Burland - Compressibility and Shear Strength of Natural Clay

Burland, J. B. (1990).
GCotechnique
40, No. 3, 329-378
On the compressibility and shear strength of natural clays
J. B. BURLAND, FEng*
The compressibility and strength characteristics of Les caracteristiques de compressibilite et de rbis-

reconstituted clays are used as a basic frame of tance des argiles reconstituees s’emploient comme
reference for interpreting the corresponding char- base getterale pour interpreter les caracteristiques
acteristics of natural sedimentary clays. The correspondantes des argiles sedimentaires naturel-
properties of reconstituted clays are termed ‘intrin- les. Les proprietb des argiles reconstituees sont de-
sic’ properties since they are inherent to the soil fitties comme des prop&t&s ‘intrin&ques,’ parce
and independent of the natural state. The proper- qu’efles sont propres au sol et independantes de
ties of a natural clay ditier from its intrinsic l’etat naturel. Les proprietes dune argile naturelle
properties due to the influence of soil structure different de ses propriMs intrin&ques a cause de
(fabric and bonding). Thus the intrinsic properties l’influence de la structure du sol (fabrique et liage).
provide a frame of reference for assessing the in Les propri&s intrinseques four&sent ainsi une
situ state of a natural clay and the influence of base gedrale pour ivaluer l’etat in situ dune
structure on its in situ properties. A new normal- argile naturelle et I’influence de la structure sur ses
izing parameter called the void index is introduced propriitb in situ. Un nouveau parametre normal-
to aid in correlating the compression character- isant appele indice des vides est introduit pour
istics of various clays. The sedimentation compres- aider dans la correlation des caracteristiques de
sion curves for most, but not all, natural clays lie compression des argiles diverses. Pour la plupart
well above the corresponding intrinsic compression des argiles naturelfes, mais pas pour toutes, les
curves. A consequence of this is that such clays are courbes de compression de sedimentation se situent
more sensitive and brittle than the reconstituted bien au-dessus des courbes de compression intrin-
material and the post-yield compression index C, &ques correspondantes. De telles argiles sont par
is usually much greater than the intrinsic value. par consequent plus sensibles et fragiles que la
This observation has important consequences for matiere reconstituee et l’indice de compression
stress-path testing of soft clays. The location of the aprLs l’ecoulement C, est normalement plus elevi
natural sedimentation curve relative to the intrinsic que la valeur intrin&que. Cette constatation a
one is shown to depend on depositional conditions d’importantes consequences pour les experiences
and on postdepositional processes such as leach- effect&es au suget du chemin de contrainte des
ing. The undrained strength of a normally consoli- argiles tendres. On dimontre comment
dated natural sediment is shown to be primarily a l’emplacement d’une courbe de sedimentation natu-
function of the in situ effective stresses and of the relle par rapport a la courbe intrinseque depend des
soil structure and not of the moisture content. For conditions de depot et des Cvenements s&ant le
overconsolidated natural clays the intrinsic com- depot, tels que le lessivage. On demontre aussi que
pression line provides a useful means of assessing la resistance nondrainee d’un sediment nature1
the degree of overconsolidation. Also the ratio of normalement consolide est en premier lieu une
the intrinsic swelling index to the undisturbed fonction des contraintes effectives in situ et de la
swelling index (the swell sensitivity) is a valuable structure du sol, et non de la teneur en eau.
measure of bonding. The strength properties of two Pour les argiles naturelles surconsolidees la ligne
overconsolidated clays (Todi Clay and London de compression intrinseque fournit un moyen utile
Clay) are presented and the intact strengths are pour ivaluer le degre de surconsolidation. Le
shown to be greater than the corresponding intrin- rapport entre l’indice de gonflement intrinseque et
sic strengths. However, both clays show brittle l’indice de gonflement non reman% (la sensibilite
behaviour with the formation of shear surfaces at au gonflement) represente une indication trb utile
peak intact strength. The strength on such a shear des liaisons. Les proprietes de resistance de deux
surface drops rapidly to a well defined post-rupture argiles surconsolidees (argile de Todi et argile de
strength after a few millimeters relative displace- Lo&es) soot present&es, et on demontre que les
ment. The post-rupture strength must be clearly resistances intactes sont superieures aux r&s-
distinguished from the residual strength which tances intrin&ques correspondantes. Cependant les
requires much larger relative displacements to deux argiles se comportent de facon fragile avec la
formatfon de surfaces de cisaillement a la r&s-
tance intacte de pit. Sur une telle surface de cis-
* Imperial College of Science, Technology and Medi- aillement la &istance d&it rapidement a une
eine, London. resistance biendefinie apr&s-rupture apres quelques
329
Downloaded by [ University of Alberta] on [03/12/15]. Copyright © ICE Publishing, all rights reserved.
330 BURLAND
develop. Evidence is given which indicates that the millimetres de d&placement relatif. I1 faut dis-
post-rupture strength may be relevant to many sta- tinguer clairemeot entre la resistance apr&s-rupture
hility problems in stiff clays and may also control et la resistance rksiduelle, qui ne se developpe que
the in situ stresses during geological unloading. For pour des deplacements relatifs plus importants.
Todi Clay and London Clay the post-rupture Des don&s sont p&se&es que indiquent que la
strengths at low con&ring stresses are close to the resistance apA+rupture peut btre importante pour
intrinsic critical state strengths. More study is beaucoup de problemes de stabilid dans les argiles
required before this can be accepted as a genera1 raides et peut aussi controler les contraintes in situ
result for most clays. pendant le dechargement geologique. Darts les cas
de I’argile de Todi et de I’argile de Londres les rk-
sistancee apks-rupture d des valeurs basses de con-
trainte avec etreinte laterale sont trb prb des
resistances intrinskques de I’etat critique. Des
etudes approfondies seront nkcessaires pour contir-
KEYWORDS: clays; compressibility; fabric/structure of mer la validiti! de ce rbultat pair la plupart des
soils; sedimentation; shear strength; sailproperties argiles.
NOTATION natural soils or artificial materials such as kaolin-

A Skempton’s pore pressure coefficient ite or illite. These studies have been of outstand-
c’ effective cohesion ing importance-perhaps the two most notable
ccf compression index

intrinsic compression index
being those of Hvorslev and Rendulic, both in the
mid 1930s. It is on these, and similar later studies
2% swelling index that the framework of what has come to be called
c,* intrinsic swelling index critical state soil mechanics has been built. In
e void ratio recent years this phrase has become generic in its
eL void ratio at liquid limit use with some of the precision of the original
eh void ratio on ICL for (I,’ = 100 kPa critical state models being lost.
G, specific gravity The critical state framework, which was formu-
Iv void index defined by equation (1) lated so elegantly by the Cambridge soil mecha-
ICL intrinsic compression line nics school under the late Professor Roscoe, has
log logarithm to the base 10 provided a coherence which the subject pre-
P’ effective mean normal stress viously lacked. It also provides a logical frame-
Pt’ capillary pressure, isotropic swelling pres- work for incorporating theories of plasticity, yield
sure and flow for the mathematical modelling of soil
XL sedimentation compression line behaviour. Over the last twenty years critical
S” undrained strength state soil mechanics has been widely taught and
S uTC undrained strength in triaxial compression increasingly applied to the solution of engineering
W percentage water content problems.
liquid limit But natural soils differ from reconstituted soils
plastic limit in a number of important ways. These differences
angle of slip plane to horizontal stem from the influence of micro- and macro-
effective normal stress structure. Following Mitchell (1976) the term
effective axial stress ‘structure’ means the combination of ‘fabric’
effective horizontal stress (arrangement of particles) and interparticle
effective radial stress ‘bonding.’
effective vertical stress When I was invited to deliver this lecture I
equivalent stress on the ICL corresponding quickly came to the conclusion that it would be
to the void ratio, or void index, of the soil both timely and appropriate to undertake a
effective overburden pressure review of the basic compressibility and shear
effective vertical yield stress strength properties of some natural sedimentary
shear stress clays and to compare these with the correspond-
intrinsic angle of shearing resistance ing properties of the reconstituted material. Only
results from the highest quality undisturbed
samples have been used. The longer term objec-
INTRODUCTION tive of this lecture is to stimulate efforts to bring
Much of modern soil mechanics has developed to natural soils the same unity and coherence
from the results of careful, comprehensive studies which critical state soil mechanics in its broadest
of the properties of remoulded or reconstituted sense has brought to reconstituted soils. Signifi-
ON THE COMPRESSIBILITY AND SHEAR STRENGTH OF NATURAL CLAYS 331
cant progress has already been made in this deposits all plot above the A line on a plasticity
respect (e.g. Leroueil et al., 1979; Leroueil & chart.
Vaughan, 1990; Hight et al., 1987; Wood, 19.90). The curves in Fig. 1 show the progressive
The logical starting point is to examine the changes in void ratio from recently deposited
compressibility of some normally consolidated muds on the sea floor, to Quaternary clays at
natural clays followed by their shear strength depths of several tens of metres to hard clays and
properties. The corresponding properties of some mudstones of Pliocene and late Pleistocene age
overconsolidated natural clays are then con- extending to about 3000m. Each curve is termed
sidered. the ‘sedimentation compression curve’ for the
natural material-a term first used by Terzaghi
(1941). Skempton drew the following conclusions
SEDIMENTATION COMPRESSION OF from the results given in Fig. 1.
NATURAL CLAYS
In 1970 Skempton published an important (4 The relationship between e, and log eve’ (i.e.
paper on the consolidation of natural clays by the sedimentation compression curve) is
gravitational compaction. Curves relating in situ essentially linear for any particular clay.
void ratio e, to effective overburden pressure e,,’ (4 At a given value of (T,,’ the void ratio of a
were presented for twenty deposits representing a normally consolidated natural clay depends
wide range of lithologies as shown in Fig. 1. The on the nature and amount of clay minerals
void ratios were corrected to allow for changes in present, as indicated by the liquid limit. The
liquid and plastic limits with depth (Skempton higher the liquid limit the higher is the void
1944). In all cases the deposits are normally con- ratio.
solidated in the sense that the strata have never (4 A most striking observation is the converging
been under greater effective pressures than those pattern formed by the various compression
existing at the present time. Excluded from the curves.
study were quick clays, diatomaceous clays, clays (4 When plotted in terms of liquidity index,
containing more than 5% organic matter as well rather than void ratio, the results lie within a
as clays with a carbonate content of more than moderately narrow band. Clays with a high
25%. The average Atterberg limits for each of the sensitivity lie towards the upper part of the
e Avonmouth 71
Depth y T YiT ia? Yiiz V$o 3Eo m
Fig. 1. Sedimentation compression c~~rvcsfor normally consolidated argillaceous sediments (Skempton 1910)
332 BURLAND
band while those with low sensitivity lie limit and the void ratio corresponding to the
towards the lower part of the band. liquid limit (er) are given for each clay. Note that,
(e) For sea-bed deposits the depositional water although Kleinbelt Ton and Argile Plastique
content in the uppermost 250mm is equiva- have the same liquid limit, Argile Plastique has a
lent to a liquidity index of about 1.75 while lower specific gravity and hence a lower eL. It
that for tidal mudflats is about 1.0. appears that eL is a more fundamental parameter
than wL. At any given value of 0”’ the void ratio
How do these sedimentation compression
is related to er, increasing as eL increases. Note
curves relate to the corresponding laboratory also the converging pattern of the various com-
compression curves on reconstituted material?
pression curves as 6,’ increases. It is evident from
Do these sedimentation compression curves rep- Fig. 2 that the compression curves are all similar
resent the in situ compressibility associated with
in shape being slightly concave upwards. It is
the loading of the stratum over a timescale
useful to normalize these laboratory compression
associated with normal construction activities?
curves with respect to the void ratio.
More generally, how do the properties of these
naturally sedimented clays relate to the properties
of one-dimensionally consolidated reconstituted Intrinsic properties
clays? Answers to these questions will help to At this stage the concept of intrinsic properties
extend our generalized understanding of the of a given clay is introduced. The term ‘intrinsic’
properties of reconstituted soils to natural soils.
is used to describe the properties of clays which
have been reconstituted at a water content of
between wL and 1.5~~ (preferably 1.25~~) without
COMPRESSIBILITY OF RECONSTITUTED air drying or oven drying, and then
CLAYS consolidated-preferably under one-dimensional
A reconstituted clay is defined as one that has conditions. Ideally the chemistry of the water
been thoroughly mixed at a water content equal should be similar to that of the pore water in the
to or greater than the liquid limit (wr). Fig. 2 clay in its natural state. It is very important to
shows the one-dimensional compression curves distinguish clearly between the properties of a
for some reconstituted natural clays covering a natural soil and its intrinsic properties. The term
wide range of plasticities. Values of the liquid intrinsic has been chosen since it refers to the
basic, or inherent, properties of a given soil pre-
3.5- pared in a specified manner and which are inde-
w eL pendent of its natural state.? An asterisk is used
o KleinbeltTon 127.1 3.521 to denote an intrinsic property (e.g. C,* is the
o Argile Plastique 128.0 3.302
o London Clay 67.5 1.629 intrinsic compressibility, and 4* the intrinsic
A Wiener
Tegel 46.7 1.288 angle of shearing resistance of a soil).
II Magnus Clay 35.0 0.956
+ LowerCromerTill 25.0 0.663
The compression curves plotted in Fig. 2 rep-
resent the intrinsic compression curves for the
various clays since they were all reconstituted at
water contents such that wL < w < 1.5~~. Fig.
3(a) shows the intrinsic compression curve for a
given clay. The quantities e:,,c and e:,eo are the
intrinsic void ratios corresponding to 6,’ =
100 kPa and 1000 kPa respectively. The intrinsic
compression index C,* is defined as e:e,, - efooo.
Following Terzaghi (1925) the parameters e:,,c
and $t are called the constants of intrinsic com-
pressibility.
Void index
The curves in Fig. 2 may be normalized by
assigning fixed values to e:,c and eTooo. The nor-
t Leroueil et al. (1985) define four states of structure:

0,?4 uv’: kPa
intact, destructured, remoulded and resedimented. A
close examination of their definitions indicates that
Fig. 2. Onedimensional compression curves for various ‘reconstituted’ is a fifth important state of structure
reconstituted clays which is used here as a reference state.
e is the ordinate. When e = eToo, I, = 0 and when
e = eTooo, I, = - 1. The void index may be
thought of as a measure of the intrinsic com-
lntriwc compresson
pactness of a sediment. When I, is less than zero
the sediment is compact and when I, is greater
than zero the sediment is loose.
Clearly there is a close analogy between void
index (= (e - e:oo)/Cc*) and liquidity index (= (w
- w,)/(w,_ - w,)). It is of the utmost importance
to be clear about the difference between these two
w
indices. The void index is defined in terms of two
log u,” kPa
(a) directly measured mechanical properties (efoo
and C,*) derived from a one-dimensional com-
pression test. In contrast liquidity index is defined
in terms of two essentially empirical tests (the
liquid limit and plastic limit tests) both of which
subject the soil to extremely complex physical
processes.
Intrinsic compression line

