1976 - The Influence of High Pore-Water Pressure On The Strength of Cohesionless Soils

[ 91 ]
I
THE INFLUENCE OF HIGH PORE-WATER PRESSURE
ON THE STRENGTH OF COHESIONLESS SOILS
Bv A. W. BISHOP AND A. E. SKINNER

Civil Engineering Department, Imperial College of Science and Technology,
South Kensington, London SW7 2B U
(Communicated by A. W. Skempton, F.R.S. - Receiued 12 August 1975 - Diagramsfor

reproduction receiued 8 March 111;,7,6),
[PIa te 1]
CONTENTS
PAGE
1. lNTRODuCTION 92
2. EXPRESSIONS
FORINTERGRANULAR
STRESSANDEFFECTIVESTRFSS 93
3. TESTINGPROGRAMME,
APPARATUSANDTESTINGTECHNIQ.UES 98
4. MATERIALSTESTED 106
5. TEST RESULTSANDTHEIR IMPLICATIONS 107
(a) Ham River Sand 107
(b) Lead shot 121
(c) Crushed marblc 125
(d) Braehead Silt 126
6. CONCLUSIONS 128
REFERENCES 129
The influence on the mechanical properties of saturated particulate materiaIs of the

component of stress carried by the water filling the pore space is fundamental to both
theoretical and experimental studies in soil mechanics.
The rle of pore pressure in controlling compressibility and shear strength is
expressed in Terzaghi's principie ofeJfective stress to a degrec of accuracy which is sufficicnt
for most engineering purposes. However, the precise significance ofthe small but finite
area of interparticle contact has remained uncertain in the application of this equation
to shearing resistance.
In the present paper the possible erro~s associated with the use of current expressions
for intergranular stress and effective stress are examined. These errors are ofsignificant
magnitude at high values of pore pressure and low values of the yield stress of the soIid
forming the particles. A very accurate experimental investigation has been carried out
into the sensitivity ofshearing resistance to large changes in pore pressure (up to 41.4
MN 1m 2), using particuIate materiaIs ranging in strength from Quartz sand to lead shot.
The results indicate that the sim ple Terzaghi effective stress equation (J"' = U - u is
consistcnt with all the observations, though for Quartz sand a range of pore pressure
changes an order of magnitude higher is desirablc for additional confirrnatory evidence.
12 [Published 4 January 1977

t J .~ "., .
."
92 A. W. BISHOP AND A. E. SKINNER \ r
t

1. INTRODUCTION
During the five decades which have elapsed since Terzaghi (1923) first stated the principIe of
effective stress the physical basis of the principIe and the equations used to express it have been
the subject of periodic review and occasionally of lively controversy. Throughout this period
thcre has been a tendency to ignore the intention underlying Terzaghi's definition of effective
stress and to identify effective stress with intergranular stress (for example Bruggeman, Zangar &
Brahtz 1939; Taylor 1944, 1948; Scott 1963)' While the consequent error is small for the relatively
low values of pore pressure and interparticle contact arc usually encountered in engineering
practicc, significant errors arise at the high values of pore pressure found in occan bottom
scdiments, in oil and gas reservoirs and in various geophysical studies. Likewise in concrete and
rocks, where thc arca of contact is largc evcn at low stresses, the use of a correct effective stress
equation is particularly important.
Terzaghi (1936) restatcd the principIe of effcctive stress in the following terms (in current
terminology) :
'The stresses at any point ofa section through a mass ofsoil can be computed from the total
principal stresses (TI' (T2, (T3 which act in this point. Ifthe voids ofthe soil are filled with water under
a stress u, the total principal stresses consist of two parts. One part, u, acts in the water and in the
solid in every direction with equal intensity. It is called the neutral stress (or pore-water pressure).
The balance (T~ = (TI - U, (T; = (T2 - u and (T; = (T3 - u represents an excess over the neutral
stress u and has its seat exclusively in the solid phase of the soil.' T
'This fraction of the total principal stresscs will be called the efJective principal stresses .... A
change in the neutral stress u produces practically no volume change and has practically no
inllucncc on the stress conditions for failurc ... Porous materiaIs (such as sand, clay and concrete)
react to a change in u as ifthey were incompressible and as iftheir internal friction were equal to
zero. All the measurable effects ofa change in stress, such as compression, distortion and a change
in shearing resistance are exclusively due to changes in the effective stresses (T~, (T;, (T~. Hence
every investigation of the stability of a saturated body of soil requires the knowledge of both the
total and the neutral stress' .
It will be noted that in this statement the emphasis is on the exclusive relation between all
the measurable effects of a change in stress (compression, distortion and a change in shearing
resistance) and changes in the effective stresses (T~, (T;, (T~. Terzaghi's simple expression for
effective stress
(T' = (T - U (1)
has been shown (Bishop & Eldin 1950; Bishop 1955; Skempton 1960) to hold rigorously for the
case of volume change if the two conditions in the statement quoted are put in the form
(1) the soil grains are incompressible,
(2) the yield stress of the grain material, which controls the contact area and intergranular
shearing resistance, is independent of the confining pressure (as in the theories offriction due to
Terzaghi (1925) and Bowden & Tabor (1942)).
It was also inferred by Bishop (1955) and demonstrated more rigorously by Skempton (1960)
that the simple expression for effective stress (T' = (T - u should hold for changes in shear strength
if condition (2) were satisfied.
Actual soils do not fully satisfy either condition. Skempton (1960) has examined the significance
r
"
..
INFLUENCE OF HIGH PORE-WATER PRESSURE 93
of the departures from both conditions (1) and (2) and has derived modified expressions for
effective stress. In the next section the various expressions for effective stress and intergranular
stress will be briefly presented and discussed.
2. EXPRESSIONS FOR INTERGRANULAR STRESS AND EFFECTIVE STRESS
(a) Intergranular stress

It has bccn considered by a number of investigators that, a priori, the mcchanical behaviour
of a particulate mass is controlled by the forces acting at the partide-to-partidc contacts. Taylor
(1948, p. 126) states: 'In concepts ofstresses ... the surface that must be considered is the one
containing the points of grain-to-grain contact, in order that it may indude the points of action
of the forces which make up intergranular stress. Thus the unit area should be visualized as a
wavy surface which is tangent to but does not cut through soil grains, and which at all points is as
dose as possible to a flat surface.'
FIGURE 1
Consideration of the equilibrium of the forces on such a flat surface (figure 1) leads to the
rclation betwcen total stress (T, the intergranular stress (Ti (dcfined as the average intergranular
force per unit area normal to the surface) and the pore-water pressure u
(T = (Ti+ (1- a)!u

or . (Ti = (T-(l-a)u
= ((T-u) +au, (2)
where a denotes the contact area of the soil particles per unit area of the surface (and projected
onto the surface).
Taylor (1944) uses this expression in a discussion of the significances of pore pressure changes
in undrained triaxial tests on Boston Clay and estimates the value of a on this basis, obtaining
a value of about 3 %. Taylor suggcstcd that the term 'neutral stress ' should be applied to the
product (1- a) u.
Most investigators have, however, agrecd that a is small in cohesionless soils, and probably in
days, at the stress levels commonly encountered in engineering practice. Thus, provided u is of
the same order of magni tude as ((T - u) the tcrm au ma y be neglccted with little loss in accuracy.
However, ifu is very large compared with ((T-u), the significance ofthe term au is radically
12-2
~-
1
94 A. W. BISHOP AND A. E. SKINNER
changed. For u = 100((T~u) a value of a of 1 % would lead to a product au equal in magnitude

to ((T - u) and a doubling of the magnitude of (Ti'
Since the magnitude of a cannot be measured directly and is a matter of some uncertainty in
many soils, the validity or otherwise of the intergranular stress concept is cIearly a matter of
considerable importance.
intergranular force P
p
u
u
u
/
/// p
forces at an interg ranular contact
..
\
\
\
\
-
,
,
,,
r
r-A{/
I
,,"
"
/
I
I.
U
P-Au
+
u
u ,
u
+ ~A1
equivalent forces on a soil grain

FIGURE 2
(b) Effective stress for changes in volume

Bishop & Eldin (1950) consequently examined the forces acting at an interparticIe contact in
the presence of pore-water under pressure (figure 2) and concIuded that the distortion of the soil
particIe was not a function of the intergranular contact pressure, but of the excess of this pressure
over the pore pressure acting on the adjacent surface of the particIe. If the compressibility of the
material comprising the soil particIe was neglected as smaIl compared with the bulk compres-
sibility of the porous granular mass, then, irrespective of the magnitude of the interparticIe
contact area a, the equation for effective stress reduced to
(T' = (T-U.
Effective stress in this context ] is defined as the function of total stress and pore pressure which
controls volume change.
t The terrninology used in the original paper hy Bishop & Eldin (1950) reflected the current identification of
the term effective stress with intergranular stress.
I
l
__:J
lNFLUENCE OF HIGH PORE-WATER PRESSURE 95
For this demonstration that the Terzaghi effective stress equation was independent of the
magnitude of a, the two conditions stated in section (1) were necessary and sufficient.
The effect on the effective stress equation of a departure from condition (1), the incornpres-
sibility ofthe soil grains, was examined by Bishop (1953), who derived the expression
a' = u - (1 - CaIC) u (3 a)
or a' = (u-u)+(CsIC)u, (3 b)
where Cs denotes the compressibility of the solid material forming the soil grains and- C denotes
the bulk compressibility of the porous mass for the relevant stress range.
For soils in the low stress range the bulk compressibility C is very large compared with the
value of Cs. Skempton (1960) has tabulated data showing that for soils ranging from norrnally-
consolidated clay to dense sand the ratio CsIC lies in the range 0.00003-0.0015 for a consolidation
pressure (u - u) ofapproximately 100kN/m2 (table 1). For a consolidation pressure of20 MN/m2
Skempton suggested that the ratio CsIC is unlikely to exceed 0.01 in the case of days. More recent
test data from high pressure triaxial tests on London Clay (Bishop, Kumapley & El-Ruwayih
1975) indicate that at a consolidation pressure of 62.1 MN/m2, CalC may ris e to 0.04.
TABLE 1. (AFTER SKEMPTON 1960.)

