-
EFFECT OF RATE OF LOADING ON THE STRENGTH OF CLAYS AND SHALES AT
CONSTANT WATER CONTENT
Professor A. CASACRANDE and S. D. WILSON
SYNOPSIS Investigations performed at Harvard Univer- Des
recherches effectuees B 1UniversiM de Har-
sity during the past three years show that some vard, au tours
de ces trois dernieres arm&es, mon- types of brittle
undisturbed clays and clay shales trent que quelques types dar es
creep under a sustained load, and that they ulti- !?
et de schistes argileux friables non remues c eminent sous
une
mately fail under a sustained load appreciably less charge
sontenue, et quils finissent par seffondrer
than the strength indicated by a normal laboratory sous une
charge soutenue bien inferieure a la r&&t-
compression test. The shear strengths of six such ante indiquee
par un essai normal de compression
soils were found to be reduced to 40-30 per cent. en
laboratoire. On a trouve pour six de ces sols
of their normal values in thirty days. This may que leur
resistance au cisaillement se trouvait 16 duite de 40 B 80 pour
cent de sa valeur normale en
explain some slides which have developed on slopes 30 jours.
Ceci peut expliquer quelques glissements that stood for many years
without noticeable move- qui se sont prod&s sur des pentes qui
&Gent ment. demeur&s pendant de nombreuses arm&es
sans
In contrast, it was found that two laboratory *uvement
appreciab1e. compacted soils, and one undisturbed soil which Par
centre, il a et6 trouve que deux sols com- was not fully saturated,
tended to become stronger presses en laboratoire, et un sol non
rem& qui
and stiffer under sustained loads, even though n&ait pas
completemerit satme, avaient tendance
water content was kept constant. These results B devenir plus
r&&ta&s et plus fermes sous des
may prove of value in connexion with the design chargas
soutenues, quoique la teneur en eau ait
of embankments. et6 maintenue constante. Ces r&hats peuvent
avoir une certalne valeur pour letude des remblais.
INTRODUCTION
Investigations of the effect of rate of loading on the
compressive strength of clays and shales at constant water content
have been in progress at Harvard University since 1947. The first
two years of these investigations were devoted to a study of the
stress-deformation and strength characteristics of soils and soft
rocks under very rapid loading and unloading (Casagrande and
Shannon, 1949 and 1949).
Since the autumn of 1949 the effect of slow rates of load
application has been studied under a cooperative research contract
between the Waterways Experiment Station and Harvard University
(Casagrande and Wilson, 1949 and 1959). This investigation has
shown that some types of brittle undisturbed clays and clay shales
creep under a sustained load, and that they ultimately fail under a
sustained load appreciably less than the strength indicated by a
normal laboratory compression test. In contrast, it was found that
two laboratory-compacted soils, and one undisturbed soil which was
not fully saturated, tended to become stronger and stiffer under
sustained loads ; that is, when loaded at very slow rates of load
application, they failed under stresses considerably higher than
would be indicated by normal laboratory tests.
This Paper describes apparatus and test procedures developed to
perform such long-time tests at constant water content, and
summarises the results for nine different soils.
DESCRIFTION OF TESTS
Two types of tests were used to study the effect of long-time
loading on the compressive strength of soils : (1) creepstrength
tests, and (Z), long-time compression tests.
A creepstrength test is one in which a load is built up quickly
and maintained constant until the specimen fails. For such tests,
time to failure refers to the elapsed time between application of
the load and failure. The following nomenclature is used to
describe these tests :
251
-
252 A. CASAGRANDE AND S. D. WILSON
U denotes unconfined creep-strength test. 51 8) quick triaxial
creep-strength test in which the specimen is first subjected to
hydrostatic pressure and then loaded. No drainage is permitted
during the entire test.
QC *I consolidated-quick triaxial creep-strength test, in which
the specimen is first consolidated under a given hydrostatic
pressure and then, without permitting further consolidation, is
subjected to the sustained load.