Three of the intrinsic compression curves from
Fig. 2 covering a wide range of liquid limits and
e log (7”‘: kPa

of pressures have been replotted in Fig. 4 in terms
of void index I, versus log a”‘. It can be seen that
a reasonably unique line is achieved which is
(b)
termed the intrinsic compression line (ICL). The
Fig. 3. The USEof void index I, to normalise intrinsic
compression curve
co-ordinates of the ICL are given in Fig. 4 and
may be represented with sufficient accuracy by
the cubic
malizing parameter chosen is defined as the void
index I, such that I, = 2.45 - 1.28% + 0.015x3 (2)
where x = log a”’ in kPa.
I, =
e - eh =-e - 4io
* (1) The intrinsic compression line may either be
eToo- elOOO CC* measured directly for a clay or, if the values of
Thus the compression curve in Fig. 3(a) may be eToo and C,* are known for the clay, the ICL may
transformed to the normalized curve in Fig. 3(b) be constructed using Fig. 4 or equation (2). In the
where the void index I,, defined by equation (I), latter case, if it is required to plot the ICL in
_ Arglle plastique LL = 128
-_- London Clay LL = 67.5

----- Magnus Clay LL = 35
0,’ (kPa) I’,

10 1.18
40 0.46
100 0
400 -0.63
1000 -1.0
._
a,‘: kPa
Fig. 4. Normnlized intrinsic compression curves giving intrinsic compression line (ICL)
334 BURLAND
terms of e versus log uV’, then the values of e cor- between the curves for each clay. If anything the
responding to various values of log 0,’ may be curves for the longer duration lie slightly above
obtained from equation (1) those for the shorter durations. Northey (1956)
obtained similar results from oedometer tests on
e = I,C,* + efOo (3) three reconstituted New Zealand clays. Prelimi-
where, again, the values of I, may be obtained nary results from tests carried out at Imperial
from Fig. 4 or equation (2). College indicate that the ICL is also insensitive to
The available experimental evidence suggests load increment ratios in excess of unity. These
that the ICL is insensitive to the test conditions. and other data lead to the conclusion that, pro-
Fig. 5(a) shows the results of some oedometer vided the soil is reconstituted at a water content
tests on three clays in which each clay was recon- of between w,_ and 1.5~~) and provided the dura-
stituted at various water contents (Skempton, tion of each load increment is sufficiently long to
1944; Leonards & Ramiah, 1959). The number allow primary consolidation to occur, then the
against each curve gives the mixing water content ICL is well defined (i.e. it is ‘robust’) for pressures
expressed as a proportion of the liquid limit of equal to or greater than 100 kPa.
the clay. At pressures less than about 100 kPa the There is much evidence to show that ageing
compression curves for each soil tend to diverge, significantly influences the compressibility of
but for (T”’2 100 kPa the differences are less. Fig. reconstituted clays. Leonards & Ramiah (1959)
5(b) shows some results by Leonards & Ramiah studied the influence of ageing on the one dimen-
(1959) in which the influence of load increment sional compression of a reconstituted residual
duration was investigated for two clays which clay and their results are given in Fig. 6. The top
were reconstituted at water contents equal to the curve is for a standard test with a load increment
liquid limit. Clearly there is little difference ratio of one and a load duration of one day. The
2.5(-
- Gosport estuarme clay (wL = 76)
-- - - - Residual clay (We = 59)
- -. Glacial silty clay (w, = 28)
- Load wrement duration = 1 day

- -. Load increment duration = 1 week
------Load mcrement duration = 4 h
Fig. 5. Influence of (a) mixing moisture content on compression curves for

reconstituted clays (load increment duration 1 day); (b) load increment dura-
tion on compression curves for reconstituted clays (initial moisture content
IV,)(Leooards & Ramiah, 1959)
preconsolidation pressure’ to describe this critical

pressure. It is recommended that the term ‘yield
stress’, or more precisely ‘vertical yield stress’
should be used and be denoted by aVY’.The term
‘preconsolidation pressure’ should be reserved for
situations in which the magnitude of such a pres-
sure can be established by geological means.
Similarly the term ‘overconsolidation ratio’
should be reserved for describing a known stress
history. Where a yield stress has been observed
then the ratio between it and the effective over-
burden pressure (Q,~‘/u~,,‘) could be termed the
‘yield stress ratio’.
eeks rest at 40 kPa
CORRELATIONS BETWEEN THE CONSTANTS

OF INTRINSIC COMPRESSIBILITY AND THE
AlTERBERG LIMITS
rest at 40 kPa
The ICL is not, at present, routinely measured,
although it is easy enough to do so. Hence it is
necessary to make use of empirical correlations
between the Atterberg limits and the intrinsic
constants of compressibility e:,, and C,*. Skemp-
ton (1944) tabulated the results of numerous
Lucite oedometer
12 weeks rest at 40 kPa
oedometer tests on reconstituted natural clays,
with creep permitted many of them carried out at the Building
Research Station. These data have been supple-
mented by other published results and are given
in Table 1. In Fig. 7 the data are plotted on a
plasticity chart and it can be seen that all except
the results for Whangamarino Clay lie above the
A line.
Fig. 6. Influence of ageing on compression character- Figure 8 shows the correlation between e,_
istics of a recoustituted residual clay (Leonards & (void ratio at the liquid limit) and e:,, and C,*.
Ramiah, 1959) Regression analyses have been carried out and
the best fit regression lines are given by the fol-
lowing equations
second curve shows the effect of 12 weeks rest at
40 kPa followed by small load increments. It is eYoo = 0.109 + 0.679e, - O.O89e,’ + 0.016er3
evident that creep occurred during ageing but
that the ‘preconsolidation pressure’ lies well to (4)
the right of the standard virgin compression line. and
The third curve shows the effect of 12 weeks
C c* = 0.256e, - 0.04. (5)
ageing with creep prevented. Again the preconsol-
idation pressure lies well to the right of the virgin The coeflicients of correlation for equations (4)
compression line. The bottom curve is a repeat of and (5) are 0.991 and 0.985 respectively. These
the second but using a lucite oedometer for which equations should of course only be used for
the side friction was known to be very small values of eL within the range 0.6 to 4.5 (i.e. wL =
(Leonards & Girault, 1961). 25 to 160). Moreover these correlations only hold
These results demonstrate that the micro-fabric for soils with Atterberg limits lying above the A
of a clay can develop increased resistance to com- line. It has been found that when the Atterberg
pression during ageing and that this resistance limits lie below the A line the values of e:,, and
does not depend on volume reduction due to C,* do not fit the correlations well-an example
creep. It can be seen from Fig. 6 that when an being Whangamarino clay, which is shown as a
aged clay is loaded the structural resistance full point in Figs 7 and 8.
breaks down at a critical pressure and the sub- The broken lines in Fig. 8 are derived from the
sequent compression curve is initially significantly work of Nagaraj 8~ Srinivasa Murthy (1986) who
steeper than the standard virgin line. Leonards established a relationship between the ratio e/e,_
and others have used the term ‘quasi- and 0”’ based on considerations of physical chem-
336 BURLAND
Table 1. Intrinsic coustmts of compressibility for reconstituted uaturnl clays
Soil G, eL CC* Reference
Lower Cromer Till 25 13 2.65 0.663 0.503 0.154 Gens (1982)

Boulder clay 28 14 2.69 0.753 0.52 0.12 Skempton (1944)
Silty clay 28 20 2.12 0.762 0603 0.136 Ramiah (1959)
Magnus Clay 35 17.2 2.13 0.956 0.16 0.27 Jardine (1985)
Grangemouth 35 21 2.78 0.913 0.659 0.229 This study
Ton V 36 18 2.71 0.916 0.14 0.25 Skempton (1944)
Weald clay 39 19 2.73 1.065 0.17 0.24 Skempton (1944)
Boston blue clay 39 23 2.78 1.084 0.80 0.21 Skempton (1944)
Red soil 45.3 22 2.661 1.208 0.785 0.27 Nagaraj et al. (1986)
River Severn alluvium 46 25 2.59 1.191 0.80 0.21 Skempton (1944)
Wiener Tegel 46.7 22 2.16 1.288 0.859 0.297 Hvorslev (1937)
Oxford clay 53 21 2.51 1.362 0.96 0.30 Skempton (1944)
Ton IV 58 26 2.85 1.653 0.97 0.32 Skempton (1944)
Residual clay 58 27 2-14 1.589 1.024 0.337 Ramiah (1959)
London Clay 62.3 24.3 2.73 1.707 1.200 0446 Jardine (1985)
Belfast estuarine clay 61 30 2.66 1.782 1.00 0.32 Skempton (1944)
London Clay 67.5 26.5 2.71 1.829 1.227 0.494 Som (1968)
Ganges delta clay 69 28 2.11 1.911 1.22 0.42 Skempton (1944)
Gosport clay 16 29 2.61 2.029 1.20 0.48 Skempton (1944)
London Clay 77 28 2.71 2.087 1.28 0.49 Skempton (1944)
Brown London Clay 88 32 2.65 2.332 1.32 0.56 Skempton (1944)
Black cotton clay 91.3 32 2.13 2.656 1.744 0.69 Nagaraj et al. (1986)
Kleinbelt Ton 127 36 2.17 3.518 2.18 0.91 Hvorslev (1937)
Argile plastique 128 31 2.58 3.302 1.82 0.81 Skempton (1944)
Whangamarino clay 136 61 2.78 3.74 244 0.791 Newland & Allely (1956)
SAIL 159.3 46 2.826 4443 2.769 l-05 Nagaraj et al. (1986)
istry. It can be seen that the two approaches give correlated with plasticity index, or its void ratio
similar correlations over a wide range of eL values equivalent, instead of eL . A statistical analysis has
but that at low and high values there are signifi- shown that equally good correlations are
cant differences, particularly for e:,, . If, for a achieved at high values of plasticity index but at
given clay, the intrinsic constants of compress- low values the correlations are significantly
ibility eToo and C,* have been measured then it worse. This is because small errors in wL and wP
would be appropriate to allow for small changes become significant when one is subtracted from
in eL between samples of that soil by correcting the other.
e:,,,, and C,* in direct proportion to the changes For all the data listed in Table 1 the soils were
in eL (or wL). reconstituted at water contents of between wL and
The question might well be asked as to why the 1.5~~. Recently Nakase et al. (1988) published an
intrinsic constants of compressibility were not independent data set for reconstituted marine
clays from a number of locations in Japan. The
key difference between the two data sets is that
Nakase et al. reconstituted the soils at very high
water contents to form liquid slurries. Fig. 9
shows a comparison between the results of
Nakase et al. and equations (4) and (5). It can be
seen that there is excellent agreement for C,*.
However, the experimental values of eToo lie a
little above the regression line. This is consistent
with the very high mixing water contents and
serves to emphasize the need to standardize these
when determining the intrinsic constants of com-
pressibility. It is encouraging that the two entirely
independent sets of data are in reasonable agree-
ment. Note that the values of C,* and e:,,, for the
01
0
1
20 40 60 60 100 120 140 160 soil lying just below the A line in Fig. 9 tend to be
Liquid limit: % displaced from the other results in the same
Fig. 7. Plasticity chart for reconstituted clays in Table 1 manner as in Fig. 8.
3r
Void ratlo at the liquid limit e,

b)
Fig. & Relationships betweeo Q ad constants of iotrinsic compressibility P:@,, and C,*
(broken line given by Nngarnj & Srinivnsa Murthy, 1986)
In concluding this section it is important to an element of normally consolidated clay with a

appreciate that wherever possible the ICL should void ratio e, under an effective overburden pres-
be measured directly. The correlations between eL sure o,,‘. The void index I,, of the clay element is
and e:,, and C,* provide an indirect method of given by equation (1)
obtaining the ICL which is less reliable than its
direct experimental determination. I,,
e. - Go0
= ~
CC*
COMPARISON BETWEEN THE The values of eToo and C,* are preferably mea-
SEDIMENTATION COMPRESSION OF sured by means of an oedometer test on the
NATURAL CLAYS AND THE INTRINSIC reconstituted soil, but for the present purposes
COMPRESSION OF RECONSTITUTED CLAYS they are obtained from equations (4) and (5).
Using the void index I, as a normalizing Thus successive values of e, and CT,’down a soil
parameter, it is possible to compare the sedimen- profile may be used to plot a graph of I,, against
tation compression curves obtained by Skempton log 0,’ to give the sedimentation compression
(see Fig. 1) with the corresponding ICL. Consider curve which can then be compared directly with
BURLAND
been derived for most of the profiles referred to in

Fig. 1. The geology of each site has been
Reconstituted marine clays
Artificially mixed clays
described by Skempton (1970) and will not be
Below A line repeated here.
00 Figure 10 shows the sedimentation compres-
/
sion curves for three of the Pliocene deposits
plotted on axes of I,, versus log crVO’.The results
show marked scatter which is due to in part to
errors in the determinations of water content and
liquid limit but is also believed to be due to varia-
tions in depositional conditions as the profiles
were being formed. The extreme variations have
been removed by taking the average of successive
(4
pairs of points, thereby preserving trends but
eliminating extreme fluctuations. All three curves
lie well above the ICL. The results from Baku are
of particular interest because of the wide range of
overburden pressures. Note the ‘saw-tooth’ shape
of the sedimentation curve which is also a feature
of the other two curves.
There is no reason to anticipate a smooth sedi-
mentation compression curve. Rates and modes
of deposition are likely to vary considerably
during the formation of a sedimentary soil profile
and in these circumstances a wavey curve must be
expected (Edge & Sills, 1989). Thus each element
will retain the imprint of the conditions under
which it was deposited.
Figure 11 shows the sedimentation compres-
Fig. 9. Comparison of correlations from Fig. 8 with inde- sion curves for three British post-glacial clays
pendent data set given by Nakase et al. (1988) from widely differing locations and having a wide
range of liquid limits. The curves all lie above the
ICL. The results from Shellhaven are for the
the ICL which is uniquely defined in Fig. 4 or by lowest layer of clay at the site. The results from
equation (2). Professor Skempton has kindly higher up the profile will be described later.
made his files available to the author and the Figure 12 shows the sedimentation compres-
detailed sedimentation compression curves have sion curves for two Scandinavian post-glacial
1
o San Joaquln Valley (wL = 64)
* Mlllazzo (wL = 62)
l Baku (w, = 40)
o-
2
2 -
E
% _
>
-l-
Fig. 10. Sedimentation compression curves for Pliocene and early Pleistocene clays and
modstones
qShellhaven layer C (wL = 82)

eAvonmouih (wL = 71)
o Grangemouth (wL = 41)
Fig. 11. Sedimentation compression curves for some British post-glacial clays
clays and once again they lie well above the ICL. The sedimentation compression line
The profile at &+ingen in Sweden is unusually Having considered some of the individual sedi-
uniform and gives relatively smooth compression mentation compression curves the data from
curves-note the high liquid limit. The profile for most of the sites considered by Skempton (1970)
Drammen was referred to by Bjerrum (1967). It are assembled in Fig. 13 including the results for
consists of an upper plastic stratum (shown as the shallow marine deposits. It can be seen that
circles) underlain by a lean stratum (shown as the various sedimentation curves all lie in a well
diamonds). In spite of the differences in liquid defined continuous band when plotted on a graph
limit between these two strata it can be seen that of I,, versus log oVO’.A regression line has been
the sedimentation compression curve is reason- fitted to the data as shown and is called the sedi-
ably continuous. This implies that the upper mentation compression line (SCL), the co-
plastic layer has not undergone substantially ordinates of which are tabulated in Fig. 13. Most
more delayed consolidation than the underlying of the data lie within the range I,, = kO.3 of the
lean layer as was suggested by Bjerrum. SCL.
q Alvangen (wL = 95)