(Compressibilities at p = 1 kgfcm 2; water C., = 48 X 10-8 per kgfcm2)
compressibility
cm2kg-1
C.
material C C. C
quartzitic sandstone 5.'8 X 10-6 2.7 X 10-6 0.46
Quincy granite (100 ft deep) 7.5 X 10-8 1.9 X 10-6 0.25
Vermont marble 17.5x 10-6 1.4 X tO-6 0.08
concrete (approx. values) 20 x 10-6 2.5 X 10-6 0.12
dense sand 1 800 X 10-8 2.7 X 10-8 0.0015
loose sand 9000 X 10-6 2.7 X 10-8 0.0003
London day (over-cons.) 7500 X 10-6 2.0 X 10-8 0.00025
Gosport Clay (normally cons.) 60000 X tO-6 2.0 X 10-8 0.00003
Tests to examine the validity of the effective stress equation for volume change have been
carried out by Laughton (1955) using high pore-water pressures in a sealed oedemeter. Laughton
tested both lead shot, to ensure large values of interpartide contact area a which could be
measured on unloading the sample, and Globigerina ooze from the bed of the eastern Atlantic
oceano
ln both cases the test data demonstrate the validity (to within experimental error) of the
effective stress equation a' = a - u, although in the case of the lead shot the observed value of a
rose to 95 % at the highest effective stress (100 MN [ti: 2). In a re-examination of the test data
Skempton (1960) has shown that the effective stress equation (3) gives a marginally better fit in
the case oflead shot when the value of CsIC rises to 0.05 at the highest consolidation pressure.
Skempton (1960) also examined the consequence of a departure from condition (2) on the
effective stress equation for volume change and found it to be ~umerically unimportant in most
cases.
It is ofinterest to note that, since the bulk properties ofa granular mass subject to a change in
stress are not those ofan ideal eIastic material, the value ofCis not a unique parameter for a parti-
cular soil (as is Cs) but depends on the stress leveI, the previous stress path, the sign and magnitude
of the stress change, and the rate of loading (Bishop & Blight 1963). Furtherrnore, although
effective strcss as defined by equation (3) determines the overall volume change, it is the com-
ponent (T - U which determines the change in compressibility C (Bishop 1973).
Whilc it is apparent from the preceding discussion that the more rigorous expression for
effective stress with respect to volume change need only be used for soils at very high consolidation
pressures, for concrete and for porous rocks the term CsIC is of much greater significance even in
the low stress range. Data presented by Skempton (1960) indicate values of CsIC of around 0.12
for concrete and from 0.08 to 0.46 for various rocksItableL).
(c) EiJective stress for changes in shear strength

Terzaghi (1936) considered that the same effective stress equation (T' = (T-U applied equally
to both changes in volume and to changes in shear strength.
Bishop (1955) considered that, if an analogy could be drawn between interparticle friction
and metallic friction, then the contact area would control the frictional forces and this area, like
the deformation of the soil particles, would be controlled by (T' (= (T - u). Hence it might be
inferred that the expression for effective stress (T' = (T - U would be valid for shear strength
irrespective of contact arca, provided condition (2) was satisficd.
Skempton (1960) presented a formal analysis ofinterparticle friction in the presence ofpore
water under pressure, arriving at the same conclusion for the special case represented by condi-
tion (2), but also extending the analysis to include the more general case of grain materiaIs whose
strength is a function of confining pressure of the form: T
Ti = k+(Ttan1fr, (4)
where Ti denotes shear stress at failure, a denotes normal stress, k denotes intrinsic cohesion,
and 1frdenotes the angle of intrinsic friction of the solid.
For thcse materiaIs Skempton obtained an expression for effective stress with respect to change
in shear strength:
(T' = (T _ (1_ tantan rp~)
a U (5 a)
tan 1jr
= ((T-u) +au~, (5 b)
tan 'I'
where a denotes interparticle contact area, as before, and rp' denotes the angle of shearing
resistance of the granular mass (in terms of effective stress).
Skempton (1960) examined published test data fromjacketed and unjacketed triaxial tests on
Marble and Solenhofen Limestone and concluded that the areas of contact given by equation (5)
(0.15 and 0.45 respectively) were consistent with the changes in strength with confining pressure
obtained with the jacketed samples. He also concluded that area of contact deduced from concrete
was about 0.2, and, though not subject to any independent check, was not unreasonable. For soils
no critical test data was available, and Skempton concluded that Terzaghi's equation would be
a valid approximation due to the small value of a.
ln deriving eq uation (5) Skempton assumed that there is a direct rclation bctwcen rp', the
angle of shearing resistance of a non-cohesive granular material, and 1'-, the coefIicient offriction
at an interparticle contact, quoting the expression due to Caquot (1934) for constant volume
shear:
tan rp' = tltjt. (6)
#
------------
Similar though not identical relations have been obtained by Bishop (1954) and Horne (1969),
but involve simplifying physical assumptions and, in the case of the earlier expression, a
mathematical approximation.
This Iorrn ofrelation is, however, not supportcd by a series ofvery careful tcsts carried out by
Skinner (1969, 1975) on particIes of almost identical shape but of widely differing values of the
coefficient fl (figure 3). Skinner's results are supported by independent tests carried out on rock
fragments and gr~vel by Tombs (1969) and discusscd by Bishop (1969). t
50 --- ---------
40f-----j-
10 30 40
::.-.
/::,. 1 mm diam. glass bailo tini (dry)
1 mm diam. glass bailo tini (flooded)
1
O 3 mm diam. glass ballotini (dry) .
3 mm dIam. g Iass b a IIoum
.. (fi00 d e d) shear box tests after Skinner (1969)
+ t in. diam. steel ball bearings (dry)
x 3 mm diam. lead shot (dry) .
cp. chert l
8:> mucI:>toneJ Rockfill triaxial tests after Tombs (1969)
O gramte
FIGURE 3. Theoretical and experimental relations between 1>p and 1>cv (Skinner 1969).
Lack of direct relation between cjJ' and fl is associated with the complexity of particle movement
in a particulatc mass subjcct to a shear strain, which involves not only interparticle slip but
particle rotation and out-of-plane displacements even under an overall plane strain displace-
rnent.j It suggests, furthermore, that one of the steps in the derivation of the effective stress
t Experimental support for the relation comes primarily from the tests in which Jl has not in fact been measured
directly and where its physical significance is open to serious criticism (see Horne 1969; Skinner 1969; Bishop 1969;
Procter & Barton 1974). .
t A detailed study using marked particles and X-ray stcrco tcchniques has bcen carried out by Y. Sharrna and
is in course of preparation for publication.
,,'
equation (5) may be based on an assumption of doubtful validity, at least for cohesionless soils.
This places added emphasis on the need for direct experimental verification of the effective stress
cquation for shear strength.
3. TESTING PROGRAMME, APPARATUS AND TESTING TECHNIQ.UES
The testing programmc was designcd generally to explore the influence of high pore-water
pressures on the strength of cohesionless soils and in particular to discriminate between the
intergranular and effective stress equations.
Thc difference between the three equations lies in thc term involving the product au where
the Terzaghi effective stress equation is
0" = O'- u, (1)
the intergranular stress equation is
O'i = (O'-u) +au, (2)
and Skempton's effective stress equation is
0" = (O' - u) + mau, (5)
where m is tan 1frltan ifJ' and is ofthe order 0.3 for sand consisting of quartz particIes and 0.03 for
lead shot (Skempton 1960). The significance ofthe au term in influencing the strength, which,
by definition, is controUed by 0", wiUdepend on the magnitude of au relative to (O' - u). In seeking,
for reasons of experimental accuracy, to maximize the ratio r, where r = au/( O' - u), we may note
that, for a given particulate material, the magnitude of a depends almost linearly on (O' - u)
(Bishop & Eldin 1950; Skempton 1960). Hence we have
. au
r=---
,. (O' - u)
n(O'-u)u
(O'-u~
= nu, (7)
where n is a constant depending on the strength parameters ofthe material. Thus the percentage
difference in strength between the predictions of the three expressions wilI depend on the magni-
tude ofthe change in pore pressure (for (O' - u) = constant), but will be independent ofthe actual
magni tude of the consolidation pressure (O' - u) and of the associated value of con tact area a,
depending instead on their ratio as represented by the parameter n. This means that the testing
programme must involve the highest pore pressures (and consequentIy the highest cell pressures
in the triaxial apparatus) consistent with the accurate measurement of small changes in strength,
and must include materials in which the yield stress ofthe particIes is relatively low.
The accurate measurement of small changes in strength at high cell pressures in the conven-
tional triaxial apparatus is rendered almost impossible by (a) the friction on the loading ram
and (b) the magnitude of tle load due to theceIl pressure acting on the inner end ofthe ram, this
load being very large compared with that due to the strength change to be detected.
The error due to friction on the loading ram can be avoided by measuring the load inside the
cell with an electric load transducer. However, the problem then arises of the sensitivity of the

,,
transducer itselfto 1arge changes in ceIl pressure.As the purpose ofthe present series oftests was
not merely to measure the strength changes correctly, but to demonstrate incontrovertibly that
they had been measured correctly, the aIternative method ofrotating the bushing enclosing the
loading ram was adopted (figure 4). Since the frictiona1 force opposes the relative motion between
the ram and the bushing, a rotary motion of about 2 revfmin is sufficient to reduce the vertical
component of friction to negligible proportions at normal rates ofaxia1 displacement. The
problems of excessive oilleakage and 'wobble' due to the loss of a common axis to the inner and
outer cylindrical surfaces 'of the bushing during machining and honing call for a very high
standard ofworkmanship in manufacture.
axial load
yokc' ------~ I
rotuting hush
stccl cylindcr
samplc ------tt--tt---k:I-'#.'.
conncction to
prcssurc slIpply
"
drainugc or
--'.-' pore prcssure
conncct ion
pedestal base
rot.uing hush
FIGURE 4
The relative magnitudes ofthe uplift on the end ofthe loading ram and the axialload required
to shear the sample can be readily estimated for a maximum ratio of cell pressure (1'3 to consolida-
tion pressure (1'3 - U of ca. 100 and a ram to sample diameter ratio of i (as used on the high pressure
cell). From the geometry ofthe Mohr circle (figure 5) it follows that, for a cohesionless soil, the
stress difference at failure (on the basis ofthe Terzaghi expression for effective stress) is given by
the expression
(8)
13
Hence, for a value of <jJ' = 35, the ratio ofthe axialload due to uplift to the axialload on the
sample at failure (neglecting the area change during compression) is
(i3Ar _ 100X((i3-U)X 1.02X1t
((iI - (i3) As - ((i3 - u) x 2.69 X 1.52 x 1t'
= 16.5, (9)
where Ar and As denote the cross-sectional areas of the ram and sample respcctively.
o~--------~--------~--------~----~
o' effective normal stress
, 0"
I.__ a:_J-_u_--..!. f-o._~=-(-,Oj,----_a:_3 )
. A' AO' h () 2sin1>'

= 00" ence =
SID'f' 0'1-0'3
" O'a-U
l-sin1>'
FIGURE 5. The geometry of the Mohr diagram.
Thus to detect a change of 0.5 % in shear strength it would be necessary to measure the total
axialload to an accuracy of O.03'%, which is beyond thc confidence limit of most Ioad measuring
devices.
The difficulty was resolved by balancing ~ut hydraulically the load due to uplift with an
opposed ram ofidentical diameter in a similar rotating bush (figure 4). A prototype triaxial cell
was built (figure 6a, plate 1) mainly from componen~ts currently in use at Imperial College
(Bishop, Webb & Skinner 1965) having a maximum cell pressure capacity of 6.9 MNJm2
(1000 IbfJin2). With carefully matched rams (15.88mm or 0.625in nominal diameter) the
maximum change in axialload for a change in cell pressure of6.9 MNJm2 was found to be 0.27 N
(0.061bf) and was thus negligible relative to the load changes to be measured.
The success of the prototype led to the adoption of the same principIe for a high pressure cell
with a capacity of69 MNJm2 (10 OOOIbfJin2). The technical problems to be overcome in actually
manufacturing a rotating bush cell to operate in this pressure range were found to be formidable
for three reasons in particular.
Firstly, the principIe of rotating the bushing to eliminate the vertical component of friction
was found to be ineffective if conventional oil seals were used. Hence control of the 10s5of the
pressure fluid between the ram and the bushing depended solely on thelength ofthe leakage path
and the fineness of the fit. The length also necessitated a relatively stiff ram to avoid buckling in
the higher load range. The dimensions chosen were aram diameter of 25.4 mm (1.Oin) and an
external bush diameter of 381 mm (1.5in) with minimum and maximum radial clearances of
0.0038 and 0.0051 mm (0.00015 and 0.0002in) internally and 0.0051 and 0.0064mm (0.0002
and 0.00025 in) externally.
Phil. Trans. R. Soco Lond. A, volume 284
Bishop & Skinner, ptate 1
FIGURE li. (a) 1i.90 MN m-~ triaxial appuratus. () liH.!l!i MN Iriaxial ccll.
111-~
(f.'ucil/!f IJ. IUO)

