
weights of the observations alters the calculation of the standard error
and alters the degrees of freedom used in making statistical inferences.
These complications can be avoided by making the same number of
observations for each different diameter orientation and employing the
same number of measuring directions for each hole. Then if the solution
is made from the individual U measurements, each U measurement
will have equal weight, each measurement direction will have equal
weight, and each hole will have equal weight.

Transformation to a New Coordinate System

Having calculated the least squares estimates of the six independent
components of the stress tensor with respect to the x, y, and z coordinate

system, one may wish to calculate the values of one or more stress
components with respect to a different coordinate system. For example,
the stability of the rock about a tunnel or other underground opening
depends primarily on the same four components of stress that give rise
to the change of diameter of an overcored drill hole: ai, a2, as , and
Ti3, where an arbitrary set of coordinate axes /zi, /?2, and hs are chosen
according to the same convention as for a drill hole (h^ parallel to the
length of the tunnel, hi horizontal to the right, and hi in a vertical plane).
These stress components can be calculated from Eqs 13-16, substituting
therein the least squares estimates of <TJ: , 0-^, 0-2, T^^ , Ty,, and T^X along
with the direction cosines for the tunnel axes hi, hi, and h^ with respect
to the X, y, and z axes.
Another transformation that may be of interest is the calculation of
the principal stresses and their directions from the least squares estimates
bi. Mathematically this consists of finding a new set of coordinate
axes 1, 2, and 3, such that the shearing stresses 512 = 5'23 = 'S'31 = 0.
The axes 1, 2, and 3, are then the principal axes, and the normal stress
components Si, S2, and St parallel to these directions are the principal
stresses. It can be shown [75] that the principal stresses are the three
roots of the following cubic equation, the coefficients of which are
functions of the six components of the stress tensor:

S' + 3BS^ + 3CS+ D = 0. .(47)

where

B = —{cTx+(7p + (7 2)73
C == {(Txay + (TyO-^ + a^(Ti — ,2)/3 .(48)
ly — (TjO'yO'z ^TxyTyzT zx "T ^x"^y z \ O'yTzx \ ^z^xy

The three roots Si^ Si, and Sz of Eq 47 are found by calculating \14\

E ^ {E' - cyi^ ^
F = -B' + (35C - D)/2
cos 3 ^ = ± F/E^
(49)
Si = - 5 ± 2EC0S A
52 = - 5 ± 2EC0S (A + 120 deg)
53 = -B ± IE COS {A + 240 deg)J

(where ± appears, use + if f is + , use — if f is —). The direction

cosines for principal Direction 1 with respect to x, y, and z are found
from [75]

(Si - (Tx)// - TxyWl' - T^^ni = 0

—T^Ji' + {Si — (jy)mi - Ty^ni = 0 i
— Tzxh' — Ty,mi' + (Si - az)ni = 0
( / / ) ' + (W:T + (ni'y = 1,
The direction cosines for principal Direction 2 are found from the
same equations, except that each subscript 1 is replaced by a 2, and simi-
larly for principal Direction 3.
The bearing and inclination of any principal stress can be found as
follows. It is easily shown that, for Direction 1,
[cos (angle of inclination of Si)] X [cos (bearing of Si)] = m / . ,(51)
[cos (angle of inclination of 5i)] X [sin (bearing of 5'i)] = h'. . .(52)
from which
tan (bearing of Si) = / / / m i ' (53)

Given the bearing angle, the inclination can be found from Eq 51 or 52.
In order to calculate the standard errors of the principal stresses,
one must find a new C matrix and the standard deviation s about the
new regression, which can be obtained by repeating the method of solu-
tion as given above for the six components of the stress tensor. That is,
since the choice of the x, y, and z directions was entirely arbitrary, one
could just as well have taken them to be parallel to the principal direc-
tions. The least squares solution for the stress components with respect
to Directions 1, 2, and 3 yields the new C matrix. The procedure is the
same as for the solution with respect to directions x, y, and z. New values
must be calculated for the Ji, the gy and the a^ , using the direction
cosines for the h, h2, and h^ axes of each drill hole with respect to the
Directions 1, 2, and 3, which are given by

Li = hh' + mimi -t- nini\

Ml = lih' + minni' + niUi
Ni = Ilk' + W1W3' + nirii
L2 = hh' + mitni + «2«i'
Mi = hW + 17121712 + «2«2' \ (54)
A'^2 = hh' + niim/ + «2«3'
Li = hh' + mmi' + mni
Ms = hh' + rnznii -f ns"?'
Ni = hh' + Warns' -f rtiUi'
where Li is the cosine of the angle between the hi and 1 axes. Mi is
the cosine of the angle between the hi and 2 axes, • • • , and Ni is the
cosine of the angle between the hi and 3 axes; Ii, rrii, • • • , «3 are the

Copyright by ASTM Int'l (all

Downloaded/printed by
Pontificia Universidad Catolica de
130 TESTING TECHNIQUES FOR ROCK MECHANICS

direction cosines of the drill hole with respect to x, y, and z; //, m/,
• • • , Ms' are the direction cosines of the principal directions with respect
to X, y, and z.
The following principles may help to explain puzzling or seemingly
inconsistent results that are obtained from stress calculations. One
should clearly understand that the least squares solution and the calcu-
lated principal stresses constitute a valid stress system only if the same
stress system exists, or can be conceived to exist, for all the U measure-
ments that are included in the calculations. If all U measurements were
made at a single point, then the stress system necessarily would be the
same for all measurements. However, one must exercise judgment in
pooling U measurements that were made at separated points. In a zone
where the stress system is expected to change significantly over a short
distance, as in the zone of stress concentration around a mine opening,
the U measurements to be pooled should in principle have been obtained
as close as possible to each other.
Alternatively, one may elect to pool only those measurements that
were obtained at distances greater than about L/2 from the edge of the
opening (where L is the long dimension of the cross section) in order
to exclude stress concentration effects [75, p. 59]. However, such a rule
of thumb, which is based on theoretical stress analysis of a perfect
opening, is far from precise for mine applications, as actual U measure-
ments [3,6,15] in many instances fail to exhibit the decreasing trend with
distance from the opening that is indicative of a stress concentration
effect.
The best procedure is to examine the plots of U versus distance from
the collar of the hole, a separate graph for each orientation 6 in each
drill hole. It is usually not difficult to identify the zone in which there
is no apparent trend with distance and for which pooling of the data is,
therefore, permissible.

Summary of Conclusions
A general equation is derived that can be used to calculate the change
of ground stress or the relief of the existing state of stress, primarily
from measurements of the diametral deformation of a drill hole in an
isotropic elastic rock body. Some of the implications of this equation are:
1. The two-dimensional stress components that act normal to the drill
hole axis can be determined from measurements of diametral deforma-
tion in a single drill hole only if additional information, such as the axial
strain, is provided.
2. A set of diametral deformation measurements in a single drill hole
provides a maximum of three independent conditions for solution of the
six components of the stress tensor.
3. Complete solution for all six components of three-dimensional

Stress can be obtained from diametral deformation measurements in a

pair of nonperpendicular drill holes. No additional measurements are
required.
Appropriate directions for the drill holes and for the directions of
diameter measurement therein can be determined by inspection of the
orientation coefficients / , . In general the drill holes should be oblique
to the important structural directions, rather than perpendicular or
parallel to them. The sampling is improved by measuring diametral
deformation in at least four directions in each drill hole.
A least squares solution can be obtained for the six components of
the stress tensor, irrespective of the number of drill holes, the directions
of the drill holes, the number of directions of measurement in any drill
hole, or the number of measurements in any direction, provided there
are six independent directions of measurement.
The least squares solution is a systematic and objective method
for calculating the average values of the three-dimensional stress com-
ponents, in any arbitrarily chosen x, y, and z coordinate system, directly
from the observed diametral deformations, without any intermediate
steps such as interpolating among the measurements or calculating
secondary principal stresses in a plane. The usual methods of statistical
analysis can then be used to calculate the standard errors for all stress
components and to develop statistical interpretations of the least squares
solution.
Finally, equations are given: (1) for transforming the calculated stress
components with respect to x, y, and z coordinates into components
with respect to any other coordinate system of interest and (2) for de-
termining the three components of principal stress and their directions.

References
[/] E. R. Leeman, "The Measurement of Changes in Rock Stress Due to Mining,"
Mine and Quarry Engineering, July, 1959.
[2] R. H. Merrill and J. R. Peterson, "Deformation of a Borehole in Rock," Report
of Investigations 5881, Bureau of Mines, 1961.
[3\ Leonard Obert, "In Situ Determination of Stress in Rock," Mining Engineering.
Vol 14, No. 8, August, 1962, pp. 51-58.
[4] Nils Hast, "The Measurement of Rock Pressure in Mines," Arshok 52, Sveriges
Geologiska Undersokning, Stockholm, Sweden, No. 3, 1958, p. 24.
[5] T. A. Morgan, W. G. Fischer, and W. J. Sturgis, "Distribution of Stress in the
Westvaco Trona Mine," Report of Investigations 6675, Bureau of Mines, 1965.
[6] D. W. Wisecarver, R. H. Merrill, D. O. Rausch, and S. J. Hubbard, "Investiga-
tion of In Situ Rock Stresses, Ruth Mining District, Nevada, With Emphasis on
Slope Design Problems in Open-Pit Mines," Report of Investigations 6541, Bureau
of Mines, 1964.
[7] S. Timoshenko, Strength of Materials, Part I, 2nd edition, Van Nostrand Co.,
New York, 1940, p. 55.
[S] I. S. Sokolnikoff, Mathematical Theory of Elasticity, McGraw-Hill Book Co.,
New York, 1956, Sections 70-78.
[9] M. Ezekiel, Correlation Analysis, 2nd edition, John Wiley and Sons, New York,
1941, pp. A&i-4n\.

[!0] A. J. Durelli, E. A. Phillips, and C. H. Tsao, Introduction to the Theoretical and

Experimental Analysis of Stress and Strain, McGraw-Hill Book Co., New York,
1958, p. 16.
[//] R. L. Anderson and T. A. Bancroft, Statistical Theory in Research, Chapters
13-15, McGraw-Hill Book Co., New York, 1952.
[12\ B. Ostle, Statistics in Research, Chapter 8, Iowa State College Press, An\es, Iowa,
1954.
[13] S. Timoshenko and J. N. Goodier, Theory of Elasticity, 2nd edition, McGraw-Hill
Book Co., New York, 1951, p. 217.
\14] O. W. Eshbach, editor. Handbook of Engineering Fundamentals, Section 2, John
Wiley and Sons, New York, 1936, p. 13.
[15\ L. A. Panek and J. A. Stock, "Development of a Rock Stress Monitoring Station
Based on the Flat Slot Method of Measuring Existing Rock Stress," Report of
Investigations 6537, Bureau of Mines, 1964.

Deformation Moduli of Rocks

REFERENCE: "Deformation Moduli of Rocks," Testing Techniques for

Rock Mechanics, ASTM STP 402, Am. Soc. Testing Mats., 1966, p. 133.
ABSTRACT: There is a real need for more precise definitions of physical
properties of rocks, particularly the deformation moduli of both labora-
tory specimens and rock masses in situ. A vital initial step in establish-
ing standards is to recognize existing problems of engineering applications.
One of the prime purposes of this paper is to furnish background infor-
mation on deformation moduli so that further steps in setting up standards
can be taken. More basic and applied research will be required before
satisfactory standards can be established.
KEY WORDS: rock (material), rock mechanics, physical properties,
elastic modulus, creep, dam sites

In the development of rock mechanics many types of tests have been

devised to yield physical property data for rock. One difficulty in the
use of these data is that a given property is a function of the method of
testing. In many cases results cannot be directly compared, and their ap-
plication may be questionable.
Standardized tests (1) must be susceptible of duplication and (2) yield
data representative of the property tested. To illustrate, the end condition
of a simple compression specimen seriously affects the manner of fail-
ure of many rocks. Many compressive strength data may be of little
value because of this factor.
While the behavior of rocks and earth materials can often be described
by elastic moduli for engineering purposes, the broader term, deforma-
tion moduli, may be taken as inclusive of deformation of elastic, visco-
elastic, viscous, or other character which may be observed in rock ma-
terials.
Both measured and effective moduli are almost without exception de-
pendent upon: (1) rate, (2) duration, and (3) magnitude, as well as (4)
geometry of loading. Hence, static elastic moduli are usually different
than those determined by sonic or dynamic methods. Rocks under long
term loads often exhibit creep, which is a combination of elastic deforma-
1
Director, Rock Mechanics Research Group, University of Missouri, Rolla, Mo.
Personal member ASTM.
133

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
1 34 TESTING TECHNIQUES FOR ROCK MECHANICS

tion and some type of nonreversible deformation. These and other types
of yielding of rock are classified as deformation moduli.
Several groups of investigators in the United States have initiated re-
search programs in rock properties and their use in design of rock struc-
tures [1, 2].2 The standardized laboratory tests employed by the Bureau
of Mines [3] fall into two groups, in each of which results appeared to be

TABLE 1—Deformation moduli.

Although tests and procedures for deformation moduli should be considered
to include all elastic and viscoelastic properties, as listed below, consideration will
be limited only to those which have been successfully applied.
A. Deformation moduli—static
1. Elastic—Young's modulus, shear modulus, Poisson's ratio
(a) Laboratory methods
(1) Standard compressive and tensile methods
(2) Sonic methods
(6) In situ methods
(1) Load bearing methods
(2) Chamber method
(3) Sonic methods
(4) Strain relief methods
(5) Other
2. Bulk modulus, compressibility
3. Uniaxial, biaxial and triaxial tests
4. Viscoelastic moduli
(a) Creep
(6) Viscoelastic models
(1) Voigt
(2) Maxwell
(3) Other
B. Deformation moduli—dynamic (high magnitude and high rates of strain)
1. Elastic—Young's modulus, shear modulus, Poisson's ratio
(a) Laboratory
(b) In situ
2. Viscoelastic
(a) Complex moduli (steady state)
(6) Moduli for transient stress fields
3. Shock moduli
C. Effects of rock structure and sampling
1. Grain size
2. Fractures, geologic structures, etc.
3. Alteration

related in varying degrees. In the first group, of the strength, hardness and
toughness tests, the abrasive hardness was found to be the best property
from which the others of the same group could be estimated with some
degree of accuracy. In the second group, the elastic properties, the
measurement of the longitudinal velocity of sound, that is. Young's mod-
ulus, was found to be one of the most reliable and simple tests to make.

^The italic numbers in brackets refer to the list of references appended to this
paper.

Most of these tests measure either a strength of rock or properties

under stresses of short duration and may be correlated to only a limited
extent with properties of rock mass. Some relationships have been found
between physical properties of rocks and pneumatic and diamond drill
rates, but no exact correlation factor has been obtained. Studies of rock
types and their geological structure, in connection with various methods
of mining, have revealed limited design factors. Evaluation of the re-
sistance factors of rock to primary and secondary blasting is progressing
rapidly, and these are related to one or more of the deformation moduli.
Bearing plate and chamber tests, as well as seismic techniques, establish
useful values of in situ deformation moduli for design purposes.
The above physical properties data of rocks are of real value in im-
provements in excavation operations and engineering design problems
only when employed in close conjunction with local geological data, prin-
cipally geologic structures. Consideration must also be given to prop-
erties, such as plasticity, viscosity, brittleness, creep, fatigue failure, and
others.
Engineering operations in rock deal with forces which range in time
from a few microseconds in the case of blasting operations to several
years in rock structures around permanent openings. They include the
effects of masses varying a few microns in diameter to hundreds and even
thousands of feet in their dimensions. Problems consist of evaluating
the properties and interaction of materials composed of single homogene-
ous crystals or small amorphous masses, but primarily of very large masses
whose structure may be very complex. Analysis and application are fur-
ther complicated by temperature effects, alteration, residual stresses, and
other factors.
Thus, deformation applications depend not only on rock structure, and
composition, and upon the length of time that it is subjected to load, but
also on the correlation of physical properties of small laboratory speci-
mens of rock with the effective properties of the same rock in place in the
earth's crust, including scale effects.

Classification of Properties and Testing Mettiods

In the past two general methods of measuring properties of rock have
been classed as [4]: (1) static and (2) dynamic, (a) resonance and (b) ultra-
sonic (pulse). The term static is appropriate for tests where the rate of
strain is not higher than that of a standard testing machine. While the
rates of strain in the resonance and pulse methods may be high for high
frequencies, the maximum strain is usually small except very close to the
pulse source, and these methods are employed to measure properties
under conditions where the effective stress in the test rock is very small.
Hence, these methods either measure effective static properties or values
at zero load.

Tangent Value

Secant Value

Strain

FIG. 1—Representation of three types of Young's modulus. The average value

would be represented by a tangent at the midpoint of the stress-strain curve.

^Rate of Loading
6075 Ib/sq in/mn

40 80 120 160 200 240 280

Deflection — in. x 10

FIG. 2—Stress-strain curves for sandstone for two rates of loading.

True dynamic moduli are defined as those which describe the behavior
of rock where both the rate and magnitude of strain are appreciable. It is
proposed that a classification of methods and properties be adopted as
outlined in Table 1, including properties such as bulk modulus, creep
properties, shock parameters, and others.
The theories of strength of materials, including deformation at the
yield point, have been summarized by Nadai [5]. Similar materials will
behave in different manner depending upon the stresses, that is, tension,
compression, or torsion, some of which were closely related to deforma-
tion moduli. Strength theories include the Maximum Strain Theory, the
Theory of Constant Elastic Energy of Deformation, the Theory of Con-

stant Elastic Strain of Distortion, the Maximum Shearing Stress Theory,

and Mohr's Theory of Strength [6].

Elastic Properties
The elastic properties of rocks differ [7] from those of structural metals
where constants are nearly independent of the applied stress. However,
the elastic moduli of rocks usually are dependent upon the magnitude of
stress, particularly at low values where the relation between unit strain
and unit stress is not linear. Young's modulus, defined for small displace-
ments of an equilibrium displacement about either zero or equilibrium
stress, often increases or decreases with increase in stress, sometimes by
large factors. The other elastic constants also may vary with stress, due

1 1

1 L N O Failure
30 Green -
Stone-J / 0 Failure
.
25 /
V)
A—Lithostone
a.
I 20 / ^
v>
V
i 15
V)
/"''^Shale
0
//
-j/ ^ O i l Shale
1 • 1 1 1 1 i 1 1

0 2 4 6 8 10 12 14 16 18

Strain p in / in
FIG. 3—Stress-strain curves of rocks.

to initial porosity. In compression pore spaces are partially closed by the

initial load. In sonic testing the pore spaces and heterogeneity of rocks
also cause deviations from ideal behavior. Thus, some values of Poisson's
ratio are negative when calculated from sonic moduli. Elastic constants
are meaningful only when defined in terms of test conditions and what
they really represent. This illustrates only one of the needs for standardi-
zation of tests and interpretation of results.

Young's Modulus
If sonic specimens are vibrated unstressed, the resulting value of
Young's modulus is the tangent value at zero load. Secant values and
average values are sometimes employed (Fig. 1). Stress-strain curves [7]
for different loading rates for sandstone (Fig. 2) indicate that the "elastic
constants" are also dependent upon time as well as magnitude of stress for
the rocks tested.

Illustrative types of stress-strain curves (compressive) (Fig. 3) have

been obtained from static tests for sandstone, oil shale, lithostone, and
greenstone. The flat initial portion of the curve for sandstones is due to an
initial stage of compression of little elastic strain but consists largely of
forcing grains closer together. The flattening of the oil shale curve in its
upper portion might be due to a failure of some of the weaker constituents
of the rock.
Other investigators [7] have stated that the various determinations of
modulus of elasticity of rocks are unreliable. Values of E for a granite-
gniess of a breaking strength of 800 kg/cm^ varied according to the dif-
ferent methods as follows:
1. Simple compression with a base of deformation measurement of
about 15 cm; value of E from stress-strain curves ranged from 250,000
to 350,000 kg/cm^.
2. Simple compression with a base of deformation measurement of
1 cm; E = 800,000 kg/cm^.
3. By vibration method; E = 200,000 kg/cm^.
4. By sound velocity method; E = ±700,000 kg/cm-.

Static Method [1]

For Young's modulus in the compressive phase cylindrical specimens
of rock are prepared with the ends lapped to provide flat surfaces per-
pendicular to the main axis. These are cut to a length of approximately
2.5 times their diameter. Electric resistance strain gages are cemented
on the opposite sides of the specimen parallel to the longitudinal axis.
The same test procedure is followed as in that employed to determine
compressive strength. The load is applied at the rate of 100 psi/sec to ap-
proximately 50 per cent of the compressive strength, with five or six
readings of load and strain being taken at evenly spaced load increments.
One minute halts are made for each reading. A similar set of readings is
made as the load is released. Six complete cycles of loading and unloading
are then taken to the 50 per cent load value, but normally data for these
additional cycles are not recorded. During the next cycle, which is the
seventh, readings are taken as on the initial portion of the first cycle,
and then the loading is continued until failure takes place.
A graph of the average stress and strain values is plotted. The secant
value at the initial load is an approximation of the tangent value at zero
load and, therefore, may often be compared with the value obtained by
the sonic method.
In relation to Young's modulus in tension, several methods have been
devised for measuring the tensile strength of rock including the Brazilian
method, modulus of rupture test, and use of "brickette" type specimens
similar to those employed for concrete. None of these are amenable, how^
ever, for measuring Young's modulus in tension.

Grosvenor [8] has suggested a method of applying an axial tensile

load to a cylindrical specimen which will also permit the measurement
of both stress and strain. A set of steel plates is glued to the ends of the
specimen, the plates being the same diameter as the cylinder. The tensile
load is applied through a set of universal joints affixed to each plate.
Particular care is taken to obtain alignment to avoid nonaxial loading
and twisting.
Although this or similar methods appear to offer a means of determin-
ing a Young's modulus for the tensile phase, no data are available.

Resonance Method [1]

The resonance method is not now extensively used, but Young's modu-
lus may be calculated by determining the frequency of induced longi-
tudinal vibrations in a specimen (drill core) of rock which is suspended by
a clamp at its center, and the vibrations are measured in their funda-
mental frequency.
The longitudinal velocity is given by:
ci = 2fiL (1)
where:
ci = longitudinal velocity of sound,
Ct = torsional velocity of sound,
fi = fundamental longitudinal frequency,
ft = fundamental torsional frequency, and
L = length of specimen in feet.
From the velocity determined by Eq 1
E = cip

G = c^p
r2
. = A - 1 = =/L - 1 (2)
2G 2P '
where:
E = Young's modulus,
G = modulus of rigidity (shear modulus),
p = density of rock, and
V = Poisson's ratio.
The theory of resonance experimentation is discussed by Bancroft
[9] and Pickett [10], their results indicating that the shape of the specimen
may have some influence on the accuracy of the results.
Modulus of Rigidity
The modulus of rigidity, defined as the shearing stress divided by the
shear deformation, is also a function of confining pressure. Very large

changes in rigidity are observed in many rocks upon the appUcation

of the first few hundred atmospheres of confining pressure. Where the
confining pressures are very high the change or rigidity with pressure be-
comes nearly linear and much smaller.
Rigidity might also be defined as the property that solid matter has to
offer which is instantaneous resistance to deformation. Sandstone, then,
has a low rigidity, while that of limestone, granite, marble, slate, and
greenstone vary increasingly in the order named. In a single crystal,
rigidity is largely determined by interatomic forces. Stress distribution in
most rocks is concerned with intercrystalline and intergranular bonds, as
well.
One type of modulus of rigidity may be determined by the sonic res-
onance method. A cylindrical specimen is clamped as in the test to deter-
mine Young's modulus, and the fundamental frequency of torsional
vibration is measured. The torsional velocity is then calculated from ap-
propriate equations. Statically determined values are not easily measured
as no completely reliable direct method appears to have been devised to
make measurements, although specimens in torsion have been utilized
to measure shear parameters. In situ methods have also been employed,
that is, utilizing velocities of shear and compressional waves. Panek [77]
has used a cylindrical pressure cell method for determining data for static
modulus of rigidity.

Sonic In Situ Moduli Tests

In the determination of elastic constants of rock in situ by sonic meth-

ods [72], longitudinal and shear waves are generated in rock by small
explosive charges in shallow drill holes. Accelerometers and strain gages
measure arrival times for both waves. From wave velocities and measured
density, the values for Poisson's ratio, modulus of elasticity, modulus of
rigidity, Lame's constant, and the bulk modulus are calculated. The same
types of waves and values of elastic constants may be determined by
laboratory methods and the results compared.
The procedure employed above utilizes a series of 8 by 1 Vi-m. shot
holes at distances of 50 to 450 ft from gages. Gage mounts of steel on 6-
in. studs were grouted in gage holes. The gages, particle velocity and
acceleration, were mounted with the proper orientation to measure
motion perpendicular to the line from the gages to the shot hole. At-
tenuator settings were made on trial shots, and then shots were fired and
results recorded on magnetic tape to determine optimum charge size,
resolution of shear wave arrivals, effect of variation of wave paths, and
finally to establish reproducibility. Test rocks were salt and granite-gniess.
The equations employed are:

V =
m-
2
}'''/
1
= Poisson's ratio (3)

(0 1

E = pv: ^

m 1
Young's modulus

G = pV^ = modulus of rigidity

(4)

(5)

X = pVs \\TT) ~ ^\ ~ Lame's constant (6)

k = p F / U p ? j - 11 = bulk modulus (7)

where:
p = density,
Vp = longitudinal velocity, and
Vs = shear velocity.
The calculations from field data are primarily from the ratio of longi-
tudinal and shear velocities. The field studies gave higher values for
E than laboratory tests and more reasonable values for Poisson's ratio.
The most popular method of measuring shear and longitudinal wave
parameters in the laboratory is the single pulse method. Gregory [13]
describes methods of exciting both longitudinal and shear waves, the
latter being derived from the compressional mode, utilizing j:-cut quartz
crystals. Shear wave velocities may also be excited utilizing y-cut quartz
crystals.
Pyrex wedges may also be employed to excite shear waves. The
theory of P-wave energy conversion to reflected 5-wave energy was in-
vestigated by Gutenberg [14] and later by Arenberg [75]. It has been
found that at a certain critical angle of incidence on a solid-free air in-
terface all of the P-wave energy is converted to 5-wave energy. Use of this
type of transducer [76] may be extended to anisotropic materials. The
ability of the material to convert reflected energy depends upon Poisson's
ratio which must have a value ot v < 0.262, Pyrex glass having a value
of 0.24. From the basic equations it is found that for v = 0.25 a shear
wave is totally converted at an angle of 30 deg. Pyrex wedges are located
as shown in Fig. 6 with ceramic transducers attached as shown with a
phenyl salicylate sealant.
The pulse source used by Jamieson and Hoskins [76] for measuring
wave velocities in brass was a pulse generator triggered with a sawtooth

of a waveform generator 162. An incident P wave from tlie driving trans-

ducer is reflected into the specimen as an S wave, transmitted and in turn
reflected in the receiving wedge as a P wave for detection by the receiv-
ing transducer. Wedges must be calibrated and travel times calculated
from equations, taking into account the total travel distances of the
waves.
Ultrasonic methods have also been employed on very small labora-
tory rock specimens [17], utilizing procedures similar to those for metals
[18].
Compressive and shear wave velocities of ultrasonic pulses are meas-
ured across a small cuboid specimen. Accurately parallel surfaces are
polished, and a thin coat of silver is applied to them for an electrical
ground. For most compressional waves a 50-v pulse of Vk /xsec duration is
applied across a 2 mc/sec BaTiOs transducer. For shear (rotational)
waves a 1500-v pulse of 10 ;usec duration is applied across a 2 mc/sec
y-cut quartz transducer. Transit times are measured to 0.01 fx sec with
conventional electronic means. The shear and bulk moduli are calculated
utilizing the relationship between shear and compressional velocities and
the elastic constants.
Comparison of Static and Sonic Measurements
Since laboratory sonic determination of Young's modulus is usually at
low stresses, a comparison of static and sonic values is meaningful only if
the values of the static modulus are taken at comparable stress levels,
that is, using initial or zero stress tangent modulus.
Values obtained by static techniques are in general lower than those
obtained by sonic methods, the difference in the constants so deter-
mined varying from 0 to 300 per cent. The difference has been ex-
plained [19, 20] as being due to the presence of fractures, cracks, or
cavities, with the static yielding being increased by deformation of cracks
and cavities and the sonic measurements being less influenced. A stress
wave is transmitted by the matrix of rock, and high frequency com-
ponents are reflected and refracted from the crevices and cavities. The
greater the degree of compactness of the rock, the more nearly static
and sonic constants may agree, as illustrated by the foflowing:

Young's Modulus
Poisson 's Ratio
Static Dynamic
static Dynamic
(dynes / cm^)

Quincy granite 3.5 X 10" 4.3 X 1 0 " 0.10 0.33

Sudbury norite 8.36 X 1 0 " 8.82 X 10" 0.22 0.27

The higher value and the closer agreement of the static and sonic data in-
dicate that Sudbury norite is more compact.

ROCK NAME YOUNG'S MODULUS RATIO

Chalcedonic limestone

Limestone ^ — I

Oolitic limestone

Quartzose stiole 1.33

Monzonite porptiyry 1.36

Quartz diorite 1.42

Stylolitic limestone 1.46

Biotite sctiist 1.48

Limestone 1.70

Limestone 1.86

Siltstone 2.05

Subgraywaciie 2,11

Sericite sctiiit 2 36

Subgraywacke 2.37

Quortzose pliyllite 2.45

Calcareous stiole 2.56

Subgraywaclte 2,57

Granite (slightly altered) 2.75

2.78
Graptiitic phyllite

2.90
Subgraywacke

^ • • • 1 Static Young s modulus

I I Dynamic Young s modulus

F I G . 4—Static and sonic values of Young's modulus.

The results obtained by the Bureau of Reclamation [21] investigation

on sonic and static moduli of the Davis Dam foundation rock agree with
the conclusions of Zisman and Ide [19, 20]. The mean value of Young's
modulus in sonic measurements was found to be 10^ psi and that in static
measurements was 5 by 10* psi. A longitudinal pulse transmitted across
waterfiUed gaps in concrete or across a crushed concrete specimen
showed little attenuation of velocity. Sonic measurements, therefore,
could indicate a fictiously high value of modulus for fractured rock or
concrete. Similar results were obtained by Dvorak [22] for a medium un-

der pressure, but differences did not exceed 50 per cent because fractures
were partially closed by pressure.
The ratio of the sonic to static moduli may vary between 0.85 and
2.9 [23], according to data determined by the Bureau of Reclamation
[24]. The description of rocks and the values of Young's modulus and
Poisson's ratio for static and dynamic tests does not detail open frac-
tures, porosity, alteration, or boundaries between crystals. The differ-
ence between static and sonic values is also less for rocks which have a
larger elastic modulus (Fig. 4).
Sonic constants are frequency dependent in a range from 40 to 140 cps
[25] for different rocks, that is, the higher the frequency, the higher the

24,000 24,000

Stress vs. E Stress vs E'

20,000 20,000

. _. . Incrementol stress . / , _._ Totol stress ,

Incremental strain Total strain
£ 16,000 / 16,000

12,000 12,000
/
o / Specimen A5 Specimen A5
" 8,000 / (Ouartzite) 8,000 (Ouartzite)
/ "
4,000 / 4,000
X' Sonic E = Sonic E =
L^-^ l2-9xl0' lb/in* l2-9xl0' \\>/\n.
-J L- -L-
4 6 8 10 12 4 5 6 7 8
Modulus of deformation,(E)xlO, lb/in,

FIG. 5—The effect of stress on values of Young's modulus.

modulus, with values being 2.1 to 2.6 per cent higher for various rocks.
For higher frequencies [26] of 140 to 4500 cps the variation is less than
1 per cent, and the increase does not exceed 4 per cent.
Eight other types of rock were tested by Sutherland [27] using static
and sonic methods, and he also found that static values for E and G were
consistently lower than sonic values, while statically determined values of
Poisson's ratio were higher. Young's modulus also varies with the stress
level (Fig. 5).

Viscoelastic Properties
Creep
Griggs [28] performed a series of experiments on rocks to determine
their creep characteristics. While experimentation on rocks has been

O
1^-
•¥•

— —
+
»-' •^

00 (0
O §
't
1
1
II 1 \ II
>>•
3
- s
<s> o
c<no U- ro
to u> m 10 •S
*
- lO Ci.

o u ^
c c c c CM ^J

CO ro fe
^ 0) CJ
— If)
CM
- S
II II II
CM

II
oo
1
a
>- >• >• V " CM
I ^-^ S—*
«ox ^-^ ^.10^ (0 >4
T mm (0 CM
c c ^
1- -Tg
\ o o o o CM C/)
\ Q. Q. a. Q. >- ."^
*- ~
CM
CM <
Q
12
O ^
h- O CM
&a K i. ^
1 1
OO ^
U
~~ s
\
•4
\
V
-^ o X 4

- 5^
(0
1
•H-v
0

CM §

- g
1
*i^ V
00
i
(0

^r
I
2
<C
1 \ 1 r 1

V. O ° O o o o O o o O O
CM —
>•> = o cR 00 CD in ro
3

primarily concerned with the flow of rocks subject to tectonic forces,

the results obtained from these studies reveal important facts of interest
in the study of engineering rock structures.
When rocks are subject to stresses for long periods of time, the terms
"elastic limit," "set point," and "strength" lose their ordinary meaning be-
cause they usually define properties in terms of short period tests. Creep
may be analyzed, according to Griggs, as an aggregate of two types of
flow: (1) "pseudo-viscous flow," which is deformation at a constant rate
and (2) "elasticflow,"which decreases logarithmically with time.
Shale and limestone shorten rapidly during the first part of the test
period, and then the curve levels off rapidly with time. The time for
limestone to shorten is well within the range of periods during which

"1 rn \ f—\ \ \ \ TT-

I I I I I I I I I

PRESSURE - BARS X 10^

FIG. 7—Compressibility of granite, gabbros, and diabases as a function of
pressure.

rocks around openings are required to support stresses which may be

close to their ultimate strength. It is also possible that creep may tend
to increase the zone of high stresses in such a manner as to hasten rock
bursts, especially if resistant layers or bands of different creep character-
istics are present. Flow or creep of rocks is partially a result of rearrange-
ment of atoms within crystals (intracrystalline slip). For example, evi-
dence points to the hypothesis that creep in lead is largely due to
intergranular motion, whereas rapid deformation occurs largely by intra-
crystalline slip.
In an application of creep analysis to the deformation of a circular
shaft in salt, Barron and Toews [29] found that for points 6 in., 4 ft, 7 ft,
and 10 ft, in the solid wall of the shaft that the displacement may be ex-
pressed in the form

Ur = —pB
© log (1 + bt). (8)

where:
Ur = radial displacement,
p = pre-excavation pressure in the rock around the shaft,
B = arbitrary constant,
t = time in days,
b = & constant,
a — shaft radius in inches, and
r = radial distance to the point in the solid.
The above equation fits data points well except for displacements at the
surface of the shaft (Fig. 6).