o Drammen (wL = 54)
o Drammen (wL -- 38)
Fig. 12. Sedimentation compression curves for two Scandinavian post-glacial clays
340 BURLAND
5- LL ‘ir_ L;’
q Oslofjord 98 - Alwlgen 95 l S.Joaquin 64

e A-33 80 9 Shellhaven 82 l M~lazzo 62
m A-31 63 0 Avonmouth 71 + Baku 40
8 B-87 58 o Drammen 54
4- 9 C-18 46 . Grangemouth 41
0 Drammen 38
= Detroit 28
e
Co-ordinates of
3- the
SCL
IT’“& 1”
kPa
0.4 3.84
-$T Sedimentation compression’ 1 3.24
02- 4 2.42
f t
10 1.92
40 1.22
yl_
100 0.77
.F - 400 0.13
2
E l- -
n -
8 - Intrinsic compression line
O-
-1 -
-21 ’ ’ ““‘1 ’ ’ ’ ““‘1 ’ 1 ’ ““‘1 1 1 I11111’ 1 1 1 ~ult.l

lo--’ 1 10 102 103 104’
u’“~: kPa
Fig. 13. Relationship between IlO and log uvO’ for many of the normally consoli-
dated clays designated in Fig. 1: best-fit regression he through the data is
termed sedimentation compression line (SCL)
Over the range of uV’ = 10 kPa to 1000 kPa the Shellhaven ~7 0 m (wL = 115)
Shellhaven 7-5 m (wL = 85)
ICL and the SCL can be seen from Fig. 13 to be Shellhaven 10.4 m (w, = 72)
approximately parallel. Over this region, for a Gosport (w, = 80)
Sault Ste Mane (wL = 55)
given value of I,,, the effective overburden pres- (PreSence of haematlte grves red colour)
sure carried by the natural clay is approximately
five times that carried by the equivalent reconsti-
tuted clay. This figure is a measure of the
enhanced resistance of a naturally deposited clay
over a reconstituted one and results from differ-
ences in the fabric and bonding (i.e. the structure)
of the soil skeleton. The influence of the natural
structure was first recognized by Terzaghi (1941)
and confirmed by Skempton (1944). At pressures
in excess of 1000 kPa the ICL and SCL tend to
converge.
Not all normally consolidated natural clays lie
close to the SCL. Fig. 14 shows the sedimentation
compression curves for three such clays. The open
circles are for a freshwater glacial lake clay from
Sault Ste Marie, near Chicago (Wu, 1958). The
reason for these data laying well above the SCL is
a’,,: kPa
not difficult to find. The clay is reddish in colour
due to the presence of haematite which has Fig. 14. Sedimentation compression carves for three
undoubtedly given rise to cementation between clays which are remote from SCL
the particles. Wu carried out a study of the fabric satisfactory. The compression curves for the
by means of a polarizing microscope and found undisturbed samples are very different from the
that it was essentially random. In contrast, the Sault Ste Marie and Shellhaven clays as they do
sedimentation compression curve for a nearby not exhibit a high post-yield compressibility and
glacial lake clay at Detroit, shown as crosses in the curves more or less coincide with the ICL.
Fig. 13, lies on the SCL. The fabric of this clay, In summary it appears that for normally con-
which contained no haematite, was shown to solidated clays whose natural states lie close to or
exhibit some horizontal orientation. above the SCL, the post-yield oedometer com-
Also shown in Fig. 14 are the sedimentation pression curve is much steeper than the SCL. It
compression curves for two British post-glacial crosses the SCL and converges slowly on the
clays-the upper clay layer at Shellhaven ICL. In contrast, for normally consolidated clays
(Skempton & Henkel, 1953) and Gosport whose natural states lie on or close to the ICL the
(Skempton, 1970). Both these clays lie well below oedometer compression curves are essentially
the SCL. The reason for this is not immediately parallel to this line.
obvious but evidence will be presented later
which supports the hypothesis that it is due to the
deposition conditions. The deeper clays at Shell- Resultsfrom the Mississippi delta
haven lie on the SCL (see Figs 11 and 13) and the Some work published by McClelland (1967)
triangles in Fig. 14 are for samples from depths of on the clays from the Mississippi delta provide
7.5 m and 10.4 m-both lie a little above the SCL. some important clues about the factors influ-
Oedometer tests were carried out on the three encing the in situ state of sedimentary clays rela-
clays referred to in Fig. 14 and the results are of tive to the SCL and the ICL. The continental
considerable interest. shelf in the Gulf of Mexico off the coast of south
eastern Louisiana is blanketed by clay sediments
of Late Quaternary age. These clays have the Mis-
Results of some oedometer tests sissippi river as a common source and consist
Figure 15(a) shows the results of four oedome- essentially of a common suite of minerals.
ter tests on undisturbed samples of Sault Ste However the depositional environments differ sig-
Marie Clay from various depths. The void index nificantly as a result of sea level changes and
I, has been used as a normalizing parameter (in changes in the course of the river.
conjunction with equations (4) and (5)) so that the Figure 16 shows the sedimentation compres-
oedometer compression curves can be compared sion curves for two locations remote from any of
with the intrinsic compression line and the sedi- the deltas associated with the present standing-
mentation compression line from Fig. 13. It can sea period. The clays are continental shelf depos-
be seen that the post-yield compression curves for its more than 15000 years old. It is evident that
the three deepest samples are significantly steeper the data lie close to the SCL. Oedometer tests on
than the SCL. The curves cross the SCL from undisturbed samples from these two boreholes
above and then flatten, converging slowly on the give post-yield compression curves which are
ICL. Note that the shallowest sample from 3.51 m steeper than the SCL and which tend to converge
depth is lightly overconsolidated due to desicca- with the ICL in accordance with the behaviour
tion. depicted in Fig. 15(b).
The oedometer results for Sault Ste Marie A borehole was also sunk through the present
Clay, which lies well above the SCL, may be con- delta front of the river. It revealed about 85m of
trasted with those for Shellhaven lying close to recent delta deposits underlain by continental
the SCL, and for Gosport lying below the SCL. shelf deposits. The top 60m have been deposited
Fig. 15(b) shows the results for the latter two so rapidly over the last 400 years that they are
clays. The triangular points are for Shellhaven. largely unconsolidated. Since the in situ effective
The full points are for a reconstituted sample of stresses within this top layer are not known the
the clay and the reconstituted compression line is sedimentation compression curve cannot be con-
seen to lie very close to the one derived from structed. However, oedometer tests on undis-
equations (4) and (5) and labelled ICL. The agree- turbed samples from the top layer and deeper
ment is encouraging. The post-yield compression layers give interesting results as shown in Fig. 17.
curves for the two undisturbed samples, the initial The open circles are for samples from the overly-
states of which are given in Fig. 14, are steeper ing rapidly deposited underconsolidated clays. It
than the SCL crossing it from above and again can be seen that the compression curves lie on the
converging slowly with the ICL. ICL. In contrast the compression curve for the
The circles are for Gosport clay. The full points sample from 86.6 m depth in the continental shelf
are for reconstituted samples and lie slightly deposit (closed circles) drops from the SCL down
below the ICL but the agreement is nevertheless towards the ICL. The sample from 119.6 m depth
342 BURLAND
2 I-
\
-----4_
(J “0 1 -
---
----
-----
Sample
Sample
Sample
Sample
l-l-4;
l-3-4:
l-5-6;
l-7-5,
3 51
6.55
9 75
12 8
m. WL = 44.9
m.
m. wL
= 48.1
We : 47.0
m: w, = 66 2
(a)
2-
l-
-?
G
E -
0
0 O-
>
b 0.17 m; wL = 75
-l- 0 5.2 m, We = 61
l Reconstituted at w = 96, We = 76
* Reconstituted at w = 76; We = 76
I I I I I I I I I 1 I I II,1111 I I Lll,,,,
1 10 a “: kPa 102 103
(b)
Fig. 15. Oedometer compression carves for (a) Sault Ste Marie Clay, site 1, and (b)
Sbellhaveo and Gosport clays
has almost certainly suffered some disturbance. water leads to an open random fabric with high
Nevertheless the compression curve lies well to values of void index laying on or above the SCL.
the right of the ICL. On the other hand rapid deposition from a dense
These results confirm that the deposition con- suspension, possibly with significant currents, will
ditions profoundly affect the fabric of the sedi- give rise to a more oriented fabric which is conse-
ment which is then not easily changed by quently more compact with a lower void index.
subsequent increases in effective overburden pres- For a soil whose state lies on or above the SCL
sure. The two most significant depositional the rate of application of load in an oedometer is
factors are likely to be the rate of deposition and sufficient to disrupt the interparticle bonding and
the stillness of the water. Slow deposition in still fabric such that the compression curve is signifi-
Resultsfrom Bothkennar tesf bed site

Recently the UK Science and Engineering
0 Location 10 Research Council (SERC) selected a soft clay test
0 Location 11
bed site at Bothkennar in the upper Forth
Estuary, Scotland. Details of the ground condi-
tions are given by Hawkins et al. (1989). It is of
considerable interest to establish whether the
ground conditions at this site fit the general
pattern portrayed in Fig. 13.
Figure 18 is a summary of the basic properties
for borehole Dl at the Bothkennar site. The clay
is of medium to high plasticity, the yield stress
ratio (otherwise referred to as the OCR) is about
1.7 and the undrained strength from vane tests
shows a linear increase with depth with a sensi-
tivity of about 4 to 6. These results indicate that
the clay is normally consolidated although
Hawkins et al. point out that there is some evi-
dence to suggest that the top 1 m or so may
Fig. 16. Mississippi Delta: sedimentation compression
have been removed by erosion. The sedimenta-
curves for late Quaternary continental shelf deposits
tion compression curve for borehole Dl is shown
in Fig. 19. The curve is somewhat jagged due to
significant variations in water content but it can
cantly steeper than the SCL and it falls towards be seen to lie very close to the SCL. The broken
the ICL. However, if the state of the soil is line is for the top 2m which is overconsolidated
already on the ICL due to its deposition condi- due to desiccation.
tions, the fabric will already be oriented and com- High quality samples were obtained by means
pression in an oedometer will n?t change things of a Lava1 sampler (La Rochelle et al., 1981) and
significantly. standard incremental oedometer tests were
Leroueil et al. (1979) have termed the post-yield carried out on them. Fig. 20 shows the results of
disruption of the clay structure as ‘destructur- two oedometer tests on a sample from a depth of
ation’. The results given, for instance, in Figs 15 6.5m plotted as void ratio against log 0”‘. The
and 17 imply that this process is a gradual one full circles are for a sample which was reconstitut-
and that the precise definition of a ‘destructured ed at the liquid limit to give the experimentally
state is not clear. There are clear advantages in determined ICL. This compares very well with
using the ‘intrinsic’ state as a reference state. the broken line which was obtained from equa-
0 15.6 m depth Recent deltw

0 30.6 m depth Recent deltalc
86.6 m depth Late QuaternarY shelf
G I 19.5 m depth Late QuaternarY shelf
I I III , I I IIIII,
103 104
Fig. 17. Mississippi Delta: results of oedometer teats on underconsolidated deltaic

deposits and uaderlying Quaternary shelf deposits
344 BURLAND
Moisture content: % a’vO: kPa S,: kPa Sensitivity
r
0 50 100 150 200
I. I r I
I
t
, = I l Remoulded vane strength

wp wo w o Peak vane strength
Fig. 18. Botkkeonar: profile for korehole Dl (Hawkius et al., 1989)
tions (4) and (5) knowing wL and hence er. The tions of in situ compressibility which may be
open circles are for an undisturbed sample. It can compared with the pattern of oedometer com-
be seen that the compression curve drops steeply pression curves presented in the previous sections.
from the SCL eventually converging with the The project is the phase II development of Sura-
ICL. Thus the Bothkennar test bed site appears baya Port and Rendel Palmer and Tritton were
to conform to a typical normally consolidated the consulting engineers for the client-the Direc-
sensitive clay profile. torate General Sea Communications, Govern-
ment of the Republic of Indonesia. The work
involved the construction of a container stacking
CASE RECORD FROM SURABAYA,
yard on land reclaimed from tidal mud flats, just
INDONESIA
to the west of the existing port of Surabaya. The
Field measurements on a land reclamation
site consists of about 5 m of silty sand overlying a
project in Indonesia provide valuable observa-
deep soft clay layer which is underlain by stiff
‘r
clay and sand. The soft clay is derived from local
volcanic clays and is highly plastic with an
c average liquid limit of about 100. A typical profile
through the soft clay as given by two boreholes is
shown in Fig. 21(a).
Accelerated consolidation by wick drains was
adopted for the reclaimed area. A number of sec-
tions were instrumented by installing settlement
plates at various depths and piezometers between
the drains. Inclinometers were used near the
slopes. The fill consisted of hydraulically placed
sand. The soft clay settled considerably more
than was predicted on the basis of normally con-
solidated behaviour using C, values from oedo-
meter tests on samples obtained by Shelby tubes.
These values of C, were consistent with the estab-
lished correlations with I, and hence approximate
to c,*. Settlements were measured at various
depths and the vertical effective stresses were esti-
mated from the unit weights from the typical
borehole profile and the measured pore pressures.
I I I I llllll I
Figure 21(b) shows a typical instrumented
10 102
(,vO: kPa section. The settlements and vertical compres-
Fig. 19. Botkkennar: sedimentation compression curve sions one year after completion of loading are
for borehole Dl shown in Fig. 21(c). Also shown in Fig. 21(c) are
2.0 -
1.6 -
1.6 -
.;1.4-
L
P
gil.Z-
0 Undisturbed sample
1-o - @ In situ state
. Reconstituted at WL
-- PredIcted ICL
0.6 -
0.6
t
Fig. 20. Botbkeaoar: oedometer tests on undisturbed nod reconstituted soil from 65m
depth (wL = 85-4, w, = 419)
the predicted settlements and compressions-the year after completion of loading. Fig. 22 shows a
differences from the measured values are large. plot of in situ values of I, versus log uV’. The
The measured compressions at various depths closed points and corresponding open points rep-
and locations can be used in conjunction with the resent the initial and subsequent values of I,
initial void ratios to calculate the void ratios one respectively. The full lines are the in situ compres-
/
Sand fill
1 I ,
Water content: % Settlement: m Verhcal compression: %
50 100 0 3 5 10 15
I 3 1 I I
Plezometer
1
Settlement plate - )
Silty sand
-----
T -7
2
-t
I
Softclay
I
I
i
I I
-t
---
!
FiF Ly-
-----
Stiff clay
----- - - - Predicted
Dense sand - Observed
- Datum
(a) (b) cc)
Fig. 21. Sorabaya, Iodowsia: (a) protile of soft clay from hvo adjaceot boreboles; (b) typical iastromented section; (c)
Observed settlements nod compressions 1 year after completion of loading (band drain at 153 m centres)
346 BURLAND
I
100 500
0’“: kPa
Fig. 22. Sarabaya: in situ relationship between I, aad log a,‘-closed points rep
resent values of I,, giving the sedimentation compression curve; open points give
the correapoading values of I, 1 year after completion of loading
sion curves for the section shown in Fig. 21(b). process resulted in a reduction of liquid limit
There was a threshold stress change of about from an average of 48.8 to 28.1. In Fig. 23(b) the
20 kPa up to which settlement was negligible. results are plotted in terms of I, so that they can
The results plotted in Fig. 22 show that the be compared with the ICL and the SCL. The
sedimentation compression curve, as given by the unleached samples lie just below the SCL. The
full points, lies well above the SCL and is steeper reductions in wi, due to leaching cause the values
than it is. The in situ compression curves of I, for the leached samples to increase substan-
resulting from the placement of the fill are very tially so that the results lie well above the
much steeper than the ICL and it is clear that SCL-a characteristic of quick clays.
they will all drop below the XL at higher values
of 0”‘. These observations are consistent with the
oedometer compression curves given in the pre-
vious section.
.
LABORATORY SEDIMENTATION STUDIES gl.O-
. . o
m
An interesting and important question is 0 0. 0
.
whether or not it is possible to reproduce the 8 0.9 - . ‘.*
natural sedimentation compression line in the - o Sedimentation into salt water
(31.7 g/l NaCI; WL = 48.8)
laboratory. The results of the classic studies of 0.8 - 0 Leached after sedimentation .
Bjerrum & Rosenqvist (1956) and Leonards & (5.0 g/l NaCI; WL = 28.1)
Altschaelll (1964) can be used to examine this o-7

