Professional Documents
Culture Documents
FUNDAMENTAL CONCEPTS
EFFECTIVE STRESS
The fundamental principle underlying all work on the shear strength of soils is that
the mechanical properties are controlled by the intergranular forces. If “” denotes
the total normal stress acting on any plane in a soil and if “u” the pressure in the pore
space, then, to a close approximation, the effective stress ’ is given by the equation:
’ = – u …. (1)
In this expression ’ is the sum of the intergranular forces resolved normal to the
plane per unit area of the plane.
An examination of this equation is given in a recent paper by Bishop and Eldin (1950),
but here it is necessary only to note that equation (1) has been substantiated
experimentally by Terzaghi (1936) for sand clay and concrete, by Rendulic (1937) and
Taylor (1944) for clays, and by Bishop and Eldin for sands. There is, therefore, no
reason to doubt the general validity of the principle of effective stress.
TRUE COHESION AND INTERNAL FRICTION
In 1773 Coulomb suggested that the shear strength characteristics of a soild could be
represented by the expression:
s = c + tan
where, in modern nomenclature,
“s” denotes shear strength
“c” denotes apparent cohesion
“” denotes total pressure normal to shear plane
“” denotes angle of shearing resistance
In Coulomb's equation “c” and “” are empirical constants the values of which, for any
given soil, depend upon the conditions of test or the conditions under which the soil is
stressed in the field; in particular, whether or not the soil is allowed to drain during
the shearing process.
In a more fundamental form, first put forward by Hvorslev in 1937, Coulomb's
equation is rewritten
s =Ce + (s — u) tan e
where Ce= denotes true cohesion
e= denotes true angle of internal
The cohesion and internal friction will, in general, depend upon the void ratio of the
soil at the instant of shear failure.
For purposes of simplicity the rate of strain is taken as an independent variable and
its efíect on the shear strength will be considered separately. Existing data do not
justify any more elaborate treatment, and, so far as the authors are aware, no serious
error is thereby introduced.
In any isotropic soil the cohesion is a non-directional property and it may be regarded
as the resultant of the physico-chemical forces acting between the particles. In clean
sands these forces are negligibly small, whilst in clays they are important. Indeed,
there is evidence to suggest that in bentonite the shear strength is composed almost
wholly of true cohesion (Skempton, 1948).
Internal friction is derived principally from the actual friction of grain on grain. It is,
however, also taken as including the resistance to shear developed as a result of the
work which has to be done when the soil changes volume during shear. Internal
friction is not itself a directional property, but in the general case of an element under
unequal principal stresses the shear resistance along different planes will vary in
accordance with the variation in , and hence the internal friction imparts directional
properties to the soil.
Whether or not this view of the physics of shear strength is wholly acceptable, the
Coulomb- Hvorslev equation has been confirmed experimentally, at least as a first
approximation, by Hyorslev (loc. cit), Peynircioglu (1939) and recent research at
Imperial College. In the Authors' opinion it must form the basis of any fundamental
consideration of shear strength.
ANGLE OF SHEARING RESISTANCE
The measurement of “ce”, and e is, however, a matter of some difficulty in many soils
and, moreover, an analysis of stability based on these basic properties is likely to be
rather elaborate in practical use. Ultimately such analyses will be carried out, as has
been done already in a provisional way for certain relatively simple cases by
Skempton (19485) and Hansen and Gibson (1949). But at the present time the usual
method of procedure is:
(i) To determine the values of “c” and ““in Coulomb's equation under definite
conditions of water-content change during shear; namely, no water-content
change (un-drained test), or full water-content change (drained test).
(ii) To use these values of c and “” as if they denoted the cohesion and internal
friction of the soil : where practical conditions approximate to zero water-content
change the undrained test parameters are used, with respect to total stresses:
where practical conditions approximate to full water-content change the drained
test parameters are used, with respect to effective stresses : where practical
conditions cannot be even approximately represented by either drained or
undrained tests, both tests should be made and the appropriate strength deduced
from a knowledge of the degree of consolidation under field conditions.
(c) small errors as compared with a more rigorous analysis based on “ce”, and “e”,
and a knowledge of the pore pressure “u”.
Of these (a) is undoubted, and there is some evidence for (b) ; but principally from
that type of problem which has been shown by Skempton and by Hansen and Gibson
to conform also to (c)—namely those problems where a saturated clay is stressed
without any change in water content and “ = 0” with respect to applied stresses. One
of the most urgent needs in the subject at the present time is thus for a theoretical
examination, and practical results, concerning the reliability of the present procedure
in those cases where a water content change precedes the stage for which it is desired
to calculate the factor of safety (typical cases include the foundation stability of slowly
constructed earth dams, rapid draw-down, and the stability of roads and runways).