Compressibility
Compressibility [30] is defined as the relative change in volume per
unit change in pressure. This may be represented in terms of partial dif-
ferentials or in terms of volume and pressure,

5 =1.^ or 5 = 1^^1—^ (9)

the latter being for small or linear changes and both defined for changes
at constant temperature. Here B is the compressibility, V the volume of
the specimen at pressure P, and Vo the volume at an initial pressure,
Po. It is also assumed that the chemical composition and physical state
of the specimen remain constant and that the pressure is hydrostatic. The
units of measurement of compressibility are the reciprocal of the pres-
sure units employed.
Compressibilities of common minerals were obtained in a heavy-
walled cylinder under high pressure. Volume changes were measured
by appropriate electrical devices selected for their accuracy. The concept
of compressibility can be applied to anisotropic materials which are made
up of grains or crystals, while elastic properties such as rigidity, Young's
modulus, and Poisson's ratio can be applied with accuracy to only iso-
tropic materials or those which approach isotropy, so that elasticity may
be assumed with only small error.
Comparison of compressibilities of rocks with those of constitutent
minerals shows that acidic rocks have lower values and, in general, con-
sist of less compressible minerals. For pressures above 2000 bars the
compressibility of a rock can be calculated with satisfactory accuracy
from the individual compressibilities of the various minerals. Below this
pressure the agreement between calculated and measured results varies
as much as 5 per cent. To calculate the compressibility the volume per-
centage of each mineral is multiplied by its compressibility, and the sum
of these products is then the calculated compressibility.
At about 2000 bars the compressibility of various rocks of a given
type is nearly the same (Fig. 7). The width of the shaded areas indicates

AMPHIBOLITE
ARGILLITE
BASALT
BORAX ooagje •^
CHERT
CONGLOMERATE
DIABASE
DIORITE
DOLOMITE J!il_
EPIDOSITE e <%) o o G
GABBRO
GNEISS
GRANITE fW''<iT°%fS'
GREENSTONE
HEMATITE ORE
HORNSTONE
JASPILITE

MAGNETITE ORE
MARBLE
MARLSTONE 4°°°*° •cl»?r4'*
META-RHYOLITE
MONZONITE
PEGMATITE
PHYLLITE
PORPHYRY
PYROXENITE
QUARTZ
OUARTZITE
SALT
o STATICALLY DETERMINED
SANDSTONE
. DYNAMICALLY DETERMINED
SCHIST
SHALE
•Sip's-'
SHONKINITE
SILTSTONE
SKARN
SLATE
SYENITE
TACTITE
TUFP
-I L I I I I i I
0 5 10 15 20
YOUNG'S MODULUS (psixlO*)

FIG. 8—Relationship between rock type and Young's modulus.

the degree of variation for ordinary rocks within a given class, in this
case granites, gabbros, and diabases. Pore spaces probably account for
the high compressibility at low pressures. Gabbroic rocks may have a
porosity of about 0.1 per cent, while that of granite rocks is higher but
seldom exceeds 1.0 per cent.
A large initial change in rigidity with the application of pressure in-
dicates high porosity, similar to the anomaly in Young's modulus in
static tests. Granitic rocks sometimes exhibit anomalous elastic behavior
because: (1) they are composed of both highly compressible and less

compressible mineral granites, which result in a complicated set of stress

conditions in the mineral grains, or (2) they possess a looseness of struc-
ture which allows elastic behavior only after the application of sufficient
pressure to lock the mineral grains tightly together.

Correlation of Standard Physical Properties

Judd and Huber [31] used a statistical analysis to correlate physical
properties of rocks with geological structure minimized. The fourteen
physical properties were compared: apparent specific gravity, per cent

Rie SG
YM SG
^.gz^
LINEAR

CURVILINEAR

MODULUS OF RUPTURE ( P S l )
MR SH
IMPACT TOUGHNESS (PSI)
I T SH
SCLEROSCOPE HARDNESS ( s )
I T LV . ^ LONGITUDINAL VELOCITY ( F T / S E C )
MODULUS OF RIGIDITY ( P S i )
COMPRESSIVE STRENGTH ( P S i )
YOUNG'S MODULUS OF ELASTICITY (DYNAM IC)(PS|)
POISSON'S RATIO
APPARENT SPECIFIC GRAVITY
APPARENT POROSITY (7.)
SLIGHTLY CURVILINEAR SPECIFIC DAMPING CAPACITY

RIG
CS
^ .
RANDOM GROUPED

FIG. 9—Sketches of comparative scattergrams.

porosity, shore hardness, abrasive hardness, specific damping capacity,

moduli of rigidity and rupture, static and dynamic moduli of elasticity,
compressive strength, impact toughness, longitudinal sonic wave velocity,
tensile strength, and Poisson's ratio. Tensile strength was finally omitted
because of a lack of data, as well as the results of triaxial tests because
of a lack of standardization of test methods. Poisson's ratio data were also
omitted because they were directly calculated from the other measured
elastic properties.
These studies were prompted by the lack of usable quantitative values
for proper design of foundations and tunnels, and discrepancies between
seismic, sonic, and static measurements. This presents a problem in the
design of an arch dam as to whether the value obtained by sonic or
seismic method is the Young's modulus required in dam design formulae.

Judd and Huber compared the fourteen properties by means of linear

regression with each other by pairs, and then in increasing combinations;
there was no experience to indicate if any one of the variables could be
considered as completely independent or dependent of the others. The
data most nearly satisfying the necessary criteria of homogeneity of
preparation were those of the U. S. Bureau of Mines.
A comparison of rock petrographic descriptions with Young's modulus
values obtained by static load tests was inconclusive (Fig. 8). However,
in a plot of some results as cumulative percentages, sufficient discernable
trends developed to warrant further future study.
The x-y plots or scattergrams of pairs of data were visually compared,
permitting a rough match of as many as four pairs of data (Fig. 9), even
though all ordinates were not of the same magnitude. Some of the plots,
indicating apparently definite linear relationships, were discarded because
the X values were derived from the y values. Some individual plots in
one category have a distribution somewhat similar to those in the other.
For example, longitudinal velocity versus impact toughness yields a dis-
tribution similar to shore hardness plotted against modulus of rupture
or impact toughness. Comparable anomalies also appear where specific
damping capacity distributes similarly to some other values, and rigidity
and Young's modulus compare to the dissimilar characteristic of specific
gravity. The similarity in distributions between dissimilar rock charac-
teristics is a hopeful indication that some correlations might be obtained,
although not necessarily by linear regression analysis.
The computer was programmed to find which properties were most
related, and compressive strength was compared with all of the other
properties, first in all possible combinations of two, then three, and
finally eight variables. On a purely linear basis, the best relationship
was between compressive strength and the combination of impact tough-
ness, scleroscope hardness, and modulus of rigidity. However, the devia-
tion is too large for practical application.
The best indications of linearity of results are in the scattergrams of
Fig. 9 which demonstrate that the dynamic modulus of elasticity varies
directly as the static modulus of elasticity when both are obtained by lab-
oratory methods.
On the basis of this somewhat limited study, the following general con-
clusions were made by Judd and Huber.
1. There appears to be a direct linear relationship between modulus
of rigidity and Young's modulus, between compressive strength and
Young's modulus, modulus of rigidity and modulus of rupture, and be-
tween the laboratory values of the static and dynamic moduli of elasticity.
2. There is also a direct curvilinear relationship between modulus of
rupture and impact toughness with scleroscope hardness and longitudinal
velocity.

3. Specific damping capacity possesses an inverse curvilinear rela-

tionship with impact toughness, compressive strength, modulus of rigid-
ity, and Young's modulus.
4. A direct slightly curvilinear relationship between specific gravity
and modulus of rigidity and Young's modulus and between modulus of
rigidity and longitudinal velocity is indicated.
5. Other comparisons indicate possible trends, but additional data
will be required for more complete definition.

Deformation Moduli and In Situ Stresses

For ideal rock, the radial deformation of a circular hole is employed
to calculate the stresses around the hole for plane stress and plane strain
[32-35]. Both Young's modulus and Poisson's ratio must be known.
The applicable theory is reviewed by Merrill and Peterson [32] which
uses the equations for stresses around a circular hole in a plate which is
subject to uniaxial or biaxial loads:
For uniaxial load, plane stress

u = ~{l+ 2cos2e) (10)

E
uniaxial load, plane strain

u = 5 (1 - v-)(l + 2 cos 29) (11)

E
biaxial load, plane stress

u = i[is + T) + 2(s - T) cos 26] (12)

E
biaxial load, plane strain

„ = ^ ( i ^ L l ) [(, + T) + 2(s - T) cos 26] (13)

E
where:
u = radial displacement,
d = diameter of hole,
s, T = mutually perpendicular applied stress, T vertical,
6 = angle clockwise from s,
E = Young's modulus, and
V = Poisson's ratio.
The relationship between the principal stresses and the other variables
is

s + T = ^iui + U2 + m) (14)
3a

^-T= ^ [(«i - u,f + («2 - u,Y + im - u^)T. •. (15)

tan2g, = y 3 ^ " ^ ~ " ^ ^ (16)

2MI — M2 — MS

where:
ui, U2, Us = borehole deformation at 60-deg angles,
M is + for increasing diameter, and
01 = angle from 5' to MI , counterclockwise.
if
U2 > Us 90 deg > ei> 180 deg
M2 < "3 0 deg > ^1 > 90 deg
ui = Ms and I' (17)
Ml > W2 ^1 = 0 d e g
Ml < U261 = 9 0 d e g

The theory was checked by using 5 by 5 by lO-in. blocks of hydrostone,

limestone, and marble with a 1-in.-diameter hole. The stress-strain curves
for hydrostone were hnear, and the radial deformation curves were also
straight lines. However, the stress-strain curves for hmestone and
marble were curved, and the radial deformation curves were concave
upward.
For values of Poisson's ratio about 0.25 the difference between plane-
stress and plane-strain results are small; hence, Eqs 10 and 11 are usually
employed.
Obert [34] reports results of stress-relief methods in trona, salt, and
tuff by the overcoring method or cutting a stress-relief block using a
rosette gage [36] to measure the deformation at 60-deg angles in a plane
normal to the axis of the borehole. Conclusions resulting from this
experimentation were:
1. The success of the borehole deformation method requires that the
rock be relatively isotropic and have a linear stress-strain curve with a
closed hysteresis loop.
2. Gage sensitivity permits measurements in rock with E = + 1 0 X
10« psi.
3. Accuracy is decreased by nonisotropic condition and creep. Meas-
urements may be sufficiently accurate for engineering purposes if these
effects and the hysteresis of the stress-strain curve is not too great.
4. Predicted (theoretical) stress concentrations at the surface of the
hole were either absent or smaller than calculated from stress equations.
5. Maximum and minimum stresses were compressive, and a Poisson's
ratio of +0.33 was indicated.
6. Fractured and jointed rock gave erratic values, indicating a large
number of measurements for accurate determination of magnitude and
direction of stresses.

TUNNEL ROOR CEMENT-SAND

MORTAR
AIR OPERATED
HYDRAULIC
PUMP

I ] ^ ^ ACCUMULATOR

I , , rn^-"^— 1 ^CONTINUOUS PRESSURE

^-T-' V\ RECORDER
HANDPUMPJ U I 1

FIG. 10—lacks and equipment employed by U.S. Bureau of Reclamation for

jacking tests.

One of the apparent sources of error of the method is that the values
of V and E must be measured on intact laboratory specimens, and, in
some cases, these may differ from the in situ properties.

Shock Parameters
Shock-wave parameters are defined in terms of quantities involved
in the equations of the hydrodynamic theory, which give the "Rankin-
Hugoniot" conditions [37\,

p = pouU .(18)

V/Vo = 1 - u/U .(19)

E - Eo= P{V, - V)/2. .(20)

where:
P = pressure,
V = specific volume,
E = internal energy,
p = density,
U = shock velocity, and
u = particle velocity.

The subscript o refers to undisturbed conditions ahead of the shock, and

Po is assumed to be negligible.
Shock waves are usually induced in rock specimens by use of ex-
plosives either directly to the rock or through a "standard" material
where equation of state is known. The values of po, U, and u can be
measured, and from these the specific volume (reciprocal of density) of
the material in the shock front, the internal energy difference, and the
shock pressure can be calculated.
Relationships between certain of the parameters may be plotted, the
curve obtained being taken as an expression of the equation of state.
Such plots, for example, might be particle velocity u versus shock veloc-
ity U.
In some materials the shock velocity is lower than the elastic-wave
velocity, and in this case the shock wave is preceded by an elastic pre-
cursor. Shock data in such cases may be subject to considerable scatter,
particularly at lower pressures.
Other related shock properties of importance are [38] the value of
the Hugoniot elastic limit, release adiabats, and elastic-stress relaxation.
Further analysis of results may be undertaken in synthesizing the Hu-
goniot equation of state for multiple component solids, such as rocks
[39]. This assumes that the shock pressure in the solid is equal to the
sum of the partial shock pressures of the constituents.
The standardization of test procedures for shock parameters is not
as urgent, perhaps, as many of the structural properties of rock. How-
ever, there will be an ultimate need for standardization of tests for solid
materials, including metals and rocks.

Field Jacking Tests for Dam Site

Tests of this type used by the Bureau of Reclamation [40] are designed
to measure the deformation of rock at the periphery of a tunnel when
it is subjected to local pressure with a hydraulic jack (Fig. 10). Tests
are conducted in exploratory drifts in each abutment along approximate
lines of principal stresses. Data are used to calculate the in-place deforma-
tion modulus of the rock.
The basic procedure is to apply vertical and horizontal loads of known
magnitude to the rock inside the abutment drifts and to measure the re-
sulting rock deformation. Hydraulic jacks of 200-ton capacity and auto-
matic pressure maintaining systems are used to apply loads. Loads are
transmitted to the roof of the tunnel through a jack shoe, timber block-
ing, a 24-in.-diameter oak pad, and a 3-in.-thick mortar pad. Downward
loads are transmitted through a jack shoe, a hydraulic cushion, and a
concrete pad 24 in. in diameter and 5 in. thick.
The site is prepared by removing all loose rock that has been visibly
damaged by blasting operations to provide a relatively plane surface so

1'
^ "^Concrete /
Pod ' - •

""-Rubber wosher

^ " ^ - - F e l t , cotton waste

or sponge rubber
"^Rubber washer

-Special j o i n t
meter

-^.
-Tubing i " o D.
Tr^v X ,186 wall

-Cut wosher i g'dia

-Inside tube coupling

-Half
Coupling

': Approx- k-

0
_L _L J
SCALE OF INCHES

FIG. 11—Rock deformation gage for foundation jacking tests.

the equipment may be properly installed. Site criteria are: (1) the site
should be at least 35 ft from the tunnel entrance, and (2) major joints
or other conditions not representing average abutment rock should be
avoided. As surfaces are prepared newly disclosed fractures may require
"maneuvering" of the load position. Experience had shown that blasting
fractures in biotite schist and quartzite normally extend to depths of 6
to 12 in. into the rock; hence, 18 in. was excavated for loading surfaces.
For blocky and jointed conditions in the left abutment 2 ft of rock was
removed. Rock was removed with pneumatic and hand drilling tools.
An AX diamond drill hole is then bored 16 ft into the rock normal to the
prepared surface.
A rock deformation gage (Fig. 11), whose sensing element is a
Carlson joint meter, is grouted into the drill hole, and the loading equip-
ment is installed in steps to insure proper alignment and load distribution.
Cold roUed steel anchors are embedded in line with the jack in the ceiling
and the floor, to which are attached invar rods, their ends being about
12 in. apart. The deformation of the tunnel is measured by measuring
the increase in this gap with a micrometer as the load is applied.
Loads are applied as follows:

Load, psi Duration, days

200 6
0 1
400 6
0 1
600 6
0 1

Strain is read from the Carlson meters and tunnel deformation recorded,
as well as air temperature in the tunnel.
The deformation modulus for surface readings is calculated from

£ =i L ^ (21)
law
where:
P = applied load,
a = radius of bearing plate,
w = displacement, and
V = Poisson's ratio.
The deformation at depth is calculated by use of Habib's formula [41]:

^ _ f (1 + '') - ^ - ^ + 2(1 - .) tan-^(a/Z) .(22)

27ra fl^ + Z^
where Z = vertical depth.

-' (oz«,.i-3nvos)
S3H0NI
30VJ MSOU JO
'ISd ' 30VJ )O0tl I V 3linSS3lld • NOI1031d3a
N0I133S SSOUS

Shannon and Wilson [42] utilize much the same general procedure;
distances between horizontal and vertical tests are kept less than 10 ft
if possible. Rock sites tested must be: (1) representative and (2) under
proposed dam. Loading areas are chipped until deviations are less than
2 in., and within 3 in. to 90-deg plane with centerline of bearing surfaces.
These requirements are modified to fit local conditions.
Loads are applied with 38-in.-diameter Freyssi flat jacks set in a 1:1.5

TABLE 2—Comparison of two recent field-jacking tests.

Morrow P o i n t Dworshak

Totalload 200 tons 425 tons

Size of shoe 24 in. diameter, round 34 in. diameter
Unit load 995 psi 1000 psi
Equipment hydraulic jacks 34 in. diameter Freyssi
flat jacks on grout and
adjustable struts
Direction horizontal and vertical horizontal and vertical
Rock surface below blast damage below blast damage Less
than 2 in. deviation of
surface
Gage rock deformation, Carl- Carlson
son
Length 15 ft 3 in. 15 to 18 ft
Grout Portland cement, 24 hr
set
Surface deformation .. micrometer extensometer dial gage extensometer
between steel anchors 0.0001 in. reading
Temperature control.. . . none given 0.5 F, corrections for
more than 1 F
Pressure control pressure recorder and preset-electronic control
regulator
Testing 200 psi, 6 days 5 cycles, 5 to 10 min, 250.
0 psi, 1 day 750, 1000 psi
400 psi, 6 days
0 psi, 1 day
600 psi, 6 days
0 psi, 1 day
Strain r a t e . none given less than 0.001 i n . / h r
Creep not measured Checked with 2 to 3 day
max load

sand grout. Instrumentation is similar to that of the Bureau of Reclama-

tion, consisting of extensometers for surface deformation and Carlson
joint gages for deformation at depth.
The test procedure involves shorter periods than those used by the
Bureau of Reclamation, as follows:
1. The load was applied in five cycles to pressures of 250, 500, 750, and
1000 psi; the maximum loading cycle was repeated. A cycle consisted of load-
ing to the specified load in 250 psi increments; then unloading again at 250
psi increments to 0 psi. A pressure of 1000 psi produced a total load at the
rock face of 425 tons.

2. The maximum and zero load in each cycle was maintained until the
strain rate at the center extensometer (two way movement) was less than
0.001 in. per hour.
3. Intermediate deformation measurements were taken during loading and
unloading cycles at pressures of 250, 500, and 750 psi. These intermediate
pressures were maintained only for the length of time required to take the
measurements (approximately 5 to 10 min).
4. Deformation of the rock at constant stress (creep) was investigated, at
selected locations, by maintaining the fifth cycle load (1000 psi) for 2 or 3
days.
Results are graphed (Fig. 12) showing rock deflections for both the
center extensometer and the Carlson joint meter. A rosette of modulus
values is plotted for reference.
The modulus of elasticity is calculated using the following equation:

E =
w. ^ L(«i' + Z^yi^ {ai + Z2)i'2j
+ ^^—^ q \jai + Zr' - W + ZY'^] (23)

where:
Wz = normal deflection,
q = surface stress,
Oi, Qi — radii of load applied by Freyssi flat jack,
Z = depth below surface, and
V = Poisson's ratio.
Results of two or three day creep tests are also plotted.
A comparison of test details used by the Bureau of Reclamation and
Shannon and Wilson (Table 2) shows a number of marked differences
in procedure. The magnitude of maximum load is different, for example,
and creep is measured in one case and not in the other. Tests must be
adapted to a local situation, including physical properties and structure
of rock. The interpretation of data for application needs further investi-
gation.

Chamber Tests
The method used in the chamber tests [42] consists essentially of seal-
ing off a portion of a tunnel, placing a membrane lining over the rock,
and measuring the rock deflection due to internal water pressure. It has
two advantages in that the pressure may be applied over larger areas and
the deflection can be measured in any radial direction. The zone of in-
fluence may be considered a cylinder, analysis is usually made on the
basis of elastic stress distribution, and some effects of anisotropy may be
investigated. It is, however, the most expensive of the adit type tests.
To avoid construction of two bulkheads, tests are conducted in the

o
q

headings or dead ends of adits in representative rock. Rectangular cross

sections were modified by only minor shaping which consisted of:
1. Removal of broken rock from tunnel floor to expose firm, in-place
rock.
2. Reduction by drilling and blasting of irregularities to less than 6 in.
over a 10 ft length.
3. Excavation for a key for the concrete bulkhead, minimum dimen-
sions depending upon adit dimensions (Fig. 13).
4. Application of a gypsum plaster to form a smooth surface.
A concrete bulkhead with water and air conduits is then formed in
place (Fig. 13), a butyl rubber membrane (%« in- thick) stapled and
vacuumed into place, and the whole sprayed with 15 coats of natural
latex rubber.
Instrumentation consisted of diametral gages to measure deformation
at the surface of the tunnel and 8 and 16 ft in the roof and floor rock.
Two sets of four Carlson gages were set in the surface and one each in
the roof and floor at distances of 14 and 22 ft from the bulkhead (Fig.
13). The embedded (deep) joint meters were grouted with epoxy resin
rather than nonshrink grout. Constant temperature is maintained by heat-
ing the water pumped into the chamber.
Test procedure is to load the chamber to a maximum pressure of 350
psi at staged values of 60, 120, 180, 240, 300, and 350 psi, with the load
cycled after each increase. When the membrane ruptures, this procedure
is modified.
Pressure versus deflection is plotted with a modulus rosette and the
value of E calculated from

^ = ^ ^1 + ") ^24)
r{Ur)
where:
E = modulus of elasticity,
Pi = internal pressure,
a = radius to rock face—assuming circular chamber,
r = radius to point where deflection is measured,
Ur = change in radius due to pressure, and
r = Poisson's ratio.

Tunnel Deformation—Austrian Method

In a paper [43] presented at the Seventh Congress on Large Dams,
1961, a method was described similar in effect to the pressure chamber
technique, utiHzing "Freyssinet" flat jacks adjacent to each other on the
circumference and over a 2 m length of tunnel. The mechanical pressure
section can be placed anywhere along the tunnel, rather than utilizing a
dead end of a tunnel as in the American method.

For a thin steel tunnel liner along the tangential force z per unit length
is z = p^, where p^ is the internal hydraulic pressure and r is liner radius.
The cost of a steel lining to resist operating pressures is usually prohibi-
tive. Where rock supports the steel, according to formulas by Foppl

P, = i—T (25)

where:
u = radial elastic displacement,
Er = Young's modulus for rock,
rrir = (inverse) Poisson's ratio for rock = I/vr,
Pr = pressure resistance by rock, and
Ps = pressure resistance by steel.
For steel the displacement is given by
M _ cr
(26)
r Esnis^
m,2 - 1
where:
Es = Young's modulus for steel,
Ms = (inverse) Poisson's ratio for steel = l/v^, and
a = stress in the steel.
It follows that

P, = 'ni+l (27)
m/ — 1
which gives the relationship between the elastic properties of steel and
rock, and the permissible stress and pressure in the liner. If the liner is
thick the permissible stress must be reduced by the ratio

7i -2 (28)
{b + a) +

where:
a = outside radius of liner and
b = inside radius of liner.
This correction is necessary because the stress is higher at the inner radius
of a thick liner.
When the rock is not fully elastic, the modulus of elasticity is replaced
by the deformation modulus Vr, assuming the behavior of the concrete
backfill is the same as that of the rock. From Eq 27 it can be seen that
the maximum supporting action of the rock, for given values of Poisson's

0 A B CD E

Dl Al "01

1—True loading showing radial deformation f/i.

2, 3—Fictive loading representing deformations [/. and Us.
L—Length of loading.
u—Radial deformations caused by the jack.
u„—Radial deformations for infinite length of loading.
FIG. 14—Effect of short length of loading of radial jack analyzed by super-
position of radial deformations measured in Sections O, A, and D.

666666 .i.
0 A B CDE
(a) Theoretical distribution for an ideal-elastic and noncrack rock.
{b) Average values of measurements carried out.
FIG. 15—Comparison of theoretical and measured longitudinal distribution of
radial deformations.

•?

.^
C4.

i!
« (w

5-S

ill! I!

<i It.

«l 3

si
y3

-Hi
10 20 30°C

4000 KG/CM

FIG. 17—Graphical calculation for sleet linings.

ratio for rock and steel, depends only upon the permissible stress of the
steel and the deformation modulus of the rock.
A test length of tunnel, 2.5 m in length is excavated and lined with
15 cm of grout over 2 m of its length. The load is applied by 16 flat
jacks, 1.90 m long and 0.38 m wide, placed around the circumference of
the tunnel, with their length parallel to the length of the tunnel. The
jacks rest on hardwood planks and are supported on rings of high tem-
pered steel. The jacks are capable of developing pressures up to 100
atmos within their limits of expansion, and the over-all testing device is
thus capable of developing a total of 10,000 tons against firm rock.
The reference axis is a central steel pipe fixed to supports well out-
side of the test length. Displacement of anchor bolts placed in the rock
at a uniform radial distance of 1.40 m is measured by means of spring
loaded wires connected to dial gages.
The effect of grouting the rock internally is determined by two ad-
jacent test sections, one of which is ungrouted and one grouted. It has
been found that in a cracked calcarious schist the deformation modulus
was increased by grouting from 60,000 to 120,000 kg/cm^ for loads
from 0 to 50 kg/cm^. Measurements are taken outside of the loaded area,
as well as within, and measured results compared with theoretical results
(Figs. 14 and 15). Data are plotted (Fig. 16) on radial coordinates, the
data for phyllite showing divergence from symmetry due to foliation s.
Young's modulus may be determined from Eq 25 using either the
maximum displacement or the average. However, an effective modulus
of an ideally elastic rock is used which would give the same deformation
as the fractured rock around the tunnel. That is, it is assumed that the
true condition can be approximated by a fractured zone of depth c meas-
ured from the center line of the tunnel. Then the effective modulus is
given by

E.^Pff^nL^A + in^) (29)

u \m, a/
One of the principal difficulties in the use of Eq 29 is that the value
of c, the depth of fracture, is not known, and seismic or ultrasonic meth-
ods are proposed as a future (as of 1962) possibility. Seismic methods
of determining the "destressed zone" have been developed by the U. S.
Geological Survey (see Destressed Zone).
Creep effects are determined from repeated load effects, and the de-
formation caused by increasing the pressure in a long time test.
Design calculations are made with empirical curves (Fig. 17). Pressure
intensities are plotted against relative displacements (upper curves) for
measured displacements for ungrouted and grouted rock (limestone). The
temperature effect is calculated from

- = 1.5 X A r X 10"' (30)

Copyright by ASTM Int'l

Downloaded/printed by
Pontificia Universidad Catolica de
CLARK ON DEFORMATION MODULI OF ROCKS 11)7

Where AT is the expected temperature difference and whose value is

taken as 20 C. These values are used for choosing the temperature scale
and the construction of curves la and b which represent the deformation
of the steel plate including the effect of temperature drop. The dotted
lines then give the deformation for various pressures for constant values
of deformation modulus, V^. The lines in the lower part of the chart give
the values of deflection for various relative thicknesses of steel plate t/a
corresponding to pressure Pa in the steel.
Destressed Zone
The U. S. Geological Survey (USGS)^ has utilized seismic methods to
determine the depth of the destressed zone and the effective elastic moduli
of the rock around tunnels. Objectives were: (1) to determine seismic
velocity layering, (2) to determine in situ (effective) elastic constants, (3)
to determine compressional wave velocities between selected points, and
(4) to compare in situ test results with laboratory measurements.
It is postulated that a low velocity zone may be formed by an increase
in porosity and secondary fracturing in naturally porous rocks, and en-
largement of existing fractures and formation of new ones in rocks of
low porosity. Other investigations [44 and 45] have shown that the
depth of destressed zone is quite variable, depending upon the homo-
geneity of the rock and the size of the tunnel.
Both inhole and linear array measurements have been made by the
uses.
Inhole Measurements—Compressional (uphole) velocities are meas-
ured by placing detectors at various depths in holes drilled in the roof,
floor, and ribs at right angles to the axis of the tunnel, detonating small
charges at the collar and measuring arrival times. Crosshole velocity
measurements are also made between holes parallel to the tunnel axis.
Signals from one low sensitivity accelerometer for zero time, and
three high sensitivity accelerometers mounted in a plastic flexible tube
held in place by an inflated rubber tube, were recorded on a 4-channel
oscilloscope. The low sensitivity accelerometer was mounted in a V'2-in.-
diameter hole near the drill hole collar, and a blasting cap, with a few
grains of explosive when needed, was detonated in another shallow 1/2-
in.-diameter hole near the drill hole.
Crosshole velocity measurements were made in holes 3 to 15 ft apart,
with two single accelerometers and a blasting cap at the same depth in
each of these holes.
Linear Array Measurements—Accelerometers were spaced along tun-
nel walls at intervals of 25 ft, with as many as 10 detectors in an array,
the purpose being to determine the depth of the destressed zone by the
refraction method, as well as the compressional and shear wave velocities
''J. H. Scott, U.S. Geological Survey, Denver, Colo., private communication.

for computation of elastic moduli. Accelerations and arrival times were

recorded on a high-speed recording oscillograph.
Long Range Measurements—Equipment for longer range measure-
ments up to a few hundred feet was identical with that for seismic (linear
array) measurements; for larger distances, 0.25 to 1.5 lb of dynamite were
employed for a wave source.

EXPLANATION
30
rr—I 1 1 1 Milium
ZONE OF LOW VELOCITY
f^ DRILL HOLE

o 2.0

2 0

FIG. 18—Results of uphole velocity measurements in extensively fractured

quartz monzonite.

Representative results of velocity results obtained by the uphole

method, indicate a destressed zone of 3 to 5 ft around the drift (Fig. 18).
Results of the two methods were correlated and compared. Zero thick-
ness of the destressed zone was found in some places where the rock was
extensively fractured. However, moderate to extensive fracturing is a
prerequisite for the development of a destressed zone. No destressed
zone was found where the rock was relatively free from fractures. The
thickness of the zone cannot be estimated from observed fractures on the
tunnel walls. Laboratory measurements of elastic properties "confirm

TABLE 3—Deformation moduli standardized tests.

Type of test; Young's modulus, static, compressive, tensile and torsion, laboratory
A. Items to be specified or standardized
1. Length of specimen
2. Diameter of specimen
3. Shape of specimen
Cross section
Vertical section
Symmetry, etc.
4. End conditions
Platen
Specimen
Lubricant
5. Measurement of load
6. Rate of loading
7. Number of cycles
8. Test to failure
9. Sensors
Nonelectronic—test length
Resistance strain gages
Type
Length
Number
Placement
10. Lateral extension—Poisson's ratio
11. Coordination of above with rock properties
Relative length of gage and grain size
Relative size of specimen and grain size
12. Instrumentation
Accuracy, etc.
B. Basic theory
1. Test
2. Instrumentation

TABLE 4—Deformation moduli standardized tests.

Type of test; Young's and shear moduli, sonic, pulse method, laboratory
Items to be specified or standardized
1. Length, size, and shape of specimen
2. Method of exciting
Longitudinal waves
Shear waves
3. Characteristics of pulse
Magnitude
Length
Shape
4. Method of measurement
5. Characteristics of driver and pickup
6. Attachment to specimen
7. Calculation of results
Corrections to readings
8. Instrumentation
Calibration, accuracy, etc.
Basic theory
1. Test
2. Instrumentation

TABLE 5—Deformation moduli standardized tests.

Type of test: Young's and shear moduli, in situ
A. Items to be specified or standardized
1. Configuration for tests
Surface placement of sensors
Borehole placement of sensors
2. Source of excitation
Explosive
Hammer
3. Instrumentation
Frequency response
Amplification
4. Interpretation of records
Arrival times
Shear wave
Compressive wave
5. Theory
Ideal rock
Fractured rock

TABLE 6—Deformation moduli standardized tests.

Type of test: Seismic determination of destressed zone and effective moduli
A. Items to be specified or standardized
1. Sensors
Sensitivity
Frequency response
2. Energy source
Explosive
Type
Amount
3. Instrumentation
Resolution
Characteristics
Location and length of cables
Transducer
Frequency
Power
4. Interpretation
5. Correlation laboratory experimentation

that field values are correct within limits of error associated with uncer-
tainties in determining the compressional and shear-wave velocities."

Recommendation for Standardization

Any successful attempts to standardize tests of physical properties of
rock must be based upon a usable definition of physical properties and
their application. Ultimately this will include size, shape, structure, and
enviroimient.
Hence, although the items which should be standardized vary with
the particular test, they include such details as: (1) shape of specimen
or test site, (2) instrumentation, (3) method of applying load or excitation.

(4) method of measuring response, and (5) interpretation of results and

use of background theory (Tables 3-6).
References
[/] L. Obert, S. L. Windes, and W. I. Duvall, "Standardized Tests for Deter-
mining the Physical Properties of Mine Rock," RI 3891, U.S. Bureau of Mines,
1946.
[2] "Physical Properties of Some Typical Foundation Rocks," Concrete Labora-
tory Report SP-39, U.S. Bureau of Reclamation, 1953.
[3] S. L. Windes, "Physical Properties of Mine Rocks," RI 4459, U.S. Bureau of
Mines, 1949.
[4] P. LeComte, "Methods for Measuring the Dynamic Properties of Rock,"
Rock Mechanics Symposium, Queen's University, Kingston, Ont., Dec. 6-7,
1963.
[5J A. Nadai, Theory of Flow and Fracture of Solids, McGraw-Hill Book Co.,
Inc., New York, 1956.
[6\ G. Balmer, "A General Analytic Solution for Mohr's Envelope," Proceedings,
Am. Soc. Testing Mats., Vol 52, 1952, pp. 1260-1271.
[7] D. W. Philips, "Tectonics of Mining," Slieffield Mining Magazine, 1948, p. 47.
[S] E. N. Grosvenor, "New Method for Determining the Tensile Strength of a
Rock," Transactions, Am. Institute Mining, Metallurgical, and Petroleum
Engrs., Vol 220, 1961, pp. 447-449.
[9] D. Bancroft, "The Velocity of Longitudinal Waves in Cylindrical Bars,"
Physical Review, Vol 59, 1941, pp. 588-593.
[10] G. Pickett, "Equations for Computing Elastic Constants from Flexural and
Torsional Frequencies of Vibration of Prisms and Cylinders," Proceedings,
Am. Soc. Testing Mats., Vol 45, 1945, pp. 846-865.
[//] L. A. Panek, E. E. Hornsey, and R. L. Lappi, "Determination of the Modulus
of Rigidity of Rock by Expanding a Cylindrical Pressure Cell in a Drill Hole,"
6th Symposium on Rock Mechanics, University of Missouri, Rolla, Mo., Octo-
ber, 1964.
[12] H. R. Nicholls, "In Situ Determination of the Dynamic Elastic Constants of
Rock," International Symposium on Mining Research, Missouri School of
Mines, Rolla, Mo., February, 1961.
[13] A. R. Gregory, "Shear Wave Velocity Measurements of Sedimentary Rock
Samples Under Compression," 5th Symposium on Rock Mechanics, 1963,
University of Minnesota, Minneapolis, Minn., Pergamon Press, New York,
1963.
[14] B. Gutenberg, "Energy Ratio of Reflected and Refracted Seismic Waves,"
Bulletin, Seismic Society Am., Vol 34, 1944, pp. 85-102.
[15] D. L. Arenberg, "Ultrasonic Solid Delay Lines," Journal Acoustic Society
Am., Vol 20, 1948, pp. 1-26.
[16] J. C. lamieson and H. Hoskins, "The Measurement of Shear Wave Velocities
in Solids Using Axially Polarized Ceramic Transducers," Geophysics, Vol 28,
February, 1963, pp. 87-90.
[17] A. T. Woeber, S. Katz, and T. J. Ahrens, "Elasticity of Selected Rocks and
Minerals," Geophysics, Vol 28, August, 1963, p. 658.
[18] D. S. Hughes and C. Maurette, "Dynamic Elastic Moduli of Iron, Aluminum
and Fuzed Quartz," Journal of Applied Physics, Vol 27, p. 1184.
[19] W. A. Zisman, "Comparison of the Statically and Seismologically Determined
Elastic Constants of Rocks," National Academy Science Proceedings, Vol 19,
pp. 680-686.
\20] J. M. Ide, "Comparison of Statically and Dynamically Determined Young's
Modulus of Rocks," National Academy Science Proceedings, Vol 22, pp. 8 1 -
92.
[21] "Effect of Cracks in Concrete upon Dynamic Measurements of Elastic Modu-
lus," Materials Laboratory Report No. C-383, U.S. Bureau of Reclamation,
1948.