I / I I I ,,/I I I I I / I III
question. Bjerrum & Rosenqvist carried out a W
.
series of experiments in which a late glacial 3
marine clay was artificially sedimented into a salt .
. .
water solution over a two month period and the
l
sediment was then left for 6 weeks. Small l .
. -a l
increments of pressure were then applied, after

which the samples were left for a further three
months. At this stage a number of the samples
were subjected to leaching over an 18 month
period in which the salt concentration was
reduced from 32g/l to 5g/l. The whole process
took about 24 years.
Figure 23(a) shows the equilibrium void ratios
for the unleached samples (open points) and the uv kPa
leached samples (closed points). Clearly the (b)
process of leaching, involving the application of Fig. 23. Results of laboratory sedimented marine clay in
an hydraulic gradient across the sample, has terms of (a) e against log a,’ and (b) Z, against log u,’
resulted in reductions in void ratio. The leaching (Bjerram & Rosenqvist, 195%)
In the experiments carried out by Leonards & clays in the laboratory but the preparation of the
Altschaelll(l964) a flocculated slurry of a residual samples involves considerable lengths of time.
clay was slowly loaded first by means of a
hydraulic gradient and then by applied load
through a plunger. The rate of change of load was SHEAR STRENGTH OF NORMALLY
controlled by syphoning oil from a counter- CONSOLIDATED CLAYS
balancing tank. The resulting compression curve The discussion on the compressibility of nor-
is shown in Fig. 24. When a,’ had reached mally consolidated natural clays was preceded by
48.7 kPa the pressure was held constant for 90 summarizing some basic properties of reconstitut-
days resulting in some creep. Unloading then ed clays. These properties are termed the intrinsic
took place and a further rest period of 90 days properties. Similarly, before examining some
was allowed. The sample was then loaded in daily aspects of the shear strength of normally consoli-
increments and the results are shown by the open dated natural clays it is important to establish a
circles in Fig. 24. clear picture of the intrinsic shearing behaviour of
It can be seen that the compression curve for one-dimensionally consolidated reconstituted
slow loading falls steeply towards the SCL and clays. For simplicity only the behaviour in triaxial
appears to be converging with it. The compres- compression is considered.
sion curve for incremental loading shows a sharp
yield point at cr,’ = 64.5 kPa (giving a yield stress
ratio of 1.32) after which the curve drops steeply Intrinsic shear strength of normally consolidated
through the SCL and converges on the ICL. The clays
full circles are for a test on a specimen which had Figure 25(a) shows the one-dimensional intrin-
been sampled after unloading. The process of sic compression line for a reconstituted clay
sampling resulted in a slightly reduced yield stress plotted on a graph of e versus Q~‘. Point 0 lies on
(= 60,7 kPa). the ICL and Fig. 25(b) shows the corresponding
The results given in Figs 23 and 24 bear a Mohr’s circle of effective stress. The maximum
striking resemblance to the measured compress- shear stress is given by point A’ which lies on the
ibility of natural clays. Locat & Lefebvre (1986) K, effective stress path. Point A’ projects as point
describe similar tests on Grande-Baleine Clay and A in Fig. 25(a) which lies on a compression line
refer to a number of other studies on artificially for the average of the axial and radial stresses (es’
sedimented clays. Contrary to the views expressed + a,‘)/2 shown as chain dotted.
by Casagrande (1932) it can be concluded that it A standard drained triaxial test entails increas-
is possible to reproduce the behaviour of natural ing o*’ with c,’ constant. Fig. 25(c) shows the
initial and failure Mohr’s circles of stress for a
sample initially consolidated to an axial effective
stress 6,,‘. The Mohr’s circle at failure is tangen-
tial to the intrinsic failure line and AD’ rep-
resents the effective stress path for the test. The
stress-strain and volumetric strain behaviour is
Sedlmented and then loaded
contmuously at - 1 kPa/day shown in the adjacent diagram. It can be seen
that the sample contracts and that at failure the
rate of contraction is approximately zero. Thus
failure corresponds to a critical state condition
and in recognition of this the intrinsic angle of
Incremental loading
shearing resistance is designated I$=“* where the
asterisk denotes an intrinsic property. The stress
path AD’ plots as the path AD in Fig. 25(a)
where D lies on the projection of the critical state
line shown as a broken line.
The undrained behaviour of the clay is shown
in Fig. 25(d). Most one-dimensionally consoli-
dated reconstituted natural clays show brittle
stress-strain behaviour with the peak undrained
compressive strength being reached at very small
strains, as shown by the point B” in Fig. 25(d).
Thereafter the contractant behaviour of the soil
skeleton results in a falling stress-strain curve
Fig. 24. Laboratory sedimented residual clay (Leonnrds coupled with large increases in pore water pres-
& Altschaeffl, 1964) sure. The effective stress path for an undrained
348 BURLAND
m u a and (u’, + n’,)/Z

(a)
(d)
Fig. 25. Ideal behaviour of onedimeasionnlly consolidated reconstituted clay in
triaxial compression: (a) void ratio changes; (b) K, stresses; (c) drained test; (d)
undrained test
triaxial compression test is of the form given by along the path B’C’ in Fig. 25(d) the stress ratio is
ABC’ in Fig. 25(d) were B’ corresponds to peak actually increasing and the soil skeleton is there-
strength and C’ to the critical state strength. The fore strain hardening. It can therefore be antici-
corresponding path in Fig. 25(a) is AC where C pated that the sample will deform in a
lies on the critical state line. The broken line CD homogeneous manner as the stresses move from
is the projection of the intrinsic critical state line B’ to C’. The significance of this will become
since it relates to a reconstituted soil. Note that apparent later.
the critical state line lies well to the left of the It was shown previously that the effect of
ICL. Although the strength of the soil decreases ageing during one-dimensional compression is to
200 r
I I
200 300 400
(u a + a J/2: kPa
(b)
Fig. 26. Influence of ageing on undrained et&dive stress paths for triaxial compres-
sion tests oo reconstiMed soils for (a) Magnus Clay (wL = 35) (Jardine, 1985), and
(b) Gullfaks clayey sand (Georgiannou, 1988)
increase the vertical yield stress uVY’. Similar Resultsfrom the Trollfield in the North Sea
behaviour takes place in undrained compression. The Troll field is located in the Norwegian
Fig. 26(a) shows the effect of ageing on reconsti- sector of the northern North Sea. Extensive site
tuted Magnus clay from the North Sea giving rise investigations have been carried out for the
to a significant increase in peak undrained design of offshore gravity oil production plat-
strength. There is also an increase in brittleness. forms. The data presented here are for block 31/2
Similar results are shown in Fig. 26(b) for recon- and high quality samples were obtained using
stituted clayey sand from the Gullfaks field in the thin walled tube samplers pushed into the ground
North Sea. In this case the volumetric strains at a steady rate. A comprehensive programme of
during ageing were negligible so that the gain in laboratory testing was carried out jointly by
strength must have been due primarily to inter- Fugro-McClelland and the Norwegian Geotech-
particle bonding. nical Institute.
In the next two sections the results of Figure 27 shows a typical soil profile. It con-
undrained triaxial tests on high quality undis- sists of 23 m of a medium plasticity clay overlying
turbed samples of some normally consolidated low plasticity clay to a depth of about 65m.
clays are compared with the the framework given Results of oedometer tests and anisotropically
in Fig. 25. consolidated undrained (CAU) triaxial compres-
350 BURLAND
water content % uvo: kPa S,: kPa

20 40 60 60 0 200 400 600 600 0 50 100 150 200
11 I I I I t 1 1 I 0""""""""""
-0
w" 0
- 0
_ B
0
0
Ooo
@a
80
0
0
Fig. 27. Troll field, block 31/2, North Sea: soil profile
sion tests show that the soils are normally con- The sedimentation compression curves for the
solidated with a yield stress ratio of about 1.3. Troll profile (Fig. 28) are particularly interesting.
The upper clay is a glacial marine deposit laid The upper clay, shown by the open circles, lies a
down between 10000 and 13 000 years BP (Sejrup little above the SCL while the lower clay (open
et al., 1989). There is some uncertainty about the triangles) lies around the ICL. These results
mode of deposition of the lower clay but it is suggest that the deposition conditions for the two
thought to be a glacial marine deposit or a lodge- layers were entirely different.
ment till or a combination of both. The upper The results of oedometer tests on samples from
part of it was probably laid down during the the two layers confirm the differences in the depo-
retreat of the Scandinavian ice sheet about 13 000 sitional environments. In Fig. 29 the open circles
years BP. are for two oedometer tests on undisturbed
I I I I11111 I 1 I I11111
10 102 103
ova: kPa
Fig. 28. Troll field: sedimentation compression curves
o- 28.4 m (We = 33.2)

2 -
8
.c
0
41.2 m (wL = 35.9)
P -
-l-
-2’ I I I111111 I I I111111 I I I I,,,,,

10 102 103 104
0’“: kPa
Fig. 29. Troll field: oedometer tests on upper and lower clays
samples of the upper clay. The compression line the effective stresses would have to reduce
curves follow the well established pattern of enormously such that a constant void ratio path
falling steeply through the SCL and then flat- would travel to the left of the ICL. Thus the
tening off and converging slowly with the ICL. In behaviour would be predicted to be very brittle
contrast the compression curves for the lower and sensitive. In contrast, since sample 27G lies
clay (open triangles) remain close to the ICL. on the ICL its behaviour would be expected to be
Although a variety of types of shear test were similar to an aged reconstituted sample with low
carried out during the investigation the results sensitivity and little or no brittleness.
from the following two types will be considered The SHANSEP procedure has caused sample
here. 1OC to move from well above the SCL to some
distance below it. Thus it would be expected to be
(a) CAU triaxial compression and extension tests
much less brittle than sample 22C. On the other
in which the samples were consolidated to
hand, since sample 27E has remained on the ICL
their estimated in situ effective stress state
during consolidation, its behaviour would be
prior to undrained shearing.
expected to be similar to sample 27G.
(b) SHANSEP tests in which the samples were
The results of the undrained triaxial tests are
compressed anisotropically to well beyond
given in Figs 30 and 31 for the upper and lower
their in situ states of stress and then unloaded
clays respectively. These figures should be studied
a little to model the apparent preconsolida-
in conjunction with Fig. 28. It can be seen from
tion. This procedure was introduced as a
Fig. 30(a) that sample 22C shows brittle behav-
method of overcoming sampling disturbance.
iour as predicted. The effective stress path (Fig.
Figure 28 shows the void ratio changes associ- 30(b)) rises to the ultimate failure line and then
ated with the two types of test. Tests 22C and travels down it towards the origin with the
27G were CAU tests and it can be seen that small average effective stress reducing to about 65 kPa.
reductions in void ratio took place when the in If the intrinsic critical state had been reached the
situ stress state was re-established. Tests 1OC and average effective stresses would have reduced to
27E were SHANSEP tests and it is evident that about 5 kPa. Clearly shearing in triaxial compres-
large reductions in void ratio took place during sion does not induce sufficient destruction of the
the consolidation phase. microstructure to bring the soil to the intrinsic
The broad framework of behaviour shown in critical state.
Fig. 25 may be used to assess the likely behaviour As expected the SHANSEP test lOC, shown by
of the samples referred to in Fig. 28. Sample 22C the broken lines in Fig. 30, is very much less
lies above the SCL. If undrained shearing were to brittle than the CAU test. Moreover the stress
cause its state to reach the intrinsic critical state path does not rise all the way to the ultimate
352 BURLAND
Axial strain: %
.. .. . . . . Test u’VO & u’, max