But even in the absence of such results the present conventional procedure will
probably continue to be used ; and the purpose of the more advanced theoretical
examination will be to provide information on the corrections which must be applied
in the conventional analyses. Consequently the Authors will consider the various tests
by which the values of “c”, the apparent cohesion, and “” the angle of shearing
resistance, can be measured on samples of the principal soil types.
SAMPLING
No matter how correct a theory of shear strength may be, and no matter how carefully
the laboratory tests are carried out, the results will be misleading unless samples have
been taken which are closely representative of the soil as it exists in Nature.
Consequently it is of the utmost importance to obtain undisturbed samples, or in the
case of fills and sub-grades to predict with reasonable accuracy the state of the soil
during and after construction,
Unless this is possible, the testing programme is likely to yield results of very. limited
value. For example, it appears that it is difficult, or even impossible, to obtain
undisturbed samples of sensitive silty clays from depths of more than about 50 feet
(Carlson, 1948 : Skempton, 1948c). For such cases the vane test has been developed
(see Section VI) and it would be inadvisable to carry out any elaborate laboratory
shear tests to determine the in-situ strength. As a second example, the problem of
predicting the moisture contents in a clay subgrade some years after construction is
still a matter of practical difficulty, although recent work (Croney, Lewis, and
Coleman, 1950) at the Road Research Laboratory shows great promise in this
direction. In highway and airfield work, this is a far more important matter than any
refinements of testing technique or stability analysis.
Nevertheless there is a wide range of practical problems where the conditions are
known with sufficient accuracy and adequate samples can be obtained. lt is with these
cases that this Paper is mainly concerned. The techniques of sampling lie outside the
scope of the present discussion.
PRINCIPAL SOIL TYPES
From the point of view of studying shear strength, soils may be classified into five
principal types. Any one of these types may be either fully or partially saturated ; and
its properties have to be considered in either the undisturbed or remoulded states.
Broadly speaking, undisturbed soils beneath ground-water level are fully saturated;
and partially saturated when above water level. Soils used in the construction of dams,
embank- ments, or subgrades are generally remoulded and, at least initially, only
partially saturated.
Sands and gravels. - In this type of soil only a small percentage of the particles are
smaller than 0.06 millimetre. The true cohesion is negligible and the permeability is
relatively great. Most sands and gravels, unless in a loose state of packing, exhibit
positive dilatancy during shear.
Clay-gravels. - This type of soil contains sufficient silt and clay fractions to impart
appreciable cohesion, but not so much of these finer fractions as seriously to reduce
the angle of shearing resistance under normal conditions. Hoggin forms a familiar
example. But in many boulder clays, the clay content is so high as to place the soil in
the class of “ clays.”
Silts. - In its most characteristic form this type of soil has a liquid limit of less than 35
and contains not more than about 15 per cent. clay fraction (less than 0.002
millimetre). Unless very soft, silts exhibit positive dilatancy.
Non-fissured clays.—This class of soil is plastic and smooth to the touch, with a liquid
limit typically greater than 35, and without any obvious macro-structure. The clay
fraction is usually in excess of 15 per cent. Unless highly over-consolidated most clays
exhibit negative dilatancy.
Fissured clays.—This type of soil is identical with that described above, except for the
presence of a network of fissures. These fissures can be readily seen if a sample is
broken apart. They are of greatest significance in over-consolidated clays, and require
special con- sideration in predicting the effects of softening.
THE BASIC SHEAR TESTS
GENERAL
The outstanding importance of the drainage conditions during shear has been
recognized since 1932, when Terzaghi published the results of two series of triaxial
tests on a saturated clay. In the first of these no drainage was allowed and the angle of
shearing resistance “” was found to be approximately zero; whilst in the second
series the clay was allowed to consolidate during the test, when the value of “” was
found to be 23 degrees.
At the present time the tests normally carried out may be classified as follows (after
Casagrande 1941) : *
No drainage during shear—
Undrained (Quick) Test : the samples are placed in the testing apparatus in any given
state (undisturbed, compacted to a specified density, etc.), subjected to an applied cell
pressure under conditions of no drainage and then sheared, also under conditions of
no drainage. At least three samples must be used, each being tested under a different
cell pressure. ,
Consolidated-undrained (Cons.-Quick) Tests: several (not fewer than three) samples
are allowed to consolidate, or soften, under a defined pressure and are then subjected
to undrained tests; each sample being under a different cell pressure. The procedure
is then repeated on further sets of samples, each set being consolidated under
different pressures.