[22] A. Dvorak, "Field Test of Rocks on Dam Sites," Proceedings, 4th Interna-
tional Conference on Soil Mechanics and Foundation Engineering, Vol 1, But-
terworth, London.
[23] J. S. Rinehart, "Propagation Velocity of Longitudinal Waves in Rocks, Ef-
fect of State of Stress, Stress Level of the Wave, Water Content, Porosity,
Temperature, Stratification and Texture," Research Foundation Report, Colo-
rado School of Mines, Golden, Colo., January, 1961.
[24] "Physical Properties of Some Typical Foundation Rocks," Concrete Labora-
tory Report No. SP-39, U.S. Bureau of Reclamation, 1953.
[25] J. M. Bruckshaw and P. C. Mahanta, "The Variation of the Elastic Constants
of Rock with Frequency," Petroleum, Vol 17, No. 1.
[26] F. Birch and D. Bancroft, "Elasticity and Internal Friction in a Long Column
of Granite," Seismic Society Am. Bulletin, Vol 28, 1938.
[27] R. B. Sutherland, "Some Dynamic and Static Properties of Rock," 5th Sym-
posium on Rock Mechanics, University of Minnesota, Minneapolis, Minn.,
May, 1962, Pergamon Press, New York, 1963.
[28] D. T. Griggs, "Deformation of Rocks under High Confining Pressures," Jour-
nal of Geology, Vol 44, July-August, 1936, pp. 541-547.
[29] K. Barron and N. A. Toews, "Deformation Around a Mine Shaft in Salt,"
Rock Mechanics Symposium, Queen's University, Kingston, Ont., Dec. 6-7,
1963.
[30] B. Gutenberg, Internal Constitution of the Earth, McGraw-Hill Book Co., Inc.,
New York, 1939.
[31] W. R. Judd and C. Huber, "Correlation of Rock Properties by Statistical
Methods," International Symposium on Mining Research, Missouri School of
Mines, RoUa, Mo., 1961.
[32] R. H. Merrill and J. R. Peterson, "Deformation of a Borehole in Rock,"
RI 5881, U.S. Bureau of Mines, 1961.
[33] T. A. Morgan and L. A. Panek, "A Method for Determining Stress in Rocks,"
RI 6312, U.S. Bureau of Mines, 1963.
[34] L. Obert, "In Situ Determination of Stress in Rock," Mining Engineering,
August, 1962.
[35] R. H. Merrill, J. V. Williamson, D. M. Ropchan, and G. H. Kruse, "Stress
Determinations by Flatjack and Borehole Deformation Methods," RI 6400,
U.S. Bureau of Mines, 1964.
[36] L. Obert, R. H. Merrill, and T. A. Morgan, "Borehole Deformation Gage for
Determining the Stress in Mine Rock," RI 5978, U.S. Bureau of Mines, 1962.
[37] D. B. Lombard, "The Hugoniot Equation of State for Rocks," UCRL 6311,
University of Cahfornia Radiation Laboratory, February, 1961.
[38] T. J. Ahrens, "Dynamic Properties of Rocks," Technical Summary Report,
SRI Project No. GSU-4816, August, 1964.
[39] A. J. Chabai, "Synthesis of Shock Hugoniots for Rock Materials," 5th Sym-
posium on Rock Mechanics, C. Fairhurst, editor, Pergamon Press, New York,
1963.
[•^0] "Morrow Point Dam and Powerplant Foundation Investigation," Bureau of
of Reclamation, October, 1965.
[41] M. P. Habib, "Determination of the Elastic Modulus of Rocks In Situ,"
Record of the Technical Institute of Construction and Public Works, France,
September, 1950.
[42] Shannon and Wilson, "Report on In-Situ Rock Tests Dworshak Dam Site,"
U.S. Corps of Engineers Report, December, 1964.
[43] H. Laufer and G. Sieber, "Rock Mechanics and the Seventh Congress on
Large Dams," Rome, 1961, The Engineer, August, 1962.
[44] D. Wantland, "Geophysical Measurements of Rock Properties in Situ,"
RM 3583, International Conference on State of Stress in Earth's Crust, Rand
Corp., 1963.
[45] B. Kujundzic and B. Colic, "Odredjivanje modula elasticnosti stene i dubine
rastresene zone u hidrotehnickim tunelima pomocu refrakcione seismicke
metode," Transactions No. 8, Hidrotehnicki Institut, 1957,

DISCUSSION

D. U. Deere^ and A. J. Hendron, Jr}—As Professor Clark has pointed

out, we are concerned both with the determination of the modulus of in-
tact specimens in the laboratory and of in situ masses in the field. Ac-
tually, in applied rock mechanics, the laboratory values will be of little
value unless they can be correlated with the field values, presumably
through the use of modifiers or constants which will be functions of the
amount and kinds of geologic structures and alteration. The field values
may be obtained from: (1) large-scale tests or (2) observation of full-
scale performance. In this latter category are included observations on
the rebound of the bottom of large excavations for spillways, highway
cuts, etc.; the settlement of the foundation rock below heavily loaded
areas (dams, storage piles, large buildings); and the measurements of the
radial displacements of the walls of large underground openings for
powerplants, military installations, and vehicular and diversion tunnels.
A complicating factor in the latter procedure—and to a certain extent in
all the procedures—is the necessity for knowing the initial state of stress
before the computation of the modulus can be made.
The large-scale load tests must be sufficiently large in area so as to
stress significantly a representative specimen of the in situ rock with its
inherent discontinuities. The test normally conducted in civil engineering
projects (for the design of concrete dam foundadons and abutments and
pressure conduits) are: (1) the pressure-chamber test and (2) the plate-
jacking test. In the pressure-chamber test a 15 to 20 ft length of tunnel
is bulkheaded off, lined with an impermeable lining, and pressurized to
values ranging from 100 to 500 psi or so. Changes in diameter, or radial
strains in the rock mass, are measured as a funcdon of pressure, and the
modulus is computed by elastic theory and the equations for thick-walled
cylinders, or by some modificadon thereof. Although the volume of rock
mass that is significanUy stressed is considerable and certainly sufficient,
the fact remains that the rock surrounding the tunnel on which the test
is run may be more severely cracked and destressed with more opened
joints than the deeper rock. Thus, the modulus obtained by using diam-
eter changes alone may give a lower modulus than the rock mass actually
possesses. Such a probability must be considered in the interpretation of
the test data and in applying the data to a specific rock engineering design.
The plate-jacking test is also normally run along the walls of the tunnel
where the rock is destressed; the modulus obtained may be too low. The

' Professor of civil engineering and geology, and assistant professor of civil en-
gineering, respectively, University of Illinois, Urbana, 111.

modulus may be low for a second reason. Severe shearing stresses will
exist in the rock beneath the edge of the bearing plate. Under cyclic load-
ing, sustained loading, and high intensities of stress there may be de-
formations caused by shearing strains which will lead to a computed
modulus lower than the rock mass may actually possess. The shearing
strains will only be important in closely jointed or altered rock and where
the test plate is small (and, unfortunately, the test is often run with plates
as small as 10 in. diameter).
The plate-jacking test suffers from another defect which is related to
the size effect. Where the joint spacing is large, say 3 to 5 ft, and the load-
ing plate is small, the rock mass that is significantly stressed in the test is
too small to incorporate all the effects of the joints. Thus, the modulus
computed from the deformation may be more nearly that of the intact
joint block than that of the whole rock mass—an erroneous value that
may be much too high. Most organizations used loading plates 2 to 3 ft in
diameter to minimize the size effect.
Professor Zienkiewicz of the University of Swansea, Wales, has pro-
posed tests on the floor of a rock excavation using a large concrete slab
several feet on a side which is loaded by cables anchored at depth in the
rock mass. This test has many favorable aspects, and we believe that it
will come into more common use in the future. The proposed test (the
writers would suggest slabs at least 15 ft^) does not have the drawbacks
of giving modulus values that are too low because of severe shearing
strains, or too high because of an inadequate volume of rock being in-
corporated in the tests—as the normal jack tests may have.
G. B. Clark {author)—The remarks by Dr. Deere and Dr. Hendron are
well taken. In situ moduli tests are in a relatively early stage of develop-
ment which will require a searching examination of procedures for estab-
lishment of standardized tests and methods which are reliable and ac-
ceptable for design purposes. Many problems related particularly to
geologic structure must be solved before the ultimate goal is reached.

The Role of Microstructure in the Physical

Properties of Rock

REFERENCE: "The Role of Microstructure in the Physical Properties of

Rock," Testing Techniques for Rock Mechanics, ASTM STP 402, Am. Soc.
Testing Mats., 1966, p. 175.
ABSTRACT: The paper describes an approach to understanding rock prop-
erties through study of their relation to the microstructure of the rock.
Petrofabric analyses of rock specimens, subjected to various physical prop-
erty tests, have shown a strong correlation among several properties and
some of the microstructural features. The equipment and techniques
being used in these rock physics studies are discussed.
As an example of this work, the results of an experimental study of
anisotropism in Salisbury granite are given. Oriented drill cores were cut
in three mutually perpendicular directions from large blocks of the rock.
Axial and diametric pulse velocity measurements were made on the cores
to determine the direction and magnitude of elastic anisotropism. The cores
were cut into disks and subjected to axial (point-load) and diametric
(line-load) tension tests to determine direction and magnitude of strength
anisotropism. The broken specimens were reconstructed, cut into thin sec-
tions, and examined petrographically. There was a strong correlation
among the direction of maximum and minimum sonic velocity, preferred
direction of failure, and a number of microstructural features, such as
defect structures and grain orientation.
KEY WORDS: microstructure, rock (material), rock mechanics, anisot-
ropy, granite

The relationships between microstructure and physical behavior have

been explored with considerable success for materials such as ceramics
and metals. As the field of rock mechanics and rock physics matures and
becomes more sophisticated, it becomes necessary to develop a more
fundamental understanding of the material with which we are concerned.
A number of persons including Patterson, Weiss, Birch, Brace, Donath,
Griggs, Handin, and many others have already contributed toward this
goal. This paper describes some of the work presently underway at the
Twin Cities Mining Research Center of the Bureau of Mines.
The primary objective of the present work is to identify and examine
1
Mining methods research engineer, Twin Cities Mining Research Center, Bureau
of Mines, U.S. Department of Interior, Minneapolis, Minn.
175
Copyright^ 1966 by ASTM International www.astm.org
Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
176 TESTING TECHNIQUES FOR ROCK MECHANICS

relationships between gross behavior and microstructure. The general ap-

proach can be compared to a case history type of study in which all avail-
able information is developed on the particular specimen under investiga-
tion. In the work to be described here, the principal tools are sonic velocity
measurements and point- and line-load strength measurements to estab-
lish the patterns of physical behavior, followed by petrographic analysis
of thin sections of the same specimens in an attempt to correlate micro-
structural elements with observed behavior. The tests were conducted on
three sets of oriented drill core, drilled in mutually perpendicular direc-
tions from a single block of granite to provide a three-dimensional frame
of reference. Sonic velocity measurements were made across the diameter
of the cores in a number of directions to identify sonic anisotropy; the
cores were cut into disks, and the directions of maximum and minimum
tensile strength were determined by an exploratory adaptation of the so-
called line-load or Brazilian indirect tension test; the preferred direction
of fracture under concentrated load was determined by a point-load test;
and thin sections of the reconstructed specimens were studied.

Experimental Procedure

Material
Salisbury granite from Rowan County, N. C , was chosen for this test
for three reasons: (1) mineralogically, it is an uncomplicated, moderately
fine-grained granite with well-defined euhedral crystals; (2) it appeared to
be structurally uniform with a slight amount of gneissic banding; and
(3) it was readily available as a flat, easy-to-core slab from a local tomb-
stone dealer. Previous experience had shown the desirability of working
with material whose average grain size is small compared to the size of the
specimen. Well-defined crystal structure was desired to avoid ambiguity in
petrographic analysis that might result from indistinct grain boundaries.
A reasonable degree of structural uniformity was desired to avoid compli-
cation of test results from extraneous factors such as pre-existing fractures
and prominent banding or layering.
The mineralogic composition of the granite is based on macrogrid
study of the surface of a number of specimens and standard microscopic
modal analyses of several thin sections. Although the granite block con-
tains the same mineral suite as noted by Councill,- the relative percentages
of mineral components are somewhat different. The principal difference
is a greater amount of sodic plagioclase; accordingly, it should be classified
as an adamellite; however, the common name of Salisbury granite will be
used for convenience. The modal composition is as follows:

-R. J. Councill, "The Commercial Granites of North Carolina," Bulletin No. 67,
North Carolina Department of Conservation and Development, Division of Mineial
Resources, 1954.

Quartz 28 per cent

Orthoclase 38 per cent
Oligoclase 31 per cent
Epidote <3 per cent
Magnetite <1 per cent

Feldspar occurs as elongate, pinkish-gray to light-gray grains interspersed

with light-gray elongate clumps of quartz grains; the resultant fabric is that
of a lenticular gneiss. Swarms of microscopic quartz grains are scattered

X D i r e c t i o n d i s k s lie i n t h e Y - Z plane.
Y Direction disks l i e i n t h e X - 2 plane.
Z Direction disks l i e i n t h e X - Y plane.
FIG. 1—Notation and core orientation in three-dimensional frame of reference.

along boundaries and pre-existing fractures of larger feldspar and quartz

grains. In general, the larger grains exhibit a shattered appearance en
masse and in thin section. This appearance is due to cleavage in the
feldspars and to myriads of microfractures in quartz grains oriented paral-
lel and normal to the direction of gneissosity. Epidote occurs as granular or
fibrous greenish-gray grains occupying microscopic fractures or as an alter-
ation product of plagioclase along cleavage planes.
Grain size of the rock ranges from 0.1 to 10 mm, whereas the average
size is 0.5 to 1 mm. This difference suggests a coarser texture than actually
exists because of prominent feldspar and quartz grains.

The slab, a typical rectangular headstone obtained from a commercial

source, was 4 by 12 by 24 in. in dimension. It was cut at the quarry in
such a manner that the smallest surfaces, that is, the 4 by 12-in. surfaces,
were parallel to the direction of visible gneissosity. The large surfaces,
that is, the top and bottom, were approximately parallel. The top surface
was polished, and the bottom surface was flat and smooth but not pol-
ished. The remaining surfaces were cleaved and rough textured.

Specimen Preparation

For the purpose of this investigation, the block was oriented in such a
manner that the direction normal to the top surface became the Z direc-
tion, the direction parallel to the longest edge became the Y direction,
and the direction parallel to the edge of intermediate length became the
X direction. The top and bottom surfaces are in the XY plane, the shortest
sides are in the XZ plane, and the longest sides are in the YZ plane. The
visible gneissosity mentioned previously is in the XZ plane.
The cores, 2 in. in diameter and from 4 to 6 in. in length, were drilled
from the slab in each of the three mutually perpendicular directions by a
thin-wall diamond coring bit mounted by means of a water swivel to a
large machinist-type radial arm drill. The massiveness and precision of
the machine permitted drilling accurately sized cores requiring a minimum
of dressing to obtain ±0.025-in. diametric tolerance. For convenience
and ease in handling, diametric pulse velocity measurements were made
on cores rather than disks. After the sonic measurements were completed,
cores were cut into disks having length-to-diameter ratios of approxi-
mately 0.5 to 1, that is, about 1 in. in length to 2 in. in diameter. The flat
surfaces were ground on a surface grinder and lapped to within 0.001 in.
of flatness, and 0.002 in. of parallelism. Density was determined by vol-
ume and weight calculations and by the loss-of-weight in water method.
The two methods agreed within 0.6 per cent. Cores and disks were air-
dried for at least 2 weeks prior to testing.
The orientation of the cores and disks was very carefully maintained
throughout the test. The cores and disks were marked and oriented as
shown in Fig. 1. A pair of parallel red and blue lines was drawn on each
core in indelible ink to provide an azimuth reference mark. The two-color
reference line was used to minimize the chance of turning the core or
disk end-for-end and making false readings. A total of 29 cores were pre-
pared from which 33 disks in the XY plane, 35 disks in the YZ plane, and
36 disks in the XZ plane were obtained for a total of 104 disks. Fifty of
these were assigned to the line-load test and 24 to the point-load test.
The remainder were held in reserve for other purposes. The disks were
each marked with a coded number which identified the core from which
it came, its position in the core, and the location of that core in the original
block.

Sonic Anisotropy
Diametric Sonic Velocity—Prior to cutting the cores into disks, pulse
velocity measurements were made at 15-deg intervals around the circum-
ference of the core at the position of each disk. The resulting pulse velocity
profile provided a measure of transverse sonic anisotropy of each disk. The
AVERAGE VELOCITY
o eoch direction
each plane
all planes

YZ Plane I 20

5.000

- 0.90

6.000

5.000

O
_l
UJ

> <
Q
UJ
6.000

4.000
90 180
PULSE DIRECTION, d e g r e e
FIG. 2—Planar and orthogonal sonic velocity anisotropy.

pulse velocity anisotropy for the block as a whole was determined by nor-
malization of the data from each of the orthogonal directions.
Normalization of Data—In order to provide convenient and meaning-
ful comparisons between pulse velocity data in each of the orthogonal
planes and between pulse velocity data and other measurements, all meas-
urements were reduced to dimensionless ratios by normalization.
All measurements of a particular parameter, sonic velocity, strength,
etc., regardless of direction or orientation, were summed and averaged
to give an average value of that parameter for the body as a whole. The
ratio of an individual measurement to the average value then indicates
the degree of anisotropy present in the individual measurement. Similarly,
other measurements, such as the average of all measurements in a plane.

AVERAGE STRENGTH
o—• each direction
- ——' each plane
all planes
- 1.40
YZ Plon«

160
1.10
MO XZ Plone
- 1.20
120
UJ
(t
I- 100
a>
- 0.80
UJ 80
< 60 <
Q: cn
LLI
>
<
160
1.40
140 - XY Plane
1.20
120 I. I 0 o
1.00
100 0 912
0.80
80

60
0 90 180
LOADING DIRECTION, degree
F I G . 3- -Planar and orthogonal tensile strength anisotropy (diametric loading).

can be related to the over-all average to show the deviation or anisotropy

of the plane. Planar and orthogonal sonic velocity anisotropy is shown
graphically in Fig. 2.
Tensile Strength Anisotropy
Line-Load Test—The directions and relative magnitudes of the maxi-
mum and minimum tensile strengths were determined by an exploratory
technique based upon the so-called diametric or line-load indirect tension
test.^' * In this test, a disk of the specimen to be tested is placed on edge

^E. Addinall and P. Hackett, 'Tensile Failure in Rock-Like Materials," Pro-

ceedings, 6th Symposium on Rock Mechanics, University of Missouri, Rolla, Mo.,
1964, pp. 515-538.
' A. Rudnick, A. R. Hunter, and F . C. Holden, "An Analysis of the Diametral-
Compression Test," Materials Research & Standards, Vol 3, No. 4 April, 1963, pp.
283-289.

between the platens of a testing machine, and a compressive load is ap-

plied across the diameter. Tensile strength is determined from the applied
load at failure by the formula

= ^
•wdt
where:
a = tensile stress normal to loaded diameter,
P = applied load at failure,
d = diameter of disk, and
t = thickness of disk.
Specimen failure nearly always occurs along the diameter through
which the load is applied; thus, the apparent breaking strength of an

I .rj ,
' •»<.'lrtf''i.

-..•#1 "J . y - V,l

FIG. 4—Typical failure patterns for line-load test.

anisotropic rock is strongly dependent upon the orientation of the direction

of loading with respect to pre-existing "strong" or "weak" directions.
Conversely, the directions and magnitudes of maximum and minimum
tensile strengths of a particular rock can be determined by applying the
test to a number of oriented disks, representing a particular plane and
varying the direction of loading. Variations in breaking strength in dif-
ferent directions disclose the existence of tensile strength anisotropy and,
when plotted, give the orientation and magnitude of both minimum and
maximum tensile strength for the plane being examined. A three-dimen-
sional representation of tensile strength can be obtained by applying the
procedure to specimens having three mutually orthgonal orientations.
In the present test, average tensile strength was determined at each 30-
deg interval of azimuth for each of the three orientations. Two or four
disks were tested for each azimuth direction in each plane; four at the 0
and 90-deg positions, since these directions coincide with matching direc-
tions in corresponding orientations, and two each at the intermediate

F I G . 5—Jig for point-load test.

F I G . 6—Typical failure patterns for point-load test.

points at 30 and 60 deg. Planar and orthogonal tensile strength anisotropy

are shown graphically in Fig. 3.
The mode of failure of the line-load test is shown in Fig. 4. The direc-
tion of fractures indicates the position of the testing machine platens with
respect to the azimuth. The position of the dark line indicates the orien-
tation of the specimen. Note the complex fracture patterns in several disks.
Addinal and Hacket regard these as secondary fractures caused by
"platen bounce," that is, at the instant of initial fracture a momentary
acceleration of the platen occurs due to the sudden release of load that
causes the newly broken halves to be reloaded. This opinion is strength-
ened to some extent by several force-deformation curves plotted during

these tests. These curves show a momentary release of load, slight reload-
ing, and then complete failure.
Point-Load Test—The point-load test" is somewhat similar to the line-
load test in that a compressive stress is applied to a specimen to produce
an indirect tensile failure. In this test, a load is applied through the central

Scale, inches

FIG. 7—Point-load failure pattern in a reassembled group of oriented disks.

axis of a disk by means of a pair of opposing hemispherical indentors. A

specially designed jig (Fig. 5) holds the test specimen and indentors in
proper position while a compressive load is applied. Because of the geo-
metrical symmetry of the system, a stress field is developed that is sym-

" D. R. Reichmuth, "Correlation of Force-Displacement Data with Physical

Properties of Rocks for Percussive Drilling Systems," M.S. thesis, University of
Minnesota, Minneapolis, Minn, June, 1962, pp. 52-56.

X-direction Y-direction Z-direction

(Y2 plane) (YZ plane) (XY-plane)
(J 30
UJ 20
a:
3
o ; k. .M^ J X J
D i r e c t i o n of m a x i m u m and m i n i m u m sonic v e l o c i t y .

>-
o Ih
z
UJ 10
3
o 5
UJ
IE
U- U _^
P r e f e r r e d d i r e c t i o n of f r a c t u r e - p o i n t - l o a d test.

130

0° 60° 120° 180° 0° 60° 120° 180° 60° 120° 180°

D i r e c t i o n a n d m a g n i t u d e of t e n s i l e s t r e n g t t i l i n e - l o a d test.

I I IVIaximum soni c v e l o c i t y
^ ^ M i n i m u m sonic velocity

F I G . 8—Comparison of sonic velocity, preferred direction of failure, and tensile

strength.

TABLE 1--Point-load strength

N u m b e r of Average Strength, Standard
Plane Specimens kg/cm2 Deviation

YZ 8 68.55 6.33
xz 8 66.38 5.53
XY 6 67.43 2.40

metrical about the loading axis. Assuming a homogeneous isotropic ma-

terial, the direction of failure would be expected to be completely random.
However, if a weak direction exists, a tendency for failure to occur in this
direction would be expected. The photographs of Fig. 6 show that an
anisotropic material, such as Salisbury granite, generally fails by a single
fracture across the diameter of the specimen. Furthermore, comparable
specimens, that is, representing the same plane, fracture in the same gen-
eral direction (Fig. 7). The fracture surface is perpendicular to the direc-
tion of least strength; thus, in Fig. 8, the preferred direction of fracture is
90 deg from the preferred direction of failure. This "preferred direction
of failure" coincides in general with the direction of minimum tensile

Strength as determined by the line-load test. Also, the preferred direction

of failure concides with the direction of minimum sonic velocity. Inasmuch
as sonic velocity is related to Young's modulus, it would appear that the
direction of least strength is also the direction of lowest Young's modulus,
which is consistent with the usual experience.
Table 1 shows the average point-load breaking strength for each plane
as determined by Reichmuth's formula
p
S = K-
dt
where:
5" = point-load breaking strength,
P = applied load,
d = diameter of specimen,
t = thickness of specimen, and
^ = a shape factor = 0.70/(t/d).

Petrographic Analysis
Specimen Preparation—Failed specimens from the point-load tests
were reconstructed using an epoxy resin adhesive, and circular thin sec-
tions were prepared through the midsection. In this manner the actual
fracture path and adjacent rock fabric are revealed for study under the
petrographic microscope.
Only point-loaded specimens were used for petrographic examination,
since these generally failed by a single clean fracture and could be recon-
structed without difficulty. The thin sections were standard in all respects
except size and thickness. Since a full cross section was required from the
specimen, the thin sections were about 2 in. in diameter and mounted on
21/2 by 25/2-in. oversize glass slides without a cover glass. The sections
were made approximately 2 to 3 times standard thickness, to avoid the
risk of causing unnecessary damage to the microstructure.

Deject Structure Analysis

Theoretical Considerations—Defect structure analysis consists of the
identification, tabulation, and statistical analysis of a variety of actual or
potential sources of weakness in a rock fabric in an attempt to relate their
frequency of occurrence, orientation, etc. to observed physical behavior.
Defect structures range in size from microscopic to macroscopic. How-
ever, for the present purpose, only structures of the microscopic or near
microscopic scale are considered. These include grain boundaries, cleav-
age planes, twin planes, inclusion trains, clustering or banding of assem-
blages of minerals, elongate grains, pre-existing microcracks, preferred
orientation of one or more constitutents, etc. The philosophical foundation
for defect structure analysis can be stated very simply; given a uniform

YZ p l a n e X Z plane XY p l a n e

400 r —
O
O
«.
— *-
^ D
M —

^ 5 5^ » _2 — o
llJ
300
o > o — o I
—2 — > *- 1
^ ^^~ —
2 00
a:
Q:
' •

()
O 100 •

o
Ll. , , , ,
0 60 120 180 0 60 120 180 60 120 180

Orientation DEFECT STRUCTURES

z>
o
UJ

60 120 ISO 0 60 120 180 0 60 120 180

P R E F E R R E D D I R E C T I O N OF F R A C T U R E , point l o a d test
F I G . 9—Comparison of frequency of occurrence of defect structures and pre-
ferred direction of failure.

F I G . ] 0—Normalized defect frequency data showing planar anisotropy.

stress field superimposed on a heterogeneous material such as rock, well-

defined defect structures may serve as stress risers to: (a) limit the ultimate
strength of the material in the manner of Griffith flaws and (b) perhaps
act as surfaces of weakness to dictate the direction of failure.
Experimental Procedure—Nearly 4000 defect measurements were
made by petrographic microscope on six thin sections representing the
three orthogonal planes. A large number of traverses were made across
the diameter of each thin section in directions corresponding to the direc-
tions of sonic velocity measurements made previously. In addition to
identifying the mineral composition and cataloging the various defect
structures, the orientation of each defect structure encountered in travers-
ing was recorded in azimuth degrees clockwise from the red-blue reference
line. The defect measurements were subsequently blocked into 30-deg
sectors according to orientation. For example, all defects oriented between
0 and 30 deg were assigned to one block, and those between 30 and 60
deg were assigned to another block. Because of point symmetry, a total
of six sectors, from 0 to 180 deg, are sufficient to accommodate the com-
plete azimuth.
A histogram showing frequency of occurrence of defects in these 30-deg
blocks is shown in Fig. 9. Note that the defects arrange themselves into
the two groups labeled visible foliation and obscure foliation. The first
group includes the readily visible features, such as elongate grains of
feldspars, prominent cleavage planes, mineral aggregates, especially quartz
grains, and larger microcracks, all of which collectively contribute to the
visible gneissic banding. The second group includes the more obscure
features, such as poorly defined microcracks, less well-defined grain
elongation, secondary cleavages, and fluid inclusions in the quartz grains.
It is interesting to note that the histograms representing the frequency of
occurrence of point-load fractures in each of the 30-deg sectors strongly
favor the visible foliation in the YZ and XZ planes but tend to favor the
obscure foliation in the XY plane.
Normalized polar plots of frequency of occurrence of defect structures
are given in Fig. 10. The cross-shaped pattern results from the separate
maxima corresponding to the visible and obscure foliation of Fig. 9.
Microfracture Analysis—Because of the presence of conspicuous mi-
crofractures in some of the quartz grains, a side study was conducted to
see whether it could be determined whether the density of microfractures
increased as a result of the test work described herein or whether they
were pre-existing and perhaps contributed to the failure process. Six ad-
ditional thin sections, representative of the rock fabric in each of the
orthogonal directions, were prepared by our usual procedures. In order
to make "before-and-after" comparisons, one thin section in each direc-
tion was prepared from a reconstructed test specimen, and the other thin

section was prepared from an oriented but untested specimen. The sec-
tions were examined in essentially the same manner as previously de-
scribed except that, in this case, only quartz microfractures greater than
2 mm in length were tabulated.
With reference to Fig. 9, the general orientation of microfractures in
the YZ and XZ planes coincide with the orientation of visible foliation
and with the preferred direction of fracture. In the XY plane, the quartz
microfractures are almost perpendicular to the plane of prominent folia-

F I G . 11—Fracture plane model in X-Y-Z coordinate system.

tion, indicating that the microfractures developed independently and later

than the foliation. Furthermore, the preferred direction of fracture in this
plane tends to coincide with the orientation of these quartz microcracks,
even though the plane of foliation would seem to be a favored direction
because of high defect frequency. In general, the population density of
microcracks indicates they are pre-existing; however, there seems to be
an increase in the number of microcracks in the vicinity of the fracture.
This would not be unexpected inasmuch as the propagating fracture could
easily induce secondary fractures by the sudden release of strain energy by
elastic rebound.
Summary and Discussion
We have shown strong correlation between sonic anisotropy, maxi-

mum and minimum tensile strength, preferred direction of failure, and the
orientation of certain features of the microstructure. These correlations
have been extended into a three-dimensional frame of reference through
the use of oriented specimens obtained on mutually perpendicular axes
from a common source. Figure 11 shows the major fracture planes as de-
termined by the point-load test. The fracture plane associated with the X-
direction disks is approximately coincident with a prominent plane of
foliation. The fracture planes associated with the Y- and Z-direction
disks are nearly coincident with each other and with the quartz micro-
fractures mentioned previously, but perpendicular to the plane of promi-
nent foliation. The microfractures themselves are a feature of the sec-
ondary foliation which is visible in the XZ plane. Thus, it appears that
the fracture pattern in the X direction is associated with the defect struc-
ture of the prominent foliation, and the fractures in the Y and Z directions
are related to the secondary foliation and the associated quartz micro-
cracks.

Influence of Rock Anisotropy and Time-

Dependent Deformation on the Stress-
Relief and High-Modulus Inclusion
Techniques of In Situ Stress Determination

REFERENCE: Influence of Rock Anisotropy and Time-Dependent De-

formation on the Stress-Relief and High-Modulus Inclusion Techniques
of In Situ Stress Determination," Testing Techniques for Rock Mechan-
ics, ASTM STP 402, Am. Soc. Testing Mats., 1966, p. 190.
ABSTRACT: Two of the most popular techniques used in the determina-
tion of in situ rock stresses are the stress-relief and high-modulus inclusion
methods. Deformations and strains observed in the measuring instruments
are usually interpreted on the assumption that the rock behaves as a
homogeneous, isotropic, linearly elastic medium, a situation which is not
even approximately correct in some practical cases.
This paper presents a theoretical analysis of the influence of: (a) rock
anisotropy, of the transversely isotropic kind, on the accuracy of values of
in situ stresses determined from the usual isotropic analysis and (b) time-
dependent rock deformation, represented as linear, isotropic, viscoelastic
creep, on the stresses developed in high-modulus plugs. It is shown that sig-
nificant errors (over 50 per cent for one stress-relief example chosen) may
be introduced by neglecting these factors.
KEY WORDS: rock (material), rock mechanics, anisotropy, stress reliev-
ing, creep, inclusions, in situ stress measurement, viscoelasticity

Considerable effort has been devoted in rock mechanics research to the

development of techniques for the accurate determination of in situ rock
stresses [I].3 Two of the most popular are the stress-relief method and the
high-modulus inclusion method. Both of these involve the measurement
of strain or displacement which results from a change in the state of stress
in the rock. The observed values are related to the stress in the rock on the

1
Research fellow, Department of Mining Engineering, University of Nottingham,
England; formerly visiting professor, School of Mineral and Metallurgical Engineer-
ing, University of Minnesota, Minneapolis, Minn.
2
Professor and associate head, School of Mineral and Metallurgical Engineering,
University of Minnesota, Minneapolis, Minn. Personal member ASTM.
3
The italic numbers in brackets refer to the list of references appended to this
paper.
190

Copyright" 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
BERRY AND FAIRHURST ON IN SITU STRESS DETERMINATION 191

assumption that the rock, at least in the vicinity of the measurement, is a

homogeneous isotropic linearly elastic material for which the two elastic
constants, such as the modulus of elasticity and Poisson's ratio, are known.
Frequently, the additional assumption is made that the axis of the borehole
in which the deformation sensing element is located is a principal stress
direction.
Although these assumptions may be reasonably valid in some cases they
are often invoked under circumstances where they do not hold, resulting in
significantly erroneous conclusions. The deformations which occur when
the principal stresses are not parallel and normal to the axis of the borehole
have previously [2] been computed for stress relief of a hole in isotropic
rock. This paper draws attention to the errors which can arise for two
further practical situations. Consideration is first given to the problem of
interpreting the results of stress-relief measurements in anisotropically
elastic rock. Attention is then turned to the effect of viscous creep on the
stresses developed in a cylindrical inclusion in an isotropic rock. It will
be assumed in both cases that the hole is drilled parallel to a principal stress
direction.

Stress-Relief Technique in a Transversely Isotropic Elastic Medium

Many rocks, particularly those of sedimentary origin, are known to be
markedly anisotropic. Frequently, this anisotropy is similar to that ex-
hibited by the so-called transversely isotropic medium. A transversely
isotropic medium may be defined as one possessing elastic properties
which are invariant with respect to rotations about an axis but whose
properties along the axis differ from those in the plane of isotropy. Such
a medium may be shown [3] to have five independent elastic constants,
compared to two for the isotropic medium. The five constants are illus-
trated in Fig. 1. Co-ordinate axes Oxi, Ox^, and Ox^ are chosen so that
0x2 is the axis of elastic symmetry.
£ i and E2 are the moduli of elasticity (for tension-compression) with
respect to directions lying in the plane of isotropy and perpendicular to it,
respectively; vi is the Poisson's ratio which characterises the transverse
reduction in the plane of isotropy for tension in the same plane, that is,
ei/es ; V2 is the Poisson's ratio which characterises the transverse reduction
in the plane of isotropy for the tension in a direction normal to it, that is,
€2/63 ; and M is the shear modulus for planes normal to the plane of iso-
tropy. (Note that M', the shear modulus for planes parallel to the plane
of isotropy, is given by M' = £1/2(1 -I- vd, that is, it is not independent.)
Shale, for example, appears to be fairly well described by the trans-
versely isotropic model with the axis of symmetry chosen normal to the
lamination or bedding planes.
A detailed analysis of the stress-relief technique in a homogeneous,
transversely isotropic medium is presented in Appendix I for the two

' <3

r--' u-,= —

2MX23
FIG. 1—Definition of independent elastic constants Ei, E^, vi, v^, and M for
a transversely isotropic medium.

cases: (1) where the axis of the hole is normal to the axis of elastic sym-
metry and (2) where the axis of the hole is parallel to the axis of elastic
symmetry. In the general case we should assume the principal stresses
to be arbitrarily inclined to the Oxy, Oxz, and Ox^ axes, but in the present
analysis we shall restrict our attention to situations in which the axis of
the hole coincides with a principal stress direction.