.~ .
:.
insct
sccrion
insct
scction
FIGURE 7. Triaxial ceIl with hydraulically balanced rams suitable for sample confining pressures up
to 69 MN m-2
..., . . "
- . .'~~'
..
102 A. w. nISHOP AND A. E. SKINNER
Honing to these small tolerances and the avoidance of seizing were made possible by the use
of a special (spheroidal) cast iron bushing in conjunction with a stainless steel ramo The interna!
and external drainage paths along the bushing were 165mm (6.500in) and 143mm (5.625in)
rcspectively, Thc oilloss from the pair ofbushings was only about 0.7 51/h using Germ Dynobear L
oil (rel ative density 0.892 at 15.5 C; viscosity Redwood No. 1 at 21.1 C 342 s; light machinc tool
lubricant with increased oiliness characteristic which prevents stick-slip behaviour) at the
maximum cell pressure of69MN/m2 (100001bf/in2).
oil prcssurc
lransfcr tubc
oil at
ccll prcssurc
t inductance
transduccr d.~~;jr--watcr
110at
rncrcury
t
back-pressurc
return line
to porous
stonc at lhe
base of lhe
~t=j1t-~_-mercury sample
.~._.
FIGURE 8. Apparatus used for (T3 - U = consto tests at high confining pressures, incorporating
a volume gauge for tests involving high back pressures.
The second problem was the uplift force due to the cell pressure acting on the inner end of the
bushing itself. Even using the minimum wall thickness convenient for manufacture and for trans-
mitting the torque under operating conditions (6.35 mm or 0.25 in), the axialload at maximum
cell pressure is 43.6 KN (98121bf). This load had to be carried by a thrust bearing with the
minimum ofvibration and friction. A tapered roller bearing (figure 7) was used for this purpose
and was mounted on the bossofthe bronze worm wheel which served to transmit both. the torque
to the rotating bush and the axialload from the bush to the inner race ofthe bearing. The friction
betwcen the bush and the worm wheel due to the axialload was sufficient to carry the torque
necessary under most operating conditions without any form ofkey.
I
I
i
.~ '-'"
. -~.:~.'::-'
:~t.l'~i-
.~
4
"' ..... .. \, ~
I
. ....
.'
The third problem was the distortion of the loading head itself, due to the high ceIl pressure,
which could result in unacceptable changes in diameter ofthe bore within which the bushing had
to run. Because of the difficulty of determining these changes in bore diameter analytically,
a modeI ofthe head was made in brass, strain gauged and tested. As a result ofthese observations
the loading head was redesigned so that the major thrust on the head passed directly across to
the clamping rings via the ceIl waIl or pedestal base (figure 7).
The success of the mechanical system and the accuracy of the workmanship involved is
indicated by the observation that the difference in axialload for a change in confining pressure of
69 MN/m2 (10 000lbf/in2) was less than 0.27 N (0.06Ibf).
One other aspect ofthe apparatus is of'special interest. ln the main series oftests the difTerence
between the ceIl pressure and the pore-water pressure (0"3-U) had to be maintained very accu-
rateIy at a constant value while the ceIl pressure and pore-water pressure were varied through
a range some 100 times the magnitude of (O"-u). This was achieved by building a system
equivalent to the constant pressure seIf-compensating melcury control system described by
Bishop & HenkeI (1962) and immersing it in fIuid at the operating pressure of the celI, contained
in two pressure vesseIs separated byan appropriate vertical distance (figure 8).
As in the widely used low pres~ure version, the difference between the pressure (0"3) in the fIuid
above the mercury surface in the suspended cylinder in the lower vesscl and in the return line (u)
from the fIuid above the mercury in the fixed cylinder in the upper vessel is the difference in leveI
multiplied by the difference in the unit weights ofmercury and the operating fIuid in the return
line R (figure 9). As the return line is connected to the porous eIement at the base ofthe sample,
the change in the mercury leveI in the upper cylinder provides a measure ofthe volume change in
the test specimen. This measurement is based on the volume ofpore water moving in or out of
the specimen and is subject to corrections for membrane penetration when (0"3-U) is changed,
and for the compressibility of the pore water in the.sample, of the soi! particles, of the water in
the cell base and return line and for the expansion of the thick-walled t li bing when u is changed
with (0"3 - u) heId constant.
The mercury leveI was sensed with an inductance transdueer, the core ofwhich was attached
to a fIat conical stainless steel fIoat (figure 8). The capacity ofthe system as a volume gauge was
20 em", The overalI discrimination ofthe system was 1xl 0-4 em", but the repeatability was only
ofthe order of 0.01 em", As the initial volume ofthe specimens was ofthe order of90cm3 this
degree of accuracy was adequate for studying the shcar strength and dilatancy charactcristics of
the soil as a granular mass, but not for investigating in dctail the influence ofgrain compressibility
on overall volume change.
The change in the pre-set value of 0"3- u, which would resuIt from a change in the mercury
levels consequent on a volume change in the sample, is automatically compensatcd for by the
calibrated spring in the lower pressure vessel. As the weight ofmercury in the cylinder suspended
from this spring changes due to the change in mercury levcl in the upper cylinder a spring of
suitable characteristics adjusts the lower mercury leveI to maintain constant pressure to an
accuracy of better than 1kN 1m 2 (ca. 0.11bf/in 2).
The high pressure ceIl is ilIustrated in figure 6 b, plate 1.
The test programme itsclfwas in principIe very simple and consisted ofthe observation ofthe
strength changes resulting from large changes in 0"3 and u, the difference (0"3-U) being held
constant to a high degree of accuracy. Since the natural scatter of a series of separate tests on
individual samples might mask small strength changes, advantage was taken of the relatively
K _.0.:.'10.c ,," r--'::--"':"
::.
..
-:~~,S"':i>~.~"",\._ . ~". '
~;:'"'~~~\i~
~"~i~;. .~~:.:.,:;..
...;,\.. ~:.~r-f :'7.
\'
/t~,
"',
'~.-~
"
'"
-
O
~
1.0:1 M:-; m-2

watcr prcssurc
ccll prcssure back (pore wutcr) lincs
pulse-damping guugc prcssure guugc
coil high pressure
'" : ~ volume gauge
?>
~
t::C
hydruulic: 11..., ....
tr:
resistors burst-disk
safcty dcvice
return ~
oil-wuter
line R O
'"d
intcrchanue
inductuncc
vessels ~ :>
transdJcer for
Z
generating
tj
flow scnsitivc
oil
servo-
safetv valvc ?>
cont rol signal
coolcrs fI/-' I~high prcssurc ~
balunced rum rJl
cdl ~
....
non-rcturn 11 11 I~ 1-1 Z
valve .
Z
cooling
water
---== ..~ ..",,,,,,.,
-..-~.'~L8t:S ,
~
:;d
dcad-weight prcssure monitoring vacuurn system

servo-corurolled hydruulic pressure
ruanuully controlled
systems for waste oil
generalor hydraulic rum
removal
FIGURE 9. Diagrammatic layout of triaxial apparatus used for tests with cell pressures up to 69 MN m-2

,
small rate of change of mobilized strength with strain in loose granular materiaIs to pcrform
multistage tests on a limited number ofspecimens. Ifthe au term were found to have an effect on
strength, this could then be detected with an accuracy ofabout 0.5 % from the discontinuitics
in thc stress-strain curve (as indicated diagrammatically in figure 10).
stress-strain curve for stress-strain curve for 0'3 - u constant

0'3 - u and 0'3 both constant and 0'3 increased if au term is significant
b
..
I
b'"'
v~
u
s:::
...v
~
:a
~
t:
Vl
0'3-U
constant
throughout
O
axial strain, e1
FIGURE 10. Effect on shearing resistance, in a drained triaxial compression test with 0'3 - u constant, of the same
large change in both confining pressure 0'3 and pore pressure u ir the term au is significant.
All tests were performed on saturated material and the samples were sedimented under water
within a rubber membrane enclosed by a split former as described by Bishop & Henkcl (1962).
After the sample cap had been placed in position and the membrane sealed to it a small negative
pressure was applied to the pore water. This consolidated the sample by drainage through the
porous ceramic disk set in the pedestal, and gave it sufficient strength for the former to bc removed
and the initial dimensions mcasured. The remaining components ofthe cell were then assembled
and the balance of the pressure differcnce (CT3 - u) applied, the volume change being mcasured
as described above.
During the application ofthe stress difference (CTI - CT3) the tests were run as controIled rate of
strain drained tests with constant (CT3 - u), the axial load, axial displaccment and volume change
bcing recorded. The test procedure in general foJIows Bishop & Hcnkcl (1962) and is dcscribed
in detail by Skinner (1975).
4. MATERIALS TESTED
Four materiaIs were tested and their particle size distribution curves are given in figure 11.
Theyare
(1) Ham River sand. This is a sieved fraction of a naturally occurring graveI and is composed
largeIy of quartzo The detailed mineral composition is given in table 2.
Ham River sand was selected as its mechanical properties have been extensiveIy studied both
at Imperial College and at the Building Research Establishment over the past 25 years in reIation
to strength and deformation, and to the performance of model foundations.
100
U R
/' c .........
a
/
----
/
80
/ J
1/ /
60
/
/
40
d'x/ Ij
/b
/ I
20
/ a
It c
/.
I---
y ~
'1
-1/ I
.~
o
0.002 0.006 0.02 0.06 0.2 0.6 2.0
fine
I medium
I coarse fine
I medium
I coarse
silt fraction sand fraction
particle size/mm
FIGURE 11. Particle size distribution curves for (a) Ram river sand; (b) lead shot;
(c) crushed marble and (d) Braehead silt.
(2) Lead shot. This material had a uniform particle size of 1mm, and was selected as a con-
venient and readily available material with a very low yield stress. The interparticle contact area
was thus almost the maximm obtainable for a given stress level, in contrast to that of quartz
sand, which is a1most the minimum.
(3) Crushed marble. Since lead shot has a very low value of the intrinsic angle of friction 1fr
(ca. ia), its behaviour does not provide a very critical test ofSkempton's effective stress equation
"
(equation (5)), since even when the productau is large for agivenstresslevel, thefactor tan l/rltan ifJ'
is very small. Calcite has the advantage of a relatively low yield stress k associated with a value of
l/r of 8 (i.e. about one half of that of quartz) and can therefore provide more critical test data.
(4) Braehead si/to This material is a naturally occurring silt, composed mainly of quartz, but
having an average particle size more than an order of magnitude smaller than the Ham River
sand. The detailed mineral composition is given in table 2. Since the force per contact, for a given
geometrical arrangement of particles, is proportional to d2, where d is the equivalent diameter of
the particle, this force is likely to be at least two orders of magnitude smaller for the silt than for
the Ham River sand. Furthermore, about 7 % of the particles are smaller than 211m and the
material has a significant plasticity index (6 %). It might be expected to indicate the beginning
of any trend in behaviour as the grading' moves towards that of materials classified as clays.
TABLE 2
mineral composition of Ham River sand mineral composition of Braehead silt
quartz 96.22 % quartz 95.20 %
limonite 3.66 % illite 2.80 %
zircon 0.10% kaolinite 1.60%
staurolite 0.02 % chlorite 0.40 %
analysis by Midgley (Building Research Station)
The time scale offully drained tests on clay is so very much longer (involving weeks rather than
hours) that a different pressure control and data logging system is required. Clays were therefore
not included in the present programme.
5. TEST RESULTS AND THEIR IMPLICATIONS
The most extensive series of tests was run on the Ham River sand, involving substantial
variations in the value of ((1"- u) as well as in the coupled values of (1"and U. Individual tests were
then run on each of the selected materials,
(a) Ham River sand