Full drainage during shear—
Drained (Slow) Test : the samples are allowed to consolidate, or soften, under a
defined pressure and are then sheared under conditions of no excess pressure in the
pore space, the cell pressure (or vertical pressure in the shear box) being held
constant.
RATE OF STRAIN
Recent research by Casagrande and co-workers has shown that the strength of a sand
is virtually uninfluenced by the rate of strain but that, on the other hand, the strength
of a clay depends to an appreciable extent on this factor.
Undrained tests are usually completed within a period ranging from 5 minutes to 30
minutes. But in the subgrade of a road or runway, for example, the shear stresses are
applied at a rate of the order of 1,000 or even 10,000 times greater than that used in
the laboratory test. With special equipment Casagrande and Shannon (1948) have
carried out shear strength measurements under such transient rates of loading and
they find that the results may be 30 to 60 per cent., or in extreme cases 100 per cent.
greater than those obtained in the normal tests. Clearly some allowance for this effect
should logically be made in subgrade design.
In contrast, the rate of loading in most constructional operations is measured in terms
of weeks or months: a rate of the order of 10,000 times smaller than that in the
routine tests. Taylor (1943) and Casagrande and Wilson (1949), have investigated the
influence of very low rates of strain on clays and it appears that the shear strength
typically falls to values about 15 to 30 per cent. lower than that normally measured.
Here again some allowance for this effect should be made in an analysis which
attempts to be thoroughgoing. From the theoretical considerations put forward by
Hansen and Gibson (1949), it is probable, however, that the “ = 0” analysis contains
an implicit factor of safety of the same order as the decrease in strength due to slow
rates of loading. It may well be that the two effects tend-to balance out and there is, at
any rate, a considerable body of evidence pointing to the reasonable accuracy of the
simple “ = 0” analysis, as shown by Skempton and Golder (1948) and Cadling and
Odenstad (1950).
So far as laboratory practice is concerned the variation in rates of loading in
undrained tests are of little consequence. A considerable number of experiments in
the Authors' labora- tory indicate that the differences in strength obtained in tests
carried out 5 times faster or 5 times slower than the standard rate (about 10 minutes
to failure) do not amount to more than about 5 or, in extreme cases, 10 per cent ; even
at water contents equal to the liquid limit. But in drained tests on clays the time of
loading varies from about 8 hours to 3 days ; and consequently for research work it is
necessary to apply corrections to those results before comparing them with
undrained tests.
LABORATORY TESTING PROCEDURES
GENERAL
The choice of apparatus for carrying out a shear test is primarily determined by the
conditions of drainage under which it is desired to carry out the test.
Table 1 indicates the procedure which is generally followed in testing the main soil
types. For example, the use of the small shear box for undrained and consolidated-
undrained tests is limited to clays of low permeability tested at a rapid rate of shear,
owing to the relatively short drainage paths in the apparatus.
Other factors also have to be considered, such as- whether samples can be
satisfactorily prepared (in very soft, friable or coarse grained soils), and the degree of
operational skill required to carry out the test.
Tests are normally carried out under a controlled rate of strain, as the conditions at
failure are then completely specified.
TRIAXIAL APPARATUS
The triaxial compression test is carried out under a constant cell pressure on a sample
3 inches long x 1 ½ inch in diameter in the apparatus shown diagrammatically in Figs
1 and 2 (for sand, silts and clays) or 9 inches long X 4 inches diameter (for gravelly
soils). The cell pressure is maintained constant in long duration tests by the mercury
control which, once it has been set at the correct level, automatically compensates for
any leakage of water out of the triaxial apparatus. For short duration tests (normally
about 10 to 20 minutes) the compressed air in the cylinder is sufficient to maintain a
constant pressure.
Table 1
Apparatus used tor shear-strength measurements in relation to main soil types
Soil type Test conditions: Undrained; Consolidated-Undrained; Drained
Sand: - ; Triaxial ; Triaxial or shear box, deep sounding in situ.
Gravel; - ; - ; Large shear box
Clay-gravel; Triaxial or large shear box (if of low permeability); Large shear box or
triaxial, depending on grading and permeability.; Large shear box or triaxial,
depending on grading.
Silt; Triaxial vane in situ ; Triaxial; Triaxial or shear box.