Case 1, Hole Drilled Parallel to the X^ Direction

The radial deformation M,'', which occurs due to stress relief by over-
coring of the (instrumented) hole, is given by the expression
h
— = /I -h 5 COS 26> + C sin 29 (1)
a
where:
a = original radius of the hole,
6 = angle measured counterclockwise from the positive direction of
axis Xi and the point on the wall of the hole at which w/' is de-
veloped, and
A, B, and C = functions of the principal stresses and the elastic param-
eters of the medium, as given by Eq 17 in Appendix I
Where the hole is instrumented to detect axial strain «» along the hole
generated by stress relief (overcoring), Eq 18 for ^4, B, and C may be
used to solve for Ni and N2 directly without N:i. Ni and A'2 are the prin-
cipal stresses in the plane containing Xi and X2 , and N^ is the principal
stress taken to act in the Xs direction.

Material A, state of stress Material A, state of stress 2

X
Anisotropic case
Isotropic case

Material B, state of stress I Material B, state of stress 2

F I G . 2—Comparison of radial deformations due to stress relief in isotropic and

transversely isotropic rock. Case 1, hole drilled parallel to the X^ direction.

Example
In order to illustrate the errors introduced by neglecting anisotropy
Eq 17 has been evaluated for two hypothetical stress conditions in two
transversely anisotropic materials as follows:
Material A Material B
Moderately Anisotropic Markedly Anisotropic
E = 4.0 X 10* psi E = 4.5 X 10« psi
£2 = 3.0 X 10« psi Et = 2.0 X 10« psi
^1 = 0.25 VI = 0.2
vi = 0.20 V, = 0.225
M = \.1S X 10' psi M = 0,71 X 10« psi
State of stress \: Ni = N-, = 500 psi; M> = 1000 psi
State of stress 1: Ni = N3 = 2000 psi; Nt = 1000 psi

The radial deformation w,'' as calculated from Eq 16 is shown graphically

for each case in Fig. 2. Curves are drawn for a = 0 and 90 deg only (see
Eq 17) for the sake of clarity.
Since it is frequently assumed that the rock is isotropic and that radial
deformation results from the relief of stresses normal to the axis of the
hole only, that is, A'l No, neglecting the effect of A^3, the radial deforma-
tion for this case is shown dotted for comparison. The isotropic curves
are calculated from the plane-stress expression [4]:

'^ = i [.Vi + A'2 + 20Vi - A'2) cos 2(0 - a)] (2)

a ho
in which £0 is taken as the mean of the anisotropic values.

£„ =^^L+^ (3)
It is seen that the differences are appreciable.
TABLE 1—Comparison of actual values of principal stresses Ni and N2 with those
computed from measured values ofur'' assuming the medium to be isotropic and
neglecting N3 .
Actual, psi Computed, psi
Material Sta_te^»'
.Vi N2 Ni Ni

A 1 0 1000 500 1008 421

A 1 90 500 1000 463 821
A 2 0 1000 2000 835 1535
A 2 90 2000 1000 1927 726
B 1 0 1000 500 1612 643
B 1 90 500 1000 765 985
B 2 0 1000 2000 1325 1919
B 2 90 2000 1000 3141 1330

If we consider that the deformations u,'' given by Eq 15 are actually

measured but incorrectly attributed to the relief of stresses Ni and A'^2,
in an isotropic medium with a modulus of elasticity £0 as given by Eq 3,
then the error in the values thus calculated for A^i and A'2 may also be
large, as is shown in Table 1.
It is realised that the discrepancies shown in Table 1 are due in part to
the neglect of the axial stress Ng in the plane stress expression (Eq 2), but
a direct comparison using the proper isotropic stress-relief expression,
that is, the plane-strain expression including the influence of N3 ;

~ = W [^^'1 + ^^'2 - "A'a + 2(1 - ;'')(A'i - A'2) cos 2(0 -a)].. (4)
a -&0

which may be obtained from Eq 17 by putting £1 = £2 ; vi = V2 ", etc., is

not practicable unless some assumption about ^3 is made.
Case 2, Hole Drilled Parallel to the Xo Direction
The radial deformation M," which occurs due to stress relief by over-
coring is given by the expression

?^ = £ + F c o s 2 e (5)
a
where a and B are as previously defined and E and F are functions of the
principal stresses and the elastic parameters of the medium as defined by
Eq 24. If the axial strain £3 is known, Eq 25 may be used to solve for Ni
and A'2 directly without N^. The deformation u," predicted by Eq 23 for
the case where a = 90 deg is shown graphically in Fig. 3.^ The isotropic
curve is calculated, as in Fig. 2, on the assumption that the effective iso-
tropic modulus may be determined from Eq 3. Since the hole is drilled
parallel to the axis of elastic symmetry, however, it seems more reason-
able to substitute Ei for Eo in the isotropic equation. This change effec-
tively decreases the isotropic deformation in the ratio Eo/Ei, or 87.5
per cent for Material A and 72.2 per cent for Material B, which consid-
erably reduces the difference between the isotropic and anisotropic curves.
It seems reasonable to suggest, therefore, that the radial deformations
measured in holes drilled parallel to the axis of symmetry will, in most
cases, not differ appreciably from the values predicted by the isotropic
solution, Eq 2, taking E equal to £1 .
Where it is necessary to drill parallel to the Xi axis for example, hori-
zontally in flat lying bedded deposits, then the isotropic approximation
can lead to highly erroneous values for the in situ stresses.
Effect of Viscous Creep of the Rock on the Stresses Developed in an
Elastic Plug
The use of elastic plugs or inclusions for the determination of stresses
in rock is confined mainly to the observation of long-term stress changes
which occur subsequent to installation of the plug. The pre-existing in situ
state of stress is generally unknown.
The plug, which is usually cylindrical, is inserted into a borehole and
set at the desired location, which must be at least several diameters away
from the end of the hole to avoid the influence of unknown stress effects.
Cement is forced between the plug and the sides of the hole to ensure
intimate contact. The principal stresses Pi and P^ developed in an isotropic
elastic plug due to the application of principal stresses A'l and A^2 to the
surrounding isotropic elastic medium in the plane normal to the axis of
the hole (see Fig. 4) are given (see Appendix II) by the expression
Pi
„ . = a(A\ + N.) ± KA'i - ^'2) (6)

where a and b are both functions of the elastic constants of the inclusion
and surrounding medium, as defined by Eq 29 in Appendix II.

' The case of a = 0° is not shown since it is simply a circle radius u" where
«r" = {a/Ei)[2N:L — vaAfz]; obtained by putting a = 0 and M = W3 in Eqs 41 and 42.
This is identical to the isotropic case taking £1 for E,^ and including the deformation
due to the stress iVs parallel to the drill hole.

It is of interest to note that the principal stress directions in the plug

coincide with those in the surrounding medium. Both plug and surround
are subjected to homogeneous states of stress.
It may be shown [5] that when the modulus of elasticity of the plug is
approximately five times larger than the corresponding modulus of the

Material A, state of stress I Material B, state of stress

Material A, state of stress 2 Material B, state of stress 2

FIG. 3—Comparison of radial deformations due to stress relief in isotropic and

transversely isotropic rock. Case 2, hole drilled parallel to the X2 direction (stress Ni
inclined at 90 deg to Xi direction, that is, a = 90 deg).

surrounding medium then a and b are virtually constant. Use of such

high-modulus plugs thus eliminates the need for accurate determination
of the elastic modulus of the rock. The modulus of the plug is limited
practically, however, by the desirability of high sensitivity, that is, a
given change of stress should produce a measurable strain in the plug.
Plugs are thus particularly favored for use in rocks having a fairly low-
elastic modulus. Such rocks also frequently exhibit time-deformation be-
havior, particularly at the elevated temperatures of rock at depth.

When the possibility of creep is coupled with the frequent use of

plugs to monitor stress changes over periods of months or years, it
seems pertinent to examine the effect of rock creep on the state of stress
in the plug. We will consider the practical situation to be that in which
an isotropic elastic plug is placed so that it just fits into a cylindrical
hole drilled into a stressed viscoelastic medium. The elastic deformation
of the hole occurs coincidentally with the drilling of the hole and so
produces no stress in the plug, which is inserted later. Assuming that the
state of stress in the medium remains constant, any stress in the plug is the
result of subsequent viscous-delayed elastic deformation of the rock.
This situation may be considered as a specialization of the more general
case in which the plug is inserted into an initially unstressed medium.
The medium is then subjected to a constant load. Stress develops in the
plug due to both elastic and time-dependent deformation. We assume
the practical situation to be similar to the case where the instantaneous
elastic modulus is infinitely high.

FIG. 4—Symbols used in analysis of cylindrical elastic inclusion in isotropic

elastic medium.

Detailed analysis of the problem is given in Appendix II. This reveals

that the principal stresses P(r) in the plug at any time subsequent to initial
loading are
P(t) = A(t) (Ni + No) + B(t) (Ni- N-d (7)
where Ait) and Bit) are time-dependent parameters, defined in Eqs 35
and 37, related to the creep properties of the rock medium. The rate at
which the stress develops in the plug depends on the time-dependent
behavior of the uniaxial creep function S^it) and the shear creep function
Sneit), which must be determined experimentally under the appropriate
environmental conditions. From Eq 41 it is seen that the ultimate stress
in the elastic plug is independent of the elastic constants of the plug
if the viscoelastic medium is such that it can continue to creep viscously
both in uniaxial compression (or tension) and in shear. The only con-
stants required are the coefficients which describe the viscous creep be-
havior of the rock, that is, bi and be,.
However, if the viscoelastic medium is not of this kind, Eq 43 shows
that the ultimate stress depends on the ultimate values of the creep

functions 5ii(oo) = Hi ; SeeC^o) = He ; and, in addition, the elastic con-

stants (/ = (A' + / ) - ^ d' = fi'-i of the inclusion.
The important general conclusion to be drawn is that the stress in an
elastic plug may change appreciably, even though there is no change in
the state of stress in the surrounding rock. Stress variations determined
from long term measurements, using elastic plugs, must be considered
suspect unless the possibility of stresses resulting from time-dependent
deformation has been evaluated.
The fact that the stresses in the plug eventually achieve the values
given by Eq 41 suggests that it may be practicable, in rocks which ex-
hibit appreciable viscous creep, to determine the stresses Ni and N2 in
the medium by observation of the ultimate values Pi and P2 which the
stresses in the plug tend to reach with time.

APPENDIX I
Theory of Stress Relief Technique in Transversely Isotropic Ground
Coordinate axes Oxi, 0x2, and Ox, are chosen so that 0x2 is the axis of
elastic symmetry. Elastic constants £1, £2, vi, V2, and M are defined as in
Fig. 1, so that the stress-strain relations take the form:
-El*! = CTl — V2O2 V\(Ji ; 2M723 = T23

£1
E\t2 = —ViCi + — 0-2 — ''20'3 ; Eijn (1 + J'i)ri3 (8)
E2
£'163 = —nfl — C2(T2 + 03 ; 2M712 = T12

where the 7i; are the tensor components of strain, 7ij = }i{dui/dxj + dUj/dxi).
Consider a primitive or undisturbed stress system in which the direction of
one of the principal stresses, —N^, lies in the plane of elastic symmetry and can
be taken to coincide with the xi direction. It is quite likely that the horizontal
direction, lying in a dipping stratum, is a principal stress direction, and it will
sometimes be convenient to refer to the X3 direction as horizontal.
The directions of the principal stresses —Ni and — A^2 make an arbitrary angle
a with the Xi and x^ direction (Fig. 5).

FIG. 5—Directions of the principal stresses —Ni and —N2 making an arbitrary
angle a with the Xi and X2 directions.

Stress Relief of Borehole With Axis in x^ Direction

Analysis of the stress-relief process presents a plane problem which is one
of neither plane-strain nor plane-stress, since both stress and strain in the X3
direction are thereby changed. It is reasonable to suppose that, when the hole is
first bored, it causes no strain in the xs direction (so long as attention is confined
to cross sections remote from the ends of the hole). The change in stress, there-
fore, satisfies the plane-strain condition 0-3 = VI(TI + vicru everywhere in the cross
section. It follows that this relation is violated only to the extent that the primi-
tive stress does not satisfy it. This means that, when the stress is relieved, a
nonzero €3 can be related to the removal of a homogeneous stress 0-3 = — A^o
given by
No = N3 — Clffl.O — I'2O'2,0 (9)

where -0-1,0 and —o-j.o are the normal components of the primitive stress in
the Xi and xi directions. Put in terms of Ni and Aô, Eq 9 becomes
No = Ni — vi(Ni cos^ a + N2 sin^ a) — v2(Ni sin' a -f A^2Cosâ)
= ^ 3 - }iin + i'2)(Ni + N2) - yziî - v2)(Nr - N.) cos'a. . (m
If Aô happened to be zero, there would be no strain in the X3 direction when the
stress is relieved.
The removal of the principal stresses at infinity, — Aî and —N2 (Fig. 2), is
equivalent to the application of stresses Aî and N^ in the same directions, and
it can be shown that the resulting plane-strain deformation of a hole r = aixi =
r cos &, Xi = r sin B) is given by

— = ^ ' -f 5 ' cos 20 + C ' sin Id. (11)

a
where
.1
[(1 + c' - / ) ( M + Ni) + 2c(Afi - Ni cos 2a]
^' = {\-c + dY

[2c(A^i + Af2) + 2( 1 + d){N^ - N2) cos 2a] . . (12)

B' = (1 - c + d)

C' = ; 2 ( 1 - d)(Ni - N2) sin 2 a

{l-c+dy
s, c, and d being elastic constants related to those used in Eq 8 as follows;

s =
E2 El

2(M- I
c =
A:i + 1 + 2'i\ki + k^yi
/ci + 1 - 2"\ki + k,)\Vi
d = .(13)

(1 - >'iy"
ki =
iEi/E2 - V2'y"
Ei/2M - 1/2(1 + vi)
ExlE2—ji

In addition to this deformation, there is some due to the relief of the stress
(73 = — A^o which causes the deviation from plane-strain. Since no stress in the
lateral directions is involved, this will cause a homogeneous strain in the medium,
unaffected by the presence of the hole. Treating it as the application of a stress
0-3 = A^' o and substituting it into the stress-strain relations, Eq 8, we find that it
causes the following strains:

(14)

The corresponding radial strain in the cross section is

0 0 2 n 1 0 * 2 / 1

fr = n COS 5 + £2 sm d

= — ^ (n cos'^ 6 + v-z sin' d)

+ V2+ (vi — V2) cos 2B\.

2£i in

Since this is independent of r, it will equal the fractional radial displacement of

the hole and so we can write

= - ^ W + v,+ in - v-z) cosie] (15)

a 2Ei
Elimination of A^o from Eq 15 can be performed using either Eq 10 or 14.
The final result for the deformation u/ = u/ + Ur° of a horizontal borehole
can be written

Ur
= A + Bco^le + Csin 26. (16)

where, using Eq 10
's{\ + c' -d') {v\ + ^2)
(A^i + N2)
(1 - c + dy + 4£i
2sc "1 V2 Vl + ^2
(Ni — No) C0S2Q: —
- c + dy +
A^3
+ .(1 4Ei
2sc (17)
(M + N2)
B .(1 ~ c + dy + 4£i
2s{\ + d) ^ in - "2
+ (1 - c + dy 4£'i
iNi - N2) cos 2a -
2Ei
\^N3

2sil - d)
C = (A^i - A^2) sin 2a
(1 - c + dy
or, using Eq 14,

A =
(1 — c + ay
+ 2c(Ni - N2) cos 2d - Min + ^2)63

B (18)
ii-c + ay [2^^^^ + ^^^
+ 2(1 + d){Ni - N2) cos 2a] - Hin - P2)i3
25(1 - d)
C = (JVI - #2) sin 2a
(1 ~ c + dy-
5//-ej:5 Relief of Borehole With Axis in X2 Direction
In this case, only one of the principal stresses of the undisturbed ground lies
in a cross section, xi = constant, of the hole. Instead, we use the normal com-
ponents —Ni and —A'^2' referred to the coordinate system (xi, x^, and X3),
with the principal stress in the ^3 direction, —N^. Thus

A^i' = Ni cos' a + #2 sin' a; N-/ = Ni sin' a + N2 cos' a. . . . (19)

The tangential components are neglected as they do not affect the radial deforma-
tion of the hole.
As with the horizontal hole, so here with the hole normal to the plane of
symmetry it is necessary to allow for the fact that the strain in the borehole direc-
tion, (2, will not be zero, in general. If it were zero, the normal stress components
would satisfy the relation 0^2 = {Ei/Ei)vi{<7i + crz). By arguments resembling
those of the previous section, it is easily shown that ^2 is due to the removal of a
homogeneous stress (j^ = —No' such that

^0' = ^2' - t >'•AN^' + N-d (20)

El
the remainder of —Ni being accounted for by the plane-strain part of the analy-
sis.
The removal of the stresses —Ni' and —N3 (equivalent to applying TV/ and
N3), under plane-strain conditions, leads to the following deformation of a
hole r = a:

- = ~s[Ni + N,+ 2(N,' - N,) cos 20]\

a El t (21)
Xi = /• cos 9; x-i = r sin d \
while Eq 20 leads to an additional strain
Aô' Aô' Aô'
El El El

and hence a radial displacement

Ur No (22)
a El

Addition of Eqs 21 and 22 and substitution from Eqs 19 and 20 leads to the total
radial displacement of a borehole normal to the plane of transverse isotropy:
n
^ = E + Fcosld (23)
a

where:

E = ~ wax - .'2)(iVi + N^) + 1.^(1 + V2)

• (JVI - TVj) cos 2a + A^s] ^ (24)

F = ^ s{N^ + A'2 + (Wi - A^2) cos la - 27V3]

or, if the second equality in Eq 22 is used, with the expression for E replaced
by
E = \& s{N^ + 7V2 + (iVi - }i^) cos la + 2Af3] - vit<i. . (25)
E\

APPENDIX II
Development of Stresses in a Cylindrical Elastic Inclusion Situated in an In-
finite Viscoelastic Medium
Because of the relation between solutions of elastic problems and those of the
corresponding viscoelastic problems [6], it is possible to use the known solution
for an elastic inclusion in an elastic medium for the purpose of solving the prob-
lem of a viscoelastic medium.
Savin [7] has the solution of the plane elastic problem, which may be easily
put into the following form (see Fig. 4):
2^1] 1+ V I + X
\ = ^ 7 ^ (#1 + N,) ± - ^ (TVi - 7V2). . . (26)
IPA 2+ M ^ ^ X + '^,

where /i is the modulus of rigidity of the surrounding medium, and x = 3 — Av

for plane-strain, x = (3 — i')/(l -|- v) for plane-stress, and the corresponding
primed quantities refer to the inclusion. We assume the cyhndrical inclusion to
be infinitely long and under plane-strain, so that x = 3 — 4v.
Then,

^ 4 = a(Afi + 7V2) ± ^(M - Afj) (27)

where:

a= ^"^ ; 6 = -ili^zL (28)

1 + ^ j ^ , 3 - 4 . + f-,
A + M M

Copyright by ASTM Int'l

Downloaded/printed by
Pontificia Universidad Catolica de
BERRY AND FAIRHURST ON IN SITU STRESS DETERMINATION 203

Since v = J.^ 1, Eqs 28 are equivalent to

.(29)

where c' = (x' + nT' • d' = M'"'

Viscoelastic Equations
There are various ways of formulating constitutive equations for viscoelastic
media. For the present purpose it seems better to use the integral relations based
on creep functions. They are, if the medium is unstressed prior to / = 0 ,

UU) = Si,(0)a,(O + I Si/{t - T),T,{T)dT (30)

where the Suit) are creep functions, and the Sij'{t) are their first derivatives.
In the elastic case 5,7' (?) = 0 and
6.(0 = e5,,;(r,(0 (31)
In the above u and o-, are normal strains and stresses for / = 1,2, and 3, and
shear strains and stresses for / = 4, 5, and 6. In this paper we consider only
isotropic media and then

eSii = - for 1=1,2, and 3

eSij = — - for i,j = 1, 2, and 3 i 9^ j

eSii = - for / = 4, 5, and 6

all other eSa being zero.

The viscoelastic problem is solved by considering the solution of the equations
which are derived by applying the Laplace transform to each equation and
boundary conditions. The equation corresponding to Eq 30 is
€i{s) = s 'Sij{s)aj{s) (32)
where j is the Laplace transform parameter. The elastic component has the
transformed equations
ii{s) = eSijdijis) (33)
Viscoelastic Solution
AppUcation of the correspondence principle to the present problem leads to
the conclusion that the Laplace transforms Pi{s) and P^i^s) of the principal
stresses in the elastic inclusion are obtained by substituting Ni{s) and N^is),
the transforms of Ni{t) and N-!.{t) in Eq 27, together with a{s) and b{s) ior a
and b, where a{s) and b{s) are the result of substituting sSnis) for 1 / £ and sSf,(,{s)
for l/;u in Eq 29. If we assume that

mit) = N,H(,t)
that is, are constant stresses applied at ? = 0,
then
Niis) = Ni/s
Ms) = Ni/s
and the result is
Piis)]
^iN^ + N,) ±^{N,-N,) (34)
P2(s)\
where:
Si6[4Sn — SM)
ais) =

(35)
566(4511 - SK)
bis) =
d'
SitilSn — 2SM) + >Sii

The time-dependent behavior will be in the form

Piit)]
= Ait){N, + Ni) ± Bit){N, - N2). .(36)
PM\
where:

^(r)=£-[^]: 5(0=£-'[^] (37)

£ - ' denoting the Laplace inversion operation.
Behavior as t —* x
In order to find the detailed behavior of Pi(t) and Pi{t), the time-dependent
behavior of the creep functions of5ii(/) and566(') must be known. However,
it is possible to determine the behavior after infinite time, with limited assump-
tions about 5ii(/) and 566(/)-
Fairly general forms are given by assuming:

Suit) =a^+b,t+ f GI(T) (1 - e""! dr

Jo
> • ,(38)
S^it) =a,+ht+ f Geir) (1 - e-'") dr
Ja

where the Ui give the instantaneous components of strain, the bi the continuous
creep, and the Gi are spectra of retardation times of creep. The Laplace trans-
forms of Eq 38 are

Giir)
Snis) = -'s +
s^ s Jo \ + ST
.(39)
G,(T)
S S' S Jn I + ST

The behavior as / -> « can be found by means of a Tauberian theorem about

Laplace transforms [8] which may be stated thus:
\imA(t) = lim ais). .(40)

(and similarly for B and b) if each limit exists. Applying this theorem to Eq 35
with 5ii and See given by Eq 39, we find that

^(oo) = 2 1 / '^^.
- 72 5(oo) = (41)
-r ,
bi

so long as bi and b^ are nonzero.

Result for bi = be = 0
When bi = bi = 0, it is obvious from Eq 39 that:

sSniO) = fli + / GM dr = Hi
•'0
(42)
iSmiO) = fle + / G6(T) dr = H,

We notice from Eq 38 that sSu(0) = 5ii(«) and

SSetiO) = 5'66(<X>)

when bi = bt, = 0. Substitution from Eq 42 into Eq 35 leads to the result

HMH, - H,)
^(oo) = a(0) =
2i/i(i/6 - c')
(43)
//6(4i/i - H,)
5(oo) = Z>(0) =
HeilHi - 2H,) + d'Hi

Conclusion
From Eq 41 it is seen that the ultimate stress in the elastic inclusion is inde-
pendent of the elastic constants of the inclusion if the viscoelastic medium is
such that it can continue to creep viscously both in uniaxial tension or compres-
sion and in shear. The only constants required are the coefficients describing
viscous creep, bi and b^.
However, if the viscoelastic medium is not of this kind Eqs 43 show that the
ultimate stress dependents on the ultimate values of the creep functions 5ii( =») =
Hi, SeeC"") = Hi, and, in addition, the elastic constants c' = (X' + M')~S
d' = n'~^ of the inclusion.

References
[1] E. R. Leeman, "The Measurement of Stress in Rock," Journal, South African
Institute Mining Metallurgy, Vol 65, No. 2, September, 1964, pp. 45-114.
[2] C. Fairhurst, "On the Determination of the State of Stress in Rock Masses,"
Paper No. SPE 1062, 2nd Conference on Drilling and Rock Mechanics, Uni-
University of Texas, Austin, Tex., January, 1965.
[3] S. G. Lekhnitskii, Theory of Elasticity of an Anisotropic Elastic Body, Holden-
Day, San Francisco, 1963 p. 24.
[4] R. H. Merrill and J. R. Peterson, "Deformation of a Borehole in Rock," Report
Investigation No. 5881 U.S. Bureau of Mines, 1961.
[5] A. H. Wilson, "A Laboratory Investigation of a High Modulus Borehole Plug
Gage for the Measurement of Rock Stress. Proceedings, 4th Symposium on
Rock Mechanics, 1961, Bulletin 76, Mining Industrial Experimental Station,
Pennsylvania State University, University Park, Pa., p. 185.
[6] A. H. Corneliussen and F. H. Lee, "Stress Distribution Analysis for Linear
Viscoelastic Materials," Creep in Structures, N. J. Heff, editor. Academic Press,
New York, 1962, pp. 1-20.
[7] G. N. Savin, Stress Concentration Around Holes, W. Johnson, translation editor,
Pergamon Press, New York, 1961.
[8] B. van der Pol and H. Bremmer, Operational Calculus Based on the Two-Sided
Laplace Integral, University Press, Cambridge, England, 1950, p. 122.

Laboratory Testing of Rock Strength

REFERENCE: B. B. Mazanti and G. F. Sowers, "Laboratory Testing of

Rock Strength," Testing Techniques for Rock Mechanics, ASTM STP
402, Am. Soc. Testing Mats., 1966, p. 207.
ABSTRACT: New equipment and testing techniques have been devel-
oped to investigate the basic strength of rock. The effect of varying the
intermediate principal stress on the strength appears appreciable in the
case of crystalline rocks such as granite. Failure of hollow cylinders of
rock under conditions of high ratios of the intermediate to the major
principal stress appears to be initiated as a "spalling" failure. Although
simple tests for the determination of the strength of intact rock can
furnish reliable data for certain design situations, the behavior of dis-
continuous rock masses must be evaluated by means of suitable model
studies and analytical procedures. While the results will still be only ap-
proximations of the mass properties, there are definite benefits in terms
of planning and interpreting field tests.
KEY WORDS: rock (material), rock mechanics, laboratory testing, tri-
axial test, strength

The strength testing of rock in a laboratory has two objectives: (1) the
determination of the fundamental behavior of rock under load and (2)
the measurement of strength and deformation parameters that can be
applied to the analysis or design of an engineering structure.
In the largest sense a rock mass is little different from any other ma-
terial. It is composed of particles, mineral grains, that are bonded together
by a combination of their intermolecular forces and mineral cements
and that are physically interlocked to varying degrees. These bonded
masses may be isotropic or anisotropic, homogeneous or nonhomo-
geneous, and the nonhomogeneity may exhibit a definite pattern, or it may
be random.
Systems of cracks, such as joints and bedding planes, usually divide the
rock mass into discrete blocks that may be regular in shape and size, like a
child's building blocks, or extremely irregular with nonplanar separations
and random sizes. The structure of a rock mass is, therefore, little different
1
Assistant professor of civil engineering, Georgia Institute of Technology, Atlanta,
Ga.
2
Regents' professor of civil engineering, Georgia Institute of Technology and vice
president, Law Engineering Testing Co., Atlanta, Ga. Personal member ASTM.
207

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
208 TESTING TECHNIQUES FOR ROCK MECHANICS

from that of a soil mass. In fact, any demarkation between soil and rock is
highly artificial, either geologically or from the engineering point of view.
All the qualities and defects in structure and composition in rock have their
counterparts in soil. The differences are in degree. The magnitude of
particle bonding in soil is usually only a fraction of that in rock. Therefore,
a crack in a soil mass may be of less relative importance than in the
stronger rock mass. The spacing of the separations in soil, such as shrink-
age cracks, may be much smaller than those in the rock; therefore, their
effect can often be found by laboratory tests of specimens of modest size
containing a statistically significant number of cracks. The spacing of
joints in a rock mass, however, may make such a test prohibitively large.
Otherwise, similar degrees of nonhomogeneity and anisotropy are pres-
ent in both soil and rock.
An engineering approach to the mechanics of either soil or rock masses
must consider two aspects of strength and deformation: first, the behavior
of the intact material and, second, the effect of the nonhomogeneity and
the separations. In soil mechanics, the greatest effort is being expended on
thefirst;in rock mechanics, the greatest effort is on the second. While it is
possible that the present emphasis in rock mechanics most nearly repre-
sents the need, it is the authors' opinion that both aspects must be under-
stood before a rational solution can be had to any engineering problem.
Rock research at the Georgia Institute of Technology has developed
from the interest of both authors in solving civil engineering problems of
heavy structures and dam foundations on rock where both aspects of the
mass behavior under load are significant. Four types of studies have been
undertaken simultaneously: (1) fundamental research on the strength of
intact rock, (2) simple tests for rock strength, (3) the mechanics of failure
in discontinuous rock and soil-rock systems, and (4) the behavior of
broken rock. The present paper describes the testing methods presently in
use and gives examples of the results being obtained.
Fundamental Research in Strength of Intact Rock
Conventional Triaxial Tests
There have been a number of investigations reported which were
concerned with the strength of intact rock specimens under various en-
vironmental conditions, among which are triaxial tests made at moderate
confining pressure (10,000 psi) and at room temperature on solid cylin-
drical specimens. In many such investigations, the control of pore pres-
sures was accomplished. The rock strength investigations at Georgia In-
stitute of Technology which utilized solid specimens were among such a
class of experiments and have been partially reported by Schwartz in 1964
[7].^ Triaxial tests were made on solid specimens % in. in diameter and

' The italic numbers in brackets refer to the list of references appended to this paper.

2.0 in. in height. Four different rocks were tested at confining pressures up
to 10,000 psi. Pore pressures were, in some cases, simply measured,
while in other cases, constant pore pressures were maintained during the
loading to failure of the specimens in an axial direction.
The equipment used in investigations of this type consisted basically of
a triaxial cell, hydraulic pressure maintainers, instrumentation necessary
to measure the stresses and strains involved, and a testing machine to ap-
ply the axial loads. The equipment has been described by Schwartz [1].
The principal conclusions from this type of test have been:
1. The Mohr criterion closely represents the "strength" of the rocks
when a two-dimensional failure surface is involved.
2. The shear strength of rock can be expressed in terms of three pa-
rameters: (a) cohesion, (b) internal friction, and (c) "fracture inter-
ference."
3. The concept of "effective confining pressure" is not valid for rocks of
all types because the pore spaces are not always interconnecting.

Hollow Cylinder Triaxial Tests

Among the investigations of the strength of rock, one of the most

neglected aspects to date has been the effect of the intermediate^ prin-
cipal stress on strength. The two stress conditions possible with the
conventional triaxial test, 0-2 = ai and 0-2 = 0-3, have indicated possible
varying effects of the intermediate principal stress with respect to these
extreme conditions only. It has been suggested by Brace [2] that the
apparent effect of the intermediate principal stress may be the result of
the effect of gliding flow within the rock. Handin and Fairbalrn [3]
have suggested that the apparent effect of the intermediate principal
stress may be due to anisotropy.
In order to evaluate the effect of the intermediate principal stress on
the strength of rock, a program of testing was initiated utilizing hollow
circular cylinders of rock subjected to different combinations of internal
and external pressures together with axial loads. The application of
pressures, either internal or external, to the surfaces of a hollow cylinder
will induce circumferential as well as radial stresses within the wall of
the cylinder. The magnitudes of the stresses depend upon the values of
the applied pressures, the diameter of the cylinder, the wall thickness
of the cylinder, and the radial distance from the center of the cyhnder
to the point in question. For a cylinder loaded uniformly over the inner
and outer surfaces, the following relations hold (based on elastic analy-
ses):

* As used herein, the three principal stresses will be denoted as follows: major prin-
cipal stress, cri ; intermediate principal stress, 0-2; minor principal stress, 0-3 ; and in
terms of absolute values, <ri > (72 > 0-3.

a%\Po - Pi).^ 11 aJp.

'Pi -_ A^p
b'P.
circumferential stress, ae
b"" - a" • ^ 2 b'' - a^ ' • .(1)
•
and[
.. , , ab\Po - Pi) 1 , a Pi - b'Po
radial stress, Or = —n^ -„ (2)
0^ — 0?- - a^

where:
a = inside radius,
b = outside radius,
r = radial distant to point in question.
Po = external pressure, and
Pi = internal pressure.
It should be noted that these stresses are the principal stresses, as are
the axial stresses. With the proper selection of loading apparatus, the three
principal stresses can be varied at will, and the effect of varying the in-
termediate principal stress between the two values attainable in the
standard triaxial test can be evaluated.
This type of test has been utilized in the past for soils as well as for
solid materials, such as metals and concrete. Except in the case of metals,
the tests were carried out at relatively low appUed pressures, and the re-
sults cannot be extrapolated to the high pressures which are of interest
in the case of rock. In many cases, the work did not include both internal
and external confinement, and, in addition, the specimen size was such
that extreme nonuniformity of stress made proper interpretation difficult.
Recently Handin and Heard [4] have reported using hollow cylindrical
specimens subjected to combined axial load, external and internal pres-
sures, and torsional loads. In their work, the specimens are on the order of
0.75 in. effective length, 0.5 in. diameter with a very thin wall.
In the selection of a specimen size for this study, several factors were
considered. One of the most important of these factors was the necessity
to utilize existing triaxial equipment insofar as possible. The triaxial cells
had been made to accommodate specimen sizes in the range of 2 in. max
diameter. In order to produce a specimen which could be used in these cells
and to meet other requirements, such as having a reasonable cross-sec-
tional area and, in addition, a wall thickness which would approach a thin-
walled condition, a specimen of 1.600 in. outside diameter with a 0.200-
in. wall thickness was chosen. A minor factor in the selection of size was
the fact that membrane materials of the proper size were available com-
mercially. For membranes, poly(vinyl chloride) electrical insulation sleev-
ing was selected, having been previously used in the rock studies on solid
specimens. Although, at pressures above about 2000 psi, some of the
plasticizer is squeezed from the membranes, there is no leakage of the con-
fining fluid, hydraulic oil, through the membranes.

Hollow Cylinder Triaxial Apparatus

Several schemes were explored in attempts to produce a relatively fool-
proof as well as flexible system of jacketing both the inner and the outer
surfaces of the cylinders to sustain confining pressures up to 10,000
psi with no leakage. The system finally adopted is shown in Fig. 1. A
base and a top cap, each having a "ring" bearing surface of the same

LOADING CAP

NUT

UPPER SPECIMEN CAP

0-RING

TOP 0-RING PLUNGE

INTERNAL MEMBRANE
EXTERNAL MEMBRANE

BOLT
BOTTOM 0-RING PLUNGER

SEAL

INTERNAL PRESSURE PORT

FIG. 1—Hollow cylinder triaxial apparatus.

dimensions as the specimen, are machined to obtain a fluid seal on

the inside when an 0-ring is forced downward (or upward as the case may
be) along an inclined surface, which causes the O-ring to press the mem-
brane against the inner surface of the ring. The bottom O-ring is forced
downward by the plunger as a bolt is tightened in threads in the base. The
top O-ring is actuated by a plunger arrangement pulled against the top
cap by a nut, tightened on the plunger stem which extends through the
top cap. To ensure a positive seal between the inner and outer fluids, the
top cap contains an O-ring which seats against the loading head. The outer

membrane is sealed simply by the fluid pressure forcing it against the

caps, although O-rings are used as safety devices to ensure sealing at low
external confinement.
The bottom cap has been designed to screw into the base of the triaxial
cells. This is an interchangeable feature designed to make the cells more
versatile with respect to the use of different size specimens, since the same
equipment is used for soils work. The design of the bottom cap is shown in
Fig. 2.
The confining pressure is controlled through the use of individual.

