Sunteți pe pagina 1din 14

Nishimura, S., Minh, N. A. & Jardine, R. J. (2007). Géotechnique 57, No.

1, 49–62

Shear strength anisotropy of natural London Clay


 
S . N I S H I M U R A , N. A . M I N H y a n d R . J. JA R D I N E

The paper reports the shear strength anisotropy of the Cet article rapporte les résultats d’anisotropie de résis-
natural, highly overconsolidated, London Clay from Hea- tance au cisaillement de la London Clay (argile de
throw Terminal 5 as established by comprehensive hollow Londres) du Terminal 5 de l’aéroport d’Heathrow, une
cylinder apparatus (HCA) testing. Multiple high-quality argile naturelle hautement surconsolidée, tels qu’estimés
block samples from 5.2 m and 10.5 m below ground level par des essais complets réalisés avec un appareil à cy-
provided samples for suites of undrained stress-path lindre creux. Plusieurs échantillons de bloc de haute
shear tests performed after consolidation to effective qualité, de 5,2 m et 10,5 m de profondeur, ont permis de
stress states similar to those estimated in situ. The direc- prélever des spécimens pour une série d’essais à chemin
tion of the major principal stress axis, Æ (or of the de contrainte en cisaillement non drainés réalisés après
principal stress increment, Æd ), and relative magnitude consolidation à des états de contrainte effectifs similaires
of the intermediate principal stress, b (2 2 3 )/ à ceux estimés in situ. La direction de l’axe de contrainte
(1 2 3 ), were chosen as the controlled stress-related principale, Æ (ou de l’incrément en contrainte principale,
variables and their influence on the peak shear strength Æd ) et l’amplitude relative de la contrainte principale
was investigated. Strong shear strength anisotropy was intermédiaire, liée au paramètre b (2 23 )=(1 23 ) ont
proven, and a potential effect of the parameter b was also été choisies comme variables contrôlées liées à la contra-
detected. With b 0.5, for example, a minimum was inte. L’étude examine leur influence sur la résistance au
noted in peak q/p9 at around Æ 45–678, and the maxi- cisaillement maximale. Il a été montré que la résistance
mum value, which was larger by some 40%, developed at au cisaillement est fortement anisotrope et l’on a pu
Æ 08. A limited set of data obtained with larger test observer un effet potentiel du paramètre b. Avec b 0,5,
specimens suggested a possibility that loading with Æ par exemple, on a pu noter un minimum du pic q/p9 pour
908 could also lead to relatively low shear strength. The Æ autour de 45–678 et la valeur maximale, supérieure de
pre-failure pore water pressure development reflected près de 40 %, est observée pour Æ 08. Un ensemble de
anisotropy of the deformation characteristics at smaller données limité, réalisé sur des spécimens plus larges,
strains and contributed to the total-stress undrained suggèrent qu’une charge à Æ 908 puisse également con-
shear strength anisotropy. Regarding the influence of b, duire à une résistance au cisaillement relativement faible.
the Mohr–Coulomb failure line passing through the data Le développement de la pression hydrique interstitielle
for b 0 or 1 provided a lower bound of all the data for avant fracture reflète l’anisotropie des caractéristiques de
a given value of Æ. The investigation of anisotropy was déformation pour des contraintes plus faibles et contribue
extended to deeper horizons of the stratum through à l’anisotropie de la résistance au cisaillement non drainé
profiling tests involving triaxial extension and simple totale. En ce qui concerne l’influence de b, la ligne du
shear conditions. Although the results confirmed the critère de rupture de Mohr–Coulomb s’ajustant aux
general patterns of anisotropy observed at 5.2 and 10.5 m données pour b 0 ou 1 a fourni une limite inférieure de
below ground level, the degree of anisotropy appeared to toutes les données pour une valeur donnée de Æ. On a pu
become stronger with depth. approfondir les horizons de l’étude de l’anisotropie de la
strate grâce à des essais de profil comportant une exten-
sion triaxiale et des conditions de cisaillement simples.
Bien que les résultats obtenus confirment les tendances
générales de l’anisotropie observée à 5,2 et 10,5 m en
KEYWORDS: anisotropy; clays; fabric; shear strength; stress dessous du niveau du sol, le degré d’anisotropie est appa-
path; torsion ru comme devenant plus important avec la profondeur.