Two tests run in the prototype apparatus with ((1"3-U) equal to 69.0kN/m2 (10.01bf/in2) and
a variation of (1"3between 345 kN 1m2 (50 Ibf/in2) and 6.9 MN 1m2 (1000 Ibf/in2) are illustrated in
figures 12 and 13. It is immediately apparent that for a change in the value of u of 6.55 MN/m2
(950 lbf/in2) there is no discernible change in shearing resistance resulting from a change in pore
pressure wi th ((1"3- u) held constan t.
The results ofthree tests run in the high pressure apparatus with ((1"3-U) equal to 363kN/m2
(52.6lbf/in2) and a variation of(1"3between 1.03MN/m2 (150lbfjin2) and 27.58MNjm2 (4000
lbfjin2) are shown in figures 14, 15 and 16. Substantially the same conclusion can be drawn as
before, although slight irregularities are observed while the changes in (1"3and u are actually being
made. This is attributed to small pressure gradients resulting from the fiow of the relatively
compressible pore fluid under the large pressure changes, in particular in the small bore high
pressure lines connecting the pressure vessels maintaining value of u at a constant difference
from (1"3'
These three tests are ofparticular interest since the value of ((1"3-U) is typical ofthe range
encountered in engineering practice and the change in u of 26.5 MNjm2 (38501bfjin2) exceeds
this value by a factor of73 times. Had the value ofthe contact area a cqualled the maximum value
I
.\
., :;...,..... /,
..
" .....
envisaged for sand by Bishop & Eldin (1950), its value on the relevantshear surface in the present
testswould have beenca. 0.4 %. Thevalueofthe term M (equal to a!::.u/(un - u))t would have been
of the order 0.004 x 46.8 i.e. 0.19. Both of the theories involving an au term would thus have
predicted substantial discontinuities in the strcss-strain curve when the value of u was varied
with u3-u constant.
30 I
1 - 200
~ ~
(
I
~
I ""
~ I
. ~"'<>-<
- 150 -
,f
..
1
----. I
!,:"r I i
.ii.~~~
:;ii, .S S
'-
--;
;i;i" .D 20 Z
~;;;. c
I
--;
c.
I
~~;, ~ - 100 ~
f;r I)
u
~~.' !::
c- 'o!'

, ;. ..
11)
.j:/..
~ 10
::~
~, :.a
1 ~',.
...
-
'"
....~
1 '::-:.\
j
1 <- j
'"
,~
I
,
,
o o
~
.....
~ 6
" s:-
-e
4
.;
!::~
I i
!;: ~ I I
2
'"o ~
:1. "5.., ..0-0"'"
~ ~
E o
:l

;>
,
i
I ..----.
1000
..-
r
I 6 S
.S Z
'-
-S
b'"
..:l
II)~
500
4 -
~
b'"
..
a
..~ 2, '"
...
11)
P.
==5
u
.,
P.
o 5 10 15 20 25 30
o
axial strain, tI (%)
FIGURE 12. Ham river sand tested with (cra-u) =

69.0 kN m-2 (10 Ibfin-2)
(sample porosity at the start of the shear stage =
43.4 %).
t In any state of stress other than that of isotropic stress the value of a in a particulate material and the value
ofthe ratio t:.uJ(cr-u) will have directional properties. Skempton (1960) has presented his analysis in terrns ofthe
Mohr-Coulomb failure criterion, in which the normal effectivestress on the shear surface (ca. (crn - u is (1 +sinrjJ')
times the minor effective stress (ca. (cr3-u. This point is discussed in more detail in subsequent paragraphs.
...
It is therefore pertinent to re-examine the possible range of values of the contact area a, in
particular in the light ofthe micro-indentation hardness data published, for example, by Brace
(1963).
Bishop & Eldin (1950) related CTp' the value of the average interparticle contact pressure, to S,
the value of the crushing strength of the grains, by the expression
CTp-U = bS, (10)
where the constant b will depend on the type of surface failure produced, but with a minimum
value probably not much Iower than unity.
30
.- 200
I
(
I
~
~I I
io-.-o.W,.
~ ~
..
~
I
I : ~:<>
..
~
1
.S
... - 150 za
-S
~
20
-
c
~
I
I
.f I
- 100.f
.,c i .,"o
5... .,
t::
!l':! 10 ~
:e :e
.,.
:;J - 50 ~
b b
Ul
'"
o,
o o
~
~
~ 6
s:- I : ~ 1-0
7'~. <I
4
"r
.~ ~
: 1
..... 2
I ~ I
'"o
'5., ~ ~
a::s o -

>
~
LOOO ..
~
I
1
et -6 a
.S
... z
- ::;s
-
S
b
ee
.,"... 500
-
~
-4 ;
/
~
.,...
Ul
-2
c,

-
o
O 5 10 15 20 25 30 o
axial strain, e1 (%)
FIGURE 13. Ham river sand tested with (CTa-U) =
69.0 kN m-2 (10 lbfin-2)
(sample porosity at the start of the shear stage 40.9 %). =
~ .. t
.
:..- . " .
. The intergranular stress Ui (defined as the intergranular force per unit area) was related to up
by the expression U i = aaP' (11)
" '"
Since ui is related to a and u by equation (2):
't
ui = (u-u) +au,
we have, from equations (10) and (11),
(u - u) + au = abS + au
u-u
ar a =}jS' (12)
I I
- 1.0
140
Vi I """ I
120 V ~
~ u -a-..
l ..
~!: I 0.8 ....
~ ~ I
lO
"i E
.S
<...' 100
;- Z
:2
;~~
-t; !
@.
I 80
- 0.6 -t;
~
I
:;f -f b
'-'
i .. 4.l
o
t::
GS
o
..
4.l 60 - 0.4 fi
$
"'O
~
:ti
'"~
1:: 40 5
1::
'"
- O.~ cn
20
, ....
,o
o'
~ o o
:::
"
s::-
..-
2
<1
I l
t::~
';;j 1
I ~ i
1:: ~ I
'"o I ~
o \:Q.... ~ I
54.l I ruY'".cr-- I
E ~~
::3
-' 30
:>
4000
....
"i
.S
....
@.
/ \ -'
"?
..::3
4.l~ 2000
/ ~
-
'"
~
..
e, I . 1\

o I ~
o
o 10 20 30
axial strain, el (%)
FIGURE 14. Ham river sand tcsted with (0'3- u) = 363 kN m-2 (52.61bfin-2).
Skempton (1960) gives an expression for the general case of a cohesive particulate material
with an initial contact area ao at zero stress, with the assumption that junction growth during
shcar is negligible: (J'-U
a-ao = Mk' (13)
where k is the intrinsic cohesion defined in equation (4) and
M = 2co.s1fr .
l-sm 1fr
- 1.0
140 _._---- --'-r- i
~I ~
r/- II ~ ~ I I
120
~
i
~ 1O( )
! li ~ "'-"'00()
- 0.8
@.
--; 7
I 80
- 0.6 -
-;
.f -f
I
O
u O~
u
..5 60 - 0.4 5
..
~
:a ~
:a
~ 40 rJ'"
t:
'" - 0.2 V)
20
,r
o o
~
~
'l :::
s-
<I
c~
.;
t:
'"u O~----+------H-~~-+----~----~----+----
'C
ti
8
:l
-'
30
;> 4000
~ ..
~
I
i 8
.S
'- - 20 Z
z
-e
S
~ 2000
-
b
..
0-
:l - 10 :l
'"rJ..
c,
/ ~
..c,
O

u I /
o
o 10 20 30
axial strain, ti (%)
FIGURE 15. Ham river sand tested with (O'a-u) = 363 kN m-2 (52.61bfin-2).
"Z
"
,
-I
Since it follows from the geometry of the Mohr diagram (figure 17) that,
2 cos tfr 2 sin tfr
CT1-CT3=k1 . tfr+CT31 -sm . tfr' ( 14)
-sm
then S, which is equal to CT1 - CT3 when CT3 is zero, is given by the expression
S = k 2cos tfr (15)
1 - sin tfr'
l.e. S = Mk. ( 16)
- 1.0
I "
ri ~ I
...~ ~I I
12D
r-~
!
0.8 ~
..
~
I
-;
e
.5 '
'-
z
-
~
~
I 80
I - 0.6 -
~
~
I
.f j .f
Q) Q)~
u u
.::: c
e - 0.4 ~
~
:.a ~
:.a
5
!::
40
~
'" - 0.2 '"
o o
1~------r-------r-------r-----~~T-------T---~~~~~
-
- 30
..
~
I
.S I
/ \ - 20 ze
~
-;
'-
;:3
~ 20.00
\ 6
~
\ - 10 .;
...
;:J
~
...o.
\
Q)
I 1\
o
o 10 20 30
FIGURE 16, Ham river sand tested with (0'3-U) = 363 kN m-2 (52.61bfin-2)

For a sand recently sedimented in water (as in the present case) ao may be taken as zero. Equa-
tions (12) and (13) then give the same value of contact area a ifthe coefficient b in equation (12)
is assumed to be equal to unity. t
Taking the values of k and 1/r for intact quartz from Skempton (1960), 932 MN/m2 (9500
kgfJcm2) and 13tO respectively, we have from equation (15) a value of S equal to 2255 MN/m2
This is more than an order of magnitude greater than the value of 138 MN 1m2 used by Bishop &
Eldin (1950) on the basis of the strength of granites and quartzitcs under low hydrostatic pressures.
The calculated contact are a on the shear surface corresponding to the state of stress in the tests
run with (0"3-U) ,= 363kNJm2 drops from O.4%to ca. 0.025%.
shear stress, T
--------- k
--------
O ~ ~ - - - - - - - - - - -~~-- __ !__-----7-::;--------l-_-
L k 'o, i
. A O'
J"'~I--,-a:-"-3 --!~~I~f--"4-,-(T,-'-1-_0'"3",-)--..J
2 cos 1fr 2sin 1fr
normal
stress, O'"
sm1fr = O O" hence (0'"1-0'"3)