A longtime compression test is one in which the specimen is
subjected to incremental axial loading, the elapsed time between
increments of load varying for different tests. For such tests,
time of loading refers to the time which elapses between the
application of the first load increment and failure. All long-time
tests reported herein are of the unconfined type, and are
classified according to time of loading,by using subscripts, as
shown in Table 1.
Table i
Symbol
-
L_
-
Type of unconfined compression test - Fast transient unconfined
compression test Transient unconfined compression test Fast
unconiined compression test Normal uncordined compression test Slow
unconfined compression test Very slow unconfined compression
test
Time of loading
Less than 1 second 1 to 30 seconds 0.5 to 3 minutes 3 to 15
minutes 15 minutes to 8 hours Longer than 8 hours
DESCRIPTION OF APPARATUS AND TEST PROCEDURES
It was necessary to develop simple and inexpensive equipment in
order to perform many tests, some of which lasted from 6 months to
more than a year. The apparatus shown in Fig. 1 was found blest
suited for both U and U, tests. It utilises ball bushings to guide
the piston, thus preventing tilting and eccentric loading. The ball
bushings have the advantage over other types of bushings in that
piston friction is reduced to a tolerable amount even when lateral
loads are applied to the piston.
For Q and I& tests, conventional triaxial equipment was
generally used, although this was later modified by the
substitution of ball bushings for plain bushings.
For all tests reported herein it was of vital importance that
the specimens were maintained at unchanged water content, since
even a small loss of water would cause a substantial increase in
strength. For unconfined compression tests, this was accomplished
by encasing the speci- mens in two thin membranes which were
separated by a tbin coating of silicone grease. In addition, the
membrane covered specimens were totally immersed in water for the
duration of the tests. All specimens were weighed both before and
after testing and it was found that the change in water content, if
any, seldom exceeded O-5 per cent. even if the tests lasted several
months.
Most undisturbed specimens were cut cylindrical in shape, l-4
inch in diameter and 3.5 inches long, by sti table trimming
devices.
The compacted specimens were prepared in the Harvard compaction
apparatus (Wilson, 1950). This apparatus uses a mould l& inch
in diameter and 2616 inches long and permits precise control of
both moulding water content and unit weight. For this investiga-
tion, the specimens were compacted at optimum moisture content by a
standard procedure. After compaction, the specimens were extruded
from the mould with as little disturbance as possible, and then
tested in the same manner as the undisturbed specimens.
All loads were applied directly by means of a dead weight on a
hangar which was supported by the piston. In unconfined tests of
the U, U, and U,, types, the load was applied in small increments
at regular time intervals. The increments of load were chosen such
that each.was
-
EFFECT OF RATE OF LOADING ON STRENGTH OF CLAYS AND SHALES 253
ng.1
komron A-4812
0 I2 s 4 Lt. 1
SCALE INCHES
Unconihed compremionapparatu~~withbailbumbiq guidaforpiston
about 10 to 15 per cent. of the expected failure load, and in
most tests the size of increment was kept constant.
In creep-strength tests the load was applied in from four to
eight increments at short intervals of time, 1 minute being the
most common interval. The size of the increment and the time
interval were the same for all tests in a particular series. The
load was then maintained constant until failure occurred.
DESCRIPTION OF SOILS TESTED
The following soils were investigated : Cambridge clay.-A
grey-green, medium plastic, inorganic clay from North
Cambridge,
Massachusetts ; it is medium stiff and brittle in the
undisturbed state, and soft and sticky when remoulded. The sample
tested had a liquid limit of about 42, a plastic limit of 21, and a
natural water content of about 37 per cent.
-
254 A. CASAGRANDE AND S. D. WILSON
Ohio River sandy c&y.-This is a fairly homogeneous light
brown sandy,clay of low plasticity from near Cincinnati, Ohio. It
is medium stiff and somewhat brittle in the undisturbed state, and
soft when remoulded. The sample tested had a liquid limit of about
32, a plastic limit of 21, and a natural water content of about 27
per cent. The degree of saturation is about 95 per cent.