0,437"
FIG. 2—Hollow cylinder bottom cap.

piston-cylinder arrangements where the pistons are positioned by a bolt

reaction member. The adjustment of the bolt either forces the piston into
the cylinder or allows it to move outward, thus causing the pressure in
the line to increase or to decrease, respectively. Such an arrangement al-
lows very close control of the pressures. The confining pressures are
measured by the use of hydraulic gages accurate to ±0.25 per cent of full
scale.
Specimen Production
The hollow cylinders are produced from quarried blocks of stone by
the use of a double-tube, thin-walled core bit. The bit is of the impregnated

diamond type and is made to produce a hollow core 4 in. in length, with
an outside diameter of 1.600 in. and a wall thickness of 0.200 in.
It was found necessary to use a heavy-duty precision drill press to cut
cores from granite and similar, hard drilling rock. Penetration rates of
about 2 in./min are possible in granite when using a 3-hp radial drill
press at 700 rpm and with very poorly controlled drilling water. To reduce
the possibility of bit damage, drilling is begun at 75 rpm until the bit is
seated, after which time the vertical alignment is checked and the higher
rate of revolution is utilized until the drilling is complete. It is neces-
sary to maintain a constant bit pressure to produce a smooth-sided,
straight specimen. The inner solid core is retained for conventional tri-
axial tests.
The cores, both hollow and solid, are cut to the approximate final
length of 3.2 in. with a diamond blade trim saw. The ends of the trimmed
cores are then ground plane and perpendicular to the long axis by the use
of a rotary carborundum wheel and a special holding jig mounted on a
feed table. Since the testing equipment accepts varied specimen lengths,
no attempt is made to produce a uniform length, since this would require
additional time in the grinding operations.

Preparation for Testing

The hollow cylinders are either sprayed with a Teflon spray or rubbed
with a powdered Teflon to reduce friction between the membranes and
the specimens, both during the jacketing operations as well as during the
testing. Either of the two preparations have been found satisfactory, but
the powdered material produces the smoother surface, probably due to a
thicker layer of the Teflon deposited on the specimen.
The steps involved in the placement of the specimen in the loading
apparatus are as follows:
1. The inner membrane is trimmed to approximate length and inserted
in the bottom cap with the bottom O-ring plunger and O-ring contained in
the membrane about V2 in. from the membrane end.
2. The bottom O-ring plunger is forced downward by the bottom pull-
down bolt, and the bottom seal is effected.
3. The inner membrane is trimmed to exact length by placing the speci-
men in position with a metal spacer on the top surface and trimming with
a sharp knife.
4. The "inner pressure" fluid line is then purged of air by pumping until
hydraulic fluid partially fills the inner membrane.
5. The upper O-ring plunger and O-ring are placed in the top of the
inner membrane, and the trapped air removed by manipulating the plunger
downward and allowing the air to bypass the plunger.
6. The specimen is placed in position on the bottom cap, the outer
membrane slipped over the specimen, and the top cap fitted into place.

7. The top O-ring plunger is drawn into position, the loading cap
screwed on, and the installation is complete.
Test Procedure
The testing procedure is quite similar to any other type of triaxial test
with the exception that there are two fluid pressures to be kept in adjust-
ment. In tests where the circumferential stress is a definite, constant frac-
tion of the axial stress, the external pressure is adjusted to predetermined
values as the axial load is increased. Thus far, all tests have been conducted
with the internal pressure maintained constant throughout a given test, ex-
cept in "tension" tests. All tension tests have been made with zero external
pressure.
Only axial deformations were measured. A linear variable differential
40
\
Legenc 1
S ^lid Spec. - Solid L i n e s
Hoilow Spec. - Dashed L i n e s
30
/60° ^,_ .— - ,
/ -O^
^ ^ ^
20

^b
X ^
J r ^ ' ^V
10
\ \

i
\ \
\
\
\
1 \ \
1 \
10 20 30 40 50 60
a - Ksi (Failure)

FIG. 3—Mohr diagram for granite tested with 02 = 0-3 and ai increased to failure.

transformer (LVDT) is used for this purpose, and the load-deformation

curve is traced on a drum recorder, with the axial load indication being
actuated by a push-rod arrangement from the testing machine load indica-
tion unit. Since a considerable amount of deformation occurs in the test
equipment itself, the cell has been calibrated with no specimen between the
end caps, and this correction is applied to the recorded total deformation to
determine the deformation of the specimen.
Results and Discussion
It is well known that the strength of a material is dependent upon
specimen size and shape, as well as on other factors. Therefore, it is of
interest to compare strength results from the hollow cylinder tests with
Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreement. No
MAZANTI AND SOWERS ON TESTING OF ROCK STRENGTH 215

results from solid cylinder tests conducted under comparable stress

states. One such state is for the condition of (r2 = 0^3 = constant, and
with (Ti increased to failure. The results from a series of such tests are
shown in Fig. 3 in the form of a Mohr diagram. It is evident that while
the solid specimens apparently had slightly greater strength at lower
confining pressures, the hollow specimen strength is slightly greater at
3000 psi confinement. The strength envelopes, for all practical purposes,
however, are identical.
Under "conventional triaxial stress states," the hollow cylinders fail
in a manner quite similar to that of the solid cylinders. Failure planes
are formed in both tests at about the same angles and extend completely
through both walls of the hollow cylinders, so that a side view of the

FIG. 4—Granite specimens failed under standard triaxial stress states. 0-3 = 3000 psi,
hollow cylinder on right.

hollow cyhnder looks like a solid cylinder. A comparison of failure sur-

faces for granite is shown in Fig. 4.
Up until May, 1965, there had been no tests conducted which would
provide a comparison between hollow cylinder and solid cylinder results
for the "extension test" conditions where <T2 = <n > as.
The effect of the intermediate principal stress, 0-2, on the strength can
be illustrated by Mohr circles for a series of tests where the minor prin-
cipal stress, (73, is held constant and 0-2 is increased with 0-1 (at a constant
ratio) until failure occurs. Results from such a series are indicated in
Fig. 5.
The minor principal stress, 0-3 (the radial stress at the inside face),
was maintained at zero in all tests and failure caused by increasing
(72 (the calculated circumferential stress at the inner face) and ai (the
axial stress) in the ratios indicated. The major principal stress increased

40 1 —
Curve a2^o-i
1 0.0
2 0.1
3 0.29
30 4 0.4 —
0.5
0.6
0.8
3 ^ ^

2 \ ^
10
\
" ^ 1

\
\ ^
10 20 30 40 50 60
a - Ksi (Failure)

FIG. 5—Mohr diagram for granite hollow cylinders for az = 0 and varying ai/ni
ratios.

iL
70
tri -n

y^=o
40

30
n P Legend
20
O "3 = 0 psi
D "3 = 500 psi
10
A "3 = 1000 psi

.4 .6 1.0

FIG. 6—Variation of principal stress difference with <ri/ai ratio for granite hollow
cylinders.

with increasing ratio of a-i to ci up to a point, after which it decreased.

The same results are shown in Fig. 6 as principal stress difference (o-i —
aa) versus cijax. Here, the maximum stress difference for the 0-3 = 0
series appeared to occur at o-s/ffi = 0.5. As the value of a-i is increased,
somewhat similar results were obtained, and, in addition, the value of
02/0-1 producing the maximum stress difference seemed to be slightly
less than in the case of 03 = 0; however, this may be due to having in-
sufficient data points for the case of 03 = 500 psi. (Equipment limitations
have restricted the maximum value of 02/01 which can be increased to

35
04

-Qr"' 1O.5
25 ^^ ^
0.4_c
0.6 0.8
20
/y^ 029

//
>£ 15
0.2°: ^
10
Legend:
>
QO 0 g; = 0
n rt: - f=.rin
A Oj = 1000 psi
No s. beside symbols
are ratios of O^-toQi
1 1
0
0 10 20 15 25 30 35 40 45
O;. ksi
FIG. 7—Octahedral shear versus octahedral normal stress for granite hollow
cylinders.

failure to those shown in the graphs for the various 03 values. This re-
striction is at present being modified so that additional data will be
available later.)
In terms of the octahedral shear and normal stresses, the effect of the
intermediate principal stress is indicated in Figs. 7 and 8. In Fig. 7, it is
seen that both the octahedral normal stress (oo)* and the octahedral
shear stress (TQ) increased with increasing 02 for a given value of 03.
This increase appeared to level off in the case of TO and may reach a
maximum, after which it decreased with increasing ratio of 0-2/0-1.

* Octahedral stresses are those stress components acting on an "octahedral plane,"

the normal of which has direction cosines equal to l/-\/3 with respect to the principal
planes.

Higher values of 0-3 produced increases in tiie octahedral shear strength

for a given value of octahedral normal stress and appeared to slightly
decrease the value of 0-2/0-1 at which the leveUng off of the curve took
place. As pointed out previously, the data available were insufficient to
state definitely such a result.
Fig. 8 seems to be significant, with respect to the data available, in
that there is an apparent unique relation between the ratio of the octa-

1.5

1
1.4

1.3
\ — Lege
0 cr3 = P|- = 0
\ 0 Bj = Pi = SOOpsi
0 A J3 =P =1000psi
1.2

1.1 \
\1
t

1.0 \
\>

0.9

0.8

0.7

0.6
0 0.2 0.4 0.6 0.8 1.0

FIG. 8—To/o-o ys ai/ai for granite hollow cylinders.

hedral shear to normal stress and the ratio of the intermediate to the
major principal stress. If further tests should substantiate the uniqueness
of the relationship, such a curve could provide the necessary relations to
calculate strength under conditions where certain principal stress ratios
exist and where one or more of the principal stresses are then deduced.
This type of relation could also be used to determine the magnitude of
applied minor principal stress necessary to maintain stability under con-
ditions where the changes in the principal stresses were sufficient to
cause failure. Similar relations have been tentatively found to hold for
Indiana limestone, although the curves are not quite as well defined.

The data presented here may be used to determine portions of the

three-dimensional stress failure surface. Although such has been done,
there is insufficient data to clearly define the failure surface, and the
presentation of such a graph will be reserved until a later date.
There is evidence to substantiate the premise that failure is initiated
at the inner radius of the hollow cylinders when they are loaded so that
the axial stress is the major principal stress, the tangential stress the
intermediate principal stress, and the radial stress the minor principal
stress. Under such conditions, several tests were begun with the ratio of
(72 to (Ti at values close to unity. Since the fluid pressure capacity of the
cell is 10,000 psi, the tests were terminated upon the attainment of such
external confining pressures without the specimens having failed. Such
failure is ordinarily evidenced by a loud cracking sound, and the in-
stantaneous drop off of axial load. In these cases, the specimens were

FIG. 9—Spalling inside granite cylinder unloaded prior to fracture as — 0 and ai/ai =
0.88 {view 1).

intact and apparently undamaged upon first inspection after removal

from the cell. Closer inspection revealed that there was indeed structural
damage to the cylinder on the inner surface. Spalling of the inner sur-
face had occurred to considerable extent. In some cases the wall thick-
ness had been reduced to less than one half of the original thickness and
the "pop-outs" had occurred in several different places. A photograph
of one specimen is shown in Figs. 9 and 10. This is typical of the type of
damage which resulted and corresponds quite closely to "rock bursts"
which occur in tunnelling and in mining works. In addition to such
damage taking place in the unbroken specimens, identical distress could
be observed in many of the specimens which actually ruptured, indicating
that the failures did indeed begin at the inside surface of the cylinders.
Tension tests have been made on specimens of the granite. Only the
case of pure tension has been produced by using only internal pressure.
Failure under such conditions is evidenced by vertical fractures in the
specimen, generally at least two fracture surfaces and sometimes more
than two. The type of fracture produced under this condition is shown in

FIG. 10—Spoiling inside granite cylinder unloaded prior to fracture as = 0 and

ailoi = OM (v/evf 2).

FIG. 11—Tensile fracture of granite hollow cylinder under internal pressure only.

P; = O 0.5 1.0 l.^!TOl

FIG. 12—Indiana limestone hollow cylinders ai > at > ff3 , ffff and
(Tr = <rz . Pi in ksi.

Fig. 11. Since the rock specimens are still reasonably intact except for
the fracture surfaces, it is planned to use some specimens initially broken
in tension for studies of the compressive strength of fractured rock.
A few tests have been made on Indiana limestone where the circum-
ferential stress was the major principal stress. Illustrative of such tests
are the specimens in Fig. 12. In this test series, the inner pressure was
maintained constant throughout a test and the external pressure in-
creased to failure. As the external pressure was increased, so was the

axial load on the specimens. Under these conditions, the axial stresses
were the intermediate principal stresses. Although a detailed analysis
of the failures has not been made, there appear to be distinct differences
in the modes of failure as well as in the values of failure stresses. Further
studies along these lines are planned.
Simple Tests of the Strength of Intact Rock
Simple tests that can be performed on cores obtained from core boring
in foundation investigations are useful in evaluating the bearing capacity
of intact rock and to some extent in estimating the bearing of layered rock.
The unconfined compression test performed axially on a diamond drill
core provides the maximum deviator stress at zero confinement. This is
analogous to the bearing capacity of a badly jointed rock with vertical open

A. LOADING B. CROSS SECTION

FIG. 13—Punching shear of rock.

joints, so that adjacent portions of the rock mass do not provide lateral
restraint. The only preparation is to cut or grind the ends to planes, so as
to provide a uniform loading and to make the length at least twice the
diameter to minimize end restraint effects.
A punching shear test is possibly a representation of the punching of a
concentrated load, such as a pile tip through a hard rock layer underlain by
a soft or compressible layer. A simple punch developed jointly by Law
Engineering Testing Co. and Georgia Institute of Technology utilizes a
disk cut from a diamond drill core with a thickness of VA or Vb in diameter.
This is placed in a cylindrical guide and a piston forced through it as
shown in Fig. 13. The results of such a punching shear test are com-
parable to the shear strength with zero confinement (the apparent cohesion
intercept of the Mohr diagram). Some very soft, porous rocks crush locally
instead of punching. It is presumed that they will fail in a comparable man-
ner in the field, but direct evidence of this is lacking.
A flexural strength test possibly represents the bending of a thin slab-
like layer of rock underlain by a much less rigid material. A disk of the
rock, with a thickness of Vi to Vb in diameter, is placed in a cylindrical
guide with parallel end supports, Fig. 14. A transverse load at the center
forces it to break inflexure,similar to a beam test of concrete. The flexural
strength, computed by assuming that the rock fails after deformation pro-

portional to the distance from the centerline, is approximately equal to

the tensile strength of the rock.
Simple tensile strength tests can be run by cementing cores into metal
sockets that are threaded to fit the standard spherical tension seats of a
testing machine designed to test cast iron and other brittle materials in
tension. The epoxy resins have been adequate to test the strongest rocks
with a socket depth equal to the core diameter.
A cylinder of rock loaded perpendicular to its axis fails in approximately
pure shear. Such a test provides the shear strength at no confinement, ap-
proximately (the apparent cohesion intercept of the Mohr envelope).

A. CROSS SECTION

C. PLAN OF LOAD PISTON

B. PLAN OF SUPPORT
FIG. 14—Flexural test of rock.

Bearing Capacity of Discontinuous Rock Masses

Both theory and limited very small scale model tests [5] indicate
that the ultimate bearing capacity of an intact, homogeneous isotropic
rock mass is several times its unconfined compressive strength. Although
full scale verifications of soil bearing capacity theory applied to rock in
which the shear strength can be expressed by the parameters C (cohesion)
and 0 (angle of internal friction) are lacking, it appears that such theories
may be reasonable approximations for that rare rock mass that is intact,
homogeneous, and isotropic.
The exponents of rock mechanics emphasize that most rock masses ex-
hibit none of these qualities. Instead, according to John [6] the properties
of the rock mass depend far more on the system of separations within the
mass, and the strength and deformation are governed by the inter action of
the blocks across the separations, rather than that by the properties
of the rock within the mass.
The authors are not in full agreement with so sweeping a declaration.

In many problems in which the compressive stresses inherent in rock

masses are reduced and the joints and other cracks allowed to open, such
as in tunneling and in rock excavations, this hypothesis is probably valid.
In other problems, where the rock has weathered along the joint and bed-
ding plane surfaces, the properties of the weathered material probably
govern the strength and deformation of the mass. However, when the
engineering problem involves increasing the compressive stresses, such
as where a foundation load acts on the mass, the problem is more complex,
and both the response of the separations and the intact rock between must
be evaluated.
An analysis of the deformation and failure of a nonhomogeneous,
anisotropic, separated rock begins with a simplified mathematical model

TABLE 1—Idealized rock discontinuities.

Layered Jointed

1. Thin hard (or rigid) layer over 1. Rectangular closed plane joint
thick soft (or resilient) layer planes
2. Thin soft layer over a thick hard (a) overlapping or in line
layer [b) horizontal or inclined
3. Alternate hard-soft layers 2. Oblique plane joints, horizontal or
4. Progressively harder or progres- inclined, overlapping or in line
sively softer layers with increas- 3. Rippled joint surfaces, rectangular
ing depth or oblique, horizontal or in-
5. All of the above either horizontal clined, overlapping or in line
or inclined 4. Open or separated joints, over-
lapping or in line, rectangular
or oblique, horizontal or in-
clined
5. Filled joints (separations with
weaker material between, over-
lapping or in line, rectangular or
oblique, horizontal or inclined)

whose response approximates that of the real mass. However, because lit-
tle is known of the behavior of real masses, it is difficult to propose such
representations. Physical models appear to offer a relatively simple, in-
expensive approach to discovering suitable mathematical models, provided
the effects of model scale can be evaluated. Although such model analysis
is not so well developed as in fluid mechanics, it has proved useful in such
fields as earth pressure and soil bearing capacity.
The combinations of nonhomogeneity, anisotropy, and separation in
actual rock masses appear limitless. However, two basic structures appear
significant: (1) layered masses, composed of layers of different strengths
and rigidities and (2) jointed systems, composed of independent blocks as
described in Table 1. Various combinations of layered and jointed sys-
tems occur and can be added to the scheme.
Heterogeneous masses with no regular joint patterns, as with random

inclusions of either hard or soft materials, are also possible. In their de-
parture from the simple ideal, rock masses are no different than soil
masses. Any differences appear to be in the degree of departure from the
simple idealized homogeneous, isotropic medium so widely utilized in
solving many engineering problems, including those in soils but so rare in
nature. Therefore, analyses based on the behavior of idealized non-
homogeneity should be of some guidance in the solution of similar prob-
lems in actual rock masses.
The failure of a thin, rigid layer underlain by a soft, resilient mass
is important in designing pile foundations driven onto a thin cemented
sand or rock layer underlain by soft clay or loose sand. The laboratory
models designed to represent this condition consist of loose sand or soft
clay placed in metal boxes 12 in. deep and approximately 1 to 2 ft. square.
A hard layer of plaster of paris is placed directly on the softer base. The
load is applied by a steel piston of varying diameters. Preliminary tests in-
dicate that when the hard layer is thin, compared to the load diameter, a
punching failure occurs; when it is thick, the hard layer fails in flexure.
Preliminary tests have been made of jointed systems in which the rock
is represented by plaster blocks. These tests indicate that, if the joints are
open, the rock fails in unconfined compression. If they are closed the
mechanism is more complex and depends on the shear transferred between
the blocks.
More comprehensive model tests are planned for the future.
Behavior of Fragmented Rock
Fills of fragmented rock, broken from ledge by blasting and dumped or
rolled into place, have been employed extensively in dam construction and
to some extent for highway embankments and fills on building sites. In
practically all cases, the requirements for stability and settlement have
been met by designs based on past experience, because it has been con-
sidered impractical to test specimens of the full scale of rock fragment
sizes, and because it was assumed that small scale tests would be valueless.
The results of a large scale test of broken rock up to 18 in. utilized in the
rolled rock fill portion of the Lewis Smith Dam were reported by Sowers
and Gore in 1961 [7]. Parallel triaxial shear tests were run on 4-in. diam-
eter by 8-in. long specimens of all particles finer than 1.5 in. The full-
scale direct shear test gave an angle of internal friction of 45 deg; the
triaxial compression gave an angle of 42 deg. Although the small scale
test gave an angle that was smaller, at least 1 or 2 deg of the difference
might be attributed to the difference between plane strain and triaxial
strain. This one test series indicates that the small-scale laboratory test re-
sults may be comparable to full-scale tests, if the relative gradations,
as expressed by for example, the uniformity coefficients, are the same.
The settlement of dams constructed of broken rock was discussed by

o '^N^^
LU
X
_i 2
< ^
Z r
o
\ . '
I-
z
lu 4
\ ^
u
a.
m UNIFORMITY
0. Kl T pnr-K rnPFPpriFMT
1 GRAYWACKE 3.5
A
2 SANDSTONE 3.6
J LiKATWAl. K t r
GRANITE
4 13

500 1,000 5,000 10,000 50,000

VERTICAL PRESSURE IN POUNDS PER SQ. FOOT

FIG. 15—Typical pressure-settlement relation for rock fragments.

DRY GRAYWACKE, C .3.5 1630 psf

X
M • I II Ml.
DRY GRAYWACKE, C^ . 3.5 6520 psf

DRY GRAYWACKE, C„ . 3.5 6300 psf

DRY SANDSTONE, C^ . 3.6, 16300 psf

°R^GwyiMS<ET^r•7^-J—
DRY GRAYWACKE, C„.1.3, 16300 psf

4-
DRY GRANITE, C„ . T.3, 16300 psf

SATURN

.5 .7 1 2 3 5 7 10 2 3 5 ; 100 2 3 s ; 1000
TIME IN MINUTES AFTER LOAD APPLIED

FIG. 16—Typical time settlement relations for rock fragments.

Sowers et al in 1965 [8]. The rate of settlement was found to be approxi-

mately a Straight line on a settlement-log time plot, similar to secondary
compression of soils. This suggested that confined compression tests,
similar to the consolidation tests of soils, might be utilized in predicting the
amount and rate of broken rock settlement.
Consolidation tests of broken rock up to 1.5 in. were made in a 7.5-
in.-diameter 4-in.-high cylinder, loaded with a tapered piston by a pneu-
matic bellows.
The results for typical rock in terms of the logarithm of average axial
pressure versus per cent settlement are shown in Fig. 15. They resemble
the consolidation curves for soils. Typical log time-per cent settlement
curves are given in Fig. 16. These resemble secondary compression data
for soils. Both the rates of settlement and the magnitudes are comparable
to those observed in rockfill dams. The mechanism is observed to be spall-
ing progressive fracture of the points of contact between the fragments.
The mechanism of presettlement of broken rock by sluicing has been
studied extensively and the results, in part, reported. Specimens of rock
1 in. in diameter were cut with a diamond saw to either a conical or a wedge
point having an angle of 60 deg. These were cemented on loading plates
and mounted to the points or wedge edges bore on a polished steel or
ground plane surface of a specimen of the same rock. The specimen was
loaded by a lever system that provided a constant force, regardless of de-
flection. The conclusion is that the wetting causes immediate additional
fracturing, which thereby reduces the amount of fracturing after wetting.
The cause of the rapid crushing of the points of rock contact is still
under study. The rock point under continued load crushes until the area
oi contact can sustain the stress. There are microfractures in the point,
however, that open under stress but do not lead to rupture immediately.
Instead, the point crushes additionally at irregular but increasing time in-
tervals. The wetting appears to precipitate such additional fracturing,
usually within a fraction of a second.
Mineralogical studies of the rocks by thin-section and electron micro-
scope photo have not indicated minerals that might be appreciably altered
or stressed by water. The addition of a wetting agent that reduced sur-
face tension to half the normal value of water in some cases accelerated
the crushing. The amount and rate of crushing precipitated by wetting was
increased at low pH values (acid) and in the case of limestone and sand-
stone also by a high pH (basic). While the increased setdement of lime-
stones in an acid environment was expected, the increased settlement at a
high pH was not expected. The mechanism by which this increased settle-
ment occurs is still to be explained, but it appears to be a deterioration of
the highly stressed mineral bonds.
Conclusion
The laboratory testing of rock to determine its response to stress is still

in its infancy, and some engineers even hold that it will always remain so,
because it is impossible to predict the behavior of a nonhomogeneous,
nonintact mass by tests of small specimens. The same objection was raised
40 years ago in soil testing; yet, in spite of these valid objections laboratory
testing of soils has proved invaluable in investigating the fundamental
nature of the strength and the possible behavior of simplified masses.
While prototype behavior is not always fully reflected in theoretical pre-
dictions or model tests, the laboratory results always have real value,
if properly interpreted. It is the authors' opinion that the same applies to
rock engineering, although probably not to the same degree.
Field tests of soil or rock should give a more realistic prediction of mass
behavior under load. Such tests are time consuming and extremely ex-
pensive. Furthermore, they cannot always subject the mass to the fuU
range of prototype stress and environment changes. Properly interpreted
laboratory tests of the intact materials and appropriate mass models can
point out the significant strength and deformation qualities that need
evaluation, the possible effects of joints separations, stratification, and
other departures from the idealized case, and should make it possible to
plan more effective field testing, minimize the amount of full scale testing
needed, and to interpret the results more intelligently.

References
[1] A. E. Schwartz, "Failure of Rock in the Triaxial Shear Test," Proceedings, 6th
Symposium on Rock Mechanics, University of Missouri, RoUa, Mo., 1964.
[2] W. F. Brace, "Brittle Fracture of Rock," Stale of Stress in the Earth's Crust,
Am. Elsevier Publishing Co., New York, 1964, p. 162.
[3] J. Handin and H. W. Fairbairn, "Experimental Deformation of Hasmark Dolo-
mite," Bulletin, Geological Society Am., Vol 66, 1955.
[4] J. Handin and H. C. Heard, "Rock Failure in Torsion Tests," Semi-Annual
Report No. 2, AF Contract No. AFI9(628)-2784, 1964.
[5] G. B. Sowers, "Load Required to Compress Rock," Civil Engineering, August,
1943.
[6] K. W. John, "An Approach to Rock Mechanics," Journal, Soil Mechanics and
Foundation Div., Proceedings, Am. Society Civil Engrs., Vol 88, No. SM4,
August, 1962.
[7] G. F. Sowers and C. Gore, "Preconstruction Tests of a Rock Fill Dam," Pro-
ceedings, 5th International Conference on Soil Mechanics and Foundation
Engineering, Paris, 1961, Vol II, p. 718.
[8] G. F. Sowers,, R. C. Williams, and T. S. Wallace, "Compressibility of Broken
Rock and the Settlement of Rock Fills," Proceedings, 6th International Confer-
ence on Soil Mechanics and Foundation Engineering, Montreal, 1965, Vol 11,
p. 561.

DISCUSSION

A. J. Hendron, Jr}—The authors have concluded that the plot shown

as Fig. 8 appears to be a unique relationship for Stone Mountain granite
which may be useful for application to various problems where combined
stresses may be encountered. The particular relationship in question is a
plot of To/o-o versus o-2/0-1, where TQ is the octahedral shearing stress,
(To is the octahedral normal stress, and tri and (n axe the major and inter-
mediate principal stresses, respectively. A critical examination of this
plot, however, will show that it does not show the inherent strength
properties of Stone Mountain granite but is basically a manifestation of
the transformation of stress at a point governed by statics. For example,
consider the series of points in Fig. 8 which represent the tests conducted
on Stone Mountain granite with 0-3 = 0. The octahedral shearing stress,
TQ , is given by
TO = }W{<ri - <T,y + ia2y + i<T,y 0)
and the octahedral normal stress, O-Q , is given by
<^0 = Hioi + <T2) (4)
Combining and simplifying Eqs 3 and 4 yields
To/<ro = V 2 V I - ffj/ai + MTIY (5)
1 + £^2/0-1

A plot of Eq 5 is shown in Fig. 17 which is identical to Fig. 8 in the

authors' paper and can be established without conducting tests on a real
material. In the case of 0-3 > 0, the plot begins to show the material
properties. The values of 0-3 for the test results plotted by the authors
are rather small with respect to ai and 0-2 at failure such that 0-3 did not
have much influence on the calculated ratio of TO/OO ; therefore, the
points still fall very close to the curve as given by Eq 5.
The writer also questions the use of elastic theory to compute stresses
for the condition of failure. The test results presented by the authors
show a much greater effect of 0-2 on strength than has previously been
observed in the extreme cases of triaxial compression and extension
tests on concrete and various rocks. A more critical examination of the
interpretation of these tests is needed because this apparent effect of 0-2
may be due to several questionable features of the tests. Firstly, two of
the three principal stresses at failure are calculated from elastic theory.
Secondly, even if one were to assume it was valid to calculate stresses

^ Assistant professor of civil engineering, University of Illinois, Urbana, El.

l.Sr

I.J \

\
)
T, . V ^ y 1 - cr.^<r^ + (o-j/tr, )*
1.2
/ ''" ' * ''2"'\
O' 0

o.t

0.2 10

FIG. 17—Variation ofrn/af, with ai/aifor az = 0.

at failure on the basis of elastic behavior, there are large stress gradients
across the wall of a hollow cylindrical specimen subjected to a different
internal and external pressure. The calculated stresses assumed to be
associated with the failure of the specimen only occur on the extreme
outer or inner fibers of the cylinder.

B. B. Mazanti and G. F. Sowers (authors)—Mr. Hendron is, of course,

correct with respect to the purely mathematical relationship generated
by the ratio of TO to O-Q versus a-2/o-i for the case of 0-3 = 0. He is also
correct in stating that the values of 0-3 are small in comparison to ai and
0-2 and, consequently, have little influence on the value of TO/(TO •
We do not agree, however, that this plot (Fig. 8) does not show the
inherent strength properties of the material. Figure 3 shows that the in-
fluence of confinement on ci is very great, for instance, increasing o-i to
about 55 ksi for 0-3 = 3 ksi. This has the eff'ect of producing a relation-
ship in terms of octahedral stresses and o-j/ui which appears rather
insensitive to the minor principal stress. However, the inherent strength
properties are properly represented, with respect to the confining pres-
sures actually utilized. The fact that large increases in ai are caused by
relatively small increases in 0-3 in no way detracts from the usefulness of
a relationship which incorporates such information and, which, for
practical purposes can be approximated by a single curve over certain
ranges of pressures. The equipment used in the tests limited the combi-
nations of (x^/ui and 0-3 to the values indicated. Higher pressure capacity
equipment is being designed which will allow the results to be extended
both in terms of 0-3 and o-2/o-i. It may be of interest that, for Indiana
limestone, an oolitic rock, there are generated distinct families of curves
closely paralleling the 0-3 = 0 curve for incremental values of as.
Mr. Hendron questions the validity of the use of elastic theory to
compute failure stresses. We are not satisfied, ourselves, with having to
calculate the stresses in this manner, but we are unable, at present, to do
otherwise and thereby to expect "more correct" answers.
The granite behaves in a quite brittle manner in that it fails suddenly
and after only small deviation from a linear axial-stress, axial-strain
curve. Also, the spalling which occurs on the inside face of the hollow
cylinders (Fig. 9 and Fig. 10) prior to a complete fracture is evidence
that failure initiates at the 0^3 surface.
Eff"orts are presently being expended to obtain load-strain data for the
inner and outer faces of the cylinders by the use of electrical resistance
strain gages.
Mr. Hendron lists as a "questionable feature" of the tests, the large
stress gradients across the wall of the hollow cylinders and states that
the calculated failure stresses only occur on the extreme outer or inner
fibers of the cylinder. When contrasted with the conventional triaxial
specimen, the nonuniformity of stresses is a definite drawback; however,
it appears that reproducible results can be obtained for comparable
stress states, and, in addition, it appears that comparable results can be
obtained when the relative orientation of the principal planes are inter-
changed.
This equipment and the testing techniques adopted are new in the

field of rock testing, and we are not advocating its widespread adoption
without considerable additional experimentation. There are, however,
many projects concerned with rock strength where information gained
by the use of this type of equipment could possibly be of significant
value.

A Loading System for the Investigation of

the Inelastic Properties of Geologic
Materials*

REFERENCE: H. R. Hardy, Jr., "A Loading System for the Investiga-

tion of the Inelastic Properties of Geologic Materials," Testing Tech-
niques for Rock Mechanics, ASTM STP 402, Am. Soc. Testing Mats.,
1966, p. 232.

ABSTRACT: This paper describes the theory and development of a

loading system for the laboratory investigation of the inelastic properties
of small rock and mineral specimens under a variety of loading modes,
including constant load, incremental loading, and constant rate of load-
ing. The application of basic feedback control theory to the design of
the control system is outlined, along with details of the construction
and testing of the completed system. The problem of obtaining uniform
loading of cylindrical test specimens under compressive load is discussed,
and details of a special specimen loading jig developed for this purpose
are described.

KEY WORDS: rock (material), minerals, rock mechanics, loading (force)

system, inelastic properties, feedback control

A knowledge of the inelastic behavior of geologic materials (rocks

and minerals) has become increasingly important in mining and civil
engineering, geology, geophysics, and more recently in a number of
defense and aerospace fields. Experimental determination of the param-
eters that may suitably describe inelastic behavior of these materials has
been, until recently, very limited. This has been due in large measure
to the numerous experimental difficulties associated with such studies.
The necessity for compressive loading of the test specimen has been the
basic cause of two of the major difficulties, namely: (a) developing and
controlling the high axial loads required and (b) obtaining a uniform
load distribution on the test specimen.

* This paper forms part of a Ph.D. thesis submitted to the Department of Engi-
neering Mechanics, Virginia Polytechnic Institute, Blackburg, Va., June, 1965.
1
Formerly, physicist, Physics Section, Fuels and Mining Practice Div., Mines
Branch, Department of Mines and Technical Surveys, Ottawa, Canada; presently,
associate professor, Department of Mining Engineering, Pennsylvania State Uni-
versity, University Park, Pa.
232

Copyright" 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
HARDY ON INELASTIC PROPERTIES OF GEOLOGIC MATERIALS 233

To date most of the experimental investigations of inelastic behavior

of geologic materials involving appreciable strains^ have utilized constant
load creep experiments (Griggs [1],^ Hardy [2], Matsushima [5], Phillips
[4], Price [5], etc.), although a few investigations under conditions of
constant strain rate have been conducted by Heard [6], Serdengecti and
Boozer [7], and others. A more general approach to the inelastic be-
havior of geologic materials, which involves the experimental determina-
tion of the associated mechanical model for the material, is being under-
taken at the Department of Mines and Technical Surveys (Hardy [8]).
To ensure that a unique model representation is obtained, it is necessary
to carry out deformation experiments under a number of different test
modes, for example, constant load, constant loading rate, etc. A flexible
and programmable load control system was developed to fulfill this re-
quirement.
The difficulties associated with compressive loading are further in-
creased when geologic materials are involved. These materials are usually
very weak in tension and, even under moderate confining stress, normally
fail as brittle materials. As a result, uniform loading is essential to ensure
that the material does not fail prematurely. Furthermore, if meaningful
strain values are to be determined from a finite number of attached
strain gages, it is necessary that the strain distribution throughout the
test specimen be as uniform as possible. To obtain initial alignment and
uniform loading a special specimen loading jig was developed.
The loading system, which includes both the load control system
and the specimen loading jig, will be described in this paper. The system
was employed initially, with considerable success, in the investigation
of the inelastic behavior of a typical marble, using incremental creep
experiments (Hardy [9]).
Load Control System
Programmable loading systems have been available commercially for
about ten years. However, the majority of these systems are relatively
expensive and, in general, are not suitable for the investigation of geo-
logic materials. The system finally developed by the writer (Hardy [9])
is based on the design of a constant load maintaining system described by
Hardy and Larocque [70]. It consists of a loading frame, incorporating
a single acting hydraulic ram with an attached load cell, and an associated
control system that monitors the ram thrust (or the specimen strain) and
adjusts the input pressure to the ram so that the ram thrust (or specimen
strain) varies in accordance with an established loading program. Such a

^ Numerous studies of the inelastic behavior of geologic and other similar ma-
terials have been conducted using vibratory and pulse techniques; however, the
strains developed in such experiments are extremely small.
' The italic numbers in brackets refer to the list of references appended to this
paper.

system is generally referred to as a feedback control system. The use of

compressed nitrogen gas as the pressure source in this system makes it
possible to maintain extremely accurate control of the loading conditions
and allows the system to operate continuously, even in the event of local
power failure.