1
= k
l-sm
. 1fr +0'"3
l-sm
. ifr
FIGURE 17. The geometry of the Mohr diagram for ,l! solid with intrinsic cohesion and friction.
Even with this lower value of a the value of allu/(O"n -u) is 0.012. This would imply discon-
tinuities of 1.2 % in the resistance to shear as the value of u was varied between the upper and
lower limits if the intergranular stress controlled the shear strength. These would have been
clearly discernible. In the case of Skempton's effective stress equation the factor tan 1/r/tan 1>'
reduces the implied percentage discontinuity to 0.4 % (for 1/r = 13tO and 1>' = 34.2 as in the test 0
illustrated in figure 15. )This value is at the limit ofdiscrimination ofthe testo On this basis, and
for pore pressure changes of this magnitude, it is apparent that the predictions of the Terzaghi
effective stress equation and Skempton's effective stress equation do not differ significantly from
each other or from the observed behaviour of quartz sand in a multistage test.
However, the prediction of the value of a, which forms the basis both of the intergranular stress
equation and of Skempton's equation, has so far been made without attempting to evaluate the
infiuence either of junction growth or of the actual magnitude of the interparticle contact forces.
Before examining these two factors it is necessary to look more closely at the conditions undcr
which the Mohr-Coulomb failure criterion can be used to relate the shear strength Tlon a plane
on which the normal effective stress is O"~ (to which Skempton's analysis applies) to the principal
stresses in an axially symmetrical compression test (as used in the present experimental program).
t There is evidence that b may ris e to about 3 for metaIs (Bowden & Tabor 1950). The predicted contact area
would be correspondingly reduced, a point which is discussed further in relation to the tests on lead shot whei e
there is direct evidence of the magnitude of the area of contact.
ln the absence of pore pressure the effective stresses equal the total stresses and the Mohr-
Coulomb criterion is given by the expression
(17)
The theory can be applied rigorously only ifthe parameters c and <p are invariant with respect to
changes in the inclination ()ofthe reference plane (figure 18), changes in the stresses on this plane
being given by the expressions:
(18)
(19)
where 0"1 and 0"3 denote the major and minor principal stresses.
T = t(O\-0'3) sin 2()

0' = t(0'1 + 0'3) + t(0'1 - 0'3) cos 2()
~
..'"
~r.'----+---~~
01
FIGURE 18. Stresses on the reference plane.
I t follows from the conventional analysis for isotropic materiaIs that the shear surface is inclined
at an angle O, where 0= 45 + 1<1>', and that the principal-stress difference is related to the minor
principal stress by the expression
2 cos <I> 2 sin <I>
(0"1-0"3),=C1 . <1>+0"31 -sm.
-Sln
<1>' (20)
This application ofthe Mohr-Coulomb failure criterion to effective stresses presents no difficulty
if the Terzaghi expression 0"' = O" - u is used, since u is invariant with respect to changes in the
inclination O of the reference plane. However, in terms of intergranular stress, we have the
expresslOn
= c' +{(<T-u) T,
+ au} tan <1>'. (21)
Skempton's expression for effective stress gives
T, = c' + {(<T- u) + au :~~ ~,} tan <1>''. (22)
------ _-- -----_ _----------------_ _ _._~--_ __ ._ . .... _--~_ ..__ .. _-_._ .._--_ .._-----------_. __ .~--._._---.... .~_. __ ._ .._ .._---_ ... __ ._--------~.... ~-~._--
I
"
,p't:t,;,IH
'.i.! ~.
, L
In both cases a is not invariant with respect to changes in e. The expression for a (withoutjunction
growth) given in equation (13) is
(T-U
a = ao+ Mk '
Substituting in equation (22) we have
T, = I
C + {(
(T - U
)
+ U ao + (TM- kU) tan
( tamjr} ""
cp tan 'f'
I
Assuming ao to be an isotropic property, and collecting the terms which vary with (T we obtain:
tan 1jr "" ( ){ U tan 1jr} ,/,, (23)

T, = C
I
+ uao---::i:i tan 'f' + (T - u 1 + Mk---::i:i tan e .
tan 'I' tan 'f'
ln terms of the Mohr-Coulomb failure criterion we now have a new material in which the
parameters c' and 1;' are replaced by (c')u and (1;/)U where
(
c ')
u = C
I
+ u (tan 1jr) tan ""
ao tan cp' ,
(24)
and (cp'),. = arctan {( 1 + ;k ~:~~/) tan cp} (25)
The term ((T - u) then correct1y reflects the variation in normal stress with change in the value
e,
of since u is invariant for each particular stage of a multistage test.
We may therefore replace c and cp in equation (20) by the expressions given in equations (24)
and (25) and replace (T3 by ((T3-U). For small differences between the parameters (c')u and c',
and (1;'),. and 1;', we may obtain the change in the stress differencc at failure ((TI - (T3), with
a change in u by differentiating this expression with respect to ut, ((T3-U) being constant as in
the tests. This leads to the expression
(26)
-.
:0_.
.
Now it follows from the geometry of the Mohr diagram that the component of ((T - u) normal
.
to the shear surface is given by the expression
((Tn-U) = c' cos 1;' + ((T3-U) (l+sin1;/). (27)

The component ofarea ofcontact which is a function ofnormal stress is given by the equation
(28)
Thus the expression for the change in strength with change in pore pressure, ((T 3 - u) being held
constant, reduces to
(29)
where ar is the total contact area on the shear surface and ar = ao + a,.
This expression is the same as that arrived at by rather different reasoning by Skempton (I960).
The validity of adding components of contact area may be questioned, especially in a 'complex
t Provided sinifJ' differs substantialIy from unity, a condition which is satisfied in alI particuIate materiaIs.
15
~.'' ' . ...I ~ ~.' .. ~_~.:.. ~i ...,
,\~~~~~':,r.:,,~~p~.,~'
..: '~.,> ..;:.:
;
.,., 116 A. W. BISHOP AND A. E. SKINNER
material such as concrete, but this is not relevant to a discussion of cohesionless particulate
rnaterials in which ao = O and c = o. However, two other points which have to be considered are
the influence ofjunction growth and the effect ofthe magnitude ofthe intcrparticle forces.
It should be noted that if the contact area ceases to be a linear function of the normal stress
component (CT - u) and varies with the magnitude ofthe shear stress T, which is zero on the major
and minor principal planes and a maximum when 0= 45, then the analysis presented in the
preceding paragraphs is no longer strictly applicable. However, the quantification ofthe phcno-
menon ofjunction growth is itself subject to some uncertainty (see, for example, Bowden & Tabor
1954,1964). It is clear that the stress path ofthejunction material would be ofthe form indicated
in figure 19, and, with clean surfaccs, could in the limit reach the intrinsic failure envelope ofthc
material. Some bounds can be put to the probable values ofinterparticle contact pressure under
normal pressure alone (CT p) and at slip (CT nj) from experimental observations of the coefficient of
friction Il and the results of the micro-indentation hardness testo
//,
/ I
'~:~/ /~/_/ __ /_/ I~!T ~ ~
arctan I' CT"j
normal stress, (J, at junction

FIGURE 19

lt follows from figure 19 that the lowcr limit ofthe stress (T nj corresponds to the upper limit of
the observed coefficient of friction. ln a series of test conducted so as to ensure a clcan contact
between a slider and plate a value of Il = 0.8 was obtained for quartz by Skinner (1975). Taking
values of K = 941 MNfm2 and ljf = 161 for quartz (based on a re-assessrnent ofpublishcd data),
we obtain from the geometry offigure 19 a lower limit for CTnj of 1851 MNfm2
As a simplification we may use the expression:
(30)
This gives the maximum value ofthe contact area a in the test run with (CT3 - u) = 363kNfm2 as
567 x 10-3
a = 1851
= 3.06 x 10-\
the value of (CTn -u) being 567kNfm2
The corresponding minimum values of CTp implicit in equation (12) (with b = 1) and in
equation (13) are both 2510 MNfm2 Howcvcr, depending on the geometry of thc interparticle
contact and on the yield pattcrn, the value of b may be as high as 3, leading to a value of
r
.
.. ~~ .
"
I -
O'p = 7530 MNfm2 Withoutjunction growth the value of a, from equation (12), would lie in the
range 2.26 x 10-4 to 5.53 X 10-5
These values represent the largest probable values of a. The micro-indentation hardness tests
on quartz published by Brace (1963), and illustrated in figure 20, suggest that at smalI values of
the interparticIe contact force the value of O'p required to cause yieId may be substantially
greater than at the larger Ioads used on a typical slidcr. Based on an analysis ofthe stress path the
corresponding value ofthe coefficient offriction is lower and is associated with a higher value of
O'nl. For an interparticIe force of 2gf (the minimum used by Brace (1963)) thc value of JL drops
to 0.4 and the value of O'ni rises to 5688 MNfm2 (Skinner 1975). For the stress levei under con-
sideration the value of a (from equation (30)) drops to a value of6.53 x 10-5
indenter loadfN
0.01 0.1 1.0 10
-,
, 3000 ~ 't-~ 1- -~~----+---I--+--+--+-30 ,
~.: 2000
~
I ~~-t.
I1-+---+---+-....1--+-4---+----+--+--+--+1- i 20
~
g
I 1000
:~ ---....:r"(!..!,:)..!:;_~-.l.- 61---' i------t----+-l---1lilO
j
.~"\.>---<. . .)
0+-----~--~--~4-+---~----~--~~+_--~----~--~~~0
1 10 100 1000
indenter load/gf
FIGURE 20. Hardness against load applied to indentcr: data from Brace (1963).
It is of interest to note that for a regular cubical packing of uniform spherical particles the
interparticIe force P is given by the expression (Bishop 1965):
(31 )
An equivalent uniform material having the mean particle size of Ham River sand (0.2 mm
diameter) and subjected to the present stress (567kNfm2) would thus have an average inter-
particle force P equal to 2.3 gf. This force almost certainly represents the upper limit for particles
ofthis mean size and stress leveI. It may thus be inferred that the value ofthe contact are a a tends
towards the lower limit in the present tests.
The corresponding values of d(O'l-O'a),fdu and of the percentage change in compression
strength are given in table 3, on the assumption that the error in applying equation (29) to the
cases involving junction growth is acceptable in view of the other unccrtaintics involved. As
15-2
convncing evidence of a discontinuity in the stress-strain curve requires a jump of about 0.5 %
in strength it will be seen that the interpretation to be placed on the absence ofjumps depends on
t~e si~nificance attached to the influence ofparticle contact force. For a pore pressure change of
this magnitude (26.5 MNfm2) the use of the Terzaghi equation is fully justified. The evidence
against the intergranular stress equation is just significant, but the test is not defini tive in terms
of Skempton's effective stress equation.
TABLE 3. PREDICTEDCHANGESIN STRENGTHFORA CHANGEIN PORE PRESSURE!lu= 26.5 MNfm2,