201 134.3 134.3 -
‘\ 20F 132.1 132.4 -
-4o- ‘\ 22C 145.6 144.0 -
66B 48.5 137.3 -
7c 34.7 106.0 -
-6O- 1oc 56.3 151.4 212.9
(a) (b)
Fig. 30. Troll field: CAU triaxinl tests on sampks from upper clay
1.
/ lob 200 300 400 500 i
(u a + o’J2: kPa
‘\
‘\
I
‘1 Test ova *a, (~‘amax
. 27G 206.5 204.6 -
29G 227.3 227.2 -
27E 204.5 467.6 606.2
(a) (b)
Fig. 31. Troll field: CAU trinxinl tests on samples from lower clay
failure line but bends sharply to the left before ‘age’ under K, stresses the undrained strength
reaching it-as a reconstituted soil would do. and brittleness increase.
Thus, by altering the structure of the clay, the Figure 31 shows the undrained triaxial test
SHANSEP test procedure underestimates both results for the lower clay. It can be seen that the
the peak strength and the brittleness of a clay for stress-strain and stress path behaviour of sample
which the in situ state lies on or above the SCL. 27G is reasonably well modelled by SHANSEP
It can be seen from Fig. 30 that the undrained test 27E, since the experimental curves are similar
extension tests behave in broadly the same in shape with only small brittleness. It seems
manner. Tavenas & Leroueil, 1985, draw atten- probable that the ultimate state of both samples
tion to the limitations of the SHANSEP pro- closely approach the intrinsic critical state line.
cedure due to ‘destructuration’. Smith (1990) As no tests were carried out on reconstituted
shows that if a SHANSEP sample is allowed to material it is not possible to be definite about
this. Thus for a clay which lies close to the ICL sufficiently to cause it to compress down to or
the SHANSEP procedure provides a reasonable below the SCL. More drastic mechanical dis-
normalized pattern of behaviour for the natural turbance would be required to do this.
material since the soil structure is not signifi- For each of the three sites a number of CAU
cantly changed during the initial consolidation. triaxial tests were carried out with the estimated
In summary the use of the void index, ICL and in situ effective stresses applied prior to shearing.
SCL, in conjunction with the framework for the Typical results are given in Figs 34(a) to (c). For
behaviour of reconstituted soils in Fig. 25, have the sensitive clay at Ons0y (Fig. 34(a)) the stress-
been valuable in gaining an understanding of the strain curves for samples from the upper clay
undrained behaviour of the clays at the Troll site. lying on the SCL show less brittleness than for the
lower clay which lies above the SCL. The stress
paths for triaxial compression bend to the left
Results>om three sites in Norway before reaching the ultimate failure line and travel
Lacasse et al. (1985) have published the results some distance down it towards the origin. The
of laboratory tests on three normally consoli- quick clay from Ellingsrud (Fig. 34(b)) shows con-
dated Norwegian marine clays. Two key features siderably more brittleness than for Onsnry. The
of the published data are: stress paths rise up to the ultimate failure line
before bending to the left and travelling a con-
(4 The tests were carried out on block samples
so that sampling disturbance was reduced to a siderable distance down it. The stress-strain
minimum. (Comparisons were also carried out curves for the extremely quick clay at Emmerstad
with samples obtained with a fixed piston (Fig. 34(c)) show sharp peaks but the brittleness is
tube sampler). no greater than for Ellingsrud. As remarked by
The clays from the three sites cover a wide Lacasse et al. (1985), the stress paths are unusual.
(4
spectrum from a sensitive clay at Onstay, They rise to above the ultimate failure line and as
through a lean quick clay at Ellingsrud, to an peak strength is approached the stress paths bend
extremely quick clay at Emmerstad. to the right which is indicative of dilatant behav-
iour. Beyond peak the paths drop down to the
The profiles for the three sites are given in Figs ultimate failure line and travel down it. This
32(a) to (c). The following features should be interesting behaviour might be accounted for by a
noted. The liquidity index increases significantly soil fabric consisting of ‘packets’ of particles with
for Onsey through to Emmerstad. The yield bonded contacts. During shear up to peak the
stress ratio aVY’/(TVo’ increases for Ons0y through packets behave as a granular material giving rise
to Emmerstad. The vane tests show that the two to mildly dilatant behaviour. Once peak strength
quick clays, Ellingsrud and Emmerstad, have has been reached the individual packets begin to
extremely high sensitivities. break down giving rise to contractant behaviour.
The sedimentation compression curves for the In none of the three cases do the stress paths
three sites are plotted in Fig. 33. For the Ons0y approach the intrinsic critical state. For the two
site (open triangles) the clay in the top 4 m lies on quick clays the very high values of I,, would
the SCL, but at greater depths it lies a little above require that the critical state is very close to the
the SCL. The chain-dotted line is the oedometer origin in a stress path diagram. Thus, as for the
compression curve for a sample from a depth of oedometer test, the triaxial test does not appear
9.07m. After yield, the curve plunges steeply and to provide sufficient mechanical energy to break
drops below the SCL. The full circles represent down the natural fabric and bonding of lean
the sedimentation compression curve for the quick clays completely although this might be
quick clay at Ellingsrud. The sedimentation com- achieved by remoulding with a vane test.
pression curve lies well above the SCL corre-
sponding to a void index of about 3. The broken
line is the oedometer compression curve for a
sample from 8.05m in depth. It is clear that the Peak undrained strength
curve remains well above the SCL. It has been shown that the critical state frame-
The sedimentation compression curve for the work, when used in conjunction with the void
extremely quick clay at Emmerstad is given by index as a normalizing parameter, is helpful in
the open circles and it can be seen that the void accounting for the brittleness and sensitivity of
index is very high (about 5). As for the other natural clays, although frequently their states do
quick clay site, although the oedometer compres- not reach the intrinsic critical state in a triaxial
sion curve falls steeply following yield, it remains test. However, in its present form, the critical
well above the SCL. It appears from these results state framework cannot be used to predict the
that the process of one-dimensional compression peak undrained strength 8, of normally consoli-
does not disrupt the structure of a lean quick clay dated natural sediments.
watercontent:
% 0 vn: kPa S,: kPa
0 20 40 60 10 20 30
0 1=.
T
, 1 I I I I 1 1 I
Crust
.
.
.
c-0
.
.
.
4 ..
E .
r -.
‘L
.
$
..
6
-VP .
O0 *
.
0
*
.
0 l
-.
Water content: % (7 vO: kPa S,: kPa

0 20 40
0 I I I I I
s O-
+ 0
+ 0
(b)
Water content: % o’“~: kPa S,: kPa
o CAU C
Block
0 CAU E
samples
+ DSS
0 avy x Vane: peak
l Vane: remoulded
cc)
Fig. 32. Soil profiles for (a) Oas#y, (b) Ellingsrud and (c) Emmerstad, all in Norway (Lscasse et al., 1985)
7 In summary, for the Troll and the Norwegian

D Emmerstad
I-‘\
sites, it appears that the peak undrained strength
6 - -: - Ellingsrud
is more directly related to soil fabric and bonding
U”O -e- Onssy as reflected by the yield stress u,~’ than it is to
1
5 liquidity index or void index.
5.96 m
4
COMPRESSIBILITY OF OVERCONSOLIDATED
CLAYS
Point A in Fig. 35(a) represents the in situ state
of an element of overconsolidated clay in an e
against log u,’ diagram. The locations and slopes
of the natural sedimentation compression curve
and swelling curve are unknown. In Fig. 35(b) the
void ratio has been transformed to I, and point A
plots as A’. Using this plot the position of A
relative to the ICL and the SCL is known and
this gives an immediate indication of the approx-
imate degree of overconsolidation for the soil
assuming that compression took place close to
u “: kPa
the SCL.
Fig. 33. Results of oedometer tests on block samples of In this section the location of some oedometer
three Norwegian sensitive clays (Lacasse et d., 1985) compression curves relative to the ICL and SCL
are investigated as shown in Fig. 35(c). Also the
measured swelling characteristics of some natural
It is a central tenet of critical state soil mecha-
overconsolidated clays are compared with the
nics that, for a given type of clay, S, is primarily
intrinsic swelling line (ISL) as shown in Fig. 35(d).
related to water content, or void ratio, and more
It should be noted that the intrinsic swelling
generally to liquidity index, or void index (Wood,
index C,* is defined as the slope of the ISL at an
1985). At the Troll site the upper clay has a much
overconsolidation ratio of 10. For this study only
higher liquidity index and void index than the
the results from block samples are considered in
lower clay (see Fig. 27). Thus, for a given effective
order to minimize the effects of sampling dis-
overburden pressure, critical state soil mechanics
turbance.
would predict that the upper clay would have a
lower S, than the lower clay. It can be seen from
Fig. 27 that, at the junction between the two
clays, there is little difference between the Sure Gault Clay
values above and below it. The value of SuTC/uvo’ Samuels (1975), working at the Building
is about 0.4 for both clays. Expressed as a pro- Research Station, carried out a number of oedo-
portion of the vertical yield stress rr”,,‘, the values meter tests on block samples of heavily over-
of Su~J~vy’ are approximately 0.32 and 0.28 for consolidated Gault Clay extracted from shafts
the upper and lower clays respectively. These associated with the Ely-Ouse tunnel. He also
values are within the normally expected range for carried out oedometer tests on samples that had
soft clays (Hight et al., 1987). been reconstituted at twice the liquid limit. Fig.
Referring again to the three Norwegian clays in 36 shows some typical results for a block from
Fig. 32, critical state soil mechanics would predict 85.3m depth. Values of e and I, are plotted on
that, for a given overburden pressure, the clay at the left- and right-hand vertical axes respectively.
Emmerstad would be weaker than at Ellingsrud, The intrinsic compression and swelling lines are
which in turn would be weaker than at Onsey shown as chain-dotted. The ratio between the
because the void indices and liquidity indices intrinsic compression and swelling indices
decrease in that order. In fact the reverse is the C,*/C,* = 0.398.
case. At an effective overburden pressure of For the tests on the undisturbed samples the
50 kPa the values of S,,, for Emmerstad, Ellings- swelling pressure was measured by adding
rud and Onssy are approximately 35.7 kPa, weights to the hanger to prevent swelling follow-
27.4 kPa and 22.0 kPa respectively. When ing soaking of the sample. It can be seen that the
expressed as a proportion of evY’ the correspond- swelling pressure is slightly less than the value of
ing values of S,, Jo,,,’ are 0.21 to 0.31 for a,,‘. The oedometer compression curve crosses
Emmerstad, 0.23 to 0.27 for Ellingsrud and 0.27 the ICL and then bends down. The stresses were
for Ons0y. not sufficiently high to establish whether or not
356 BURLAND
(a)
1
I Axial strain: %
&lo: ’ 2 3 4 56 7
\ ’ _-)----
\ ___---
-,o- ‘. _.v,;-, - -
-2o-
03
Axial SIram: %
5.92 m
CC)
Fig. 34. Results of CAU triaxial tests on black samples of Norwegian sensitive clays from (a) Onsq~y,(b)
Elliogsrud and (c) Emmerstad (Lacasse et al. 1985)
the curve intersects the SCL but it is clear that soil. Schmertmann (1969) defined this ratio as the
the normal consolidation line has not been ‘swell sensitivity’. Note that, after loading up to
reached. 7000 kPa, the first sample had become approx-
A swelling test was carried out on an identical imately twice as expansive as the one only sub-
sample. It can be seen that it is four times less jetted to unloading. Thus the process of loading
expansive than the reconstituted material. The must have destroyed some of the bonding
ratio C,*/C, for a soil may be a sensitive indicator although the clay is still less than half as expan-
of fabric and interparticle bonding in the natural sive as the reconstituted clay.
-10
100 1000
CJ“0
(4
(a)
I”
100 1000
(4 Cd)
Fig. 35. Comparison of compressioo and swelling properties of overcoosolidated clay with corresponding intrinsic
!
0.5
Reconstituted
-A
at2xw,
Fig. 36. Cult Clay (wL = 794): oedometer tests on block sample from Ely-0~ tuawl,
shaft 10, depth = %3 m (Samoels, 1975)
358 BURLAND
The inset diagram in Fig. 36 shows the results dation pressure of approximately 6MPa giving
of a cyclic swelling and compression oedometer an overconsolidation ratio of about 2.4. Horse-
test devised to investigate the susceptibility of the man et al. (1987) state that it is difftcult to recon-
clay to structural breakdown. Evidently the cile the preconsolidation pressure with present
bonding was sufficiently strong to resist this geological evidence which points to a much lower
process. It should be noted that the clay had a preconsolidation pressure. They suggest that the
calcium carbonate content of about 30% and this yield stress may be larger than the preconsolida-
may have been the source of strong interparticle tion pressure due to mechanisms such as creep
bonding. and diagenesis. However, the fact that the yield
stress lies below the SCL suggests that such
mechanisms were not of major significance.
Boom Clay Perhaps the geological evidence requires further
Horseman er al. (1987) have published the evaluation.
results of some high pressure oedometer tests on Unfortunately no tests were carried out on the
block samples of Boom Clay from Mol in reconstituted material so that the value of C,* is
Belgium. The tests are of interest because geo- not known. However, it is evident that as the
logically the clay is only lightly overconsolidated material is compressed the swelling index
but, because of the great depth from which the increases pointing to a progressive disruption of
samples were taken (247m), the clay is stiff. The the natural fabric and bonding.
results of a typical oedometer test are plotted in
Fig. 37. The in situ state is seen to lie between the
ICL and the XL. Moreover the swelling pressure
is considerably less than rrVO’.Both of these obser- Todi Clay
vations confirm that the soil is only lightly over- Over the last decade a programme of
consolidated. fundamental research into the properties of Todi
The compression curve exhibits a reasonably Clay has been carried out at the University of
well defined yield point and thereafter it drops Rome under the direction of Professor 6. Calab-
steeply towards the ICL appearing to join the resi. Todi is an attractive hill top city to the north
extension of it. The normally consolidated state of Rome. It has suffered from landslip problems
prior to geological unloading must have been in recent years and the main thrust of the
located fairly close to the SCL with a preconsoli- research has been directed towards understanding
0.8
0.i
0.t
.g
z 0.5
0.d
0.:
I111111 I I lllllll I I I
“.i 1 10
u y: MPa
Fig. 37. Boom Clay (wL = 65): bigb pressure oedometer test OIIblock sample from Mel,
depth = 247 m (Hoiseman ef a& 1987)
the influences of swelling and weathering on the index C, is insensitive to loading history and is
shear strength properties of the clay (Calabresi & only slightly less than the intrinsic value C,* it
Scarpelli, 1985). Todi Clay is a low to medium appears that interparticle bonding is not strong.
plasticity lacustrine clay of Pleistocene age. It is
overconsolidated and intensely fissured.
Block samples of the clay were extracted from SHEAR STRENGTH OF INTACT
vertical faces of a brick pit as and when they were OVERCONSOLIDATED TODI CLAY
required. Fig. 38 shows some oedometer compres- Figure 39 shows three Mohr-Coulomb failure
sion curves for the clay. The chain-dotted line envelopes for Todi Clay. The broken line labelled
shows the experimentally determined intrinsic intact strength is for intact samples (i.e. not con-
compression and swelling lines for the material. taining fissures) which were compressed or
The XL is shown as a full line. The open circles swelled from their natural moisture content prior
are for a compression test on an undisturbed to shearing in drained and undrained triaxial
sample starting from the swelling pressure. compression. The intact failure envelope shows
The compression curve crosses the ICL and significant curvature for confining pressures of
bends down without reaching the SCL. The sub- less than 15OOkPa. The chain-dotted line is the
sequent swelling index is a little less than the failure envelope for samples which were allowed
intrinsic value. The open triangles are for a test to swell freely for three months after which they
which was allowed to swell under a very low were reconsolidated and sheared. The free-swell
pressure in the oedometer prior to compressing. failure envelope lies below the intact strength
The compression curve appears to join up with envelope. Tests on normally consolidated recon-
that for the sample compressed from the swelling stituted samples were also carried out giving the
pressure (open circles). The closed triangles are intrinsic strength envelope. The differences in
for a sample which was immersed in saturated strength are due to two main factors: the void
loose sand and left to swell freely for three ratio at failure and the soil structure (fabric and
months prior to testing. The compression curve bonding).
crosses the ICL but lies beneath the curves for the The influence of void ratio may be eliminated
other two samples. The swelling index is about using the normalization procedure first developed
the same as for the other two samples. It can be by Hvorslev (1937). Fig. 40 shows the void ratio
concluded that most of the differences between of a number of samples prior to shearing. It can
the natural clay and the reconstituted clay are be seen that the samples which were allowed to
due to differences in the fabric. Since the swelling swell freely for three months (closed points) have
- - Reconstituted at Zxw~; r/L = 57.9

-Compressed from swelling pressure, wL = 43.4
-Swell to 3.7 kPa, recompress; WL = 45.8
--c Free swell for 3 months; wL = 45.8
-0.5
102
iv: kPa
Fig. 38. Todi Clay: oedometer tests on block samples after various swelling regimes (Rampello, 1989)
360 BURLAND
,--
c’
I’
,’
, ,r I Free swell for 3 months
/ ,‘,-
,-‘*/-
,,-I/
lntrmslc strength
Normal effective stress: kPa
Fig. 39. Todi Clay: Mohr-Coolomb failure envelopes
-o- Measured ICL
ICL from e,,

.
0 Swell from natural w
Free swell for 3 months
.
8
0.5 c
I I I IllIll I I ,11,,,, I / I1111,