SUMMING
NETWORK
INPUT c
CONTROLLER PROCESS
PROGRAMMER ^ V ^

-B

FEEDBACK

TRANSDUCER

FIG. 1.—Basic feedback control system.

PROCESS
PNEUMATIC OR

POWER
HYDRAULIC
SUPPLY
r TEST '
SPECIMEN ,

FEEDBACK TRANSDUCERS

FIG. 2—Simplified block diagram of control system for rock deformation ex-
periments.

Feedback Control Systems

The description and the analysis of feedback control systems are
presented in detail in many recent texts (Ahrendt and Savant [ii],
Truxal [72], and Thaler and Brown [73]). Figure 1 shows the arrange-
ment of the basic feedback control system. It consists of a process to be
controlled, a controller, a programmer, a feedback transducer, and a
summing network. The input programmer generates a signal, R, which
is designed to be an analog of the desired process state, and the feedback
transducer generates a signal, —B, proportional to the actual process

State. These two signals are applied to the summing network and the
error signal E = (R — B) applied to the input of the controller. In opera-
tion, the controller adjusts the process state so that the error signal, E,
to the controller is a minimum, and under optimum conditions the proc-
ess state will accurately follow the input program. Feedback control
systems have been developed that utilize pneumatic, hydraulic, mechani-
cal, and electrical signals. Using appropriate transducers, mixed systems
that incorporate a number of types of signals may be designed.
Figure 2 shows a block diagram illustrating the form of the feedback
system developed for carrying out rock deformation experiments. Here
the "process" consists of the loading of a test specimen under various
test modes, the controlled variables being specimen stress, o-, or strain,
e. By selection of the appropriate input program and feedback trans-
ducer the following test modes may be obtained:

1. Constant strain rate CER

2. Constant stress rate CSR
3. Constant strain CE
4. Constant stress CS
5. Incremental I
6. Cyclic CY

Depending on the time scale and the desired accuracy of control,

two types of controllers may be utilized. These are classed as "continu-
ous" and "on-off" types. In the continuous type controller the rate of
correction of the controlled variable is proportional to the error signal,
that is, the difference between the actual value and the required value
of the controlled variable. However, in the case of the on-ofi type con-
troller, only two states are possible, namely, fully off and fully on. The
controller is normally in the off state and can only be switched into the
on State if the error signal exceeds some definite level. Hence, the rate
of correction of the controlled variable is independent of the magnitude
of the error signal, and under this condition the actual value of the
controlled variable will tend to oscillate about the desired control point.
If rapid changes in the controlled variables, a- and e, are required, a
continuous type controller must be utilized, whereas if only slow changes
are involved a simpler on-off type may be used. Since it was decided
that initial deformation experiments would be conducted over rela-
tively long periods of time and would involve either fixed or slowly
varying controlled variables, it was decided that a pneumatic on-off type
control system would be the most practical.^ The loading system utilizing
this control system will be referred to for convenience as the Type-A

* Nikiforuk and Westlund [14] have recently published the analysis of a hydraulic
"on-off" type control system.

I
(J
o
< •-
LLI
UJ
0.

o
I—I

loading system. It should be noted that a Type-B loading system incorpo-

rating a hydraulic continuous type control system, to be used in experi-
ments at higher loading rates, is presently under initial development
(Hardy [9]).

Type-A Loading System

Operation of the Control System
Figure 3 shows a block diagram of the Type-A loading system. Since
the system has both stress and strain feedback available, it is capable of
operating in any of the test modes described earlier. Basically, the system
may be divided into a pneumatic-hydraulic section and an electronic
section. The former section consists of the pneumatic controller and its
associated regulator unit, the rapid load-rapid unload (RL-RUL) unit
containing the separator and its associated priming unit, and the load-
ing frame containing the hydraulic ram. The latter section consists of
the electronic controller, the control bridge, the programmer; and the
feedback transducers and associated circuitry.
To illustrate the operation of the system, let us consider that it is con-
nected in the constant stress (CS) mode^ and that the axial specimen
stress 0-23 (as measured by the load cell) is below the control point, azz{c).
Since (7,2 is below the control point, the pneumatic controller will be in
the "load-on" state which will allow gas to flow from the regulator unit
to the primary side of the separator. As a result, the oil pressure on the
secondary side of the separator will rise, increasing the pressure in the
hydraulic ram, and raising cz. The load cell, placed between the ram and
the test specimen, senses the magnitude of the applied load and provides
a feedback signal, —B. The summing network applies the signal, R-B,
which is proportional to <7„(c) — a^z, to the control bridge. The resulting
output voltage from the control bridge, AK, is fed to the electronic con-
troller, which in turn operates the pneumatic controller. Now as a^z ap-
proaches (Jzz{c), AF tends to zero and the electronic controller switches
the pneumatic controller into the off" state, shutting off" the flow of gas
to the separator. If Ozzic) is exceeded, the resulting feedback voltage
triggers the pneumatic controller into the "unload-on" state, which al-
lows gas to bleed out of the separator, reducing o-^^ to the control point,
<Tzz{c).
Operation in the various test modes is similar. For example, if a con-
trolled rate of straining is required, strain feedback is employed, and the
programmer provides a signal corresponding to the required strain
versus time function.

^ Note that in the CS mode the control mode selector will be set for stress feed-
back and the programmer output remains constant for a particular value of stress.
The zero balancing unit, marked ZB-1 in the stress feedback line, enables the output
of the load cell to be set to zero when the applied load is zero.

I
_ 3
a.

Pneumatic-Hydraulic Section
Pneumatic Controller—Figure 4 shows the circuit of this unit. It con-
tains two solenoid valves," the loading valve, V-5, and the unloading
valve, V-8, which control the gas pressure on the primary side of the
separator. These valves are activated by electrical signals from the elec-
tronic controller. Micrometer-type metering valves,' V-6 and V-10, in
series with the solenoid valves allow the gas flow, during loading and un-
loading, to be finely adjusted to obtain optimum control. For some oper-
ations it is desirable to have full gas flow, and for these cases by-pass
valves, V-7 and V-9, are included. The pneumatic control pressure, Pc,
which corresponds to the gas pressure on the primary side of the sepa-
rator, is monitored by a large 0 to 3000-psi gage included in the control-
ler. Figure 5 shows two views of the completed controller.
Three control modes, auto, manual, and remote, may be selected.
Normal operation is in the auto mode, in which case the loading and
unloading solenoid valves are energized directly by the electronic con-
troller. During certain experimental operations it is necessary to operate
the load or the unload valves manually. This is accomplished by putting
the controller in the manual mode and activating the loading or unload-
ing valves using switches located on the pneumatic controller. In this
mode power to operate the solenoids is obtained from the external stor-
age battery.
The operation of the solenoid valves is interlocked with other parts
of the control system so that in the event of certain system failures the
controller will "fail safe." If the interlock signal is lost, the unload sole-
noid valve will automatically be energized, and the loading solenoid
valve will be prevented from operating. A switch on the controller pro-
vides a temporary override of the interlock to allow various adjustments
to be made to the system.
Associated with the pneumatic controller is a regulator unit, which is
shown in Fig. 4. It consists of a highly stable spring-type regulator,* a
gage to indicate outlet pressure, and a number of valves. This regulator
suppUes nitrogen gas at a fixed pressure, Pm , to the input of the pneu-
matic controller. It serves two purposes. First, it provides a means for
adjusting to a minimum the differential pressure AP = (Pin ~ Pc) across
Valves V-5 and V-6. This is a requirement for optimum operation of the
control system. Secondly, it limits the maximum ram pressure in the
event of a malfunction of the pneumatic controller.
Rapid Load-Rapid Unload Unit—The pneumatic circuit of this unit
is presented in Fig. 4. The unit was incorporated in the control system

"Asco, Type 8262-99 solenoid valves, rated at 3000 psi, and fitted with 6-v d-c
solenoid coils.
' Hoke, Type 2RB280 metering valves with attached vernier handwheels.
* Grove, small volume, high pressure, combination reducing and relief regulator,
Model 15 LH.

(a) Front view.

(b) Rear view.
FIG. 5—Pneumatic controller.

to serve three purposes: (1) to provide a separation between the hy-

drauhc fluid in the ram and the nitrogen control gas; (2) to facihtate
slow, positive motion of the ram during alignment and preloading of the
test specimen; and (3) to provide an arrangement whereby rapid loading
and unloading of the test specimen could be carried out. The role of this
unit in rapid loading and unloading experiments will be discussed only
briefly here. The unit consists of: a separator;' a ballast tank; two sole-
noid valves/" V-13 for rapid loading and V-14 for rapid unloading; a
by-pass valve, V-12; and a metering valve, V-11. The hydraulic circuit
of the priming unit which is associated with the RL-RUL unit is also
shown in Fig. 4. It consists of a small hydraulic hand pump," a reservoir
for hydraulic fluid, a series of valves, and a pressure gage, Ph, which
indicates the pressure of the hydraulic oil on the secondary side of the
separator.
For normal operation the solenoid valves remain in the normally
closed state, and the by-pass valve, V-12, is opened. Any increase in gas
pressure on the primary side of the separator results in an increase in the
oil pressure on the secondary side and a resulting increase in ram thrust.
During the initial stage of a typical deformation experiment the gas
pressure in the separator is set to some low value (approximately 50 psi),
and the ram piston is slowly lowered into contact with the specimen load-
ing jig using the hand pump in the priming unit. When contact has been
made, Valve V-16 is closed, isolating the RL-RUL unit from the priming
unit. A ballast cylinder having an internal volume of approximately 35
in.' is included in the circuit to improve the load stability of the system
when it is operating in the constant stress or constant strain mode. Its
effect on the system may be increased or decreased by adjustment of the
metering valve, V-11.
For rapid load operation Valve V-12 is shut off, Valve V-13 is opened
remotely, and a low value of preload is applied to the system, as just
described for normal operation. Once the specimen has been preloaded,
Valve V-13 is shut off, and the gas pressure on the primary side of
the separator is increased to a specific level. Valve V-5, in the pneu-
matic controller, is then shut off, and the control system is clamped in an
inactive mode. To initiate rapid loading. Valve V-13 is remotely opened
causing the ram pressure to suddenly rise and come to equilibrium with
the pressure on the secondary of the separator. When the specimen stress
just reaches the control point, the control system is undamped, and nor-
mal operation of the system is re-established. The operation of the RL-

" Sprague, Model 60010 accumulator with a volume of approximately 25 in.^ and
rated at 4000 psi.
""Hoke solenoid valves. Type B90D-120C rated at 2000 psi and fitted with 6-v
d-c solenoid coils.
" Blackhawk hand pump, Model P-39.

RUL system is controlled by a number of additional electronic units

that will not be described in this paper.
Axial Loading Frame—The axial loading frame consists of upper and
lower steel platens (ISJ^ by 1 5 ^ by m in. thick), which are tied to-

FIG. 6—Front view of axial loading frame.

gether by four heavy steel rods (l^^^ in. diameter). These rods are threaded
at both ends to provide adjustment of the working space between the
platens and to allow the platens to be adjusted perfectly parallel. A modi-
fied Blackhawk 50-ton hydraulic ram is threaded into the upper platen
and locked into position by upper and lower locking rings. A strain gage

type load cell is attached to the lower end of the ram piston to monitor
the thrust. Figure 6 is a photograph of one of the axial loading frames.
A specimen loading jig is shown in position between the load cell and
the lower platen. The associated RL-RUL unit appears near the top of
the photograph, above the ram.

Electronic Section
Electronic Controller—The electronic controller performs a number of
important functions. First, it accepts the low-level unbalance signals,
AK, from the control bridge and converts these to high-level signals for
operation of the loading and unloading solenoids in the pneumatic con-

FROM
G- S. C.
GALVflNOMETEB
BRIDGE CIRCUIT
- PHOTOCELL
UNIT

'
LOGIC TRANSISTOR
CIRCUITS AMPLIF lERS

PNEUMATIC
CONTROLLER

FIG. 7—Block diagram of electronic controller.

troller. In this respect it performs the function of an amplifier. Secondly,

it contains a number of logic circuits that ensure that correct sequencing
of the loading and unloading valves is maintained, for example, the
loading valve must be closed before the unloading valve can be opened.
Thirdly, additional logic circuits are included so that the controller will
fail safe, for example, failure of a critical section of the controller will
automatically put the controller into the neutral position, where both
loading and unloading valves will be closed.
Figure 7 shows a block diagram of the controller. The low level output
signal from the control bridge enters the controller via the galvanometer
signal conditioning circuit (GSC). The output of this circuit is connected
to the coil of a light-spot type galvanometer in the galvanometer-photo-
cell unit (GPU). The latter unit allows the polarity of the galvanometer
input signal to be selected, provides a manual galvanometer sensitivity

control, and includes circuitry for manually or automatically shunting

the galvanometer system.
Details of the main section of the controller are discussed by Hardy
[9], and Hardy and Larocque [70]; hence, its design and operation will
be outlined only briefly here. The bridge unbalance signal, AF, causes the
light-spot galvanometer in the GPU to deflect an angle a, which is

sp_ O ifi^ ^^ ^

(a) Front view.

(b) Top view.
FIG. 8—Photographs showing the completed electronic controller.

proportional to the magnitude and the sign of the unbalance. For ex-
ample, if the measured signal load were above the control point, the un-
balance signal would be positive in sign, and the galvanometer would de-
flect in a counterclockwise direction. A set of three photocells sense the
position of the hght spot. If the control bridge unbalance continues to
increase to +AVff , where the angle of deflection becomes a = -1-/3, then
the high-limit photocell will be illuminated generating an output signal.
This signal is amplified and in turn closes a relay which energizes the un-

loading solenoid valve in the pneumatic controller. As the load begins to

decrease, AV decreases, and when a < P the unloading valve is turned
off. If however the load decreases too rapidly, it may overshoot the con-
trol point, AV will become negative in sign, and the galvanometer will
deflect clockwise. If it reaches a level of — Vp where a = —/3, the low-
limit photocell is illuminated; this will result in a relay closing, and the
loading solenoid valve in the pneumatic controller will open.
In effect the controller is a highly sensitive relay circuit. For
AV ^ \AV^ I,

either the loading or unloading solenoid valve is open depending on the

sign of AV. For AV < | AF^ | both valves are closed. The controller
therefore has a dead band of 2 AF^j, which places a limit on the control
accuracy of the over-all system. The relative importance of the magnitude
of the dead band depends on the gain in the feedback loop and the con-
trol bridge. For example, consider that the system is operating in the
constant stress (CS) mode, and that a change in specimen stress Ao-^
results in an unbalance signal A V^ at the input of the electronic control-
ler. If it is desired to control the stress to within ACA of a given value,
then it is necessary that AVA ^ AVg. Since AF^ = GKAa, where G is the
gain of the control bridge and K is the sensitivity of the feedback trans-
ducer in volts per psi (stress), the importance of the dead band may be
reduced by having large values of G and K.
Figure 8 shows a number of views of the completed controller. In par-
ticular Figure 86 shows a top view of the unit including the inside of the
galvanometer-photocell unit.
Control Bridge—The control bridge serves two purposes: (1) it sup-
plies an excitation voltage to the bridge circuits in the feedback trans-
ducers, and (2) it amplifies the low-level unbalance signals from these
transducers to a higher level for activation of the electronic controller.
The design and merits of various types of control bridges are described
in detail by Hentenyi [15], The control bridge used was a Baldwin Type-K
strain indicator modified for use with external four-arm bridge trans-
ducers. This was mounted in a standard relay rack panel. In order to
stabilize the bridge from short-term temperature variations, the rear of
the bridge was enclosed in a thick-walled aluminum box, which was in-
sulated by a layer of glass wool and enclosed by an outer steel case at-
tached to the rack panel. Some sections of the indicator were rewired to
enable all electrical connections to be made at the rear of the unit. A
signal denoted as the control bridge unbalance, AF, was obtained by a
parallel connection across the existing panel galvanometer in the indi-
cator. This was brought by shielded cable to a connector at the rear of
the unit.
In its original form, the Type-K indicator operated from a set of self-

A e (UNBALANCE)
-O
CONTROL
BRIDGE
TO
-O
PROGRAMMER ELECTRONIC
e (EXCITATION)
CONTROLLER
-O

CONTROL
MODE ~ I I I I
SELECTOR

ZERO Fi I I i i
BALANCE
UNIT

TRANSDUCER
BRIDGE

Z§®n

CS/CE CSR/CER/CY
FIG. 9—(a) Relationship of the programmer to the feedback loop, (b) pro-
gramming arrangement utilizing the "one-arm" method, and (c) programming ar-
rangement for CSR mode utilizing the "two-arm" method.

contained batteries. Since it was necessary that the bridge operate con-
tinuously for long periods of time, these batteries were replaced by a
regulated power supply operated from the 110 v a-c lines.
Method of Programming—The relationship between the controller,
programmer, and feedback sections is somewhat unconventional in this
system. The programmer does not produce an independent signal that is
summed with a feedback signal to give an error signal. Instead, the pro-
grammer, which is basically a variable resistance network, is connected
directly to the feedback transducer bridge. For example, one type of pro-
gramming, termed the "one-arm" method, is illustrated in Fig. 9a.
Here the program resistance, Ry, is connected across one arm of the
feedback transducer bridge. A variation in Rr will change the balance

condition of the transducer bridge and will result in a shift in the electrical
zero of the feedback transducer. The error signal, Ae, will thus be pro-
portional to the diflference in the effect on the transducer bridge due to
measured variable and the unbalancing effect due to the programming
resistor. This type of system is simple and relatively convenient for those
types of programming where Ryis either constant or varies incrementally,
for example, constant stress, constant strain, incremental loading, etc.
However, for one-arm programming, Ae is a nonhnear function of Rr ,'^
and programming of continuous variations in the controlled variable,
for example, constant rate of stressing, would require the use of a resist-
ance that varied nonlinearly'^ with time.
Figure 9b shows some of the programming arrangements that have
been investigated. In the constant stress (CS) or constant strain (CE)
mode, the stress or strain level may be set manually by the use of a ten-
turn potentiometer for Rv . If it is desired to operate in the incremental
(I) mode, that is, to vary the stress or strain over a set of fixed levels, the
simple switching circuit shown may be used, where the values of the re-
sistors are selected to provide the desired stress or strain levels. Finally,
operation in the constant stress rate (CSR), the constant strain rate
(CER), or the cyclic (CY) mode may be accomplished by using a suita-
ble nonlinear potentiometer driven by a small constant speed motor.
The one-arm method for continuous programming of the controlled
variable, although technically feasible, was not used because of the ob-
vious electrical and mechanical difficulties involved. Instead, the "two-
arm" method, shown in Fig. 9c, was employed. For a restricted range of
RA and RB it may be shown that: (a) the error signal, Ae, is a
linear function of the angle of rotation of the potentiometer RA , and
(b) for a fixed rate of rotation the rate of unbalance will be a function of
the value of iJ^ . A number of different programming modules have been
developed to date. In particular these include modules for conducting
constant-load creep experiments (CS mode), incremental creep experi-
ments (IS mode), and constant rate of loading experiments (CSR mode).
Since the IS-mode programmer is the one that has been used most to-
date, its design and operation will be briefly discussed here as an exam-
ple.
IS-Mode Programmer—The IS-mode programmer, which utilizes the
one-arm programming method, consists of a galvanometer damping
circuit, a valve control circuit, and a program circuit. A function switch
co-ordinates the operation of the unit. In Position 1 the system is inac-
tive; in Position 2 the system is active (or set) with one of the program-

" See Hetenyi [15] for the standard balance condition equations for a four-arm
Wheatstone bridge.
" Nonlinearity may be obtained either by special nonlinear resistive networks
(Chance et al [16], Greenwood et al [17], and Korn [IS]), or by linear potentiome-
ters driven by nonlinear actuators (Truxal [12]).

ming resistors connected across one of the arms of the feedback trans-
ducer bridge. Under these conditions the control system maintains the
load constant at a value determined by the particular programming re-
sistor selected.
To increase the control stress to the next higher level the following
steps are carried out: (a) the function switch is placed in Position 3,

»s«»iMMT*L LOAOtNt mauit t

(a) Front view of IS module.

(6) Side view of IS module.
(c) Plug-in program card showing program resistors.
(d) Plug-in program card showing trimming resistors.
F I G . 10—Photograplis of the IS mode programmer and the plug-in program
cards.

which damps the control galvanometer in the electronic control unit;

(b) a different programming resistor is selected and connected into the
circuit; (c) the function switch is placed in Position 4, and a push button
switch on the programmer is used to open the loading solenoid valve in
the pneumatic controller allowing the stress to rise; and (d) observation
of the panel galvanometer on the control bridge indicates when the sys-
tem comes into balance at the new level and at that time the function
switch is placed back in Position 2. The control system will now main-

tain the load constant at the new level. Figure 10 shows two views of the
I-mode programmer module.
The program or calibration resistors each consist of an approximate
resistance in series with a trimming potentiometer. Final adjustment of
the particular values are made, using a calibration chart and with the
controller connected into the actual control system. For convenience,
each set of program resistors and their associated trimming potenti-
ometers are mounted on a plug-in card, as shown in Fig. 10. Various
loading programs covering different loading ranges have been designed.
Feedback Circuit—The feedback circuit consists of a stress (load)
transducer, a strain transducer, two zero balance units for zero adjust-
ment of the two transducers, and a control mode selector which allows
either stress or strain feedback to be selected.
Stress feedback is obtained by a Baldwin Type-C (strain gage-type)
load cell with a capacity of 20,000 lb, attached to the piston of the hy-
draulic ram. The load cell employed is a double bridge type, which
provides two separate outputs. One is used to provide the feedback signal
in the control system, the other is fed to the measurement system to enable
the applied specimen load to be accurately monitored. These cells are
calibrated to give an output of 2 mv per volt excitation at an applied
load equivalent to their capacity. The 20,000-lb cell employed has a
sensitivity K = 0.6 ix v/lb with an excitation voltage of 6 v.
Strain feedback is obtained from a single resistance type strain gage
attached to the test specimen. This gage and a similar dummy gage form
two arms of a four-arm bridge arrangement, the output of which provides
the strain feedback signal. It is planned to modify this bridge arrangement
so that two arms of the bridge are strain sensitive. In the new configura-
tion the strain sensitivity of the transducer will be increased by a factor
of approximately two.

Behavior of Completed Load Control System

To date the loading system has been used in the CS, IS, and CSR
modes. A number of studies have been carried out to optimize the sys-
tem and to check on its performance in these operating modes. It should
be noted that all these studies were conducted in a special test facility
in which the temperature was maintained constant at approximately
22.5 ± 0.2 C.
Load Cell Zero Drift—Using stress (load) feedback the stability of the
control system in all modes will depend on the stability of the load cell.
Since no sufficiently accurate experimental method could be conceived
to determine this stability during actual loading conditions, indirect
means were adopted to estimate its magnitude. These consisted of: {a)
studying the zero drift of the load cell over a period of time under no-
load conditions and (Jb) investigating the zero drift produced after a

period of sustained loading. It should be noted that all results quoted

refer to a 20,000-lb, Baldwin, Type-C load cell. The zero drift under no-
load conditions (for periods of time up to 48 hr) was found to vary
randomly with a mean deviation that was approximately constant. The
maximum limits of this variation were equivalent to approximately
±8.5-lb ram thrust. The values of zero drift, observed after the apphca-
tion of various sustained loads (up to 3500 lb for 5 hr), varied between
approximately ± 1 3 lb. The variations observed in both cases are within
the experimental error of the load measurement system, namely, ± 1 7 lb.

LOAD

TIME

^MARKER
^PULSES N ,

MEAN

1^-50 SECH

(a) Typical load versus time variation (no drift).

(6) Section of a typical load versus time record.
FIG. 11- -Behavior of control system under optimum control conditions.

Optimization of the Pneumatic Controller—The pneumatic controller.

Fig. 4, contains a number of parameters that must be adjusted to provide
optimum control. These are the settings of the loading and unloading
micrometer valves (V-6 and V-10), the differential pressure
(AF = Pi Pc)
across the loading valve (V-6), and the ballast volume. Since the system
is basically quasi-static in all modes,'^ that is, the control point, Lc, is

" Although a rapid change in ram thrust is introduced initially in the CS mode
and similarly at a number of intervals in the IS mode, these operations do not di-
rectly involve the pneumatic controller.

either fixed (CS and IS modes) or varies only slowly with time (CSR
mode), the optimum settings of the various parameters are independent
of the operating mode. In all modes the output (ram thrust) varies in a
cyclic manner with time. For example, Fig. 11a shows this typical be-
havior for the case of CS-mode operation. The load fluctuation, AL{F),
is defined as one half the peak value, that is, AL(F) = (Lmax — I'min)/2.
The problem of optimization is one of adjusting the controller parame-
ters so that the amplitude of this cyclic fluctuation is a minimum. The
efi"ects of the various parameters were investigated in considerable de-
tail. A brief review of these investigations indicates the following:
(a) The smaller the value of AP the less sensitive the control system is
to the setting, LV, of the loading valve.'^ The lowest possible AP is
therefore desirable. However, when using low values of AP care must be
taken to ensure that the regulator that sets AP is sufficiently stable to
maintain the desired value.
(b) The lowest load fluctuations are attained with the ballast cylinder
in the system.
TABLE 1—Optimum control conditions.
Ballast IN
Ballast volume 35 in.'
AP 100 psi
LV 4.0
ULV 4.0

(c) In general, the lowest load fluctuations are obtained with the
lowest values of LV and ULV. The lower limit of LK is determined by the
natural leakage of the system. The response time of the system is also a
function of LK and ULV. If these settings are too low, the system will
require an undesirably long time to come back into a stable condition, if
unbalanced significantly from the control point by a random disturb-
ance." In general, an LV and ULV of 4.0 has been found to be the most
satisfactory.
{d) It has been generally found that the percentage load fluctuation
decreases with increasing load level. It is felt that this is due in part to
the fact that frictional effects in the hydraulic ram and the separator be-
come less important at higher load levels and that the over-all control
system is only sensitive to load fluctuations greater than ±AL((3)." The

" The settings of the loading and unloading micrometer valves are abbreviated
as LV and ULV, respectively. For example, the statement that LV — 4.0 means
that the loading micrometer valve is set at 4.0. Each division corresponds to a
valve rotation of 36 deg.
" This condition is particularly undesirable when incremental loading is being
carried out.
" A load flutcuation of ±AL(/3) is just sufHcient to cause the control galvanome-
ter in the electronic controller to deflect an angle ±j3, hence activating one of the
control photocells.

value of AL(/3) depends on the design of the galvanometer-photocell

unit in the electronic controller, the sensitivity of the load cell, and the
over-all gain of the system. It is, therefore, a constant independent of
load level and, therefore, its effect on the percentage load fluctuation
will be of less importance at the higher load levels.
Based on these investigations the optimum conditions listed in Table 1
are normally employed. For these conditions the load fluctuation, AL(F),
appears to be less than 10 lb. Figure 1 lb shows a section of a typical load
versus time record illustrating the behavior of the system under optimum
control conditions.

Behavior in Various Test Modes

CS Mode—The CS mode of operation is employed for conducting
constant load creep experiments. For this purpose it is necessary that the
LOAD

TIME
FIG. 12—Typical load versus time variation with linear drift.

ram thrust remain as nearly constant with time as possible. This requires
that both the load fluctuation, AL{F), and the load drift, AL{D), be small.
In the CS and IS modes the control point, Lc, for convenience is de-
fined as the average load level based on the peak values, namely.

The load drift, AL{D), of the control system is defined as the variation
of the average load level, Lc, with time, namely, AL{D) = L^ (final) —
Lc (initial). Figure 12 illustrates the case of a linear load drift.
Load drift data from a number of deformation experiments conducted
at difi'erent load levels (up to 6000 lb) and for various periods of time
(up to 24 hr) indicated that the magnitude and direction of the load drift
did not appear to be dependent on the load-level or on the duration of
the experiment.^' In general AL{D) varied in a cyclic manner within the
limits of load cell drift ( ± 1 3 lb) quoted earlier. The over-all load varia-
tion (including fluctuation and drift) of the system appears to be in the

^^ It may be significant, however, tiiat the largest load drift observed was asso-
ciated with the longest test duration. Further experiments are presently being con-
ducted to check on the drift over long periods of time.

LOAD,i

T , . - SHORT
A L ( o ) - T O O LARGE
Ts - TOO LONG

TIME

LONG
AL(o) -SMALL
T j - SHORT

C TIME
(a) Load versus time incremental loading.
(6) Fast loading rate,
(c) Slow loading rate.
F I G . 13—Load versus time variations which occur during incremental loading.

order of ±0.2 per cent for loads on the order of 5000 lb. This is an order
of magnitude better than that obtained in an earlier loading system which
utilized spring- and ballast-type gas regulator units to control the ram
pressure (Hardy [2]).
IS Mode—The IS mode of operation makes it possible: (a) to main-
tain the load constant for a period of time at any one of a number of

preset levels and (b) to raise or lower the load rapidly from one level to
another when required. To date this mode has been mainly employed in
incremental creep studies, where it is necessary to carry out a number of
experiments on the same specimen at a preselected set of load levels. To
obtain significant data from such experiments, it is necessary that the
shift from one load level, Lc{l), to the next, Lc{2), be carried out as
rapidly as possible.
Figure 13 illustrates the form of the load versus time disturbance pro-
duced during such an incremental load change. Three important parame-
ters (see Fig. 13a) are associated with this type of operation, namely, the
loading time, TL ; the setthng time; Ts ; and the load overshoot, AL(0).
Optimum operation in this mode requires that TL be as small as possible
while still maintaining Te and AL(0) within certain limits. If an attempt
is made to obtain very low values of TL (which requires the control pres-
sure to be raised very rapidly), it is found that both AL(0) and T^ be-

TABLE 2—Operating conditions, IS Mode.

Factor Units Range

AL lb 100 to 500"
Lc lb 2000 to 6000«
TL sec 4 to 6
AL(0) lb 4 to 20
T, sec 100 to 300
" Behavior outside this range has not been investigated in detail.

come excessive (Fig. 136). If an attempt is made to maintain AZ,(0) and

Ts at too low a value, TL will necessarily be excessively large (Fig. 13c).
The results of a large number of experiments have indicated that with
care it is possible to carry out incremental loading within the range of
conditions listed in Table 2.
The reproducibility of the various load levels was investigated using
the load data obtained from two separate experiments. The average
error was found to be ± 1 0 lb, which was well within the possible error,
± 1 3 lb, due to load cell drift. The over-all variation of load with time
at any of the preset levels is consistent with that observed using the CS
mode.
CSR Mode—In this mode the programmer causes the load control
point, Lc , to vary linearly with time. It has been found that the Type-A
loading system operating in this mode is particularly suitable for conduct-
ing constant rate of loading experiments at low loading rates. Two fac-
tors are important in this mode of operation, namely: (a) tracking—the
degree of correspondence, over relatively long periods of time,i* between

" In this case it is necessary that t » l/fs where /*• is the on-off frequency asso-
ciated with the normal operation of the control system.

the observed and the specified load versus time curves and (b) fluctuation
—the short period load variations, AL(F), associated with the normal
on-off action of the control system.
For slow rates of loading the rate of variation of Lc is small and, hence,
the magnitude of AL{F) should be consistent with that observed during
operation in the CS mode, namely, AL{F) ^ 10 lb. In the CSR mode
AL{F) is one of the factors that sets a limit on the minimum rate of load-
ing which may be achieved by the control system. The tracking quality
of the system depends on a number of factors, including the linearity of
the programming circuit^" and the magnitude of the load drift, AL{D).
Since, in the present system, AL{D) has been shown to be small, its ef-
fect on tracking would only be important at very low loading rates. Only
initial experiments have been carried out using this operating mode.
These have indicated satisfactory behavior of the control system for
loading rates in the range of 0.1 to 10 lb/sec. Due to the on-off nature of
the control system, it is felt that a loading rate in the range of 10 lb/sec
represents a practical upper limit. The minimum attainable loading rate
depends on a number of factors, including, in general, AL{F) and AL{D),
and experiments are presently being conducted to determine this rate.

Loading Jig
Initial Studies
In previous experiments by the writer an attempt was made to obtain
uniform loading by using accurately prepared test specimens mounted
in some form of loading jig or subpress to maintain specimen alignment
during deformation. In these earlier experiments axial strain was meas-
ured by two resistance strain gages mounted parallel to the specimen
axis at a spacing of 180 deg and connected in series to give an output
equivalent to the average strain. Such a strain measuring arrangement is
convenient in some respects, but it provides no indication of whether
bending is taking place in the specimen.
In recent experiments^^ (Hardy [9]), strain measurements were made
by three attached gages at a spacing of 120 deg. All three gages were ob-
served separately, making it possible to evaluate the uniformity of load-
ing produced by different loading arrangements. Preliminary experi-
ments by the writer indicated that compressive loading of cylindrical
specimens between rigid platens often results in an extremely nonuniform
strain distribution in the specimen. In many cases the bending was so

'° Using the two-arm programming method the over-all linearity mainly depends
on the linearity of the programming potentiometer.
^ Experiments using the present loading system have utilized right-circular
cylindrical specimens nominally 114 in. in length and Vt in. in diameter. The end
surface of these specimens were ground plane parallel to ±0.0005 in. and their di-
ameter ground to within ±0.001 in. over the length of the specimen.

•O U D si
t/5 c
-a
' - j 1/3 .s o
60 M 2 K ftj
—' "^ a" u- i ^ •t^
bO 60 « £ r
c c ti o
•-5 •3 ''S "5 o
Ofl
o o 2 -2 (U
hJ i-l h h W •^ Ci/l

a <i o Ts «

4
d

STRESS EQUALIZERS
( T EFLON INSERT )

3/8 X 0.750 + 0.001

ALL SURFACES
GROUND FINISH

FIG. 15—Details of loading heads with deformable inserts.

large that one and sometimes two of the gages indicated large tensile
strains.
An experimental program (Hardy [79]) was established to investigate
the uniformity of loading obtained with various types of loading jig
arrangements. In these experiments a ground steel cylinder 0.75 in. in
diameter and 1.50 in. long with three attached axial strain gages (120
deg apart) replaced the test specimen in the loading jigs. The jigs were
loaded over a range of stress levels using the loading system described
earlier. At each level the strains at the three gages were recorded sepa-
rately, and the average strain e^, the maximum deviation between any
two gages Ae, and the per cent deviation (Ae/ea) X 100 were determined.

ill.
m
(a) Loading jig assembled.
I

{b) Component parts of loading jig.

1. Alignment cylinder.
2. Lower jig block.
3. Upper jig block.
4. Loading head containing Teflon insert.
5. Loading head.
6. Teflon insert.
FIG. 16—Photographs showing details of loading jig.