INTRODUCTION strength anisotropy, embankment geometry, ground move-


The importance of the mechanical anisotropy of soils has ments and collapse heights.
long been recognised in geotechnical engineering. For exam- However, similar advances do not appear to have been
ple Bjerrum (1973) and Jardine & Smith (1991), among made yet with stiff overconsolidated plastic clays. One of the
others, drew attention to principal stress axis rotation in soft major hurdles in interpreting and simulating failure of stiff
clays under embankment loading, and highlighted the need clays, in addition to sampling disturbance, is their markedly
to understand better shear strength anisotropy. Zdravkovic et progressive failure processes, deriving from their brittle con-
al. (2002) simulated a full-scale embankment test success- stitutive relations (Burland, 1990; Burland et al., 1996;
fully with an anisotropic constitutive model calibrated to Jardine et al., 2004). The progressive mechanisms, with
laboratory test data, also exploring interactions between particular reference to the London Clay, were explored in
numerical analysis by Potts et al. (1997) and Kovacevic et al.
(2004). Noting these and other areas of uncertainty, the
potential effects of shear strength anisotropy have been
Manuscript received 5 May 2006; revised manuscript accepted 14
November 2006.
largely neglected. The currently available reliable experimen-
Discussion on this paper closes on 1 July 2007, for further details tal datasets regarding the shear strength anisotropy of stiff
see p. ii. plastic clays are severely limited and do not offer sufficient

Imperial College, London, UK. evidence to develop and apply anisotropic constitutive models
y in practice. The research work undertaken in the 1960s on the
Atkins Ltd; formerly Imperial College, London, UK.

49
50 NISHIMURA, MINH AND JARDINE
natural London Clay (Ward et al., 1965; Bishop, 1966; of an HCA is the four degrees of freedom in stress and strain
Bishop & Little, 1967; Agarwal, 1968) forms an important control attained through combinations of axial force, torque
part of the available database. However, these studies had to and inner and outer cell pressures. The four stress components
rely on triaxial tests on diagonally-cut cylindrical specimens induced in a specimen, z , r , Ł and zŁ , were assessed as
that are subject to several inherent difficulties, as recognised outlined by Hight et al. (1983), and zr ¼ rŁ ¼ 0 was assumed.
in later years. Such disadvantages include the generation of They are conveniently converted into an equivalent set of four
bending moments in soil specimens (e.g. Saada, 1970; Saada stress-related parameters, p, q, Æ and b, where
& Townsend, 1981; Molenkamp, 1998) and the infeasibility 1 þ 2 þ 3
of applying the in situ K0 stress conditions. p¼ (1)
Investigations connected with the Heathrow Airport Term- 3
inal Five (T5) project offered the opportunity to apply more q ¼ 1  3 (2)
 