0.4L
10 10’ 103 104
o,andp: kPa
Fig. 40. Todi Clay: void ratios prior to sharing in triaxial compression
higher void ratios than the samples which were freely swelled Todi Clay lie a little above the
swelled or compressed from their natural mois- intrinsic Hvorslev surface. This inherently greater
ture content. Following Hvorslev, the vertical strength of the natural clay is attributable to
effective pressure on the ICL corresponding to microstructural effects. It is important to note
the void ratio of the soil is termed the ‘equivalent that the natural clay can exist in states well to the
intrinsic pressure’ uve*. By dividing the strength right of the intrinsic critical state line and Rendu-
and normal effective pressure by oVc* the influ- lit surface. This is a logical consequence of the
ence of differences in void ratio are eliminated. natural SCL lying well to the right of the ICL.
Figure 41 shows a plot of (u,’ - 0,‘)/2a,,* Also shown in Fig. 41 are some typical
against (oaf + a,‘)/2ave*. The dotted lines show undrained stress paths for the natural clay. It can
the state boundary surfaces? for normally and be seen that the intact clay is strongly dilatant
overconsolidated reconstituted Todi Clay in with the stress paths moving a considerable dis-
which the initial consolidation took place under tance up the failure envelope prior to rupture.
isotropic stresses. The critical state line plots as a Even for stress paths lying outside the intrinsic
single point in this diagram separating the Hvors- Rendulic surface the clay is strongly dilatant. The
lev from the Rendulic surfaces. It can be seen that freely swelled material shows less dilatancy.
the normalized failure surfaces for the intact and
t These surfaces are termed the Hvorslev surface for POST-RUPTURE STRENGTH
overconsolidated clays and the Roscoe surface for nor- A number of tests on intact samples of Todi
mally consolidated clays (Atkinson & Bransby 1978). Clay were carried out at Imperial College by Dr
Historically it is more appropriate to call this latter Rampello and Dr Georgiannou as a collaborative
surface after Rendulic (Burland 1989). project with the University of Rome. The instru-
Reconstituted, isotropic consolidated
--o-- Swell from natural w
-‘- Free swell for 3 months
0.6 -
o . Undrained failure
0 * Drained fallure
0.5 -
Hvorslev surface
I
0 1.3
Km a + u’rPl~u’“e
Fig. 41. Todi Clay: results of trinxinl compression tests normnlixed by the equivalent
pressure o,* at failure
mentation included local strain transducers to rupture. Thereafter the relationship between
(Burland & Symes, 1982; Burland, 1989) and a the shear stress t on the slip surface and the rela-
local pore pressure probe (Hight, 1982). For all tive displacement across it is plotted.? It can be
the tests failure took place abruptly along a single seen that the shear stress drops rapidly at first but
slip surface as shown by Fig. 42. By good fortune reaches a nearly constant value after a relative
the slip surface for this test passed outside both of displacement of about 1 mm. The closed circles
the local strain transducers and close to the pore show the ratio +J,’ doing the same. The strength
pressure probe. This has made possible a detailed corresponding to the post-peak plateau is defined
and reliable study of the process of rupture. as the post-rupture strength.
In Fig. 43(a) the closed circles show the Figure 44.(a) shows the post-rupture failure
relationship between deviator force and notional envelope for Todi Clay. It can be seen that the
overall axial strain and the open circles are for envelope is bi-linear with a transition between
local axial strains. The excess pore pressures mea- low and high pressures at a normal effective stress
sured by the probe and at the base are shown by of about 1500 kPa. For high stresses the envelope
the open and closed triangles respectively. The is defined by the parameters c’ = 0, 4,,r’ = 20.2”
following important observations can be made where & ’ is the post-rupture angle of shearing
resistance. The post-rupture failure envelope is
(a) the local strain transducers show that the for-
seen to lie well below the intact failure line and a
mation of the failure plane coincides with
little above the residual failure line for which 4,’
peak strength
is approximately 17” (Calabresi, 1990). The results
(b) after peak the curve of deviator force versus
are shown to a larger scale in Fig. 44(b) for low to
notional overall strain falls steeply to a well
intermediate stresses. For these conditions the
defined plateau
post-rupture strength parameters are c’ = 23 kPa
(c) the excess pore pressure changes cease abrupt-
ly shortly after peak strength is reached and &’ = 23.7”. The cohesive intercept may
result from the fact that the failure plane is slight-
(d) prior to mak strength the local strains are less
ly wavy. The chain dotted line is the intrinsic
ihan the overall strains, as expected (Jardine
failure line from tests on reconstituted normally
et al., 1984)
consolidated samples. It is somewhat curved with
(4 after peak strength is reached the local axial $EV* = 28” at the origin decreasing to 24” at
strains decrease as a result of the unloading
0,’ = 600 kPa. Over the range of (r,’ = 100 kPa to
process; thus the post-rupture deformation
1OOOkPa the post-rupture and intrinsic failure
consists of near-rigid body sliding on the
envelopes lie very close to each other.
failure plane with very slight axial extension
in the surrounding clay.
t Chandler (1966) and Webb (1969) give expressions for
the surface area of the slip surface which is used for
Fig. 43(b) shows the relationship between calculating t and u”‘. They also give correlations for
maximum shear stress and overall axial strain up membrane restraint and lateral restraint of the end caps.
16OOr
A -Sk surface
YiYI / t , Notional overall r’-y” 1“f
z
1 800 -
,o
B
Tii
5 600-
n
Notional overall
6
Axial stram: %
(a)
600
Relative displacement: mm
0 1 2
I I 0 I 10
0 1 2
Axial strain: %
Fig. 43. Todi Clay: unconsolidated undrained triaxial test with pore pressure
measurement showing post-rupture behaviour
was carried out on intact samples with those con- stresses. These properties have been obtained
taining obvious fissures being rejected. The from a thorough re-analysis of all the data and
samples used for quick undrained testing were differ slightly from those published by Bishop et
not selected in this way and included many con- al. (1965).
taining fissures. The results have been published
in two classic papers in Gtotechnique (Ward,
Marsland & Samuels, 1965; and Bishop, Webb &
Lewin, 1965). Moreover, Webb’s doctoral thesis Intact strength
contains most of the original data (Webb, 1964). Figure 48 shows the Mohr-Coulomb failure
These data were analysed by Wroth (1972) in his envelopes for the intact clay at various depths.
study of the elastic behaviour of overconsolidated Tests were also carried out on isotropically con-
clay. Table 2 lists the basic index properties solidated reconstituted clay from level E giving
together with the estimated in situ effective the intrinsic failure line shown in the figure. At
364 BURLAND
. -.
9 c” = 28” I I I I I I I I
04,
0 200 400 600 800 1000
Normal effectwe stress: kPa
(b)
Fig. 44. Todi Clay: post-rupture failure envelope for (a) high pressures and (b) low to
medium pressures compared with intact, intrinsic and residual failure lines
200 -
m
B
N
g100-
0 100 200 300 400 500

(O a + D J/2: kPa
Fig. 45. Effective stress path for CAU triaxial compression test on normally consolidated
aged kaolin (Ninis, 1990)
250r Shp surface formed; 0 = 62
z 200 Notional overall strain

$
$
5 150
.m
?I
D
0 ,A” AU
5 100 I
Axial straw %
cn -c:
0.2 2
Relatwe displacement: mm
0 2 4 6 8
0
Axial stram: %
Fig. 46. Stress-strain behaviour for test given in Fig. 45 sbowing post-rupture behaviour
low stresses the value of 4cV* = 20.1” and this sample which was swelled isotropically to
decreases somewhat for effective stresses greater p‘ = 69 kPa and then compressed one-
than 1000 kPa. dimensionally in the triaxial apparatus. Fig. 49(b)
Figure 49(a) shows the relationship between shows the relationship between void ratio and the
void ratio and log p’ for the samples from level E log of the maximum shear stress at failure for
after swelling or compressing from the initial void drained and undrained conditions. It can be seen
ratio. The ICL and SCL are also shown. It can be that for shear strengths greater than about
seen that the isotropic compression curve crosses 1000 kPa the failure line is approximately parallel
the ICL but the applied cell pressures were not to the ICL. These data may be used to derive the
sufftcient to bring the clay to a state of normal value of gve* at failure for each test (remembering
consolidation. The chain dotted line represents that Q,,* is the pressure on the ICL correspond-
the relationship between e and log uV’ for a ing to the void ratio of the soil).
Table 2. A&ford Common-index properties and in situ effective stresses
Level Depth: wL: We: G, <2/l: W,: % eL I “0 uvO’: or’: Ko dhO’:

m % % % % kPa kPa kPa
A 9.1 58.9 23.8 2.14 42 22.59 0.619 1.614 - 1.140 117 317 3.4 400
B 15.2 68.5 28.7 2.15 59 25.68 0,706 1.884 - 1.076 179 373 2.6 469
C 20.1 70.6 28.9 2.77 53 24.82 0688 1.956 -1.151 235 448 2.3 538
D 27.1 62.3 26.6 2.72 47 22.70 0.617 1.695 - 1.191 310 524 2.0 621
E 34.8 70.0 27.0 2.17 57 23.89 0.662 1.939 - 1.200 386 690 2.1 814
F 42.1 67.8 29.0 2-14 60 23.84 0.653 1.856 - 1.184 455 159 2.0 911
l.l-
0 lsotroplc pressure
1 .o -
- - K,consolidated after
swelling top = 69 kPa
0.9 -
0.6 -
1.1
r
o Consolidated undrained
1.0
0 Consolidated drained
l Unconsohdated undrained
0.9
i
t
0.51
0.4 I I I,,,,Ll I I Illilll 1 I I Iilili I ,11!1!1!

1 10 102 103 104
(7y and (~7’~- CT,),12: kPa
W
Fig. 49. Asford Common, level E: relationship between (a) void ratio and log p’ after swell-
ing or consolidating from initial void ratio; and (b) void ratio and log (e.’ - 0,‘)/2 at failure
for drained aad undrained triaxial compression tests
stiffness (Henkel, 1972). The stress paths for the broken line which represents the one-dimensional
vertical samples bend to the right shortly before compression of a reconstituted sample. This
reaching the failure line, but, in contrast to the observation is indicative that the fabric of the
lower plasticity Todi Clay (see Fig. 41), only natural clay possesses some bonding.
travel a short distance up it prior to failure.
Marsland (1977) has published the results of
undrained tests on natural clays with a range of Post-rupture strength
plasticities and these show very clearly that the Most of the tests at low to moderately high
stress paths for overconsolidated low plasticity confining pressure exhibited brittle behaviour
clays tend to travel much further up the failure with a well defined slip surface forming at peak
line before rupture than do medium to high plas- strength (Bishop, Webb & Lewin, 1965). Fig. 52
ticity clays. shows the results for a typical undrained test and
The chain dotted line in Fig. 51 is the stress the general pattern of behaviour is strikingly
path followed by a sample which was first swelled similar to Todi Clay (Fig. 43) except that for the
isotropically to p’ = 69 kPa and then compressed London Clay the excess pore pressures remain
one-dimensionally in the triaxial apparatus. It positive throughout the test. It can be seen from
can be seen that the stress path lies well above the the bottom diagram that the shear stress on the
368 BURLAND
0.6
Level c 0 Undrained
(and A) U Dramed
P 1
g * Level A
$ 0.4 -
1
0.6r
Level E 0 Undramed
- (and D) D Drained
n Is
.t * Level D
2 0.4 -
t! -
0.6 -
0 Undramed
_ n Drained
0 0.2 0.4 0.6 0.6 1.0 1.2 1.4 1.6 1.6 2.0 22
[(U’a + 0’,)/2]lri,,
Fig. 50. A&ford Common: intact effective strength envelopes normalized by tbe equivalent pressure uVr*at failure
1000
o Consolfdated undrained
Q Consokdated dramed
l Unconsokdated undrained
600
K, consolldatlon
200 lntmx K, lme
0 200 400 600 600 1000 1200 1400 1600
(a’, + a ,)/2: kPa
Fig. 51. A&ford Common, level E: results of consolidated drained and undrained triaxial
compression teats on vertical nod horizontal intact samples
14OOr tipr’ = 152”, a somewhat higher value than the
residual angle of friction which is about 12”. The
chain dotted line in Fig. 54(a) is the intrinsic
failure line for the reconstituted soil. It lies below
the post-rupture failure line at low stresses and
above it at higher stresses. Fig. 54(b) shows a
more detailed comparison between the post-
rupture strength and the intrinsic strength
envelopes at low and intermediate stresses. Again
the general picture is strikingly similar to Todi
Clay. At low stresses the intrinsic failure line is
defined by c’ = 0; d,,* = 20.1”. Initially the post-
rupture failure line has a slightly higher angle of
friction and a cohesion intercept of about 10 kPa.
However, it bends over and drops below the
intrinsic failure line at about 750 kPa.
As mentioned previously, almost all the
samples used for effective stress testing were ini-
tially intact. However, Webb (1964) noted that a
few specimens failed on obvious pre-existing fis-
sures (marked F). In Fig. 54(b) it can be seen that
Notional axial strain, % two of the samples containing fissures have values
of &’ close to the high pressure value of 15.2”. It
is of interest to note that Skempton et al. (1969)
obtained a post-peak angle of friction of 16” for
the strength of fissures and joints in the London
Clay at Wraysbury.
In the original Ashford Common publication
Bishop et al. (1965) referred to what I have
termed the post-rupture strength as the residual
strength. Similarly Skempton et al. (1969) referred
Relatwe displacement: mm to the post-peak strength on fissures and joints as
the residual strength. Subsequently attention
0 1 2 shifted from the immediate post-peak strengths to
Axial strain: % ultimate values after large displacements. It is
Fig. 52. Ashford Common, level E: consolidated now generally agreed that the term residual
undrained test &owing post-rapture behaviour strength refers to the ultimate steady state condi-
tion, usually after large displacements. The post-
rupture strengths I refer to here are therefore not
slip surface drops to a minimum after a relative residual values. They may, however, be relevant
displacement of about 1 mm. The slight rise there- to many stability problems such as bearing
after is probably due to lateral restraint of the capacity, first-time slides in excavated slopes and
end caps. Fig. 53 shows a typical result for a retaining walls.
drained test. In this case the minimum post-
rupture strength was reached after a relative dis-
placement of about 3mm. In general the overall In-situ stresses
notional strain between peak and post-rupture Bishop et al. (1965) used the laboratory mea-
strength seldom exceeded 5%. Chandler (1966) surements of the swelling pressure pL’ to estimate
has concluded that the membrane corrections are the in situ horizontal effective stresses at the
reliable up to strains of about 12%. various levels at Ashford Common (see Table 2).
Figure 54(a) shows the results of all the post- Figs 55(a) and (b) show the resulting Mohr’s
rupture strength measurements for levels C and E circles of in situ effective stress for level A and
over the full range of stresses. The full line rep- levels B to F respectively. Also shown in these
resents the post-rupture failure envelope. This is figures are the corresponding post-rupture failure
similar in shape to that for Todi Clay (Fig. 44(a)) lines. (Note that the post-rupture failure line for
as it has an initially steep portion with a tran- level A lies above that for the other levels and this
sition to a flatter envelope at a normal effective is consistent with its lower plasticity).
stress of about 2000 kPa. For high pressures the It is important to bear in mind that a number
post-rupture failure line is defined by c’ = 0, of assumptions are involved in deriving the in situ
370 BURLAND
Shp surface formed; H = 64”