Figure 14 illustrates the various types of loading jigs and loading heads
that were investigated.
Loading Jig with Deformable Loading Head Inserts
The only arrangement investigated that provided satisfactory uni-
formity of loading was the Type-2 jig fitted with loading heads containing

7 1 1 T - 1 1 r 1 1 1 1—
600

TEST 64-11
500 - RUN 12
1. 2 , 3 -

400 -
a
300 -

200 y^ -

100 -
n . / r 1 ^ 1 i / 1 1 L
0(1) 0(2) 0(3)
COMPRESSIVE STRESS
(EACH INCREMENT= 2 0 0 0 P S I )

600

(/> 500
=1-
1 400
z
<
n- 300
1-
(O
1 200
TEST 64-11
< O-RUN II
X
< 100 • -RUN 12

10000 20000
COMPRESSIVE STRESS
(a) Individual stress-strain curves.
(/)) Average stress-strain curves and maximum deviations.
FIG. 17—Curves indicaling uniformity of loading obtained on steel specimens.

deformable Teflon inserts. This developed as a modification of the

hydraulic loading arrangement described by Handin at the Second
Canadian Rock Mechanics Symposium (Paquin [20]). Figure 15 gives
further details of this loading arrangement, and a number of views of the
completed unit are shown in Fig. 16. Two experiments using the steel
specimen were carried out using this arrangement. The results for the
individual gages from Test 64-11(12) are shown in Fig. 17a, and the
curves of average strain, e^, and the deviation, Ae, against stress for
both experiments are shown in Fig. 17^?. The curves for the individual
gages, as well as the average, were found to be linear and passed through
zero. The maximum deviation, Ae, up to a stress level of 25,000 psi

1 1 1 1 1 1 I I
1400 ,
SPECIMEN WO-3
1200 '/ V 3/
•

a. 1000 -
1
800 a
z
<
cr
1-
600 -
05

400
<
X
< 200 - / -
* 1 t< 1 1 1 1 ' '
0(1) 0(2) 0(3)
COMPRESSIVE STRESS
(EACH INCREMENT = 2 0 0 0 P S I )

=1. 100
50
0

1600

1200
<
a:
800 -
SPECIMEN WO-3
- 400 0 - L A R G E TIME
DEPENUENT
< STRAINS NOTED
_i_
4000 8000 12000 16000
COMPRESSIVE STRESS P SI
(a) Individual stress-strain curves.
(6) Average stress-strain curves and maximum deviations.
FIG. 18—Curves indicating uniformity of loading obtained on marble specimens.

was only 10 juSj^^ gsiA this appeared to be independent of stress level.

It is also important to note the very close agreement between the two
tests. Since the loading jig was completely disassembled after Test 64-
11(11) and reassembled for the following test, this agreement indicates
the excellent reproducibility and uniformity of the loading technique.
A series of experiments was later conducted on a typical marble^^
^For brevity 'Vs" is written for "microstrain," representing a strain of 10"°.
^Wombeyan marble is macroscopically homogeneous and isotropic, fairly
coarse-grained (average grain size about 1.0 mm in diameter) and of high purity.

~m—I—rn—r I I I I I I I I I I I II:,

>J (ft o m
in in J, en
CO in <D <r o
a.
O o
•?•
en _' O CM OJ CO

ol s- o OJ
z CO ffi • -

UJ
•o
fi, *+-,
K
u ^ ^;^
n Cl.
i:
fO rn
v>
OJ
N.
in
(D
^1) oS
^ o
o!- w
00 00 •—»s
N
a •c:
ot
- f O)

II
OJ rO in

u
o Q
o z
C/1
Is
o o ~ 2
o
cvi o
2^
"K
UJ
en §1
SO

5Q
3 ^

"iJ ?u

s to .ca
c -S
Ci. ^
^^sak. ao
^ S "5
1s
1—4 c
c
^S
-Q
d
I—<
5
O
b 0 •S
*+-»

_I_L

Nivdis nvixv

- 1 1 1 1 1 1 1 1 1 1 -
10

° ^ - ^ ^
-

in
8
-
a. o
.-r''^^
o

< 6
a:
(-to
-J
<
- -
< 4
/ SPECIMEN WO-5

' 10.1-10 PSI

A» 420 PSI

"/ 0 EXPERIMENTAL
Z • THEORETICAL

y s 0-37^ S

-
1 1 1 1 1 I 1 1 1 1

TIME - SECONDS

FIG. 20—Experimental results from a typical incremental creep experiment

{Wombeyan marble, Specimen WO-5) at "low" stress.

1 1 1 1 1 1 1 1 I 1 1
140

100

to
i
^^ -
Z .r^
CC 80
1-

< SPECIMEN WO-5 -

< tr =12,453 PSI

Aer = 4 4 8 PSI
40
~ ^ 0 - EXPERIMENTAL
• - THEORETICAL
S ' 1.8 ^ S

f 1 1 1 1 1 1 1 1 1 1 1

TIME SECONDS

FIG. 21—Experimental results from a typical incremental creep experiment

{Wombeyan marble. Specimen WO-5) at "high" stress.

using this jig. Figure 18 illustrates the results obtained in these experi-
ments. It was found that, although the apparent load uniformity was
not as good as that obtained earlier using the steel specimen, the per
cent deviation rarely exceeded 10 per cent. This lower load uniformity
observed is probably due, however, in part to the fact that the geo-
logical material has a larger grain size and is not as homogeneous as the
steel specimen.
A secondary problem that arises when using a loading jig, which
introduces forced alignment, is that of friction between the upper moving
piston and the walls of the alignment cylinder. This effect was initially
considered to be small and was neglected. When refined deformation
experiments were carried out, however, it was found that the loading
jig should be as free of friction as possible. This was accomplished by
modifying the loading procedure^* so that the alignment cylinder was
raised free of the lower jig block and supported from the upper block
during the critical phases of the deformation experiment. A new form of
jig, which will be experimentally more convenient, has been designed
and is presently under construction.

Conclusions
The loading system described in this paper has been in operation now
for some twelve months. The ability to select both operating mode and
range using the various programming modules and plug-in programs
has resulted in an extremely versatile system.
In the incremental loading (IS) mode it is found that incremental
loads of 100 to 500 lb may be applied between preset levels in a time
of less than 6 sec. The reproducibility of the various levels is found to be
within ±10 lb. The over-all variation of load with time at the various
levels, over a typical 24-hr test period, is found to be approximately
±0.2 per cent. Figures 19-21 illustrate graphically the results of a typi-
cal set of incremental creep experiments. Figure 19 illustrates diagra-
matically the over-all strain versus time curve for the complete set of
incremental loadings. Figures 20 and 21 are individual incremental
creep curves for two values of the applied stress.
Utilizing the specimen loading jig, incorporating Teflon insert-type
loading heads, it is possible to load test specimens extremely uniformly.

^The modified procedure is carried out as follows: (a) the specimen is placed
in the assembled loading jig and the jig located in position in the loading frame,
(Z>) a small preload is applied to the specimen to ensure alignment of the specimen
in the loading heads, (c) the set screws holding the alignment cylinder to the lower
jig block are loosened and the alignment cylinder slid upwards along the upper
jig block until it eventually becomes free of the lower block, (d) the alignment
cylinder is locked to the upper jig block in the raised position, and (e) the deforma-
tion experiment is continued.

The maximum deviation between a number of attached axial strain

gages rarely exceeded 10 per cent.
The ability to load uniformly and accurately makes it possible to ob-
serve in detail incremental creep behavior amounting to a total of less
than 10 /ts in a period of the order of 2200 sec (Fig. 20).

A cknowledgments
The author wishes to express his thanks to B. Feldstead and M.
Tienharra, Physics Section summer student assistants, P. Okulich of this
section, G. Larocque and F. Kapeller of the Mining Research Section
for their assistance in the development of this loading system. The assist-
ance of P. Hudson and L. Nadon of the Fuels and Mining Practice Di-
vision in construction of many of the mechanical components is also
greatly appreciated.

References
[7] D. T. Griggs, "Creep of Rocks," Journal of Geology, Vol 47, 1939, p. 225.
[2] H. R. Hardy, Jr., "Design of Instrumentation for the Measurement of Time-
Dependent Strain in Stressed Rock Specimens," Transactions, Instrument
Society of Am., Vol 1, 1962, p. 147.
[i] S. Matsushima, "On the Flow and Fracture of Igneous Rocks," Bulletin No.
36, Disaster Prevention Research Inst., Kyoto University, Kyoto, Japan (in
English), 1960.
[4] D. W. Phillips, "Tectonics of Mining," Colliery Engineering, Vol 25, 1948,
p. 312.
[5] N. J. Price, "A Study of the Time-Strain Behavior of Coal-Measure Rocks,"
International Journal, Rock Mechanics Mining Science, Vol 1, 1964, p. 277.
[6] H. C. Heard, "Effect of Large Changes in Strain Rate in the Experimental
Deformation of Yule Marble," Journal of Geology, Vol 71, 1963, p. 162.
[7] S. Serdengecti and G. D. Boozer, "The Effects of Strain Rate and Tempera-
ture on the Behavior of Rocks Subjected to Triaxial Compression," Proceed-
ings, Fourth Symposium on Rock Mechanics, Bulletin No. 76, Mineral In-
dustries Experiment Station, Pennsylvania State University, University Park,
Pa., 1961, p. 83.
[S] H. R. Hardy, Jr., "New Rock Mechanics Program Underway by Canadian
Government," Mining Engineering, Vol 17, 1965, p. 62.
[9] H. R. Hardy, Jr., "The Experimental Investigation of the Inelastic Behavior
of Geologic Materials," Ph.D. thesis, Department of Engineering Mechanics,
Virginia Polytechnic Inst., Blacksburg, Va., 1965.
[10\ H. R. Hardy, Jr., and G. E. Larocque, "An Electro-Hydraulic Load Main-
taining System," Mines Branch, Department of Mines and Technical Surveys
(report in preparation).
[11] W. R. Ahrendt and C. J. Savant, Jr., Servomechanism Practice, McGraw-
Hill Book Co., Inc., New York, 1960.
[12] J. G. Truxal, Control Engineers Handbook, McGraw-Hill Book Co., Inc.,
New York, 1958.
[13] G. J. Thaler and R. G. Brown, Analysis and De.'iign of Feedback Control
Systems (2nd edition), McGraw-Hill Book Co. Inc., New York, 1960.
[14] P. N. Nikiforuk and D. R. Westlund, "Analysis of a Loaded High-Pressure
Hydraulic On-Off Servomechanism," Journal Mechanical Engineering Science,
Vol 6, 1964, p. 371.
[15] M. Hetenyi, Handbook of Experimental Stress Analysis, John Wiley and
Sons, Inc., New York, 1957, p. 160-238.

[16] B. Chance, V. Hughes, E. MacNichol, D. Sayre, and F. Williams, Wave-

forms, MIT Radiation Laboratory Series, Vol 19, McGraw-Hill Book Co.,
Inc., New York, 1949, p. 434-439.
[17] I. A. Greenwood, Jr., J. V. Holdam, Jr., and D. MacRae, Jr., Electronic
Instruments, MIT Radiation Laboratory Series, Vol 21, McGraw-Hill Book
Co., Inc., New York, 1948, p. 95.
[18] G. R. Korn, "Design and Construction of Universal Function Generating
Potentiometers," Review Scientific Instruments, Vol 21, 1950, p. 77.
[19] H. R. Hardy, Jr., "A Technique for Uniform Loading of Small Cylindrical
Test Specimens of Geologic Materials," Mines Branch, Department of Mines
and Technical Surveys (report in preparation).
[20] J. A. Paquin, "Some Aspects of Uniaxial Compression in Rocks," Proceed-
ings, 2nd Canadian Rock Mechanics Symposium, Queen's University, De-
cember, 1963, p. 73, (published by Mines Branch, Canadian Department of
Mines and Technical Surveys, 1964).

DISCUSSION

B. W. Paulding, Jr}—Although the Teflon inserts may minimize the

frictional constraint on the ends of the specimen so that the deformation
moduli can be accurately measured, I think this technique may have
limitations when determining the compressive strength of rock. This is
because of the possibility of the Teflon intruding small flaws or cracks
on the ends of the specimen, which may result in fracture at stress levels
less than the true compressive strength. Do you have any suggestions
on how this possible difficulty can be overcome?
H. R. Hardy, Jr. (author)—I agree with Dr. Paulding's comments.
The loading heads using Teflon inserts were developed specifically for
determining the deformation properties at stresses below the compressive
strength of the test material. However a number of experiments have
been conducted in which sustained loading caused eventual fracture of
the test specimen. No penetration of the Teflon into the specimen ends
was noted in any of these experiments; however, increased extrusion of
the Teflon into the diametric clearance between the specimen and the
loading head was observed. In all cases the specimens failed leaving an
upper and lower conical remnant. In most experiments the maximum
stresses did not exceed 25,000 psi.
If intrusion of the Teflon becomes an important factor at higher
stresses, it is suggested that a thin disk of stainless steel placed between

^ Associate research engineer, IIT Research Institute, Chicago, 111.

the ends of the specimen and the Teflon inserts should prevent intrusion
while providing sufficient flexibility to ensure uniformity of loading.
J. R. Hoskins^—The loading system designed and built by Dr. Hardy
is a welcome addition to existing capabilities for determining rock prop-
erties. As the author suggests, a large group of problems involving en-
gineering in underground rock structures can be solved only by a better
understanding of rock properties and behavior. The system, in itself,
will not give direct answers to the problems but will enable the investi-
gator to isolate certain characteristics and parameters that will more
fully explain the in situ behavior of rock. The equation of state concept
of rock materials is a desirable though possibly an unreachable goal.
Simple equations of state such as that given in the Burgers model have
already been used to explain some rock behavior.^. '* The system under
discussion provides both the instantaneous loading and unloading and
the constant stress mode necessary for this model. It is reasonable to
believe that investigation of more complicated models to explain be-
havior of other rock types would more fully utilize the potential of the
system. In addition, each test mode can be related direcdy to some
individual structural rock behavior, particularly those of constant strain
and constant strain rate observed in mine working convergence.
Ultimate solutions to the problems will likely be a synthesis of tests
of the above type, analogs, physical model studies, and in situ tests. The
compression test in the above and possibly other modes is not likely to
be superseded, in any case, since it is simple, economical, and more
frequently represents realistic conditions than tension or shear tests.
Dr. Hardy's description of his system to provide uniform stress distribu-
tion in compression test specimens and to develop a versatile uniaxial
loading system prompts the following comments.
Since the compression test has been examined in detail by Hardy,^
a question arises concerning the use of a Teflon platen, since such a
platen will allow differential lateral movement due to the relative stiff-
ness of Teflon and rock and the lateral ±0.001-in. tolerance. Uniformity
of loading is desirable, but any deviation of test procedure from a natural
condition, that is, rock upon rock, should be suspect; therefore. Teflon

^ Mining methods research engineer, Bureau of Mines, U.S. Department of

Interior, Denver, Colo.
^H. R. Hardy, Jr. "Time-Dependent Deformation and Failure of Geologic
Materials," TM 25/59 MIN, Mines Branch, Department of Mines and Technical
Surveys Fuels Div., Ottawa, Canada, April, 1959.
' N. B. Terry, "The Elastic Properties of Coal, Part 6, Some Measurements
of Internal Damping and Some Considerations of Visco-Elastic Behavior," Report
No. 2080, British National Coal Board, Mining Research Establishment, 1957.
" H. R. Hardy, Jr. discussion of the paper "Some Physical Properties of Rocks
and Their Relationship to Uniaxial Compressive Strength" by D. F. Coates et al,
Proceedings, Rock Mechanics Symposium, McGill University, Montreal, Quebec,
Sept. 7-8, 1962.

should be used only if the benefits of using it outweigh the undesirable

effects introduced. In such a case using Teflon is definitely justified; how-
ever, uniformity alone is undesirable, if unqualified, since realistic rock
strength will statistically cover a wide range. Design strengths from a
safety standpoint should consider a weighted minimum strength-risk
factor value.
Seldenrath's study^ shows a definite tensile failure under compression
conditions similar to the author's (particularly in the case of a — 0.001-in.
tolerance). The writer has studied similar end conditions with similar
results and believes that short column platens of identical shape and
rock material are more desirable. Either increased length or altered end
conditions or both that provide a volume rather than a plane of uniform
stress is to be desired.
An excellent documented discussion of this subject has recently been
presented by Newman.'^ What he states for brittle elastic materials is
also applicable to inelastic materials, generally. The length-to-diameter
ratio suggested by Newman is also to be strongly considered, for even
a ratio of 2 or iVz restricts the volume under uniform load to a statisti-
cally small fracture incident zone.
The results of Kerper's* study on strength of glass, using the two point
versus the single point loading system, can be useful in understanding
the length-to-diameter ratio influence in compression testing. Kerper
found that lower but more realistic breaking strengths of glass resulted
from tests that did not confine the failure region to a limited area or
volume. In the compression test with a length-to-diameter ratio of two,
cross-sectional uniformity is believed to exist midway between the ends.
If the ratio is 2Vz or greater, this uniform zone is lengthened, and sta-
tistically the opportunity for unbiased specimen behavior on cross sec-
tions away from end influence is greatly increased.
Newman's testing specifications, when applied to rock, should be
considered as an empirical standard and not an applied behavior char-
acteristic. Since most underground conditions restrict behavior to a
biaxial or triaxial state of blocks under varied end effects, the uniaxial
compression test would only be a means of closely relating the various
rock types and specimens. The practicality of the data would have to
be determined by each physical situation. Even so the proposed uniaxial

" Th. R. Seldenrath and J. Gramberg, "Stress-Strain Relations and Breakage of

Rocks," Mechanical Properties of Non-Metallic Brittle Materials, W. H. Walton,
editor, Interscience Publishers, New York, 1958, pp. 84-87.
' K. Newman and L. Lachance, "The Testing of Brittle Materials Under Uni-
form Uniaxial Compressive Stress," Proceedings, Am. Soc. Testing Mats., Vol 64,
1964.
•* M. J. Kerper and T. G. Souderi, "Comparison of Single-Point and Two-Point
Loading for Determining the Strength of Flat Glass," Proceedings, Am. Soc.
Testing Mats., Vol 64, 1964.

compression tests are of great value and must precede other more com-
plicated and uncertain testing procedures. Data from these tests will be
needed to help develop the triaxial and biaxial tests of a similar nature
or of actual model configurations.
One should also note that the most accurate surfacing of test speci-
mens and platens usually does not approach 0.0001 in. in either surface
flatness or waviness. A mismatch of both surfaces represents 200 /An. of
deformation that must occur in the 1 %-in. half length of the specimen,
representing a strain that is nearly 10 per cent of the failure strain of
many rocks. Uniformity is probably achieved irrespective of this due
to the viscoelastic effects of geologic material and the St. Venant effect.
Other considerations relating to the specimen rather than the system
are: (1) objectives of the test, (2) specimen size and shape, and (3) strain
gage measurements.
The objectives of the test must always be the foremost consideration,
and the writer believes standardization versus application is the first
qualification. If, as in the author's testing system, standardization ap-
pears to be the foremost consideration, well defined conditions such as
end surfacing of the specimen, continued platen surface condition, tem-
perature, humidity, and uniformity of specimen and load are a major
part of the standardization. If application is the primary consideration,
model conditions of size, shape, end restraint, and measurement location
are of greater importance.
In geological material, the flaw factor is particularly pronounced, and
small size specimen selection eliminates both large fractures and speci-
mens having high frequencies of fracture. Most small specimens are
representative of a perfect population statistically rather than an average
population; hence, the resulting data should be regarded as representa-
tive of a high strength or a more optimum result rather than what would
be expected as an average.
Shape was discussed previously under height-to-width ratio, but a
square or round cross section is also important. Whereas strength values
appear to be constant with shape, apparent strain values seem to be in-
fluenced by the specimen corners. Unpublished work by the writer in-
dicates that stress-strain relations, which appear linear under normal
strain gaging procedures as described by the author, will show marked
nonlinearity on all boundaries near the corners with reverse trends at the
top and bottom of the specimens. While this does not seem to influence
a standard laboratory test, it is of considerable value in applications.
Several strain gaging arrangements have been suggested in the past
but the author's system of three center gages at 120-deg spacing would
appear best. The results of many tests by the writer show that opposing
lateral surface gages of two, four, and eight center-gage configurations
give little deviation (that is deviation is less than gage error) from the

mean. However, the differences between opposite gages without a jig

were greater than the author found, indicating that either a jig or balanc-
ing of the load by realignment, using an initial small load and reposition-
ing of the specimen to get equal readings of the three gages, is necessary
to conduct proper strength tests.
The following comments concern the versatile uniaxial loading system.
The author does not clarify whether the temperature control of 22.5 ±
0.2 C is a standard operating condition for all tests and for all parts of
the system. Since temperature and humidity are important variables at
the low strain values of geologic materials, elaboration of the brief
statement offered would be helpful.
The rapid changes of controlled variables mentioned by the author
must refer to a short elapsed time of recognition of deviation of the
variable from the set value rather than rapid change of the variables,
since the on-off controller produces a change in the variable at a pre-
determined set rate, whereas the continuous controller can be of several
types, most of which change the variable at a rate proportional to its
deviation from the set value and reach a maximum rate equal to the
maximum setting of the on-off unit.
Whether the controller is of the on-off or continuous type, the ±AL
(/?) band control, in some cases, is being more closely controlled by two
bucking photocells or two cadmium sulfide cells, one each in an arm
of the same four-arm bridge. The light beam from the galvanometer
impinges direcdy on both cells which can be adjusted to a neutral gal-
vanometer position. When the galvanometer is balanced, an equal seg-
ment of its light shines on each cell, and the resulting bridge circuit is
balanced. Any unbalance of the galvanometer places more light on the
appropriate cell which unbalances the circuit or bridge. This unbalance
cither activates an on-off system or provides a proportional continuous
controller unbalance. The very accurate control possible with the split
light beam shifts the primary error to other parts of the system.
The errors of the system the author describes are quite low for the
complexity and number of components involved. My experience after
designing and building a pneumatic-hydraulic control system of a less
complex nature but for triaxial loading (Hoskins") was that feedback
control often became unpredictable due to factors such as intermittent
valve heating with pressure change, load from inertia, and regulator
instability resulting in errors of 5 per cent or greater. Another error
factor I experienced was an unexplainable system change of V2 to 2 per
cent in control characteristics after a shutdown period of several days.
Both moisture and temperature were believed to be responsible.

" J. R. Hoskins, "Design and Construction of a Basic Geomechanics Laboratory,"

Ph.D. thesis. University of Utah, Salt Lake City, Utah, 1962.

I am in complete agreement with Dr. Hardy that a system of this

type must be used for better understanding of rock materials. The
writer's suggestion on specimen length and platen material is believed
to offer a better source of standardization. The two- and three-dimen-
sional compression modes would represent other systems of an applied
nature where standardization would not be likely.
The author has overcome many factors to produce a multimode test-
ing system for developing and controlling axial pressure for wider
standardization of rock tests. It is questionable as to whether uniform
load distribution has been achieved. The effort and result are to be com-
mended, and it is hoped that similar two- and three-dimensional loading
systems will not be far off.
Dr. Hardy—First, regarding Dr. Hoskins' comments on the test speci-
men and loading conditions, it is appreciated that the geometry and size
of the test specimen as well as the uniformity of loading are important
factors in the study of the mechanical properties of geologic materials.
A number of workers have investigated the effects of such parameters
on the various elastic and failure properties, particularly under condi-
tions of uniaxial short-period compression. To date, however, rela-
tively few studies have been conducted for inelastic properties. The
specimen geometry and dimensions used by the author were selected
mainly for convenience and are probably not optimum for purposes of
field application. It is felt that in studies of the inelastic behavior of
geologic material where the inelastic parameters are often highly de-
pendent on stress level, uniformity of loading is a basic necessity which
must be maintained even at the risk of introducing a loading condition
which deviates from the natural condition, that is, rock on rock. It is
further appreciated that although the tests carried out to date have been
under conditions of uniaxial stress, studies under conditions of biaxial
and triaxial stress, more common in the field, are required. Such tests
are planned for the near future.
Considering Dr. Hoskins' comments on the load control system itself,
reference was made in the paper to the fact that the behavior of the con-
trol system was studied in a test facility in which the temperature was
controlled at 22.5 ± 0.2 C. This is the temperature maintained in a
two-section test enclosure which houses the complete test facility. One
section, which is held at the control temperature and whose humidity
may be controlled, contains the loading frames. To date the humidity
in this area has rarely exceeded 5 to 10 per cent. The other section of
the enclosure, also at the control temperature, is at all times maintained
at as low a humidity level as possible. This section houses the control and
measurement facilities.
The author's statement regarding the use of "continuous" and "on-
off" type controllers should be clarified. In a control situation in which

the process (namely, one of the controlled variables u, b, e, e) is being

programmed to undergo rapid changes (for example, a high rate of load-
ing), it is necessary to employ a continuous type controller, for only
with this system will the controller output vary in proportion to the
error signal making it possible for the process to accurately follow the
programmer with a minimum of over and undershoot. If, however, only
slow changes in the process are required, the system is at all times in a
quasi-statistic state and an on-off type control may be utilized.
Dr. Hoskins suggests the use of a split photocell arrangement for
detection of the error signal. The author agrees that this type of detector
would result in a decrease in the dead-band and would in addition pro-
vide a proportional output suitable for operation of a continuous type
controller. However, the present system is not limited by the magnitude
of the dead-band but rather by the fact that simple on-off control has
been utilized. It should be pointed out that the null point of a split
photocell detector is highly sensitive to the shape and the brightness
distribution of the light spot used. Past experience by the author in
utilizing this type of detector in a recording optical extensometer indi-
cated that it would in fact add unnecessary complication in an on-off type
control system.

Foundation Testing Techniques for Arch

Dams and Underground Powerplants

REFERENCE: G. B. Wallace and O. J. Olsen, "Foundation Testing Tech-

niques for Arch Dams and Underground Powerplants," Testing Tech-
niques for Rock Mechanics, ASTM STP 402, Am. Soc. Testing Mats.,
1966, p. 272.

ABSTRACT: To ensure that foundation tests are performed on repre-

sentative specimens and field sites, geologists take advantage of regional
surveys, vicinity surface mapping, and subsurface conditions at the site
as revealed from borings and borehole camera studies. Following this,
the physical properties of the rock are determined from drill cores. Defor-
mation of the rock, for computation of arch reactions and deflections, is
determined by numerous laboratory tests of sound rock specimens and by
jacking and seismic tests in the field that show the influence of cracks
and other inherent weaknesses. The shear strength of the rock is deter-
mined triaxially in the laboratory with specimens from holes oriented at
various angles to the foliation plane. In the field, large blocks of rock are
tested to obtain the influence of joint systems on shear strength and re-
sistance to sliding under various normal loads. Strain relief techniques are
used to obtain in situ stresses on the walls and roof of an exploratory tun-
nel and within the rock mass by means of deep drill holes. Rock defor-
mation caused by excavation of a machine hall chamber is measured by
the use of multiple borehole extensometers and convergence gage points.
When jointed rock is bolted together and used as a structural material,
tests are made to determine the torque-tension relationship for bolts to be
sure that the anchors will hold and not relax their grip with time. As a
safety measure and to check design assumptions, deformation meters are
installed between the base of the dam and the foundation rock at various
depths below the structure.

KEY WORDS: rock (material), rock mechanics, foundations, dams, arch

dams, field tests, deformation modulus, shear strength

The design of concrete dams is changing from heavy structures relying

significantly on weight for stability to thin arches which make more effi-
cient use of materials. At the same time, the design of powerplants is
changing. Powerplants previously constructed at the toe of concrete dams
usually required massive foundations and restricted use of economical
1
Supervisory civil engineers, Division of Research, Bureau of Reclamation, Den-
ver, Colo. Personal members ASTM.
272

Copyright^ 1966 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de
WALLACE AND OLSEN ON FOUNDATION TESTING TECHNIQUES 273

overflow spillways. Those located along the riverbanks require long, ex-
pensive penstocks. When site conditions are favorable many engineers
favor thin-arch dams in combination with underground powerplants
which provide protection from rockfalls, freedom in spillway design, and
short waterways between the reservoir and turbines.
To achieve the advantages of the thin-arch and underground powerplant
designs, engineers are extending their knowledge of the properties of the
rock masses with which they are dealing. Present field and laboratory
tests used to determine design criteria vary widely. The so-called elastic

7300

6900
FIG. 1—Morrow Point dam site showing angle drill pattern.

modulus of a rock mass may vary by several hundred per cent when de-
termined by different test procedures currently in use.
Subcommittee 12 on Rock Mechanics of ASTM Committee D-18 on
Soil and Rocks for Engineering Purposes, has started preparing standard
test methods to determine the engineering properties of rock foundations.
An initial step in this effort is to study those techniques which have gained
acceptance and use by major design organizations. As partners in founda-
tion investigations conducted by the Bureau of Reclamation, engineers
and geologists perform exhaustive studies. The investigations usually start
with aerial surveys, then inspection of surface features, such as faults,
folds, joints, and weathering, which help to determine the extent of sub-

FIG. 2—Borehole TV camera: {upper left) cable control wheel (upper right)
mobile trailer, (bottom) monitoring equipment.

surface investigations required to adequately define the capabilities of

the rock.
Sampling and Testing
Figure 1 shows the pattern of holes drilled for investigation of Morrow
Point Dam. This double-curvature, thin-arch structure with its associated
underground powerplant is now under construction in a rugged gorge
just upstream from the famous Black Canyon of the Gunnison River in
western Colorado. Angle holes were drilled parallel to lines to thrust in
the abutments. From these as well as from vertical holes, core specimens

were extracted for logging of subsurface geological features, including

bedding, foliation, internal joint patterns, and variations in rock types.
Similar procedures are used by the Bureau in investigating all sites for
concrete dams.
As an aid to the geologist in gaining an actual look into a foundation,
the borehole television camera, shown in Fig. 2, is employed. The cam-
era, located near the center of a 4-ft-long tube, is connected by coaxial
cable through a storage reel to a mobile trailer which houses the monitor-
ing equipment. The camera can be inserted into NX holes and rotated
360 deg in search of cracks or fissures and for identification of rock types.
As the cable is fed into the hole, it passes over a wheel which keeps the
operator informed as to the exact location of the camera. The operator
can raise, lower, and rotate the camera in the drill hole from the control

UPSTREAM UPSTREAM

ABUTMENT

ARCH CANTILEVER

Tension (T) may exist in the arch

and cantilever elements as shown.
FIG. 3—Ahutmetit defunnation and arch deflection.

panel. By observing the signal from a built-in clinometer and compass,

the strike and dip of the features observed can be oriented. A stationary
camera can be used to photograph special features shown on the televi-
sion screen, or they can be recorded on video tape. By comparing closed-
circuit television observations and logs of drill holes, more meaningful
specimens and test sites can be selected than can be obtained on the basis
of drill cores alone.
After the cores have been examined and logged in the field, they are
carefully crated and shipped to the Bureau's Engineering Laboratories
in Denver, Colo. In performing laboratory tests of rock cores, first in
importance are the basic physical characteristics of the rock, such as
compressive, tensile, and shear strengths, modulus of elasticity, and
Poisson's ratio. Knowing that these properties vary with angle of bed-
ding planes or foliation of the rock. Bureau engineers exercise care to
note the orientation of these features with respect to direction of the test
loads. Tests are also made to determine specific gravity, absorption, and
porosity. Laboratory tests to determine rock characteristics cost con-

siderably less than in situ field tests, but they do not reflect the influence
of cracks and joint systems. Therefore, in situ tests are made at care-
fully selected sites to evaluate the effects of discontinuities in the rock.
As iUustrated in Fig. 3, when a reservoir loads an arch dam it spreads
the abutments and tends to straighten out the arch. The movements are
small; however, the abutment reactions and arch deflections cannot be
computed until deformation characteristics of the rock are known. Also,
shear strength of the rock must be determined to ensure that it will safely
resist the downstream force of the water transmitted through the dam to
the abutments and base. This paper focuses attention on techniques for
determining these two important design properties—deformation modu-

FIG. 4—Elasticity frame for small cores.

lus^ and shear strength. It also describes a method for determining in

situ stresses in rock masses resulting from overburden and geologic
history. As a follow-through, a technique is also described to measure de-
formation in a rock mass as excavation of an underground ©pening pro-
ceeds.
Deformation Modulus
The frame shown in Fig. 4 is used in determining static elastic prop-
erties of NX (2.12-in.) cores in the laboratory. It is capable of measuring
strain to one part in a million. Pretensioned, unbonded resistance wires,
located within the vertical and radial tubes, indicate length and diameter

" Deformation modulus as used here is the ratio of stress to total unit deforma-
tion produced by static loads on a rock mass in situ for the duration of loading,
which includes closure of joints and fractures.

changes of the specimen due to applied loads. Stress-strain values are

visually displayed and automatically printed on paper tape by an elec-
tronic digital scanner. The ratio of lateral to axial unit strain, or Poisson's
ratio, derived from this test is used in design and also to compute defor-
mation modulus from field jacking tests.
In situ tests for deformation modulus are conducted in an exploratory
tunnel, using two 200-ton jacks, shown in Fig. 5, to apply the load hy-

F I G . 5—In situ jacking test equipment for deformation modulus.

draulically. In these tests, load is transmitted through steel jack shoes to

hydraulic cushions and concrete bearing pads. The bearing pads are cast
directly against the rock on the roof and floor of the tunnel. The hydraulic
cushions are connected to a pressure control switch, so that as deforma-
tion occurs the pump supplying the jacks is actuated and accurately main-
tains the desired load. Timber blocking fills excess space between the
jacks and the tunnel roof. The rock surfaces for the bearing pads are
prepared by trimming with hand tools for a depth of about 2 ft beyond
the original tunnel wall. AX (2-in.) holes are drilled under the center
of each jack to a depth of about 16 ft. A deformation meter with steel

tube extension is anchored into the bottom of each drill hole by grouting
and concreted into the bearing pads. Horizontal jacking tests are made
with the same equipment rotated 90 deg.
As the rock is compressed under load, the deformation meters measure
over-all movement between the bearing pads and the bottom of the drill
holes. In addition, the change in tunnel diameter is measured midway be-

,--DIRECTION OF PROPAGATION

. LONGITUDINAL
COMPRESSION WAVE

TRANSVERSE ('
"SHEAR" WAVE ',~

FIG. 6—Seismic modulus determinations.

tween the two jacks with a micrometer. These measured values are ap-
plied to Boussinesq's theory to compute the deformation modulus.^
Jacking tests involve considerable expense and are, therefore,
limited to a few test sites. To determine whether or not the values
obtained from jacking tests are representative of the entire foundation,
they are supplemented with field seismic tests, as illustrated in Fig. 6.

^ S. Timoshenko, Theory of Elasticity, McGraw-Hill Book Co., Inc., New York,

1934, p. 333.

Although seismic tests represent rather low stress levels and are not as
precise as jacking tests, they can economically indicate variations in rock
quality over a greater portion of the foundation. The seismic method is
based on the principle that shock waves are transmitted through rock
at a velocity which is related to its modulus of elasticity and density. The
test equipment is mounted in a truck and is easily and quickly moved
from site to site. Several geophones, which serve as wave pickups, are

FIG. 7—Small triaxial shear test equipment.

mounted at various distances along a test line or down a drill hole, and
these are connected to a recording oscillograph. After detonation of a
small explosive charge, transverse shear waves as well as longitudinal
compression waves travel through the rock. The oscillograph records the
time of their arrival at various geophones. By accurately measuring the
time interval and distance, the velocity of the waves may be determined
and the elastic constants computed.*

* C. A. Heiland, Geophysical Exploration, Prentice Hall, Inc., New York, 1940,

p. 504.

Shear Strength
Special equipment, shown in Fig. 7, is used to determine the shear
strength of confined NX (2.12-in.) cores. The specimen is sheathed in a
rubber or plastic jacket, so that the hydraulic fluid in the pressure cham-
ber does not enter the pores of the rock. The specimen is lowered into
the chamber, confining pressure is applied from a hydraulic pump, and

FIG. 8—Large triaxial shear test equipment.

axial load is supplied by a conventional testing machine. In determining

shear strength, several similar specimens are subjected to difl'erent con-
fining pressures. The magnitude of axial load required to produce shear
failure of a rock may increase 3 to 15 times as much as the confining
pressure. Mohr's diagram, constructed from the test results, yields the
shearing resistance.'' The deformation frame shown mounted on the
specimen in Fig. 7 is not required for determining shear strength. It is

' D . P. Krynine, Soil Mechanics, McGraw-Hill Book Co., Inc., New York, 1941.