O&r be&&&--This is a fairly hard and very
brittle bentonite from a bed exposed during the driving of an
exploratory tunnel in the Pierre shale for the Oahe dam in South
Dakota. It is mottled, grey-green in colour and .has a soapy feel.
The samples tended to part rather easily along inclined planes
which appeared to be existing joints. These planes of weakness were
avoided whenever possible in cutting the specimens.
Mexico City day.-An extremely compressible, bentonitic clay of
volcanic origin, light grey-brown in colour, and having a natural
water content of approximately 490 per cent. It is medium soft, yet
brittle, in the nndisturbed state, and extremely soft and sticky
when thoroughly remoulded. It has a liquid limit of the same order
as the natural water content.
C#carrrcka c&ay-&&.-A mottled grey-green and dark
grey, soapy, slickensided clay-shale from the Panama Canal Zone ;
this contains slickensided surfaces of varying dimensions, some
apparently open and uncemented, other closed and apparently
recemented. The material in different cores, and occasionally in
the same core, varied in colour, hardness and jointing.
Bearpaw day-shak.-This is a dark grey, slightly organic
clay-shale, from the South Saskatchewan River Project, Canada, very
stiff and brittle with numerous slickensides in the undisturbed
state, of sa consistency when remoulded at natural water content.
It is highly plastic and has considerable toughness at the plastic
limit.
Mi.wissi&%gnm& .-A highly plastic alluvial clay with a
liquid limit of about 94, a plastic limit of 31, and a natural
water content ranging from about 48 to 56 per cent. It is firm and
somewhat brittle in the undisturbed state and shows little loss of
strength when remoulded.
C&yey sand.This is a subgrade material consisting of a field
mixture of a fairly uniform coarse sand with a small amount of a
highly plastic clay, from Clinton, MissiGppi. The liquid limit is
about 18 and the plastic limit 16. This material was thoroughly
investigated in the field and the results were reported by the
Waterways Experiment Station (1949).
Silty clay.-A lean silty clay from Vicksburg, Mississippi, with
a liquid limit of about 37 and a plastic limit of 23. It was also
investigated by the Waterways Experiment Station (1949).
ANALYSIS OF DATA
INSTANTANEOUS MODULUS OF DEFORMATION
The deformatron which accompanied each increment of load was
obtained by means of exteusometer readings at speci&d time
intervals following the application of load. The following sequence
was generally used : 2, 5, 10, 15, and 39 seconds ; I, 2,4,8, 15,
and 39 minutes, and so forth. These extensometer readings were then
plotted against elapsed time. Detailed study and analysis of these
time curves showed that the application of each load increment was
accompanied by a sudden deformation, indicated by a jump in the
extenso- meter readings, followed by creep deformation which
decreased rather quickly with time. The magnitude of the sudden
deformation was estimated by extrapolating the time curves back on
a smooth curve to zero time. For any one specimen this sudden
deformation was found to be proportional to the magnitude of the
load. The ratio of the total stress to the summation of the
instantaneous strains is referred to as the instantaneous moduhrs
of deformation (Ms) (Fig. 2).
Apparatus was later developed to determine the mod&s of
elasticity of soils by the dynamic method, &ii forced
vibrations to produce resonance. Comparative tests showed excellent
agreement between the instantaneous modulus of deformation as
defined above and the modulus of elasticity computed from the
vibration procedure.
-
EFFECT OF RATE OF LOADING ON STRENGTH OF CLAYS AND SHALES
255
Fig.2
STRESS: YGPLI) saCY 0. 0s ob 0,
EQQRCT OF SLOW RATES OF LOADING ON THE STRESS-STRAIN CURVE
It Was found that the stress-strain curves of materials which
exhibited loss of strength in long-time tests showed a decreasing
secant modulus of deformation with decreasing rate of loading. Fig
2 compares the stress-strain curves of three specimens of Mexico
City clay with different rates of loading.