ANDEQ.UALTO 0.363 MNfm2
THE VALUEOF (0"3-U) BEINGCONSTANT
intergranular stress Skempton's effective Terzaghi effective
equation (2) stress equation (5) stress equation (1)
"
, A
, A
,
assumption about 6(0"1-0".), (O"I-O"')' (%) (O",-O".), 6(0",-0"3)' (%) 6(0",-0".), 6(0"1-0".), (%)
contact area a value of a kN/m3 (0"1-0"3)' kN/m' (0"1-0"3)' kN/m' (0"1-0".)
without junction 2.26 X 10-' 15.4 1.65 6.61 0.71 O O
growth, and
with b = 1
without junction 5.53 x 10-6 3.78 0.41 1.62 0.18 O O
growth, and
with b = 3
with junction 3.06x 10-' 20.9 2.24 8.97 0.96 O O
growth, max.
probable value
;
s with junction 6.53 x 10-6 4.46 0.48 1.91 0.21 O O

r.;
growth, value corre-
sponding to load of
2 gf per contact
A test with a greatly increased vaI ue of 0"3 - u is the most obvious wa y of increasing the inter-
particle contact forces and thus reducing the yield stress at the contacts. Such a test is illustrated
in figure 21, the breaks in the curves at the pressure change zones indicating the difficulty of
maintaining (0"3-U) constant whiIe adjusting both 0"3 and u continuousIy once the range ofthe
seIf-compensating mercury controI was exceeded.r Here the value of (0"3-U) is 6.91 MNfm2
(1002.6Ibfjin2) and Aa = 41.4MNjm2 (6000 Ibf/in2).
,;
. '
,
If we consider the point on the stress-strain curve at which the axial strain is 7.52 %, and the
stress difference (0"1 - 0"3) = 12.8 MNjm2 (1852Ibf/in~), then the mobiIized value of 9' = 28.7.
"
Proceeding as before, ;1nd considering only the cases withjunction growth, we obtain the vaIues
given in table 4. Since the vaIue of (O" n - u) on the failure surface is now 10.2 MN/m2 and is thus
18 times Iarger than in the test discussed above, the interparticle contact force might be expected
to correspond to a contact area mid-way between the limiting values. The absence of a step in the
stress-strain curve would then constitute strong evidence against the validity ofthe intergranular
stress equation (2), and significant evidence against Skempton's effective stress equation (5).
However, samples of Ham River sand sheared at this stress leveI (i.e. 6.9 MN/m2) show a
substantial degree of particle crushing as the shear strain increases. The change in particle size
distribution observed in an earlier series oftests and described by Bishop (1966) is illustrated in
figure 22. While it is difficult to quantify the effect ofparticle fracture on the average interparticle
force, it will clearly involve a substantiaI reduction. The prediction of the influence of pore
pressure change using Skempton's equation thus returns to the range ofvaIues where discrimina-
tion is difficult, unless the magnitude of the pore pressure change is increased to a value well
beyond the range ofthe present equipment.
.....
~ ~ .. ;
'"
s.. ;"'L'
..1 l'
, .'
-
- 15
~- ---~
'r
2000 "-T
p
......
O;
.5 Vi ..
I
'- I E
--S 1500
- 10 Z
I
~
'?
I
--
'?
~ I
4)
u ~
5... 1000 4)
u
I
e
l
-a
~ -5 ~
1::
'"
500
I
o o
c::~ O
'(;j
.........
:;~ -2
u~
'E~
sS:- -4
::l<l

:> -6
.......
..
.......
I
8000 i
E
.5 , -
Z
_. '- \
o. S
-- 40 ~
--b
b
..
4)~ 4000
\ -
...

a
'"..
4)
\ - 20 ::l
~
...
c, - c,
<J
u o <J
u
o 5 10 15 20 25
axial strain, ti (%)
FIGURE 21. Ham River sand tested with (u3-u) = 6.91 MN m-2 (1002.61bfin-2).
o TABLE 4. PREDICTED CHANGES IN STRENGTH FOR A CHANGE IN PORE PRESSURE Su = 41.4 MNfm2,
THE VALUE OF (<T3 -u) BEING CONSTANT AND EQ.UAL TO 6.91 MNfm2; HAM RIVER SAND
intergranular stress Skempton's cffcctive Tcrzaghi effcctive

\ "--
assumption about 6(0',-0'3)' .6(0',-0'.),(%) . 6(0',-0'.) 6(0',-0'.), (%) 6(0',-0'.) 6(0',-0'.), (%)
contact area a value of a kNfm2 (0',-0'3)' kNfm2 (0',-0'3)' kNfm2 (0',-0',),
with junction growth, 5.53 X 10-3 423 3.31 225 1.76 O O
max. probable value
with junction growth, 1.18 x 10-3 90 0.71 48 0.38 O O
value corresponding to
".0 0
load of 2 gf per contact
100 ~------~------~--------r-------~~~
__~------ _
(a)
80 sheared at
0';=27.6 MN m?
.(4000in -Z) Ibf
60
I-----+-----+-----I---}
sheared at ./
",
40 0';=6.9 MNm -z -/,-=----1-+1 ----l------J
(1000 IbftJ."" [I
20 ~-_----I"...""''''
~ glacial til!

unshcared
O~------~--------~------~----~--~-------L------~
0.002 0.006 0.02 0.06 0.2 0.6 2
100.-------~----~----.----_._,~~~---~
(h)
!;(J 1------
shc.ircd .u
0';=27.6 MN 111-~- ~_ I
'(40()O Ihr inC)"
60 --- -------j
"~
.
..~
,
40 ----J---._-~---
.- consol ida tcd
under 0';.=27.6 MN 111-'
_ -.-.-
It-_- initial
-.----l
gruding
(4000 lbf in . ')

20 I------r---i------I
I
()~-------L---------L--------L---
O.()02 O.O()6 O.O!>
__~--L-------~--------~
(J.()2 0.2 0.('
fine I mcdium I coarsc inc

I mcdium I coarsc
silt Iraction sand Ir.rction
partcle szejmm
FIGURE 22. Changes n gradng of saturated samples of Ham Rver sand resultng (a) from shear at dfferent
pressures and (b) from consoldaton as compared wth complete shear tests (tests by Skinner, 1964-66).
(b) Lead shot

Owing to the low yield stress oflead, this material offered the prospect of defini tive results at
relativcly modest stress levels. The results ofa test run in the high pressure apparatus with (0"3 - u)
eq ual to '363 kN Jm 2 (52. 61bfJin 2) and a variation of 0"3 between 1.03 MN Jm li (150 1bfJin 2) and
27.58 MNJm2 (4000 lbfJin2) are shown in figure 23. Once again, there is no discernible change
in shearing resistance rcsulting from a change in pore pressure of 26.5MNJm2 (3850IbfJin2)
with (0"3-U) hcld constant.
.:
- 1.4
200
.....,.
~~ - 1.2
~
<j
.S
....
150
I I I I
Nu,I I
I
'-'-00
I
i-=c
~
<j
1.0 Z
E
-S;
.t\
-
I
100
/ I
- 0.8 I
-;
~
I
v
- 0.6 g
v
~
:.a
- 0.4 5
!::
'"
1/ - 0.2
,.
o o
~
.....
..,.0
s:-
"
- -e
c~
'@
~
u
'':;
ti
E;:l

:> - 30
4000
I
f
~
<j
.S
.... -
2
~ 2000
...
CI)
;:l -
~
...
c,

u I \ I ~
o
o 10 20 30
FIGURE 23. Lead shot tested with (G"3-U) = 363 kN m-2 (52.61bfin-2).
122 . A. W. BISHOP AND A. E. SKINNER
The are a ofcontact may be derived by three independent methods. Skempton (1960) gives the
values ofthe intrinsic parameters k and 1fr for lead as 9.81 MN/m2 (100kgfJcm2) and r
respec-
tively. Taking the value of (O"n - u) at an axial strain of9.24 % as 591 kN/m2 we obtain a value of
contact are a a of2.97 % using Skempton's expression (equation (13)). For b = 1 the sarne value
is obtained from equation (12), but with b = 3, which seems more appropriate in the light of
values obtained on other metaIs by Bowden & Tabor (1954), the value of a drops to 0.991 %.
Micro-indentation hardness tests on representativc lead shot from this series of tests gave a
mean normal stress H; of93.1 MN/m2 (950kgf/cm2). On the assumption that H; = 3, S = 6 Mk,
this value corresponds to k = 15.3 MN/m2 and suggcsts a slightly stronger material than the lead
quoted by Skempton. Taking the contact pressure as approximating to the indentation hardness
-,.
value we obtain a value of a = 0.635 %. The low value ofcoefficient offriction (0.1) measured in
direct observations of interparticle friction (Skinner 1975) suggests that friction is controlled by
surface films on the lead and thatjunction growth is not very significant in this case.
TABLE 5. VALUES OF CONTACT AREA a FROM OEDOMETER TESTS ON LEAD SHOTt
vertical pressure p in oedometer value of contact

A
area a on release
MNJmz kgfJcml of stress
2.65 27 0.03
12.55 128 0.11
100.42 1024 0.95
t Data from Laughton (1955).
Thirdly, Laughton (1955) measured with a microscope the facets caused by subjecting lead
shot to a series of normal stresses in an oedometer. Since thcse facets represent the inelastic
component of deformation, they indicate- the minimum values of a. Laughton's observations,
which are also quoted by Skempton (1960), are presented in table 5. It will be seen that up to
a contact area of 11 % the relation between a and the vertical pressure p is almost linear and is
given by the expression
a = 0.0120p(1- 0.0216p), (32)
where p is expressed in MN/m2
In the oedorneter the vertical pressure p is applied through a piston and the sample is confined
in a rigid-walled cylinder. The radial stress, neglecting side friction, is equal to KoP, where Ko
is the coefficient of earth pressure at rest and approximates numerically to (1 - sin </>') on first
loading (Jaky 1948; Bishop 1958). Laughton did not examine the variation ofcontact area with
the direction of the plane of reference and the average contact are a measured wiIl therefore
correspond to a normal stress of H 1 + 2Ko) p.
In the present test the peak value of </>' = 39.7 and the equivalent normal stress O" is related to p
by the expression
O" = 0.573p. (33)
For the value of (O"n -u) = 591 kN/m2 we thus obtain a = 1.21 %. It is ofinterest to note the
relatively elos e agreement of this direct experimental value with the values deduced from
equation (12) with b = 3 and from the micro-indentation hardness tests, having regard to the
fact that they refer to different samples of lead.
The magnitudes ofthe predicted changes in strength for this particular test are given in table 6.
It will be seen that the intergranular stress equation predicts an increase in strength, for the
! .
.. ,
TABLE 6. PREDICTED CHANGES IN STRENGTH FOR A CHANGE IN PORE PRESSURE u = 26.5 MNfm2,
THE VALUE OF (o-s-u) BEING CONSTANT AND EQ.UAL TO 0.363 MNfm 2; LEAD SHOT
intergranular stress Skempton's efTective Terzaghi effective

assumption about d(oo,-oo.), d(oo,-oo.), (%) d(oo,-oos), d(oo,- 00.), (%) d(CT,-oo.), d(CT,- CTo),
value of a
(%)
con tact arca a kN/m2 (CT,-CT.), kN/m2 (CT,- CTa), kN/m" (CT,-CT.),
without junction 0.991 x 1O-a 894 72.3 14.4 1.17 O O

growth with b = 3,
using Skempton (1960)
valucs of k and t/J'
wiihout junction 0.635 x 10-1 573 46.4 9.24 0.75 O O
growth, using results
of micro-indentation test
using interpretation 1.21 X 10-1 1091 88.3 17.6 1.42 O O
from Laughton (1955)
experimental values
1
1
I - 250
.: IT ~
~ 30
Mk 11
- 200 f'
"I
.5
'-
S
---;
I ~ ro E
z
--
~
.f
I
20
J - 150 -; .
.f
I
lU
U
..
t::
lU
I - 100
'"
o
~
~ ~
:a :a
<Il
<Il 10 <Il
lU
n
~ - 50 .::
<Il
':',-,
o o
~
......
~ 11 ~ -o
s:- 4
I I r-4Y
<l ~
t::~
.~ 2
,.-o- ~T
.::
<Il I~
5
o
lU o ~ ~ rr I
E
~ -1