Still under development to obtain modulus of elasticity and Poisson's

ratio under triaxial conditions.
The large triaxial shear machine shown in Fig. 8 is capable of testing
cores 6 in. in diameter by 12 in. in length. Lateral pressures up to
125,000 psi may be introduced through the top of the machine into the

Loads
Lateral: 22,100 psi
Axial: 125,100 psi (computed from original cross-sectional area)
Axial: 79,700 psi (computed from final cross-sectional area)
Duration of test: 2 hr and 5 min
Physical Characteristic Before Test After Test Remarks
Diameter, max 5.9 in. 7.4 in. 25% increase
Height 12 in. 8.8 in. 27% decrease
Volume 328 in.' 354 in.' 8% increase
Density 173 lb/ft' 159 lb/ft' 8% decrease
Dynamic modulus of
elasticity £ 10.88 X 10«psi 1.07XlO«psi Ko original mod-
ulus
Average velocity of lon-
gitudinal vibration.. . 17,070 ft/sec 5580 ft/sec \i original ve-
locity
FIG, 9—Deformed limestone core tested in triaxial compression.

chamber surrounding the specimen. Axial loads up to 8,000,000 lb may

be applied to the specimen through the bottom of the chamber. Figure 9
shows a limestone specimen before and after testing in this machine.
A test setup for determining shear strength in the field is illustrated in
Fig. 10. Essentially, the test procedures consist of applying shear and
normal loads to a prepared 15 by 15 by 8-in.-high block projecting above
the floor of a tunnel. The shear load is inclined so that its line of thrust

passes through the center of the shear zone between the projecting test
block and the rock mass, thereby eHminating overturning moments. Fol-
lowing shear failure, the rock is caused to slide under various normal
loads and the coefficient of friction determined. Similar tests, using the
same equipment, are performed on large blocks of rock (approximately

F I G . 11—Eight-point rosette for measuring surface strain relief with mechanical

extensometer.

F I G . 12—One-directional borehole gage, assembled and exploded views.

one cubic yard in volume) shipped into the laboratory. Tests are also
conducted, utilizing concrete blocks cast on rock to determine the co-
hesion and friction resistance at the plane of contact. Instrumentation
consists of dial gages measuring both vertical and horizontal deforma-
tions of the block. The results of the field tests are correlated with the
tests of large blocks in the laboratory and with triaxial tests to give a
better interpretation of the shearing resistance of the rock.

FIG. 13—Overcoring to relieve in situ stresses measured by borehole gage.

FIG. 14—Tests of donut overcore to determine elastic modulus.

In Situ Stresses
Design of underground powerplants requires tliat the state of stress
in the rock be known, including both local boundary stresses around the
opening and stresses within the rock mass. To obtain design criteria,
strain relief tests are made in an exploratory tunnel located at the arch of
the machine hall. To determine the surface boundary stresses, eight ro-
sette gage points are cemented into holes in the rock on a 10-in. circle,
as shown in Fig. 11. Preparatory work involves cutting back the rock
with hand tools and smoothing to a plane surface. Precise length meas-
urements between gage points are made before and after line drilling
around the gage points to obtain the strain due to relief.
To determine stresses in the mass beyond the tunnel wall, a borehole

Tensioned music wires between

hollow rock onchors and
cantilevers—-,

"Cover plate

-Sensor head containing 8 stainless cantilevers

with attached strain gages (anchored
at collar of d r i l l hole)

"^-Readout cables
FIG. 15—Multiple borehole extensometer.

gage is used. This gage, shown in Fig. 12, was developed by the Bureau
of Mines. An EX (1.50-in.) hole is drilled for a depth of about 20 ft
into the tunnel wall or roof. The gage, which is slightly smaller than the
EX hole, is inserted a short distance into the hole. At this time, the rock
around the gage is under stress due to the weight of overburden and its
geological history. The EX hole is then concentrically overcored, as
shown in Fig. 13, to a point just beyond the gage, with a 6-in. diamond
bit to relieve the in situ stresses. The resulting change in hole diameter
is measured with the borehole gage. The gage is then advanced at inter-
vals of about 6 in., oriented in various directions 45 deg apart, and suc-
cessively overcored. From these observations and laboratory tests to de-
termine the modulus of elasticity, the existing two-dimensional stress
field is computed in a plane perpendicular to the drill hole axis."
The 6-in. "donut-shaped" cores obtained from drilling around the

° R. M. Merrill and J. R. Peterson, "Deformation of a Bore Hole in Rock," Re-

port RI 5881, U.S. Bureau of Mines, 1961.

borehole gage are placed in a pressure sleeve, as shown in Fig. 14. The
borehole gage is again installed in the EX hole and changes in hole diam-
eter measured as pressure is applied radially to the exterior surface. A
thick-wall cylinder formula is used to determine the modulus of elasticity.
In addition to the strain relief tests made in an exploratory tunnel to
obtain in situ stresses for design, rock deformations due to adjustment
of stress are measured as the machine hall is excavated. Multiple bore-
hole extensometers (MBX's), shown in Fig. 15, are installed vertically
upward 50 ft from the crown of the exploratory tunnel. Readings are

Axia Axial
extension -J- controction
(tension) (compression)

Roof MBX
18 fixed anchors)-;.

Strain variation
curve

Underground
opening

-Wall MBX

FIG. 16—Strain variation curve for MBX.

taken daily between blasts as excavation of the machine hall proceeds.

At the completion of excavation, the gages are read on a weekly basis
until the rock has stabilized. Additional extensometers are placed hori-
zontally at the spring line as soon as excavation permits. The extensom-
eters contain eight sensing stainless steel cantilevers at the collar of the
hole, recessed into the rock for protection against blasting. The canti-
levers are attached by music wire to anchors placed at different depths
in the borehole. Each anchor is constructed so that the wires from an-
chors deeper in the hole may pass through them without restraint. Any
rock movements that occur are carefully evaluated in terms of their ef-
fect on stability of the rock surrounding the opening.
A strain variation diagram representing readings taken in a tunnel is

shown in Fig. 16. In this case the maximum thickness of rock that must
be supported by bolting or lining is indicated by distance H, as illustrated
in the figure.
To supplement readings from the MBX's, convergence measurements
are made by an invar tape extensometer between gage points on the

FIG. 17—Torque-tension test for rock bolt.

'Hollow rock bolt

Air vent
Anchor

Coupling

"Grout inlet
FIG. IS—Rock-bolt load cell.

walls, floor, and roof of underground openings. Stainless steel inserts

with bail connections are installed in the rock at selected locations. The
extensometers, tightened to constant tension over a pulley, measure con-
vergence between the steel inserts to the nearest 0.005 in.

Rock-Bolt Anchorage
The performance of rock bolts is of critical importance in construc-
tion of large underground openings. Tests are made to determine the
tightening torque required to develop the maximum potential of the bolts

in binding the rock together and strengthening it. Bolts are torqued be-
tween the tension heads of a testing machine, as shown in Fig. 17, to ob-
tain the torque-tension relationship. The amount of tension induced in a
bolt under a given amount of torque varies considerably, depending upon
the lubrication used on the threads and washers. About one half as much
torque is required to develop 20,000 psi in a bolt with a molybdenum
lubricant as is required with ordinary lubricating oil.
During the rock-bolting operations in the field, selected bolts are
tested with a center-pull jack to determine anchor slippage, or, if no
slippage occurs, to measure the load required to break the bolt. Other
bolts are initially stressed, then retightened after 8 to 72 hr to measure
the tension loss. Rock-bolt load cells, illustrated in Fig. 18, are used at
selected locations to monitor rock-bolt performance over extended pe-
riods of time.
Tests made in the laboratory and field to evaluate various anchors for
rock bolts have shown that no one anchor is suited for all types of rock.
In general, serrated bearing areas with lower lateral pressures are better
suited for soft rocks, whereas relatively smooth bearing areas with higher
lateral pressures work better in very hard rock. Both smooth and de-
formed grouted bolts have performed satisfactorily. Three-foot-long
bolts, grouted under tension, were tested in the laboratory to determine
the influence of anchor failure on bond. Strain gages, located at 6-in.
intervals along the bolts, indicated there was no bond failure when the
end grip was released and the entire load was transmitted to the grout
within 1 ft of the anchor. After release of the anchor, the collar end of
the bolt was gripped by a center-pull jack and loaded in tension. The full
strength of both smooth and deformed bars was achieved without bond
failure.

Deformation Meters
To measure movement between the base on an arch dam and its
foundation, deformation meters are anchored by extension rods grouted
into the bottom of drill holes at depths of 30, 60, and 90 ft into the rock.
The other ends of the meters are embedded in the concrete base, so that
each meter spans the rock-concrete contact plane. The meters are in-
stalled, normal to the base of the dam, in at least three locations, one each
on the right and left abutments at midheight, and one at the bottom mid-
section.

Conclusions
The following are based on the authors' experiences with the equip-
ment described above:
1. Lighter jacking equipment that can be easily and quickly moved
from site to site is needed for tests to determine deformation modulus

and shearing resistance. Test equipment has recently been designed,

utilizing flat jacks and lightweight structural aluminum members in lieu
of the heavy hydraulic rams and timber cribbing. A hole in the center
of the flat jacks will permit direct measurement of the change in tunnel
diameter.
2. The borehole gage used by the authors is limited to measuring the
change in hole diameter in a single direction. Subsequently a gage capable
of measuring diametrical change in three or four directions simultane-
ously has been developed by the Bureau of Mines. Since deformation in
at least three directions is required for computation of principal stresses
in a given plane, this development is particularly significant. Use of a
borehole camera for setting the borehole gage away from discontinuities
increases the percentage of successful measurements.
3. The elasticity frame (Fig. 7), for use in a triaxial test chamber,
produces satisfactory axial strain measurements, but the transverse
strains are too small to yield satisfactory results. A frictionless hinge or
other modification is needed to overcome this limitation.
4. The multiple borehole extensometer provides a reliable means of
measuring relative movements between points in a drill hole over limited
periods of time. To gain the fullest potential of this technique, there is
need for developing a sensing device which is free from drift and stable
over a period of years.
5. Good agreement was obtained between direct shear tests performed
in the field compared to the same type of tests on large blocks tested in
the laboratory. Therefore, when it is not practical or feasible to make
such tests in the field, they can be performed in the laboratory. Tests
on the large blocks show the effects of jointing that cannot be achieved by
laboratory tests on small cores.
6. Rock-bolt anchors which perform well in hard rock may not be
satisfactory in soft rock, and conversely those that are satisfactory in
soft rock may not perform well in hard rock. In determining the torque-
tension relationship for rock bolts, the type of lubricant used on the
threads and washers has a very pronounced influence on the results of the
tests.

A cknowledgment
These techniques were developed through the combined efforts of
many individuals on the staff of Chief Engineer B. P. Bellport of the
Bureau of Reclamation.

DISCUSSION

A. J. Barry^—The authors give a good over-all description of the mul-

tiple borehole extensometer (MBX) in the text and accompanying sketch
(Fig. 15). However, details are not given regarding the construction of
the anchors and method of setting. I would appreciate having the authors
clarify this important element of the extensometer.
G. B. Wallace and O. J. Olsen (authors)—Several anchors have been
tried, the simplest and most economical being a spring-type with re-
tractable jaws which expand against the walls of the hole when the anchor
is pulled back slightly. Coupled rods push the anchors to the desired
depth. The spring-type anchor is shown in Fig. 19 and is best suited for
soft rock. Other types of anchors that have been tried are: explosive,
groutable, and screw-expansion. The latter is retrievable and more re-
liable, especially in hard rock, but cost more.
E. F. Poncelefi—The authors refer to shock waves that are trans-
mitted through rock at a velocity which is related to its modulus and
density. The waves referred to are the elastic P-waves. Shock-wave
velocities are not so related.
The authors also refer to rubber or plastic jackets sheathing the speci-
mens being subjected to triaxial pressures. To obtain reproducible re-
sults, it is essential that the sheath material, or any coating of the
specimen inside the jacket, be well standardized, as these materials will
be forced into any incipient cracks or pores of the specimen. This is of
the greatest importance when the confining pressure exceeds the axial
pressure.
The authors also use Mohr's diagram constructed from the test re-
sults to obtain the shearing resistance. I would like to point out that
Mohr's diagram provides nothing of the kind. Mohr's diagram gives a
very clear graphical representation of the relationship between normal
and so-called "shear" stresses within the elastic limit across a virtual
plane of any possible orientation.
The envelope of various Mohr diagrams constructed from the meas-
ured stresses of various strength tests cannot represent failure conditions
in spite of the widespread assumption that it does. This is a consequence
of the fact that the limit of elasticity in each of these diagrams falls well
inside the Mohr circles, not on them. Any experimentally determined
condition either of fracture or of flow can be plotted separately and ac-

' Chief, Roof Control Research Group, U.S. Bureau of Mines, Pittsburgh, Pa.
^ Research consultant, Sunnyvale, Calif.

curately on a variety of diagrams involving the three principal stresses,

but the Mohr circle is not one of them.
The authors also show in Fig. 9 a limestone that has been severely
deformed in the triaxial compression machine in which the modulus of
elasticity E was reduced to one tenth its original value, and the F-wave
to one third its original velocity. I would suggest that this effect is due
to the limestone specimen being badly broken up and approaching the
elastic properties of a calcareous sand. It must be remembered that the
cohesive (or tensile) strength of rocks is low compared to their rigid (or

EKpansi

Setlinq

"Cover

F I G . 19—Spring-type anchor used with multiple borehole extensomeler.

flow) Strength. Were it high, as it is in metals for instance, it would have

supported such a deformation without breaking up and without important
changes in either modulus or P-wave velocity.
I wish to congratuate the authors on a most informative, well docu-
mented, and most ably presented paper.
Messrs. Wallace and Olsen—The term "shock" wave is used in our
paper in a very general sense because the source of wave is from a ham-
mer blow or dynamite charge. We appreciate Dr. Poncelet's emphasizing
that we measure the velocity of elastic P-waves, and not the velocity of
shock waves.
The type of sheath material used in triaxial testing is important as well
as its thickness and resiliency. In some instances where the specimen is

rough due to surface voids or incipient cracks, tliese are filled to prevent
entrance of the sheath material. Standardization is needed.
Shearing resistance is used in our paper in a general sense. Reference
to Mohr's diagram is only one of three techniques mentioned, the other
two being direct shear tests conducted on large test blocks in the field and
in the laboratory. All three methods are correlated "to give a better inter-

FIG. 20—Unconfined compression test assembly.

pretation of the shearing resistance of the rock." We realize that no

failure criterion fits all situations, but this does not preclude applying
Mohr's diagram to various behavior studies. Dr. Poncelet's scholarly
challenge is welcome. We realize that ours is not the final answer. It is
hoped that additional experiences with failure criteria will come forth to
stimulate further discussions.
The highly deformed limestone specimen shown in Fig. 9 is intended
to illustrate that failure under such confining pressure occurs gradually
in a material which is ordinarily brittle. From outward appearances, the

limestone is still intact, but inside drastic changes have taken place. This
emphasizes the need to define failure.
G. G. Balmer^—A compressometer similar to the one shown by the
author in his presentation has been used in the triaxial testing of soil-
cement mixtures by the Portland Cement Assn. The compressometer was
submerged in the hydraulic fluid and operated successfully. A description
of the instrument is given in Vol 58 of the ASTM Proceedings, in a paper,

FIG. 21—RTV silicone rubber membrane encapsulating procedure. Instru-

mented core specimen between upper and lower steel platens. Assembly rests on
vacuum draw-down pedestal. Plastic split mold at side.

"Shear Strength and Elastic Properties of Soil-Cement Mixtures Under

Triaxial Loading."
Messrs. Wallace and Olsen—The compressometer mounted on the
sheathed rock core in Fig. 7 measures axial strain satisfactorily while
submerged in the hydraulic fluid, but with a low modulus rock having low
Poisson's ratio, it is not sufficiently sensitive to measure lateral strains.
The authors appreciate Mr. Balmer's reference to, and development of
a similar compressometer for soil-cement specimens, which of course,
have a higher magnitude of strain.

" Senior development engineer, Paving Development Section, Portland Cement

Assn., Skokie, 111.

E. J. Deklotz*—The authors have presented a concise review of the

numerous considerations in the investigation, design, and construction of
underground structures. The paper is considered a worthwhile contribu-
tion to this symposium.
Attention is directed to the compressometers shown in Figs. 4 and 7 of
the authors' work. Although capable of measuring strain to 1 /xin., these
instruments are subject to considerable damage should the specimen fail
while these measurements are being made. Upon entry into the rock

FIG. 22—RTV silicone rubber membrane encapsulating procedure. Plastic split

mold assembled and secured to vacuum draw-down pedestal immediately prior to
pouring RTV silicone rubber.

mechanics field at the Corps of Engineers Missouri River Division Lab-

oratory, it was immediately established that axial and diametric strain
be accurately measured to failure in compression tests of cylindrical rock
specimens. This requirement essentially eliminated the use of ring and
collar extensometers of the type shown by the authors because of the
tremendous energy release at failure with consequent risk of equipment
damage. Attention, therefore, was directed to determining the suitability
of bonded electrical resistance strain gages and the development of
proper application techniques. The system finally evolved entails the

* Chief, Rock Mechanics Section, Corps of Engrs., Missouri River Division

Laboratory, Omaha, Nebr.

cementing of three axial gages equally spaced around the specimen at

the center of its height, and three diametric gages similarly located
around the circumference.
Figure 20 is a photograph of an instrumented unconfined compressive
strength specimen immediately prior to electrical hookup and test. The
wire resistance strain gages here pictured have a gage length of ^ % e in.
These permit a circumferential coverage of 36 per cent or 2^46 in- ^t
midheight. Two additional advantages of this system of strain measure-
ment are thus evident. First, strain is measured over an incremental por-
tion of the specimen at midheight, well away from the areas in which end

F I G . 23—RTV silicone rubber membrane encapsulating procedure. Plastic split

mold removed after curing of RTV silicone rubber.

eifects are manifest. Secondly, diametric strain is measured over a con-

siderable circumferential distance around the specimen, as opposed to
the usual two diameters across which strain is observed in the conven-
tional collar or ring-type extensometer. The latter item is important in
accurately calculating Poisson's ratio for rocks which do not dilate uni-
formly around the circumference.
When this system of instrumentation was first applied to triaxial com-
pression testing, some difficulty was experienced with the puncturing of
membranes due to surface irregularities caused by the wiring configura-
tion. This problem was eliminated by the development of a method of
encapsulating the core in a room temperature vulcanizing (RTV) silicone
rubber membrane before testing. Figures 21-23 show the details of the

procedure. The instrumented core specimen is positioned between the

upper and lower steel loading platens of the triaxial apparatus and
placed on the vacuum draw-down pedestal. The assembly is then en-
closed in a cylindrical plastic mold. The liquid silicone rubber is poured
over the top of the core and drawn by vacuum down the sides of the
specimen. After the silicone rubber has cured for at least 16 hr (usually
overnight), the mold is removed and the specimen tested. The mem-
branes, approximately %2 in. thick, have been found to be extremely
tough. No failures have been observed to date even at confining pressures
of 10,000 psi, the limitation of present triaxial equipment.

14
n 1
12
Voli me rrc Str ] i n - 1

lio
I J
o t ! i !
} J
b „ 1 /
Diametr c Strain—
/ f — 'ois ion's Ratio

\
" 6 ^
o
< 4 /
1 /
4
f i
2 If^ df 1 /
\a\ , n

0 1 A\0.05 1 0.10
0.20 0.15 0.10 0.05 0 0.15 0.20 0.25 0.30
Strain ( e ) , percent 0 0.2 0.4 0.6 0.8 1.0
Poisson's Ratio

F I G . 24—Unconftiied compressive strength test results.

Figure 24 is a plot showing results obtained by this system of instru-

mentation in an unconfined compression test. The uniformity of points
(observations) forming the axial and diametric stress-strain curves is
considered notable. The points and curve for Poisson's ratio and volu-
metric strain are similarly uniform, since they are derived from the basic
stress-strain data.
Foil-type gages have been used with equally good results. In the case
of rock which shows considerable axial deformation, this type of gage
has been found preferable. This is because they considerably reduce
the effects of transverse strain sometimes observed when wire resistance
strain gages are employed.
Two negative comments have been raised concerning the above sys-
tem: cost and time. Such comment does not take into consideration the

fact that conventional ring and collar compressometers cannot be oper-

ated to the failure point without risk of extensive damage. Further, it has
been found that production line techniques in the application and wiring
of gages permit an output comparable to that of the ring and collar com-
pressometer methods.
Messrs. Wallace and Olsen—Although not mentioned in our paper,
the Bureau's elasticity frame shown in Fig. 4 was a pioneer in its field,
having been developed some 15 years ago. At that time, emphasis was
on accurate strains within the elastic limit for design purposes.
For occasional failure strains, SR-4 gages were used. Each method
has its advantages and disadvantages. The frames can be installed in less
time but require a higher capital investment. Unbonded gages result in
lower unit cost when many tests are required for elastic strain only. The
bonded resistance gages, on the other hand, are better suited for small
numbers of tests, where time is not so critical.
As Mr. Deklotz so clearly points out, the present emphasis is to study
strains all the way to failure for research purposes. He also mentions
measuring diametric strain over a considerable circumferential distance,
which we agree is an advantage. His method looks very promising for
encapsulating the specimen and bonded gages for triaxial elasticity tests.
This method warrants further study and evaluation.
The results plotted in Fig. 24 appear quite uniform. However, it is
not clear why axial and diametric strain, which should have opposite
signs, are plotted on the same side of the zero axis. The graph does not
distinguish between positive and negative strains.
The authors are grateful for the varied and stimulating response which
their paper generated. They feel that through discussion and exchange
of ideas and experiences the objectives of the symposium are fulfilled.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreeme
T H I S P U B L I C A T I O N is one of many
issued by the American Society for Testing and Materials
in connection with its work of promoting knowledge
of the properties of materials and developing standard
specifications and tests for materials. Much of the data
result from the voluntary contributions of many of the
country's leading technical authorities from industry,
scientific agencies, and government.
Over the years the Society has published many tech-
nical symposiums, reports, and special books. These may
consist of a series of technical papers, reports by the
ASTM technical committees, or compilations of data
developed in special Society groups with many organiza-
tions cooperating. A list of ASTM publications and
information on the work of the Society will be furnished
on request.

Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreement.
Copyright by ASTM Int'l (all rights reserved); Mon Sep 21 11:53:35 EDT 2015
Downloaded/printed by
Pontificia Universidad Catolica de Chile (Pontificia Universidad Catolica de Chile) pursuant to License Agreement. No furth

The Complete Isrm Suggested Methods For Rock Characterization, Testing and Monitoring - 1974-2006 (2007)
100% (1)
The Complete Isrm Suggested Methods For Rock Characterization, Testing and Monitoring - 1974-2006 (2007)
644 pages
AGS Guide The Selection of Geotechnical Soil Laboratory Testing
No ratings yet
AGS Guide The Selection of Geotechnical Soil Laboratory Testing
132 pages
Laboratory Shear Strength of Soils-Townsend 1981
No ratings yet
Laboratory Shear Strength of Soils-Townsend 1981
720 pages
A277852-Rock Testing Handbook-Us Army
100% (5)
A277852-Rock Testing Handbook-Us Army
947 pages
Lecture2 Vertical Stress
No ratings yet
Lecture2 Vertical Stress
122 pages
CPT 2010 Vol 2
No ratings yet
CPT 2010 Vol 2
591 pages
Soil Mechanics Example Questions
100% (1)
Soil Mechanics Example Questions
14 pages
Piezometers: Types, Functions, & How It Works?
No ratings yet
Piezometers: Types, Functions, & How It Works?
38 pages
Assignment 4
No ratings yet
Assignment 4
3 pages
N66067216 Final Project Plaxis 2D
No ratings yet
N66067216 Final Project Plaxis 2D
11 pages
Previewpdf
No ratings yet
Previewpdf
63 pages
ASTM - Dinamic Geotechnical Testing
100% (1)
ASTM - Dinamic Geotechnical Testing
408 pages
STP 15 51 Permeable Concrete
No ratings yet
STP 15 51 Permeable Concrete
116 pages
STP 601-1976
100% (1)
STP 601-1976
593 pages
Eng-Choon Leong, Martin Wijaya - Laboratory Tests For Unsaturated Soils-CRC Press (2023) (2) - 1-150
No ratings yet
Eng-Choon Leong, Martin Wijaya - Laboratory Tests For Unsaturated Soils-CRC Press (2023) (2) - 1-150
150 pages
Week 6 - Lecture Notes - EC Leong
No ratings yet
Week 6 - Lecture Notes - EC Leong
117 pages
How To Control CPT Quality Data During Test
No ratings yet
How To Control CPT Quality Data During Test
154 pages
1971 Rock Mass Rating Estimation of The Mechanical Properties
No ratings yet
1971 Rock Mass Rating Estimation of The Mechanical Properties
54 pages
One-Dimensional Flow of Water Through Soils: Soil Mechanics
No ratings yet
One-Dimensional Flow of Water Through Soils: Soil Mechanics
19 pages
ASTM - Intruments and Apparatus For Soils and Rock Mechanics
No ratings yet
ASTM - Intruments and Apparatus For Soils and Rock Mechanics
177 pages
Electrokinetic Treatment of Desiccated Expansive Clay
No ratings yet
Electrokinetic Treatment of Desiccated Expansive Clay
34 pages
Anchors in Sand Bed: Delineation of Rupture Surface: K. Ilamparuthi, K. Muthukrishnaiah
No ratings yet
Anchors in Sand Bed: Delineation of Rupture Surface: K. Ilamparuthi, K. Muthukrishnaiah
25 pages
Influence Analysis of Overlying Soil Layer To Seismic - 2023 - Soil Dynamics and
No ratings yet
Influence Analysis of Overlying Soil Layer To Seismic - 2023 - Soil Dynamics and
13 pages
In Situ Testing Guide 2022
No ratings yet
In Situ Testing Guide 2022
96 pages
GeoTechnical Earthquake Engineering Handbook.
100% (1)
GeoTechnical Earthquake Engineering Handbook.
615 pages
Subsurface Exploration and Sampling of Soils For Civil Engineering Purposes 2 of 4
100% (1)
Subsurface Exploration and Sampling of Soils For Civil Engineering Purposes 2 of 4
142 pages
Isrm SM Quantitative Description of Discontinuities - 1978
No ratings yet
Isrm SM Quantitative Description of Discontinuities - 1978
50 pages
Livro - Advanced Triaxial Testing of Soil and Rock - Menzies
No ratings yet
Livro - Advanced Triaxial Testing of Soil and Rock - Menzies
905 pages
PLAXIS 3D Reference Manual 2018
No ratings yet
PLAXIS 3D Reference Manual 2018
516 pages
IES CE O II 2001 (Gate2016.Info)
0% (1)
IES CE O II 2001 (Gate2016.Info)
15 pages
Cone Penetration Testing 2022
No ratings yet
Cone Penetration Testing 2022
1,205 pages
Wittke 1990
No ratings yet
Wittke 1990
549 pages
Triaxial Permeability Test: Specimen Details Saturation Details
No ratings yet
Triaxial Permeability Test: Specimen Details Saturation Details
3 pages
Topic 4
No ratings yet
Topic 4
26 pages
Castro 1969 Thesis - Liquefaction of Sands
100% (1)
Castro 1969 Thesis - Liquefaction of Sands
127 pages
Geotechnics of Fills
100% (1)
Geotechnics of Fills
376 pages
Consolidation of Soils Testing and Evaluation Editor(s), Yong Townsend STP 892
No ratings yet
Consolidation of Soils Testing and Evaluation Editor(s), Yong Townsend STP 892
383 pages
TC ĐKT Nhật 1 PDF
100% (1)
TC ĐKT Nhật 1 PDF
336 pages
Seismic Performance of Two Classes of Ea
No ratings yet
Seismic Performance of Two Classes of Ea
20 pages
Testing and Specification of Recycled Materials For Sustainable Geotechnical Construction - STP1540
No ratings yet
Testing and Specification of Recycled Materials For Sustainable Geotechnical Construction - STP1540
646 pages
Geotechnics of Waste Fills
100% (1)
Geotechnics of Waste Fills
376 pages
Vibrating Wire Piezometer - Product Details
No ratings yet
Vibrating Wire Piezometer - Product Details
2 pages
The Measurement of Soil Properties in The Triaxial Test
No ratings yet
The Measurement of Soil Properties in The Triaxial Test
236 pages
Livro Propriedades Dos Solos e Suas Mediçoes
No ratings yet
Livro Propriedades Dos Solos e Suas Mediçoes
271 pages
Vibrating Wire Piezometer
No ratings yet
Vibrating Wire Piezometer
4 pages
Capillary Rise Defnition
No ratings yet
Capillary Rise Defnition
4 pages
Coduto - Soils Book
No ratings yet
Coduto - Soils Book
826 pages
Zhen-Yu Yin, Pierre-Yves Hicher, Yin-Fu Jin - Practice of Constitutive Modelling For Saturated Soils-Springer Singapore - Springer (2020)
100% (1)
Zhen-Yu Yin, Pierre-Yves Hicher, Yin-Fu Jin - Practice of Constitutive Modelling For Saturated Soils-Springer Singapore - Springer (2020)
417 pages
Caliifornia Geology Magazine Sep-Oct 1993
No ratings yet
Caliifornia Geology Magazine Sep-Oct 1993
36 pages
43 - Geotechnical Engineering 3
No ratings yet
43 - Geotechnical Engineering 3
3 pages
PLAXIS 2D 2017 Tutorial Lesson07
No ratings yet
PLAXIS 2D 2017 Tutorial Lesson07
13 pages
Field Instrumentation For Soil
100% (1)
Field Instrumentation For Soil
420 pages
Bishop Triaxial Test PDF
100% (2)
Bishop Triaxial Test PDF
202 pages
State of The Art and Practice in The Assessment of Earthquake-Induced Soil Liquefaction and Its Consequences
No ratings yet
State of The Art and Practice in The Assessment of Earthquake-Induced Soil Liquefaction and Its Consequences
297 pages
Geotechnical Instrumentation News - John Dunnicliff
100% (2)
Geotechnical Instrumentation News - John Dunnicliff
16 pages
3 - Load Testing of Deep Foundations - Carroll L. Crowther PDF
No ratings yet
3 - Load Testing of Deep Foundations - Carroll L. Crowther PDF
123 pages
Bray Olaya 2023 2022 H Bolton Seed Memorial Lecture Evaluating Liquefaction Effects
No ratings yet
Bray Olaya 2023 2022 H Bolton Seed Memorial Lecture Evaluating Liquefaction Effects
23 pages
Validation of Geotechnical Finite Element Analysis
No ratings yet
Validation of Geotechnical Finite Element Analysis
6 pages
Piezometers For Groundwater Measurement and Monitoring
No ratings yet
Piezometers For Groundwater Measurement and Monitoring
10 pages
Settle3D: Settlement and Consolidation Analysis
No ratings yet
Settle3D: Settlement and Consolidation Analysis
60 pages
Geo Pressure
No ratings yet
Geo Pressure
27 pages
Geotechnical Applications PDF
100% (2)
Geotechnical Applications PDF
325 pages
Igs Tc04 Gi Manual 2016
No ratings yet
Igs Tc04 Gi Manual 2016
45 pages
Pneumatic Piezometers: TEL 604 540 1100 RST Instruments Ltd. 11545 Kingston ST., Maple Ridge, BC V2X 0Z5 Canada
No ratings yet
Pneumatic Piezometers: TEL 604 540 1100 RST Instruments Ltd. 11545 Kingston ST., Maple Ridge, BC V2X 0Z5 Canada
2 pages
Application of Seismic Refraction Survey in Assessment of Jointing - Palmstrom
No ratings yet
Application of Seismic Refraction Survey in Assessment of Jointing - Palmstrom
8 pages
Lecture 2 - Theory of Soil Strength
No ratings yet
Lecture 2 - Theory of Soil Strength
35 pages
Soil Testing, Soil Stability and Ground Improvement: Wissem Frikha Serge Varaksin Antonio Viana Da Fonseca Editors
100% (2)
Soil Testing, Soil Stability and Ground Improvement: Wissem Frikha Serge Varaksin Antonio Viana Da Fonseca Editors
419 pages
TMW22 - Evaluation of Flow Liquefaction Susceptibility and Liquefied Strength - A Case Study
No ratings yet
TMW22 - Evaluation of Flow Liquefaction Susceptibility and Liquefied Strength - A Case Study
12 pages
The Geological Society of America: Application of Geology To Engineering Practice
No ratings yet
The Geological Society of America: Application of Geology To Engineering Practice
22 pages
An Introduction To Geotechnical Engineering PDF
No ratings yet
An Introduction To Geotechnical Engineering PDF
746 pages
Topic 4: Topic 4: Shear Strength of Clays: CVEN 7718 CVEN 7718 Engineering Properties of Soils Spring 2010
No ratings yet
Topic 4: Topic 4: Shear Strength of Clays: CVEN 7718 CVEN 7718 Engineering Properties of Soils Spring 2010
62 pages
Relevance of The Triaxial Test Bishop
100% (3)
Relevance of The Triaxial Test Bishop
69 pages
Astm D4829
No ratings yet
Astm D4829
4 pages
Nonlinear Analysis of Stress and Strain in Soils B.j.duncan
No ratings yet
Nonlinear Analysis of Stress and Strain in Soils B.j.duncan
13 pages
CE 353 Geotechnical Engineering: in Situ Stresses
No ratings yet
CE 353 Geotechnical Engineering: in Situ Stresses
12 pages
Shear Strength of Soils
100% (1)
Shear Strength of Soils
13 pages
Department of Mining Engineering: Course Title: Rock Mechanics Topic: Presentation On Field Tests in Rock
No ratings yet
Department of Mining Engineering: Course Title: Rock Mechanics Topic: Presentation On Field Tests in Rock
14 pages
D653 StandardTerminology PDF
No ratings yet
D653 StandardTerminology PDF
35 pages
Lecture 4 Effective Stress
86% (7)
Lecture 4 Effective Stress
18 pages
A New Ring Shear Apparatus and Its Applicationt o The Measurement of Residual Strength
No ratings yet
A New Ring Shear Apparatus and Its Applicationt o The Measurement of Residual Strength
58 pages
2016 - Limit Equilibrium Method For Rock Slope Stability Analysis by Using The Generalized Hoek-Brown Criterion PDF
No ratings yet
2016 - Limit Equilibrium Method For Rock Slope Stability Analysis by Using The Generalized Hoek-Brown Criterion PDF
9 pages
SHANSEP METHOD New Design Procedures For Stability of Soft Clays by Ladd
100% (1)
SHANSEP METHOD New Design Procedures For Stability of Soft Clays by Ladd
13 pages
Empirical Correlations Drained Shear Strength For Slope Stability Analyses
No ratings yet
Empirical Correlations Drained Shear Strength For Slope Stability Analyses
10 pages
ASTM D4767 (TriaxialCU) PDF
No ratings yet
ASTM D4767 (TriaxialCU) PDF
11 pages
Aashto T 310 - Ingles
No ratings yet
Aashto T 310 - Ingles
10 pages
Correl Soils Properties Carter&Bentley
100% (22)
Correl Soils Properties Carter&Bentley
136 pages
As 1726-1993 Geotechnical Site Investigations
No ratings yet
As 1726-1993 Geotechnical Site Investigations
8 pages
Bieniawaski Z. T.
No ratings yet
Bieniawaski Z. T.
11 pages
From Bedrock to Blueprint, Reminiscences of an Engineering Geologist
From Everand
From Bedrock to Blueprint, Reminiscences of an Engineering Geologist
P.L. Narula
No ratings yet
Geotechnical investigation The Ultimate Step-By-Step Guide
From Everand
Geotechnical investigation The Ultimate Step-By-Step Guide
Gerardus Blokdyk
No ratings yet

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.