The stress-strain curve for Specimen C has been separated into
elements of creep and sudden strain. The dashed line shows the
cumulative sum of the sudden strains, for which Mi has been
computed to be 53 kilograms per square centimetre. For this same
specimen the secant modulus of deformation at 50 per cent of
compressive strength was computed to be 22 kilograms per square
centimetre. The decrease in compressive strength and increase in
strain at failure with slow rates of loading, as shown by this
plot, was found to be typical for ah tests on Mexico City clay, as
well as for other clays and the clay-shales tested.
Fig. 3 shows a typical time curve for each of the three tests
shown in Fig. 2. The dashed portions of these curves show how they
were extrapolated back to zero time. For normal rates of loading
(Specimen A) the time curves were sharply bent near the origin, but
for long- time tests (Specimen C) the time curves were quite
flat.
The stress-strain curves of materials that gained in strength in
long-time tests showed an mcrease in secant modulus of deformation.
Typical of this behaviour are the results of long- time tests on
the clayey sand, for which four stress-strain ewes are plotted in
Fig. 4. No information has yet been obtained on the influence of
the rate of loading on Mi.
-
256 A. CASAGRANDE AND S. D. WILSON
w-3
ELAPSEDTIML:SECONDS
Typicaltime/deformationcurveefor Mexico City clay
STRESS: KG PER saxht.
Typical mtramm-mtrahcurvemfor U, Us, and U,,t@~ti on
compact& clayey sand
-
EFFECT OF RATE OF LOADING ON STRENGTH OF CLAYS AND SHALES 2.57
CREEP-STRENGTH TESTS
In creep-strength tests failure was invariably preceded by a
reversal of slope of the time- deformation curve, followed by
continuous deformation at an increasing rate. Through the rubber
membranes around the specimens it was often possible to see the
shear crack develop shortly after the reversal of slope in the time
curve took place. For these reasons, time to failure in
creep-strength tests has been defined as the elapsed time between
the application of the final load increment and this reversal in
slope. Typical time curves illustrating this phenomenon are shown
in Figs 5, and the corresponding stress-strain curves in Fig.
6.
Fig. 7 is a typical plot of compressive strength against time to
failure for a series of U tests on Mexico Clay. Within the range of
time to failure of from 1 minute to 36 days, the best relationship
was found to be a straight line on semilog plot. Deviations of
individual tests may possibly be explained by non-homogeneity of
the sample from which the specimens
Dial readiq/elapmed time : $jc tests on Cucaracha clay-ohab
-
258 A. CASAGRANDE AND S. D. WILSON
DKVlAlUR STRKSS:KG.PSR SO.CU. 0 t b 8 IO II I.
0
.I 0
were cut, as it was found that there were some differences in
deformation characteristics between specimens even when test
procedures were identical.
It is emphasised that, even though the relationship shown in
Fig. 7 is best represented by a straight line, straight-line
extrapolation for duration of sustained loads longer than 1 month
is not justified, as this line probably curves and approaches a
horizontal asymptote. From the results of later creep-strength
tests in which failme did not occur, and in which all creep ceased,
it is believed that the lower limit of compressive strength for
this sample of Mexico City clay is of the order of O-45 to 055
kilogram per square centimetre.
LONG-TIME UNCONFINED COMPRESSION TESTS
It was found that for some compacted materials creep would cease
in a few days or weeks, even for sustained loads only slightly
lower than the normal compressive strengths. The compressive
strengths of these materials were found to have been increased as a
result of the sustained loads. For these materials it was necessary
to run a series of long-time tests. Fig. 8 is a plot of compressive
strength against time of loading for a series of long-time tests on
specimens of the silty clay compacted at optimum moisture
content.