advanced laboratory testing techniques to high-quality block 1 2zŁ
samples of the natural London Clay from three different Æ ¼ tan 1 (3)
2 z  Ł
depths, involving stress-path triaxial cells and hollow cylin-
der apparatus’ (HCAs). Multiple block samples from two 2  3
b¼ (4)
key horizons allowed suites of tests to examine anisotropy 1  3
by imposing 1 in different directions, while other conditions
such as the reconsolidation regime and the intermediate where Æ is the angle between the major principal stress
principal stress ratio, b ¼ (2  3 )/(1  3 ), applied during direction and the vertical. The effects of Æ on material
shearing were held constant, so allowing assessment of the behaviour are therefore direct indices of anisotropy. The
net influence of anisotropy. The investigation was extended quantity b is called the intermediate principal stress ratio,
to deeper horizons by means of multiple triaxial extension and varies between 0 and 1.
and simple shear tests on rotary cores recovered from 6 m to Two different computer-controlled HCAs were employed:
35 m below ground level. Referring to existing triaxial com- the ICRCHCA (Porovic, 1995; Nishimura, 2006) and the
pression test data, the shear strength profiles in these shear ICHCA II (Jardine, 1996; Minh, 2006), which are illustrated
modes were used to gauge how anisotropy developed over in Figs 2(a) and 2(b) respectively. Both apparatus sets have
the main thickness of the stratum. the ability to perform experiments under mixed stress and
This paper focuses mainly on the peak shear strength strain control, depending on the particular test requirements.
anisotropy, and the influence of the intermediate principal The nominal outer diameters, inner diameters and heights of
stress on the effective and total-stress shear strengths avail- test specimens were 70 mm, 38 mm and 170–190 mm in the
able on shearing from in situ stress conditions. The authors’ ICRCHCA, and 100 mm, 60 mm and 200 mm in the ICHCA
experiments investigated the clay’s behaviour over a wide II. The ICRCHCA was equipped with a Hardin oscillator,
strain range. The anisotropic and non-linear pre-failure de- with which torsional resonant column tests were performed to
formation characteristics are reported and discussed in a obtain dynamic shear moduli. The sample deformation was
companion paper (Gasparre et al., 2007b). The clay’s geol- measured globally, apart from the semi-local (platen-to-pla-
ogy, composition and structure, which are the causes of its ten) torsional rotation with a proximity transducer system. A
anisotropy, are discussed in a second companion paper full suite of high-resolution local transducers was available
(Gasparre et al., 2007a), which also gives details of index for a part of the test programme conducted with the ICHCA
properties. Hight et al. (2007) synthesise the authors’ find- II. The axial displacement and torsional rotation were meas-
ings with other field and laboratory investigations, and refer ured with a system consisting of enhanced electrolevel inclin-
to the potential practical application of the results. ometers, and changes of the inner and outer radii were
measured with a linear variable displacement transducer
(LVDT) and a set of three proximity transducers respectively.
APPARATUS AND MATERIAL From these boundary displacements, the average strains in-
Hollow cylinder apparatus (HCA) duced in a soil specimen were calculated according to the
The boundary forces and pressures applied to a hollow equations given by Hight et al. (1983). Most of the triaxial
cylindrical specimen and the stresses and the strains induced in extension tests were performed on 38 and 100 mm diameter
its elements are illustrated in Fig. 1. The most attractive feature specimens with the locally instrumented hydraulic stress-path
triaxial cells described by Gasparre (2005).

σz
Axial force dθ Natural London Clay samples
dr
Torque τzθ Hand-trimmed block samples and triple-tube rotary cores
τθz were retrieved from the T5 site, at the locations identified by
dz Hight et al. (2007). The blocks were taken from excavation
σθ
bases formed at 1.0 m, 5.2 m and 10.5 m below ground level,
Outer cell σr with the latter two depths being chosen as key horizons for
pressure (b) the detailed investigation of anisotropy. The rotary cores were
Inner
σ1 retrieved from a location where the London Clay was overlain
cell α by a 6 m thick Quaternary gravel layer; this layer had been
pressure
z removed decades earlier at the block sampling location. The
σ3 cores from 6 m below ground level (and hence the top of the
London Clay) to 35 m below the ground level were tested in
the present study. As described by Hight et al. (2007), this
θ
σ2  σ3  b(σ1  σ3) range covers lithological units A3, B2 and C, established by
(a)
r
(c) King (1981); unit B1 was not recovered successfully.
Table 1 summarises the index properties of the samples
Fig. 1. Definition of coordinate system, stresses, strains and from 5.2 m and 10.5 m bgl. The standard deviations re-
stress-related variables corded indicate that the samples were sufficiently uniform at
SHEAR STRENGTH ANISOTROPY OF NATURAL LONDON CLAY 51

Displacement
transducer Bellofram cylinder
Air valve
Tie bars
Ram Perspex
Load cell chamber
Clamp Extension
Sprocket and torque Pillars
rod
transmission chain Two double-axis
Stepper motor for 1
Resin cups electrolevels
torsion Rotary tension cylinder
Drainage 2 Local LVDT
lines 3
Single-axis
electrolevel
3
Specimen Three proximity
2 4
4 transducers
Tie rod Hardin oscillator
5
Displacement
1
transducer (axial)
Proximity transducers 6 Displacement
Cam Main
transducer (rotation)
shaft
To inner and outer
Acrylic chamber wall cell pressure
transducers
Specimen 6
Bellofram cylinders
for torque
Transfer plate
Chain and
5
Load cell sprocket

Outer cell and pore water


pressure transducers
Concrete Axial
Concrete
base Bellofram 0 50 100 200 mm
base
cylinder
100 mm
(approximate)

To foundation
(a) (b)

Fig. 2. Schematic illustration of HCAs: (a) ICRCHCA (Porovic, 1995; Nishimura, 2006); (b) ICHCA II (Jardine, 1996; Minh, 2006)

Table 1. Index and other properties of block samples

5.2 m bgl (12.5 mOD) 10.5 m bgl (7.2 mOD)

Average Std dev. Average Std dev.