;>
1000
~ -6
"I
.5 -
'-
--S
b"
500
-4
..
IU~ -
~
<Il
<Il -2
..'c,
"
-

o
, o
o 5 10 15 20
FIGURE 2~. Crushcd marble tcsted with (oo.-u) = 9.0 kN m-2 (10 lbfin-2).
16 Vol. 284. A.
specified increase in pore pressure, ofbetween 46.4 % and 88.3 %, whereas no detectable increase
can be observed in figure 23. The validity of this equation is thus undisputably disproved.
Skempton's effective stress equation predicts an increase of between 0.75 % and 1.42 %, and,
though much less at variance with the observed stress-strain curve, must be considered to be
~, . significantly inconsistent with the experimental resulto
Only the Terzaghi equation predicts to within the accuracy ofthe experimental observations.
values of cell pressure, G"3/(MN rn-2)

0.35 6.90 0.35 6.90
~I
J "li'
4
"I I" "11 - 250

i i : I II
I
I I 0.05 lbf
l~
,I-t in-2
I1
0.34~cN II2 ii II
I
J '\
I
I
I
I
- 240
11
"
,
34
I1
..
.......
I
.s
j
I
\
~
11
::
"
I
I - 230
'- I
2
b'"'
~! II
,
I 32 I -
~ I EB
\
I
!li II
U
I'::
!li
":...,.... Jj I
:.a I
\
~
~
II
, ::l
t: - 210 b
'" '"
30 I;t(
N~ i 200
\
I
I
28
~~ - 190
B
.
O 5 10 15 20
I
FIGURE 25. Crushed rnarble - an enlargernent of lhe critical part of lhe stress-strain curve
shown in figure 24.
,.., _,' .. 1,
!..' INFLUENCE OF HIGH PORE-WATER PRESSUR~ 125
(c) Crushed marble

Part ofthe testing facilities required for the high pressure cell was in use on another project at
this stage ofthe programme, and the multistage test on crushed marble was therefore carried out
in the 0-6.9 MN/m2 capacity cell. Particular care was therefore taken to eliminare stray sources
of scatter from the values defining the stress-strain curve.
The test was run with (0"3 - u) equal to 68.9 kN 1m2 (10. OIbf/in2) and a variation of 0"3 between
349 kN/m2 (50 Ibf/in2) and 6.895 MN/m2 (1000 Ibf!in2). The test results are presented in figure 24
and the critical section of the stress-strain curve is given to an enlarged scale in figure 25. In-
spection of the curve in figure 25 suggests that any consistent jump in the curve of more that
0.34 kN/m2 (.O lbf/in") would be readily discerned, i.e. ajump ofO.14 %.
TABLE 7. PREDICTEDCHANGESIN STRENGTHFORA CHANGEIN POREPRESSURE/::.u = 6.55 MN/m2,

THE VALUEOF (0"3-U) BEINGCONSTANT ANDEQUALTe) 68.9 kN/m2; CRUSHEDMARBLE
intergranular stress Skempton's effective Terzaghi effcctive
equation (2) strcss cquation (5) strcss equation (1)
I \
assurnption about 11(0",-0"3)' 11(0",- 0"3), (%) 11(0",-0"3)' 11(0",- 0"3)' (%) 11(0",-0"3)' 11(0",-:0"3)' (%)
contact arca a value of a kNfm2 (0",-0"3)' kNfm2 (0",-0"3)' kNfm2 (0",-0"3)'
without junction 2.63 X 10-( 6.04 2.50 1.03 0.43 O O

growrh, and with
b = I
without junction 8.78 X 10-' 2.02 0.84 0.34 0.14 O O
growth, and with
b = 3
with junction growth, 3.39 X 10-( 7.78 3.22 1.32 0.55 O O
probablc valuc
The predicted values are based on two sets of data. The intrinsic parameters k and 1fr may be
assumed to approximate to those of calcite given by Skempton (1960) as 186.3 MN/m2 (HlOO
kgf/cm2) and 8 respectively. Taking the value of (O" n - u) at an axial strain of 5.69 % as 112.85
kNfm2 we obtain a value of the contact area a ofO.0263 %, using Skcmpton's expression (equa-
tion (13)). For b = 1 the same value is obtained from equation (12), but with b = 3 this value
would fali to 0.00878 %. Micro-indentation hardness tests on the actual material gave a value of
H" = 2.354 MNlm The value of the coefficient of friction fL may then be obtained from the
2
rclationship obtained by Skinner (1975):

k
fL = tan 1fr+ H . 1fr' (34)
vS1n
This gives a value ofO. 7, which is dose to the range ofvalues obtained for calcite under saturated
conditions by Horn & Deere (1962), namely 0.6-0.68.
Assumingjunction growth as in the case ofquartz sand we obtain a value of O"ni of330.0 MN/m2
and thus a valuc of a = 112.85-:- 333.0 x 103 (= 0.'0339 %). Thc results of micro-indentation
hardness tests on calcite by Brace (1963) indicate that the effect of the magnitude of'interparticle
force is in this case relatively unimportant,
The magnitudes of the predicted changes in strength on the basis of these contact areas are
givcn in table 7. The validity of intergranular stress equation is again disproved. Skcmpton's
cflcctive stress equation, assuming junction growth, predicts a jump of about four times the
minimum magnitude which could be detected, and must therefore bc considercd to be signifi-
16-2
,
e
cantly at variance with the experimental results. The Terzaghi equation again gives a correct
prediction,
(d) Braehead silt
Two multistage tests were carried out on Braehead silt in the 0-6.9 MNfm2 capacity cell with
(er3-u) equal to 68.9kNfm2 (10.0IbfJin2) and a variation of er3 between 349kN/m2 (50IbfJin2)
and 6.895 MNfm2 (1000 lbfjin2). The results are presented in figures 26 and 27. Inspection ofthe
curves does not show any significant change in shearing resistance following the changes in pore
pressure.
25
~ !! r
n I1
<r
~
I1 ~h_ 150
I
20
/
15
I
V I I
10
/
5
~
I ,~

--
-
o o
r\<,
1I i- __
"o... ---1_.'.
~
L! I I
,'I "'-'li
I I
-8
v....~
I / -
..I
::l a
-4
~z
l3.;:E
I I lI---
-
o
--
-~
.., "
u b
o 5 10 15 20 25
FIGURE 26. Braehead silt tested with (us-u) = 69.0 kN m-2 (10 lbfin-2).
The theoretical prediction of the magnitude of strength changes on the basis of the two
equations involving the magnitude of the contact area a is subject to some uncertainty, as the
predominant mineral is quartz and the average interparticle force is now at least two orders of
magnitude smaller than the minimum load used in the micro-indentation tests by Brace (1963).
In the absence ofthe data for evaluating (Tnj' table 8 (which refers to the test in figure 27 at 15 %
strain) is restricted to estimates based on equations (12) and (13), which do not take account of
junction growth or of the magnitude of the interparticle force. However, these values serve to
show that a pore pressure change of this magnitude (6.55 MNfm2) is insufficient to provi de
25
. !.l 11
I /jl li ~
,/ - 150
20
I
vn ..
..-..
Z
I
S
::::-
- 100 :::;;
/
b
,
.f
<)
o
c
<)
/
~
~
. - 50 '"
g
'"
5
r
o o
r-.<, .~~
vr
I1
II
II
lLr
iI 1I
-8
-
-
-
o 5 10 15 25 o
FIGURE 27. Braehead silt tested with (lT3-u) = 69.0 kN m-2 (10 Ibfin-2).
~.. ;~.
,
e
,, 128 A. W. BISHOP AND A. E. SKINNER

..
critical data for discriminating between the different theories in the case ofmatcrials composed
prcdominantly of quartz particles.
The test results, however, do serve to indicate that no other physical phcnomenon of greater
significance than the interparticle contact area effect enters into the effective stress relationship
as the particle size decreases and the material begins to show plasticity in the sense indicated by
the Atterberg limits (the liquid limit being 29 % and thc plastic limit 23 %).
TABLE 8. PREDICTED CHANGES IN STRENGTH FOR A CHANGE IN PORE PRESSURE !lu = 6.55 11N/m2,
THE VALUE OF (0'3-U) BEING CONSTANT AND EQUAL TO 68.9 kN/m2; BRAEHEAD SILT
intergranular stress Skemplon's effective Tcrzaghi effective