-
EFFECT OF RATE OF LOADING ON STRENGTH OF CLAYS AND SHALES
259
.
s
c.
i .L
compml#aive mtaagthpms to tailura : v kmtm on Mexico city
clay
SUMMARY OF RESULTS ON NINE SOILS
STRENGTH RATIO
The effect of long-time and of sustained loading on the
compressive strengths of the nine soils tested is summarized in
Figs 9 and 10. Fig. 9 is a plot of strength ratio against dura-
tion of sustained load for six undisturbed soils in creepstrength
tests, and Fig. 10 is a plot of strength ratio against time of
loading in long-time unconfined compression tests on four soils.
Strength ratio is defined as the ratio of the compressive strength
obtained in an actual test to the estimated compressive strength
corresponding to a normal rate of loading. In creepstrength tests a
time to failure of 1 minute has been used, and for long-time tests
a time of loading of 10 minutes. These times, although arbitrary,
correspond approximately to normal laboratory test procedures.
CREEP-STRENGTH TESTS
For creepstrength tests at constant water content, the shear
strength of Mexico City clay was reduced to about 90 per cent of
its normal value in about 30 days, while that of the Cucaracha
clay-shale was reduced to approximately 40 per cent. The shear
strength of four other soils was reduced in between these values,
as shown in Fig. 9.
With the possible exception of the Cucaracha clay-shale, for
which data were not obtained, these six soils are considered to be
fully saturated and therefore it is assumed that there was no
change in void ratio during these tests.
With the exception of the Mississippi gumbo (for which the data
were somewhat erratic and the results inconclusive for tests
lasting more than one day), these soils were brittle and gave
well-defined shear fractures.
LONG-TIME UNCONFINED TESTS
For long-time unconfined tests, also at constant water content,
one undisturbed sandy clay and two compacted soils showed a
substantial increase of strength with times of loading
-
260 A. CASAGRANDE AND S. D. WILSON
Compressive strength/time of loading : long-time unconfined
compression tests on compacted silty clay
Fig. 9
Strength ratio/elapsed time to failure : creep-strength tests at
constant water content on six undisturbed soils
-
EFFECT OF RATE OF LOADING ON STRENGTH OF CLAYS AND SHALES
261
Fig. 10
TlUE OF LOADING: MlNlrTES
Streqth ratio/time of loading for long-time unconfined
compression tests at constant water content on four soils
greater than about 1 day, and a small loss of strength for time
of loading greater than 10 minutes but less than 1 day. These three
soils were not fully saturated and it is probable that the increase
in strength was accompanied by a decrease in void ratio at constant
water content. This means that internal consolidation has occurred,
the gas in the voids being compressed and absorbed by the water. No
attempts were made in this investigation to measure this change in
void ratio.
Possibly the strength of the two compacted soils was increased
by thixotropic action. The effect of thixotropy was not
investigated.
CREEP EFFECTS
When analysing stress-deformation characteristics for the
various materials tested, it appeared that materials which were the
most susceptible to loss of strength in creep-strength tests also
showed a relatively low ratio of secant modulus of deformation to
instantaneous modulus. The trend of this relationship is shown in
Fig. 11 for four of the soils tested. In Fig. 11 the abscissae
represent the slope of the lines shown in Fig. 9, that is, the
decrease in strength ratio per cycle of time, and the ordinates
represent the ratio of secant modulus of deformation (at 30 per
cent. of normal compressive strength) to instantaneous modulus.
Individual tests have not been plotted in Fig. 11 and the ranges
shown are only approximate. It is interesting that, of the four
materials, the Cucuracha clay-shale was the strongest, yet showed
the greatest relative amount of creep. In contrast, the Mexico City
clay had the lowest compressive strength, yet showed the least
amount of relative creep.
CORRELATION OF CREEP-STRENGTH TESTS WITH TRANSIENT TRSTS
It was possible to correlate the results of creep-strength tests
on Cucaracha clay-shale with results of transient tests as
determined in the Panama Research Project. To do this, it was
R
-
262 A. CASAGRANDE AND S. D. WILSON
Fig. 11
/ I
I
, I I
ct 6 !
Oahc kntontte
Cucoracha clay-shok I / -
L I -.
v, . . . .