\
\
\
\
\
z - \
\ 0
\ m
‘:L
I I I I I I I I I ,
0 2 4 6 6 10
NotIonal axial strain: %
Relative displacement: mm
0123456
I I I
0 1 2 3
Axial strain: %
Fig. 53. A&ford Common, level E: consolidated drained test

showing post-rupture behaviour
values of cho’. Nevertheless, on the basis of the Figure 56 shows histograms of the unconsoli-
evidence given in Fig. 55, it is plausible that the dated undrained strength of vertical samples at
post-rupture strength limits the magnitude of the six levels. The black histograms refer to tests
the horizontal effective stresses in a heavily over- in which failure was known to take place on an
consolidated fissured clay since the fissures them- obvious fissure. Some of the low results not
selves probably result from the brittle nature of shown in black may also have resulted from the
the intact material. presence of less obvious fissures. Although the
scatter is large, the results at any level can be
broadly divided into two groups: those obviously
Quick undrained tests affected by fissures near the lower limit of
It is now widely accepted that the undrained the range and those for which the samples were
strength of a stiff fissured clay is primarily a func- more or less intact giving higher strengths. An
tion of the volume of soil being sheared and that important question is: how do the quick
the presence of fissures and joints play a major undrained strengths affected by fissures relate
role in this. At Ashford Common a large number to the post-peak and fissured effective strengths
of quick undrained triaxial tests were carried out given in Fig. M?
at Imperial College and at the Building Research As mentioned previously a few of the
Station. The results of this work have not so far undrained effective stress tests carried out by
been properly integrated with the effective stress Webb (1964) showed premature failure on pre-
testing. For some of the tests it was noted that existing fissures. Fig. 57 shows a comparison
failure appeared to take place prematurely on one between the behaviour of two such samples (C62
or more pre-existing fissures. The inclination of and C65) with an intact sample (C50). The initial
these failure planes was carefully measured. portions of the stress paths are similar in slope
- Post-rupture failure lme
--- Intrinsic failure lme
F
0 Pre-exlstlng fissure
z
LI
&ooo- -&>cII: 15.2”
(I)
2 ,e--F l
S _@
(I)
- .DflD@zSrTS- w ; i ~;~~~L?~~G%“‘,,
_--
,--- X n Drained horizontal
I I I I I I 1 1 I I
0 1000 2000 3000 4000 5000
(a)
600 -
1
0 200 400 600 800 1000 1200 1400
Normal effective stress’ kPa
(b)
Fig. 54. A&ford Common: post-rupture failure envelopes for (a) high pressures and (b) low to
medium pressures compared with the intrinsic failure line
- Post-rupture failure line, level A
400 -
m
4
;; 300 -
P
&m
$ 200-
r”
(I)
0 I I I I I I
(a)
400
- Post-rupture failure line. levels C and E
Normal effectwe stress: kPa

(b)
Fig. 55. Ashford Common: in situ MOWS circles of effective stress superimposed u.1 the
post-rupture failure envelopes for (a) level A and (b) all other levels
371
372 BURLAND
S, : kPa From the geometry of the problem it can be

0: I I
200 I I
400 , I
600 I 1 1
600 shown that the shear stress on the plane of the
l Failure on obvious fissure at failure is
fissure
rr = 7_ sin 28 (7)
and the corresponding normal effective stress is

Level A
10
/ b”f
‘=7 ,,,(l - 2A + cos 20) + pk’ (8)
Equations (7) and (8) have been used to calcu-
late the values of 7f and cnf’ for all the quick
undrained tests on vertical and horizontal
samples which obviously failed on pre-existing fis-
sures. The results are plotted in Fig. 58. It must
be emphasized that the individual values of pk
and A are not known, only the average values at
each level. Hence some of the points may be sig-
nificantly in error. Nevertheless it is clear that the
broken line for 4’ = 15.2” (taken from Fig. 54)
forms a reasonable lower bound to the data. The
E
post-rupture failure line for initially intact speci-
mens and the intrinsic failure line are also shown.
‘qw F
Up to normal effective stresses of about 600 kPa
the experimental points lie on either side of these
Fig. 56. A&ford Common: histograms of undrained lines. At higher stresses, particularly for level
strength from quick uocomolidated undrained triaxial E, the experimental points tend to lie below
compression tests 00 vertical samples these lines.
It can be concluded that the strengths from the
quick undrained tests on samples containing fis-
but premature failure on a fissure truncates the sures are consistent with the strength envelopes
stress path and in particular eliminates most or established from effective stress tests on samples
all of the dilatant portion. in which slip surfaces have formed. Rate effects
These observations assist in the analysis of the seem to be less important than they would be for
standard quick undrained tests as shown in Fig. intact strength.
58. The average value of the swelling pressure pL’
is known for each level. The slope of the stress
path is related to the pore pressure parameter A.
Average initial values of A for the tests on the OPERATIONAL STRENGTH OF STIFF
vertical and the horizontal samples are 0.67 and FISSURED CLAYS
0.29 respectively. The values of r,,, and 0 for Figure 59 shows the well known results
each of the tests on fissured samples are known. obtained by Marsland (1974) for London Clay at
600
r
60
1 -J
0 1234 5 6
(o’, + 0’,)/2: kPa Notional axial straw
Fig. 57. Askford Common, level C: results of two consolidated uodrnined tests which failed prematurely
00 preexisting fissure compared with test on intact sample
374 BURLAND
S,: kPa
100 200 300
I I I
a 98 mm diameter speclmen
(4
Fig. 59. Loodoo Clay, Hendon: comparison between operational undrained strengths back-
analysed from 86Smm dia. plate loading tests and peak strengths from quick unconsolidated
undrained triaxial compression tests oo (a) 98 mm dia. specimens and (b) 38 mm dia. specimens
(Marsland, 1974)
DISCUSSION AND CONCLUSIONS both with respect to fabric and bonding (both of
This lecture has demonstrated the value of these constituting the soil structure). The struc-
using the compressibility and strength character- ture of a natural clay depends on many factors
istics of young reconstituted clays as a framework such as depositional conditions, ageing, cementa-
for interpreting the corresponding properties of tion and leaching. These structural features pro-
natural clays. A reconstituted clay is one that has foundly affect the mechanical properties of the
been thoroughly mixed at between 1 and 1.5 natural material. One objective of this Lecture
times the liquid limit and preferably consolidated has been to show that the intrinsic properties of a
one-dimensionally. The properties of such a clay natural clay provide a robust frame of reference
are termed ‘intrinsic properties’ since these are against which to assess the in situ state of the soil,
inherent to the soil and independent of its natural its structure and the measured mechanical
state. An intrinsic property is denoted by an properties of undisturbed samples.
asterisk. Examples of intrinsic parameters are It has been demonstrated that the intrinsic
e:,,, C,* and C,* for compression and swelling compression line (ICL) is a valuable reference line
and $=“* for the intrinsic critical state angle of for studying the compression characteristics of
shearing resistance. The intrinsic Hvorslev natural normally and overconsolidated sedimen-
strength parameters 4=*, Je* and gVc* are relevant tary clays. The ICL is defined by the two con-
to the study of overconsolidated clays. Numerous stants of intrinsic compressibility eFoo and C,*
other intrinsic properties could of course be mea- (see Fig. 3(a)). Provided the Atterberg limits lie
sured including those relating to permeability. above the A line, there is a good correlation
In the past insufficient distinction has been between these constants of intrinsic compress-
made between intrinsic properties and the proper- ibility and the void ratio at the liquid limit eL as
ties of natural undisturbed soils. Natural soils shown in Fig. 8. These correlations have proved
differ from the corresponding reconstituted soil useful when e:,,, and C,* have not been directly
determined. It is recommended that, whenever overconsolidation of a natural clay particularly