equation (2) strcss cquation (5) stress cquation (1)
,----..-----, ,~------~~ ,
assumption about Il(u,- us), Il(u,- us), (o/,) ll(u,-u3), Il(u,- u3), (%) ll(u,-u3), ll(u,-u3), (%)
contact area a value of a kN/m2 (u,-us), o kN/m2 (u,-u3), kN/m2 (u,-u3),
without junction 4.24x 10-5 0.66 0.40 0.30 0.19 O O
growth, and with
b = 1
without junction 1.41 X 10-5 0.22 0.13 0.10 0.06 O O
growth, and with
b = 3
6. CONCLUSIONS
The experimental results demonstrate, with a greater degree of accuracy than has been
achieved in any previous investigation, the validity of the principIe of cffcctive stress as applicd
to the shcar strength of cohesionless particulate materiaIs. The results indicate that for the range
of effective stresses and pore pressure changes investigatcd the principIe of effective stress can be
expressed to a high degree of accuracy by the simple Terzaghi equation (7' = (7 - u. For the
Ham River sand the maximum value of the minor principal effcctive stress was 6.91 MNfm2
(1002.6Ibffin2) and the maximum change ii'l pore watcr pressure was 41.4 MNfm2 (6000Ibfjin2)
which is equivalent to a height of 4.2 km of water.
The invcstigation also demonstrates quite clearly that the often-madc assumption that shear
strength is controllcd by the intergranular stress is invalido ln the case of lcad shot this assumption
would have led to gross errors ofthe order of 46-88 % and here therc can be little doubt about the
magnitude ofthe contact area a.
The expression derived by Skempton (1960) to take account ofthe intrinsic anglc offriction of
the solid material forming the particles gives a significantly less satisfactory agreement with the
experimental data than the Terzaghi equation for lead shot and for calei te. For quartz sand the
conclusion to be drawn is less clear, owing to the probable sensitivity of the value of the contact
area a at a given value of ((7 n - u) to the magnitude of thc interparticle force. Changes in pore
pressure an order of magnitude greater than were used in the prcsent series of tests would be
necessary to give a really convincing demonstration of the status of this expression for hard
particles such as quartzo However, in as far as the expression predicts jumps in the stress-strain
curve as indicated in tables 3 and 4, it is not supported by any positive evidence.
One reason for the apparent lack ofvalidity ofSkempton's expression for effective stress may
be its dependence on the Caquot-Bishop-Home type ofrelation between cjJ'and the coefficicnt
of interparticle friction fl. More recent work (Skinncr 1969 and 1975, and unpublishcd radio-
graphic work) indicates that the mechanism of failure in random particulate materials is more
complex than assumed in any current analysis and that no uniquc cjJ'- fl rclation exists. A rigorous
analytical solution to this problem is not yet forthcoming.
, .
The only da ta available in 1960 for studying the influence of contact area on the strength of
porous materiaIs subject to high pore-water pressures was from tests on strongIycohesive materiaIs
such as intact marbIe, Iimestone and concrete. The critical data (Skempton 1960) was provided
by unjacketed sampIes (in which 0"3-U = O) tested under a range of cell.pressures in rather
cruder apparatus than used in the present series oftests. Unjacketcd samples are subject to brittlc
failure at small strains, cohesion making the main contribution to compression strength. In the
authors' view more relevant data would be obtained by running constant (0"3-U) tests in the
pressure range in which plastic failure occurs (for example 0"3 - U = 50-70 MNJm2 on the basis
of Karman's (19 I I) classic tests on marble). Herc friction plays a more importan t part in the
resistance to shear (whether considered in terms ofthe Mohr-CouIomb or the modified Griffith
failure criterion (sec, for example, McClintock & Walsh 1962; Hoek 1965)) and the shape ofthe
stress-strain curve is sufficiently flat-topped to permit multistage tests, with due allowance in the
rate of strain for the low permeability of the sample.
The early stages of the investigation were supported by the Department of Scientific and
Industrial Research and by the Civil Engineering Research Association (now S.R.C. and
C.I.R.I.A. respectiveIy). Mr F. Winser contributed to the success ofthe experimental investiga-
tion by very high precision work on the high pressure ceIls and Mr David Evans assisted in
the assembly of the various pressure systems.
REFERENCES
Aldrich, M. J. J r. 1969 Pore pressure effects on Berea Sandstone subjected to experimental deformation. Buli. Am.
Ceol. Soc. 80, 1577-1586.
Bishop, A. W. 1953 (Private communication to Dr A. S. Laughton) (see Skempton 1960).
Ilishop, A. W. 1954 Correspondence on shear characteristics o[a saturated silt, mcasurcd in triaxial compression.
Penman, A. D. M. (Geotechnique, 3, No. 8, 312-32H). Letter to Gcotechnique. Ccotechnique 4, 43-45.
Bishop, A. W. 1955 Lecture delivered in Oslo, entitled 'The principIe of cflcctivc stress'. Printcd in Teknisk
Vkcblad, No. 39 (1959), 859-863.
Bishop, A. \'V. 1958 Test requirements for measuring the coefficient of ei\rth pressure at resto Proc. Brussels Conf:
Eartli Pressure Problems 1, 2-14.
Bishop, A. W. 1965 Discussion. Proc. 61h Inl. Conf, Soil Mech. 3, 306-310.
Bishop, A. W. 1966 Strength of soils as engineering materiaIs. 6lh Rankinc Lecture. Ceolechnique 16, 89-130.
Bishop, A. W. 1969 Discussion. Proc. 71h Inl. COIif. Soil Mech. 3,182-186.
Bishop, A. W. 1973 The influencc of an undrained change in stress on the pore pressure in porous media of low
compressibility. (Technical note.) Geotechnique 23, 435-442.
Bishop, A. W. & Blight, G. E. 1963 Some aspccts of effective stress in saturated and partly saturated soils.
Geotechnique 13, 177-197.
Bishop, A. W. & Eldin, A. K. G. 1950 Undrained triaxial tests in saturated sands and their significance in the
general theory of shear strength. Geoteclmique 2, 13-32.
Bishop, A. W. & Henkcl, D. j. 1962 The measurement qf soil Jlroperlies in lhe triaxial test, 2nd ed. London: Edward
Arnold.
Bishop, A. W., Kumapley, N. K. & EI-Ruwayih, A. 1975 The influence ofpore-water tension on lhe strength of
clay. Phil. Trans. R. Soe. Lond. A 278, 511-554.
Bishop, A. \"1., Webb, D. L. & Skinner, A. E. 1965 Triaxial tests on soil at clevated cell pressures. Proc. 61h COI?f
Soil Mech, 1, 170-174.
Bowden, F. P. & Tabor, D. 1942 Mechanism of metallic friction. Nature, Lond. 150, 197-199.
Bowden, F. P. & Tabor, D. 1950 The frictipn andlubrication ofsolids. Pari I. Oxford University Press.
Bowdcn, F. P. & Tabor, D. 1954 The friction and lubrication of solids. Pari I. Corrected impressiono Oxford University
Prcss.
Bowdcn, F. P. & Tabor, D. 1964 The friction and lubrication qf solids, Pari 11. Oxford University Press.
Bracc, W. F. 1963 Bchaviour of Quartz during indentation. J. Geol. 71, 5, 581-595.
Bruggcman,"]. R., Zangar, C. N. & Brahtz,j. H. A. 1939 Memorandum to ChiefDesigning Engineer: notes on
analytic soil mechanics. V.S. DePI. of lhe Interior, Tech, Mem., No. 592.
!
r,
I '. r
I <'
f -
;
Caquot, A. 1934 Equilibre des Messifs Frottement Interne, Pulurulents et Cohrentes. Paris: Gauthier- Villars.
Hoek, E. 1965 Rock fracture under static stress conditions. Ph.D. thesis, University of Capetown. GSIR Report
MEG 383, Pretoria, South Africa.
Horn, H. M. & Deere, D. U. 1962 Frictional characteristics of minerais. Geotechnique 12, 319-335.
Horne, M. R. 1969 The behaviour of an assembly of rotund rigid, cohesionless particles. LlI. Proc. R. Soe. Lond.
A 310, 21-34.
Hubbert, M. K. & Rubey, W. W. 1959 Rle of fluid pressure in mechanics of overthrust faulting. Bull, Am. Geol.
Soe. 70, 115-206.
Jaky, J. 1948 Pressure in silos. Proc. 2nd Int. Conf. ss Mech. 1, 103-107.
Karrnan, Th. von. 1911 Festigkeitsversuch unter allseitigem bruck. Ziet, Vereines Deutsch, Ing. 55, 1749-1757.
Laughton, A. S. 1955 The compaction of ocean sediments. Ph.D. thesis. University of Cambridge.
McClintock, F. A. & Walsh,J. B. 1962 Friction on Griffith cracks under pressure. 4th U.S. Nat. Congress of Appl.
Mech. Proc. 1015-1021.
Procter, D. C. & Barton, R. R. 1974 Measurements ofthe angle ofinterparticle friction. Geotechnique 24,581-604.
Scott, R. F. 1963 Principies of soil mechanics. Addison Wesley.
Skempton, A. W. 1960 Effective stress in soils, concrete and rocks. Gonf. on Pore Pressure and Suction in Soils. London:
Butterworths.
Skinner, A. E. 1969 A note on the influence of interparticle friction on the shearing strength of a random
assembly ofspherical particles. Geotechnique 19, 150-157.
Skinner, A. E. 1975 The effect of high pore water pressures on the mechanical behaviour of sediments. Ph.D.
thesis, University of London.
Taylor, D. W. 1944 Cylindrical compression research program on stress-deformation and strength characteristics
of soils. M.I. T. 10th Progress Report to U.S. Engineers Department,
Taylor, D. W. 1948 Fundamentals of soil mechanics. New York:John Wiley.
Terzaghi, K. 1923 Die Berechnung der Durchlssigkeits-Ziffer des Tones aus dem Verlauf der hydrodynamischen
Spannungserscheinungen. Sitz, Akad. Wissen. Wien Math-natuna. Kl. Abt. lIa 132, 125-138.
Terzaghi, K. 1925 Erdbaumechanik auf bodenphysikalischer Grundlage. Leipzig: Deuticke. . - ..
.Terzaghi, K. 1936 The shearing resistance of saturated soils and the angle between the planes of shear. Proc. 1st
Int. Gonf. Soil Mech. 1, 54-56.
Tombs, S. G. 1969 Strength and deformation characteristics of rockfill Ph.D. thesis, University of London.

1976 - The Influence of High Pore-Water Pressure On The Strength of Cohesionless Soils

Uploaded by

Copyright:

Available Formats

1976 - The Influence of High Pore-Water Pressure On The Strength of Cohesionless Soils

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1976 - The Influence of High Pore-Water Pressure On The Strength of Cohesionless Soils

Uploaded by

Copyright:

Available Formats

[ 91 ]

Bv A. W. BISHOP AND A. E. SKINNER

(Communicated by A. W. Skempton, F.R.S. - Receiued 12 August 1975 - Diagramsfor

The influence on the mechanical properties of saturated particulate materiaIs of the

12 [Published 4 January 1977

2. EXPRESSIONS FOR INTERGRANULAR STRESS AND EFFECTIVE STRESS

(a) Intergranular stress

(T = (Ti+ (1- a)!u

94 A. W. BISHOP AND A. E. SKINNER

changed. For u = 100((T~u) a value of a of 1 % would lead to a product au equal in magnitude

equivalent forces on a soil grain

(b) Effective stress for changes in volume

TABLE 1. (AFTER SKEMPTON 1960.)

(c) EiJective stress for changes in shear strength

INFLUENCE OF HIGH PORE-WATER PRESSURE 97

3. TESTING PROGRAMME, APPARATUS AND TESTING TECHNIQ.UES

and Skempton's effective stress equation is

0" = (O' - u) + mau, (5)

INFLUENCE OF HIGH PORE-WATER PRESSURE 99

I.__ a:_J-_u_--..!. f-o._~=-(-,Oj,----_a:_3 )

. A' AO' h () 2sin1>'

FIGURE 5. The geometry of the Mohr diagram.

(f.'ucil/!f IJ. IUO)

INFLUENCE OF HIGH PORE-WATER PRESSURE 101

102 A. w. nISHOP AND A. E. SKINNER

K _.0.:.'10.c ,," r--'::--"':"

1.0:1 M:-; m-2

dcad-weight prcssure monitoring vacuurn system

INFLUENCE OF HIGH PORE-WATER PRESSURE 105

stress-strain curve for stress-strain curve for 0'3 - u constant

silt fraction sand fraction

INFLUENCE OF HIGH PORE-WATER PRESSURE 107

5. TEST RESULTS AND THEIR IMPLICATIONS

(a) Ham River sand

FIGURE 12. Ham river sand tested with (cra-u) =

INFLUENCE OF HIGH PORE-WATER PRESSURE 109

a-ao = Mk' (13)

where k is the intrinsic cohesion defined in equation (4) and

112 A. W. BISHOP AND A. E. SKINNER

sm1fr = O O" hence (0'"1-0'"3)

T = t(O\-0'3) sin 2()

FIGURE 18. Stresses on the reference plane.

T, = c' + {(<T- u) + au :~~ ~,} tan <1>''. (22)

tan 1jr "" ( ){ U tan 1jr} ,/,, (23)

and (cp'),. = arctan {( 1 + ;k ~:~~/) tan cp} (25)

((Tn-U) = c' cos 1;' + ((T3-U) (l+sin1;/). (27)

~.'' ' . ...I ~ ~.' .. ~_~.:.. ~i ...,

'~:~/ /~/_/ __ /_/ I~!T ~ ~

arctan I' CT"j

normal stress, (J, at junction

INFLUENCE OF HIGH PORE-WATER PRESSURE 117

TABLE 3. PREDICTEDCHANGESIN STRENGTHFORA CHANGEIN PORE PRESSURE!lu= 26.5 MNfm2,

s with junction 6.53 x 10-6 4.46 0.48 1.91 0.21 O O

INFLUENCE OF HIGH PORE-WATER PRESSURE 119

intergranular stress Skempton's cffcctive Tcrzaghi effcctive

(4000 lbf in . ')

fine I mcdium I coarsc inc

silt Iraction sand Ir.rction