Natural water content, w: % 23.7 0.3 24.9 0.5


Plastic limit 25 0.3 26 0.8
Liquid limit 69 0.3 70 0.9
Specific gravity, Gs 2.77 0.01 2.77 0.01
Clay fraction: % 51 1.3 55 1.0
Sand content: % 3.7 1.1 1.7 0.3
Unit weight, ª: kN/m3 19.9 0.2 19.9 0.3
GzŁ after consolidation: MPa – – 92.5 3.6

each depth in terms of index properties. Also tabulated is blocks from 5.2 m bgl were more heavily fissured than those
the range of shear moduli GzŁ obtained from 22 resonant from 10.5 m bgl, although the upwardly intensifying trend
column tests at p9 ¼ 323 kPa at the end of isotropic reported by many researchers (e.g. Ward et al., 1965;
reconsolidation. The relatively small deviation suggests that Chandler & Apted, 1988) was not recognised in samples
variable disturbance imposed during the sampling and pre- from the gravel-capped rotary borehole locations.
paration was minimised.
One of the dominant features in the London Clay tested
was the macrofabric represented by natural discontinuities. HCA TESTING PROCEDURES AND SCHEMES
These include fissures and joints, although no particular The HCA stress-path tests in which Æ (or Æd ; defined
distinction between them is made here. As noted by later) and b were varied as controlled parameters are
Skempton et al. (1969), many ran in horizontal to sub- described in detail in this section; briefer comments are also
horizontal and vertical to sub-vertical directions. Although given on the simple shear and triaxial extension tests.
their spacing was difficult to quantify, most of the visible The driving systems in both the ICRCHCA and ICHCA II
discontinuities appeared to be separated by 2 to 5 cm. The incorporated both stress- and strain-controlling devices. Axial
52 NISHIMURA, MINH AND JARDINE
displacements and torsional rotations can be driven by cylinder samples were formed by placing the solid cylinders
stepper-motor-controlled, fluid-filled pumps with the ICHCA in a metal mould that was transferred to a metal working
II. The ICRCHCA has a stepper-motor-driven torsional rota- lathe. While the specimen was rotated at 190 rev/min, a set
tion drive, but relies on a pneumatic stress-controlling of six drill bits was advanced from one end. The process
Bellofram cylinder for axial loading. A robust feedback involved gradually increasing the bit diameters until the
system and algorithm was developed that could maintain desired inner diameter was achieved. For drier samples
prescribed axial strain rates and avoid post-peak instability (including all the block samples and deeper rotary core
with the potentially brittle London Clay. Satisfactorily linear samples), the friction between the inner clay surface and the
strain–time relationships were obtained with all types of larger-diameter drills tended to open natural discontinuities
servo system when operating under strain control. Nishimura and disturb the specimen. For such samples, the final drilling
(2006) gives data demonstrating these features, along with a stages were replaced by careful point-edge reaming to reach
flowchart detailing the ICRCHCA control algorithm. their final diameter.
The simple shear tests were conducted in the ICRCHCA
by testing hollow cylindrical specimens (as described by
Shibuya and Hight, 1987, and Pradhan et al., 1988) to
overcome some of the drawbacks inherent in conventional
simple shear apparatus. Strain-controlled loading was Reconsolidation scheme
achieved by applying torsional rotation through the stepper After being set up and saturated to an acceptable B value,
motor and gearing arrangement while prohibiting mechani- all test samples were reconsolidated to isotropic or anisotro-
cally axial movement of the ram and a volume change of pic effective stress states similar to the estimated in situ
the inner cell and the sample. The mechanical axial restraint conditions. The final reconsolidation stresses were p9 ¼ 323
was further assisted by a servo-system control to counteract kPa and (z  Ł )/2 (¼ (z  r )/2) ¼ 83 kPa for the
the apparatus compliance. The undrained triaxial tests were 10.5 m bgl HCA samples (test Series AC), corresponding to
conducted under axial strain control in the computer- K ¼  Ł9 = z9 ¼ 1.78. For the 5.2 m bgl HCA samples (test
controlled stress-path apparatus described by Gasparre Series AM), a wider range of p9 and (z  Ł )/2 values was
(2005), working mainly with specimens of 200 mm height employed, as shown in Table 2. The estimated OCRs under
and 100 mm diameter; reference is also made to extension these conditions are 9–12. Isotropic reconsolidation was first
tests on smaller samples, 76 mm by 38 mm. conducted to reach the desired p9 values, and then (z  Ł )/
2 was reduced, keeping p9 constant, to reach the eventual
reconsolidation points. The IC and IM series of tests in-
volved only isotropic reconsolidation, and the subsequent
Sample preparation undrained shearing was initiated from p9 ¼ 323 kPa and
All samples were trimmed to the specified outer diameter (z  Ł )/2 ¼ 0 kPa in Series IC and from p9 ¼ 280 kPa
and height with a wire saw in a rotary soil lathe. Hollow and (z  Ł )/2 ¼ 0 kPa in Series IM. In test NC4505, the