Peat; Triaxial; Triaxial: Triaxial or shear box
Fissured clay, Triaxial; Triaxial or oedometer triaxial; Triaxial
In an undrained test on a cohesive soil the specimen is simply placed between two
solid end pieces (of duralumin or “* Perspex ”) and enclosed in a thin rubber
membrane which is sealed against the end pieces by rubber rings. The rubber
membranes in general use in Britain are about 0.2 millimetre in thickness, and it has
been found that these givenan apparent increase in compression strength of about 0.8
lb, per square inch. At Imperial College, when very soft clays are being tested, a
membrane only about 0.1 millimetre in thickness is used ; this being made in the
laboratory from self-vulcanizing latex. The correction is then about 0.3 lb. per square
inch.
In the consolidated-undrained test, for any type of soil, the sample is placed on a
porous disc, which is connected through the base of the cell to a burette. This permits
consolidation to take place under the applied pressure. When consolidation is
complete | the valve to the burette is closed, and the sample is sheared under
undrained conditions with any desired cell pressure. When clays are being tested, side
drains made of thin strips of filter paper are employed to increase the rate of
consolidation.
The procedure is similar in the case of the drained test, except that the valve to the
burette remains open throughout the test: and in testing clays the shearing process
must not be completed in less than about 1 to 3 days (using side drains).
In the case of saturated non-cohesive sands, the rubber membrane is surrounded by a
E while the sample is placed in it, and the cap sealed in place. The burette is then
lowered to apply a small negative pressure sufficient to hold the sample rigid while
the former is removed, the cell assembled, and the lateral pressure applied. A
consolidated-undrained or drained test is then carried out.
The same method is applied in the case of dry sands (Fig. 2), where a lowering of the
mercury level in the burette (by the screw control) sets up a negative pore pressure
while the cell in being assembled. Volume changes are measured during the test by
maintaining atmospheric pressure in the pore space (indicated by the oil manometer),
which means that the volume of air in the system is constant, and hence changes in the
volume of the sample are indicated by changes in mercury level in the burette. This
procedure if, of course, applicable only to drained test.
(2) How could one be certain that full drainage occurred in a constant rate of strain
test ?
In conclusion, he would stress his appreciation of the very excellent work on the shear
strength of soils being carried out by Dr Skempton and his staff.
Mr W. Kjellman was of the opinion that the “-theory” due to Dr Skempton was the
most important contribution to the understanding of the fundamentals of shear-
strength theory that had appeared recently. More experimental work should be done
to establish the validity of the relations between strain and effective stress change
upon which that theory was based. It would be necessary to introduce a further
parameter expressing the effect of pure shear, as had been done by Odenstad in a
recent Paper. That theory of Odenstad's, moreover, was in agreement with the
experimental findings of Professor Taylor. Had Dr Skempton any experimental results
which might be used as a check for such theories? He was not in disagreement with
anything in the present Paper, but he wished to make the following comments :
(1) He suggested that the vane should be used only for undrained tests and only in
materials which did not exhibit dilatancy. When the vane was used, samples
should be taken as a check.
(2) When the validity of the cone test was being investigated at the Imperial College
and the cone conversion-factor was found to vary with the soil type, what test
was used as a standard to measure the soil strength ? So far as he was aware, the
only method of measuring the strength of very sensitive clays was with the vane.
(4) The method of determining the porosity of sands by the use of air pressure and
volume measurements, described by Mr Bishop, was not new. There was, in fact,
a Swiss patent for it.
M. J. Florentin dit qu'il veut remercier tout d'abord les conférenciers du remarquable
exposé relatif aux méthodes expérimentales utilisées en Angleterre pour la mesure de
la résistance au cisaillement des sols. Toutefois, il y a un point qui le trouble et qui
l'améne 4 poser la question suivante: les laboratoires anglais de mécanique du sol,
ont-ils abandonné la mesure de la résistance au cisaillement avec des appareils de
torsion ? Si oui, pourquoi ?
Il pense que la comparaison entre les différents appareils actuellement utilisés aurait
été plus complete si Pon y avait inclus la mesure du cisaillement par torsion. Il pense,
comme beaucoup d'autres, que le cisaillement par torsion est le seul moyen de
produire le cisaillement sans effets parasites. A son avis, la torsion sous des
compressions différentes constitue un bon moyen de contróle de la loi de Coulomb,
puisque les résultats expérimentaux se traduisent par la détermination d'un effort
tangentiel + correspondant á une compression 4. Dans les essais de compression,
(sans ou avec étreinte latérale) on fait appel, en plus, á la théorie de Mohr sur la
rupture des corps. Ce que l'on tire en définitive de ces essais, ce sont les cercles
construits sur les contraintes principales extrémes atteintes au cours de la rupture.
Or, il croit se souvenir que la théorie de la plasticité (et l'on ne peut nier que les
matériaux essayés