possible, the constants should be determined when the yield pressure dyyr is not well defined.
experimentally. To take account of small varia- Also the ratio of the intrinsic swelling index to the
tions in liquid limit within a given clay stratum natural swelling index C,*/C, (the swell
the values of e:,, and C,* may be assumed to sensitivity) provides an important measure of
vary in direct proportion to the liquid limit. Thus bonding in the natural soil.
it would not normally be necessary to carry out a The critical state framework provides a coher-
large number of determinations of e:,,, and C,* ent model of the behaviour of reconstituted soils
down a given profile. in terms of void index, shear stress and direct
The effects of variations in soil type, as reflected effective stress. This framework has been shown
in the values of eToo and Cc*, may largely be to explain qualitatively why normally consoli-
eliminated by replacing the void ratio e with a dated natural clays lying above the ICL are more
normalizing parameter I, (void index) defined by brittle and sensitive than reconstituted soils. For
equation (1) in terms of the two constants of these clays the SHANSEP test procedure is not
intrinsic compressibility. The ICL forms an appropriate. It appears that, when sheared in the
almost unique line in a plot of I, against log uV’ triaxial apparatus, most natural clays do not
as shown in Fig. 4. This plot has proved useful for reach the intrinsic critical state. Much more
comparing sedimentation compression curves for vigorous shearing is evidently required to break
various soil profiles and for studying the one- down the natural structure of the clay.
dimensional compression characteristics of In its present form the critical state framework
natural clays in a unified way. cannot be used to predict peak undrained
The majority of normally consolidated natural strength S, of normally consolidated clays. S,
clays have sedimentation compression curves depends primarily on the structure of the clay
which, when expressed in terms of I,, lie within a and the in situ effective stresses and not on the
narrow band well above the ICL (see Fig. 13). void ratio or void index. It has been shown that
The regression line through this band has been for undisturbed natural sensitive clays
termed the sedimentation compression line (SCL). S”,& ,,Y’N 0.3, although for quick clays this ratio
Not all natural clays lie on the SCL reflecting dif- may be somewhat lower. The yield stress bVY’is a
ferences in depositional and post-depositional measure of the yield properties, or yield locus, of
environments. Moreover the SCL for most soils is the clay.
not a smooth curve and is often ‘saw-toothed’, The intact strength properties of two heavily
again reflecting temporal variations in deposi- overconsolidated undisturbed clays have been
tional conditions. Thus the SCL is not a funda- studied: a low plasticity clay from Todi, Italy,
mental line but is nevertheless useful since it and high plasticity London Clay from Ashford
represents a norm for the majority of natural Common. For both clays the intact failure sur-
sedimentary clays. faces lie above the intrinsic Hvorslev surfaces
The location of the SCL to the right of the ICL clearly demonstrating the enhanced strength of
shown in Fig. 13 implies that, for a given value of the natural microstructure. In the case of the
I “0) the effective overburden pressure carried by Todi Clay a prolonged period of free swell does
the natural clay is approximately five times that not entirely eliminate this enhanced strength.
carried by the equivalent reconstituted clay. This Both these clays exhibit brittle behaviour at
is a measure of the enhanced resistance of the low and intermediate stresses with the formation
structure of most natural clays. For quick clays of shear surfaces at peak intact strength. The
and cemented clays the enhanced resistance is strength on a shear surface drops rapidly to a rea-
many times larger than the above figure. sonably steady value after only a few millimeters’
It has been shown that, for clays whose natural relative displacement. This is termed the post-
state lies above the ICL, the one-dimensional rupture strength and should be clearly distin-
compression curve is usually significantly steeper guished from the residual strength which is
than the ICL and tends to converge with it at reached after very much larger relative displace-
high pressures (e.g. Fig. 20). This behaviour ments.
results from the progressive collapse of the For Todi Clay and London Clay the post-
natural soil structure. However, for clays whose rupture failure envelopes and intrinsic critical
mode of deposition is such that its in situ state state failure envelopes lie close together at low
lies close to the ICL, the one-dimensional com- stresses, but at higher stresses the post-rupture
pression curve will tend to lie parallel to the ICL strengths are less than the intrinsic critical state
since the structure of the natural clay is similar to strengths. Further work is required to investigate
that of the reconstituted material. the phenomenon of post-rupture strength in other
For overconsolidated clays the ICL and SCL intact materials and to carry out comparisons
provide a useful means of assessing the degree of with the intrinsic critical state strength. A prelimi-
376 BURLAND
nary study on normally consolidated kaolin gives data for levels B, D and F from Ashford
the post-rupture angle of shearing resistance $rr’ Common. During the study described here I have
somewhat less than dcv*. benefited from discussions with many colleagues
In addition to studying the strengths of initially and friends too numerous to list. I am particu-
intact samples, the results of tests on samples con- larly grateful to Professor G. A. Leonards, Pro-
taining existing fissures have also been examined. fessor A. W. Skempton, Professor P. R. Vaughan,
The results of many of the tests give strengths on Dr R. J. Chandler, Dr R. J. Jardine, Dr D. W.
the fissures close to the post-rupture strength of Hight and Dr Suzanne Lacasse, I also wish to
initially intact specimens. However, some acknowledge the support and encouragement of
strengths are somewhat lower and a well defined all my colleagues in the Soil Mechanics Section at
lower limit to the fissured strength has been iden- Imperial College.
tified for the London Clay.
It is suggested that the post-rupture strength
may be relevant to many stability problems in
stiff clays. For example the in situ effective REFERENCES
stresses at Ashford Common deduced by Bishop Atkinson, J. H. & Bransby, P. L. (1978). The mechanics
et al. (1965) are consistent with the mobilization of soils. An introduction to critical state soil mecha-
of post-rupture strengths during the process of nics. London: McGraw-Hill.
geological unloading and the formation of fis- Bishop, A. W., Webb, D. L. & Lewin, P. I. (1965).
sures. It is also demonstrated that the operational Undisturbed samples of London Clay from the
Ashford Common shaft : strength-effective stress
undrained strength of a stiff fissured clay en masse
relationship. Giotechnique 15, No. 1, l-31.
is consistent with the post-rupture strength and Bjerrum, L. (1967) The seventh Rankine Lecture. Engin-
probably results from a combination of the pre- eering geology of Norwegian normally-consolidated
sence of fissures and progressive failure due to the marine clays as related to settlements of buildings.
brittle nature of the intact material. Finally the Gtotechnique 17, No. 2,81-118.
operational effective strength envelope for first Bjerrum, L. & Rosenqvist, I. Th. (1956). Some experi-
time slides in brown London Clay deduced by ments with artificially sedimented clays. Gkotech-
Skempton (1977) lies between the lower bound nique 6, No. 3, 124-136.
Burland, J. B. (1989). The ninth Bjerrum Memorial
envelope for the strength on fissures and the post-
Lecture: ‘Small is beautiful’-the stiffness of soils at
rupture strength for initially intact samples.
small strains. Can. Geotech. .I. 26,499-516.
Burland, J. B. & Symes, M. (1982). A simple axial dis-
placement gauge for use in the triaxial apparatus.
ACKNOWLEDGEMENTS GCotechnique 32, No. L62-65.
I would like to express my gratitude to the Calabresi, G. (1990). Private communication.
British Geotechnical Society for inviting me to Calabresi, G. & Scarpelli, G. (1985). Effects of swelling
deliver the thirtieth Rankine Lecture, which I caused by unloading in overconsolidated clays.
Proc. 11th Int. ConJ on Soil Mech., San Francisco, 2,
regard as one of the highest honours not only in
411415.
soil mechanics but also in civil engineering. Also I Casagrande, A. (1932). The structure of clay and its
wish to thank Mr Thorburn for his kind and gen- importance in foundation engineering. J. Boston
erous remarks. Sot. Ciu. Engrs. 19, No. 4, 168-209.
I am grateful to the following for permission to Chan, D. H. & Morgenstem, N. R. (1989). Bearing
publish experimental data in this Lecture: Statoil capacity of strain softening soil. De Mello volume,
for the results from the Troll field; Rendel, pp. 5968. Editora Edgard Blucher Ltda.
Palmer and Tritton for the results for Surabaya; Chandler, R. J. (1966). The measurement of residual
the Norwegian Geotechnical Institute for the strength in triaxial compression. Giotechnique 16,
results from the three sensitive clay sites in No. 3,181-186.
Chandler, R. J. (1984). Recent European experience of
Norway and the Building Research Estab-
landslides in over-consolidated clays and soft rocks.
lishment for the results on Gault Clay. Dr B. State-of-the-art report. 4th Int. Symp. on Landslides,
McClelland made available the results of oedo- Toronto, 1,61-81.
meter tests from the Mississippi delta, Mr D. Edge, M. J. & Sills, G. C. (1989). The dew .opment of
Nash provided the oedometer test results and layered sediment beds in the laboratory as an illus-
other soils data from Bothkennar, and Professor tration of possible field processes. Q. J. Engng. Geol.
A. Nakase provided the data for Fig. 9 on the 22, Part 4,271-279.
marine clays from Japan. I am indebted to Pro- Gens, A. (1982). Stress-strain and strength character-
istics of a low plasticity clay. PhD thesis, University
fessor G. Calabresi and his colleagues for allow-
of London.
ing the publication of the results on Todi Clay Georaiannou, V. N. (1988). The behaviour of clayey
and for many stimulating discussions. Professor sands under monotonic’ and cyclic loading. PhD
A. W. Skempton kindly made available his files thesis, University of London.
for the data in Fig. 1 as did Dr P. I. Lewin for the Hawkins, A. B., Larnach, W. J., Lloyd, I. M. & Nash, D.
F. T. (1989). Selecting the location, and the initial Marsland, A. (1977). The evolution of the engineering
investigation of the SERC soft clay test bed site. Q. design parameters for glacial clays. Q. J. Engng.
J. Eng&. Geol. 22, Part 4,281-316: Geol. 10, Part 1, l-26.
.
Henkel. D. J. (1972). The relevance of laboratory mea- McClelland, B. (1967). Progress of consolidation in
sured parameters in field studies. Proc. Roscoe delta front and prodelta clays of the Mississippi
Memorial Symp., Foulis, pp. 669-675. River. In A. F. Richards (Ed). Marine Giotechnique,
Hight, D. W. (1982). A simple piezometer probe for the pp. 2240. Urbana: University of Illinois Press.
routine measurement of pore pressure in triaxial Mitchell, J. K. (1976). Fundamentals of soil behaviour.
tests on saturated soils. Gbotechnique 32, No. 4, 396 New York: Wiley.
401. Nagaraj, T. S. & Srinivasa Murthy, B. R. (1986). A criti-
Hight, D. W., Jardine, R. J. & Gens, A. (1987). The cal reappraisal of compression index equations. Gho-
behaviour of soft clays. chapter 2, Embankment on technique 36, No. 1, 27-32.
soft clays, pp. 33-158. Athens: Public Works Nakase, A., Famei, T. & Kusakabe, 0. (1988). Constitu-
Researcd Centre of Greece. tive parameters estimated by plasticity index. J.
Horseman, S. T.. Winter, M. G. & Entwistle, D. C. Geotech. Engng. Div., Am. Sot. Ciu. Engrs 114, No. 7,
(1987). Geotechnical characterisation ofBoom Clay in 844-858.
relation to disposal of radioactive waste. Luxem- Newland, P. L. & Allely, B. H. (1956). Results of some
bourg: Ollice for Ollicial Publications of the Euro- investigations of two sensitive clays. Proc. 2nd Aust.-
pean-Communities. NZ Con& Soil Mech., 3945.
Hvorslev. M. J. (1937). Uber die Festigkeitseigenschaf- Ninis, N. (1990). Private communication.
ten Gestorter Bindiger Boden. Danmarks -Naturvi- Northey, R. D. (1956). Rapid consolidation tests for
denskabelige Samfund. Ingenioruidenskabelige routine investigations. Proc. 2nd Aust.-NZ Conf.
Skrgter, A, No. 45. Soil Mech., 2&26.
Jardine, R. J. (1985). Investigation of pile-soil behaoiour Ramiah, B. K. (1959). Time effects in the consolidation
with special reference to the foundations of offshore properties of clays. PhD thesis, Purdue University.
structures. PhD thesis, University of London. Rampello, S. (1989). E&tti de1 rigonjamento sul com-
Jardine, R. J., Symes, M. J. & Burland, J. B. (1984). The portamento meccanico di argifle fortemente souracon-
measurement of soil stiffness in the triaxial appar- solidate. Doctoral thesis, University of Rome.
atus. Gtotechnique 34, No. 3,323-340. Rendulic, L. (1937). Ein Grundgesetz der Tonmechanik
Lacasse, S., Berre, T. & Lefebvre, G. (1985). Block sam- und sein Experementeller Beweis. Der Bauingeneur
pling of sensitive clays. Proc. 11th Int. Conf on Soil lg.
Mech., San Francisco, 2, 887-892. Samuels, S. G. (1975). Some properties of the Gault
La Rochelle, P., Sarrailh, J., Tavenas, F., Roy, M. & Clay from the Ely-Ouse Essex water tunnel. Gho-
Leroueil, S. (1981). Causes of sampling disturbance technique 25, No. 2, 239-264.
and design of a new sampler for sensitive soils. Can. Schmertmann, J. H. (1969). Swell sensitivity. Gizotech-
Geotech. J. 18, No. 1, 52-66. nique 19, No. 4, 530-533.
Leonards, G. A. & Ramiah, B. K. (1959). Time effects in Sejrup, H. P., Nagy, J. & Brigham-Grette, J. (1989).
the consolidation of clay. ASTM Special technical Foraminiferal stratigraphy and amino acid geochro-
publication No. 254, pp. 116130. Philadelphia: nology of Quatemary sediments in the Norwegian
ASTM. Channel, northern North Sea. Norsk Geologisk
Leonards, G. A. & Girault, P. (1961). A study of the Tidsskrft 69, No. 2, 111-124.
one-dimensional consolidation test. Proc. 5th Int. Skempton, A. W. (1944). Notes on the compressibility of
Conf. on Soil Mech, Paris, 1, 213-219. clays. Q. J. Geol. Sot., 100, 119-135.
Leonards, G. A. & Altschaelll, A. G. (1964). Compress- Skempton, A. W. (1970). The consolidation of clays by
ibility of Clay. Soil Mech. Dia., Proc. Am. Sot. Ciu. gravitational compaction. Q. J. Geol. Sot. 125, 373-
Engrs 90,133-155. 411.
Leroueil, S., Tavenas, F., Brucy, F., La Rochelle, Skempton, A. W. (1977). Slope stability of cuttings in
P. & Roy, M. (1979). Behaviour of destructured brown London Clay. Proc. 9th Int. Con)Y on Soil
natural clays. Proc. Am. Sot. Ciu. Engrs 105, GT6, Mech., Tokyo, 3,261-270.
759-778. Skempton, A. W. & Henkel, D. J. (1953). The post-
Leroueil, S., Tavenas, F. & Locat, J. (1985). Discussion: glacial clays of the Thames Estuary at Tilbury and
Correlations between index tests and the properties Shellhaven. Proc. 3rd Int. Conf: Soil Mech., Zurich,
of remoulded clays. W. D. Carrier III and J. F. 1,302-308.
Beckman. Gtotechnique 35, No. 2,223-226. Skempton, A. W., Schuster, R. L. & Petley, D. J. (1969).
Leroueil, S. & Vaughan, P. R. (1990). The important Joints and fissures in London Clay at Wraysbury
and congruent effects of structure in natural soils and Edgware. Gtotechnique 19, NO. 2,205217.
and weak rocks. Gbotechnique 40, No. 3. Smith, P. R. (1990). Personal communication.
Locat, J. & Lefebvre, G. (1986). The origin of structur- Som, N. N. (1968). The effects of stress path on the defor-
ation of the Grande-Baleine marine sediments, mation and consolidation of London Clay. PhD
Quebec. Canada. Q. J. Engng. Geol. 19, Part 4, 365- thesis, University of London.
374. Tarzi, A. I., Kalteziotis, N. A. & Menzies, B. K. (1982).
Marsland, A. (1974). Comparison of the results from Finite element analysis of strip footings on strain-
static penetration tests and large in-situ plate tests in softening clay. Proc. Int. Symp. on Num Models in
London Clay. Proc. European Symp. Penetration Geomechanics, Rotterdam, 194-200.
Testing, Stockholm. Tavenas, F. & Leroueil, S. (1985). Discussion on Theme
378 BURLAND
Lecture 2. Proc. 11th Int. Co& on Soil Mech., San Webb, D. L. (1969). Residual strength in conventional
Francisco, 5,2693-2694. triaxial tests. Proc. 7th Inc. Conf. on Soil Mech.,
Terzaghi, K. (1925). Erdbaumechanik auf bodenphysika- Mexico City, 1, 433-441.
lischer Grundlage. Vienna : Deuticke. Wood, D. M. (1985). Index properties and consolidation
Terzaghi, K. (1941). Undisturbed clay samples and history. Proc. Ilth Int. ConJ on Soil Mech., San
undisturbed clays. J. Boston Sot. Civ. Engrs 28, No. Francisco, 2, 703-706.
3,45-65. Wood, D. M. (1990). Soil behaviour and critical state soil
Ward, W. H., Marsland, A. & Samuels, S. G. (1965). mechanics. New York: Cambridge University Press.
Properties of the London Clay at the Ashford Wroth, C. P. (1972). Some aspects of the elastic behav-
Common shaft: in-situ and undrained strength tests. iour of overconsolidated clay. Proc. Roscoe Memo-
Gbotechnique 15, No. 4,321-344. rial Symp., Foulis, pp. 347-361.
Webb, D. L. (1964). The mechanical properties ofundis- Wu, T. H. (1958). Geotechnical properties of glacial lake
curbed samples of London Clay and Pierre shale. PhD clays. Proc. Am. Sot. Civ. Engrs s4, SM3, paper
thesis, University of London. 1732.
VOTE OF THANKS performing tests on reconstituted material. He

has introduced new definitions: the intrinsic com-
DR R. J. MAIR pression line , the sedimentation compression line
We have been privileged to hear a superbly and the void index. It is my belief that these defi-
delivered 30th Rankine Lecture by Professor nitions will be widely referred to in the future by
Burland, displaying the combination of skills of a practitioners and research workers alike.
distinguished engineer, scientist and teacher By introducing these definitions, John Burland
referred to by Mr Thorburn in his introduction. has elegantly distinguished between the behav-
John Burland has more than lived up to his repu- iour of natural soft clays and reconstituted clays.
tation for clarity of thought and ability to reduce He has demonstrated with characteristic clarity
apparently complex geotechnical problems to a the difference in their behaviour due to
simple framework. In his outstanding Nash structure-the combination of fabric and
Lecture at the Dublin Conference in 1987, John bonding. By drawing on examples from the Mis-
referred to Terzaghi’s aim to ‘maintain that vital sissippi Delta and the North Sea, he has high-
balance between idealization and reality’. In this lighted the importance of the deposition
Rankine Lecture John Burland himself has made conditions on a clay’s subsequent behaviour.
a most valuable contribution to that balance He has also covered overconsolidated clays
between idealization and reality by clarifying the and introduced the concept of post-rupture
factors affecting the compressibility and shear strength. He has provided clear insight into the
strength of natural clays. complex behaviour associated with formation of
John Burland’s career uniquely qualifies him to rupture surfaces-it must be gratifying for him to
address the behaviour of natural clays. At Cam- see the electrolevel device he originally proposed
bridge he was closely associated with the valuable for measurement of local strains lead to an
framework of critical state soil mechanics describ- improved understanding of soil behaviour. Prac-
ing idealized soil behaviour in a new and funda- titioners are constantly faced with the problem of
mental way. At Imperial College he and his selection of appropriate design parameters for
colleagues have been concerned with the reality stiff overconsolidated clays, and it is likely that
and complexities of behaviour of natural soils. In the concepts of post-rupture failure line and
between Cambridge and Imperial College, John intrinsic failure line presented by John Burland
had a distinguished period at the Building this evening will become of significant practical
Research Establishment, where he was primarily value in the years to come.
concerned with a wide range of field In this Lecture we have seen the results of a
measurements-these have enabled him to iden- thorough re-analysis of data from a wide variety
tify the strength and weaknesses of idealized soil of sources from across the world. The meticulous
behaviour and to appreciate the complexities of way in which this has evidently been done, and
real soils. the enthusiasm with which the results have been
He has introduced the important concept of presented, are both hallmarks of John Burland’s
what he terms the intrinsic properties of clays: the style. It is with the greatest pleasure that I
properties of reconstituted clays. He has empha- propose a warm vote of thanks to Professor
sized the importance of the combination of Burland for an excellent and memorable 30th
testing good quality undisturbed samples and Rankine Lecture.

Burland - Compressibility and Shear Strength of Natural Clay

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Burland - Compressibility and Shear Strength of Natural Clay

Caricato da

Copyright:

Formati disponibili

Burland, J. B. (1990).

On the compressibility and shear strength of natural clays

The compressibility and strength characteristics of Les caracteristiques de compressibilite et de rbis-

NOTATION natural soils or artificial materials such as kaolin-

ccf compression index

Depth y T YiT ia? Yiiz V$o 3Eo m

t Leroueil et al. (1985) define four states of structure:

Intrinsic compression line

e log (7”‘: kPa

_ Arglle plastique LL = 128

-_- London Clay LL = 67.5

0,’ (kPa) I’,

- Load wrement duration = 1 day

Fig. 5. Influence of (a) mixing moisture content on compression curves for

preconsolidation pressure’ to describe this critical

CORRELATIONS BETWEEN THE CONSTANTS

Table 1. Intrinsic coustmts of compressibility for reconstituted uaturnl clays

Soil G, eL CC* Reference

Lower Cromer Till 25 13 2.65 0.663 0.503 0.154 Gens (1982)

Void ratlo at the liquid limit e,

In concluding this section it is important to an element of normally consolidated clay with a

been derived for most of the profiles referred to in

qShellhaven layer C (wL = 82)

q Alvangen (wL = 95)

q Oslofjord 98 - Alwlgen 95 l S.Joaquin 64

-21 ’ ’ ““‘1 ’ ’ ’ ““‘1 ’ 1 ’ ““‘1 1 1 I11111’ 1 1 1 ~ult.l

Resultsfrom Bothkennar tesf bed site

0 15.6 m depth Recent deltw

Fig. 17. Mississippi Delta: results of oedometer teats on underconsolidated deltaic

Moisture content: % a’vO: kPa S,: kPa Sensitivity

, = I l Remoulded vane strength

Fig. 18. Botkkeonar: profile for korehole Dl (Hawkius et al., 1989)

(a) (b) cc)

Altschaelll (1964) can be used to examine this o-7

increments of pressure were then applied, after

m u a and (u’, + n’,)/Z

water content % uvo: kPa S,: kPa

Fig. 28. Troll field: sedimentation compression curves

o- 28.4 m (We = 33.2)

-2’ I I I111111 I I I111111 I I I I,,,,,

.. .. . . . . Test u’VO & u’, max

Water content: % (7 vO: kPa S,: kPa

Water content: % o’“~: kPa S,: kPa

7 In summary, for the Troll and the Norwegian

- - Reconstituted at Zxw~; r/L = 57.9

Normal effective stress: kPa

Fig. 39. Todi Clay: Mohr-Coolomb failure envelopes

-o- Measured ICL

ICL from e,,

I I I IllIll I I ,11,,,, I / I1111,

YiYI / t , Notional overall r’-y” 1“f

0 100 200 300 400 500

250r Shp surface formed; 0 = 62

z 200 Notional overall strain

Table 2. A&ford Common-index properties and in situ effective stresses

Level Depth: wL: We: G, <2/l: W,: % eL I “0 uvO’: or’: Ko dhO’:

0.4 I I I,,,,Ll I I Illilll 1 I I Iilili I ,11!1!1!

200 lntmx K, lme

0 200 400 600 600 1000 1200 1400 1600

(a’, + a ,)/2: kPa

Shp surface formed; H = 64”

NotIonal axial strain: %

Fig. 53. A&ford Common, level E: consolidated drained test

- Post-rupture failure line, level A