Table 2. Loading conditions in HCA stress-path test series

Series Depth 10.5 m bgl (tested in ICRCHCA) Series Depth 5.2 m bgl (tested in ICHCA II)

Test name Æpeak *: Æd : b Test name p90 : kPa (z  Ł )0 /2: Æ: degrees b
degrees degrees kPa

AC AC0000 0 0 0 AM AM0000 280 55 0 0


AC0005 0 0 0.5 AM4500 280 55 45 0
AC2305 23 15 0.5 AM0003 280 76 0 0.3
AC4505 48 30 0.5 AM0003B 280 80 0 0.3
AC6705 67 55 0.5 AM3003 280 82 30 0.3
AC9005 90 90 0.5 AM0005 312 80 0 0.5
AC4507 45 30 0.7 AM3005 312 80 30 0.5
AC4510 46 30 1 AM5005 312 80 50 0.5
AC6710 72 55 1 AM6005 332 108 60 0.5
AC9010 90 90 1 AM9005 312 80 90 0.5
IC IC0000 0 0 0 AM4510 280 78 45 1.0
IC2300 23 23 0 AM6010 280 70 60 1.0
IC4500 45 45 0 AM9010 280 75 90 1.0
IC4503 45 45 0.3 IM IM0005 280 0 0 0.5
IC0005 0 0 0.5 IM9005 280 0 90 0.5
IC2305 23 23 0.5
IC4505 45 45 0.5
IC6705 67 67 0.5 Series p90 : (z  Ł )0 /2: zŁ0 :
IC9005 90 90 0.5 kPa kPa kPa
IC9005B 90 90 0.5
IC9010 90 90 1 AC 323 83 0
IC 323 0 0
NC NC4505 45 45 0.5 NC 323 0 83

AM See above table 0

IM See above table 0

*The values at maximum q/p9.