. -. \ \ . \ . I
I .
-__ %
4 .
, gproximata relationship between creep and loss of strength :
creep-strength to&n on four soilll
Fig. 12
TIME 10 FAILURE : UmTES -. lo lo- I IO IO Id
Effect of rate of loading on the compramsive mtrmgth of
Cucaracha CliIy-mhalO
-
EFFECT OF RATE OF LOADING ON STRENGTH OF CLAYS AND SHALES 263
necessary to correlate the time of loading as defined in the Panama
Report with the elapsed time to failure as defined herein. The
results are summarized in Fig. 12, which covers the entire range of
time from O*OOl second to 1 year. Within this range the strength
ratio varies from about 2-O for the fastest time of loading to
about 02.5 for the slowest.
CONCLUSIONS
The following tentative conclusions can be drawn from the tests
performed during this investigation :
(1) Sustained loads at constant water content reduce the
strength of fully saturated, brittle clays and clay-shales. This
may explain, at least in part, slides which develop on slopes that
have been standing for many years without noticeable movement, such
as, for example, the slides in the Cucaracha formation along the
Panama Canal (Casagrande, 1949).
(2) The strengths of compacted soils and of undisturbed soils
which are not fully saturated increase with time, even when the
water content is kept constant. From the standpoint of earth dam
design these results have important practical implica- tions. It
would seem that most compacted soils of which earth dams are made
increase in strength with time, even without additional
consolidation.
It is realized that the results obtained will have to be
verified by additional investigations using triaxial compression
tests, rather than unconfined tests. There is also a possibility
that even at unchanged water content there is a redistribution of
water content within a sample which may change its strength
characteristics in a manner not necessarily representative for the
effects which take place within a large mass. The Authors realize
these and other limita- tions of the present work which are all
part of the great complexity of strength properties of soils, of
which probably only little is really understood.
ACKNOWLEDGEMENT
The investigations described herein were performed under a
cooperative research contract with the Waterways Experiment
Station, Department of the Army, Vicksburg, Mississippi. The tests
were performed in the Soil Mechanics Laboratory at Harvard
University by various members of the research staff. In particular,
J. M. Corso, G. J. Kling and H. B. Seed, Re- search Associates,
have contributed materially to the success of the
investigations.
REFERENCES CASAGRANDE. A.,and SHANNON,W. L., 1948. Research on
stress-deformation and strength characteristics
$~i; and soft rocks under transient loading. Pub. Haward Univ.
Grad. Sch. Eug. Soil Mech. Series, . . 132 pp.
CASAGRANDE, A., and SHANNON, W. L., 1949. Strength of soils
under dynamic loads. Trans Amer. SOC. Civ. Eng. 114 : 755-772.
CASAGRANDE, A., and WILSON. S. D., 1949. Final report to U.S.
Waterways Experiment Station on investieation of effect of
lona-time loading on the strength of clays and shales at constant
water conten;. Harvard Univevsity_ 77 pp. -
CASAGRANDE, A., CORSO, J. M., and WILSON, S. D., 1959. Report to
Waterways Experiment Station on the 1949-1959 program of
investigation of effect of long-time loading on the strength of
clays and shales at constant water content. Harvard University.
WILSON, S. D.. 1950. Small soil compaction apparatus duplicates
field results closely. Engineering News- Record. 145: 18:
34-36.
CASAGRANDE, A., 1949. Discussion of the paper on excavation
slopes by Binger and Thompson, symposium on the Panama Canal-the
sea level project. Trams. Amer. Sot. Civ. Eng. 114 : 870-874.
U.S. War Department, 1949. Soil compaction investigation. ?ortno
l*
Compaction studies on clayey sands, Tech memo. U.S. Watem. Expt.
Sk No. 271-3. 4 pp.
U.S. War Department 1949. Soil compaction investigation, report
no. 2. Compaction studies on silty clay, U.S. Waterw. Expt.
Sta.