p90 , (z  Ł )0 /2 and zŁ0 are p9, (z  Ł )/2 and zŁ at final reconsolidation points respectively.
SHEAR STRENGTH ANISOTROPY OF NATURAL LONDON CLAY 53
sample was reconsolidated to p9 ¼ 323 kPa, (z  Ł )/2 ¼ The general shear strength characteristics, and the defor-
0 kPa and zŁ ¼ 83 kPa (i.e. Æ ¼ 45o ). mation characteristics observed in HCA tests run at medium
The undrained triaxial extension and undrained simple to large strains on samples from 5.2 to 10.5m bgl, are briefly
shear tests were performed following a reconsolidation described below, presenting the key results. Because of the
scheme similar to that of the HCA samples, with the four degrees of stress and strain freedom provided by the
eventual K0 reconsolidation effective stress states simulating tests, a variety of 2D and 3D plots are required to fully
the conditions expected in situ at each sample depth describe the observed effective stress paths and stress–strain
(Nishimura, 2006). relationships. Nishimura (2006) and Minh (2006) provide
complete datasets for the tests on 10.5 m and 5.2 m bgl
samples respectively. However, it is necessary to concentrate
Shear scheme here on a restricted set of plots, covering the elements that
Slightly different shearing procedures were applied in the are essential to the subsequent discussion.
HCA test series. In each case sufficient time was allowed It is worth noting that the onset of rupture and shear
after the end of the consolidation stages for creep to stabilise. localisation could be detected during testing by inspecting
Typically this involved a pause of several days, waiting for the deformation of a grid pattern drawn on the outer
the rates of strain components to fall below 0.002%/h. In the membrane. Owing to the optical magnification effect caused
series starting from isotropic stress conditions (Series IC and by the cell water, a shear displacement of less than 0.2 mm
IM), undrained shearing was performed applying a nominal could be resolved by eye along any grid line, allowing the
shear strain rate of d(1  3 )/dt ¼ 4% per day while point of shear band formation to be identified before the
maintaining constant p, Æ and b values through the servo- shear strain had grown by 0.1%, or less, provided that
control arrangements. In the anisotropically consolidated continuous observations were made. Matching this potential
Series AC, b was changed from its initial value of 1 (K0 . 1) strain lag with the tangential stress–strain data suggests that
to a specified value under undrained conditions, keeping p, q q at rupture should not be overestimated by more than
and Æ constant. After allowing 12 h for creep, further un- 10 kPa in most cases. Continuous optical observations were
drained shearing was initiated with constant p, b and Æd made in Series AC, IC and NC, whereas inspection of the
values, where Æd is the major principal axis direction of shear surface formation was less frequent in the other series.
stress increment and defined in a similar way to equation (3). The test data before and after the onset of localisation of
It follows that Æ varied during undrained shearing in the shear deformation, or visible displacement along natural
Series AC tests except in the tests with Æd ¼ 08 and 908, as discontinuities (both are simply called rupture hereinafter),
illustrated in the inset in Fig. 3(a). The combinations of Æd are therefore distinguished only for Series AC and IC.
and b employed are listed in Table 2, which also shows the Æ The stress ratio–shear strain relationships, q/p9–(1  3 ),
values recorded at the eventual peak shear stresses. In the observed during the undrained shear stages of the HCA tests
other anisotropically consolidated Series, AM, the reconsoli- are shown in Fig. 4. Considering the data from Series AC
dation stage was followed by a drained change of b and then and IC, it can be seen that rupture generally preceded the
by an undrained change of Æ, except in tests AM0003, peak points in q and q/p9, whereas these two peaks were
AM0003B and AM3003, in which the Æ-change stage was practically concurrent in most of the tests (see Fig. 5). This
replaced by an undrained q reduction back to isotropic stress. feature may be absent with ductile materials but is often
Further undrained shearing was conducted with constant p, Æ seen in with brittle soils. Torsional shear modes (where 308
and b and increasing q. The selected combinations are shown , Æ , 608) provoked shear discontinuity formation at smal-
in Table 2. These shear schemes are illustrated in Fig. 3(a) ler shear strains than other shearing types. In many tests, the
for Series AC, IC and NC and in Fig. 3(b) for Series AM and post-peak behaviour did not exhibit a marked displacement-
IM. Note that the conditions (Æ, b) ¼ (08, 1) and (908, 0), and softening, although the latter trend is common in triaxial
their vicinities are the undesirable ‘no-go’ zones identified by tests on highly structured or plastic soils. The ‘post-rupture’
Symes (1983), where significant stress non-uniformity is strengths (Burland, 1990) were not well defined in these
expected across an HCA specimen. Note also that the appara- HCA tests, nor were residual strength conditions approached.
tus employed had wall thicknesses and height-to-diameter It should be borne in mind that an HCA test is in fact a
ratios designed to minimise stress non-uniformity (Rolo, non-uniform boundary value problem, as are most types of
2003). soil shear testing. The behaviour observed after shear surface
The triaxial extension and simple shear tests employed formation is a consequence of interactions between the
reconsolidation schemes, creep rate criteria and shearing intact soil, the soil within shear bands, and the apparatus.
strain rates similar to those outlined above. One difference The trajectories of the effective stress paths observed in
that applied to the 100 mm diameter undrained triaxial the above tests are shown in Fig. 5 in the (z  Ł )/2–p9
extension tests was the lower axial strain rate (0.02%/h) plane. In Fig. 5(b), the undrained effective stress paths
applied over the early stages of the test to help define the followed during the Æ-change stage are also included. Note
small-strain response. However, the axial strain rate was that the final reconsolidation effective stress in AM6005
increased to 0.1%/h (equivalent to d(1  3 )/dt ¼ 3.6% per (Æ ¼ 608) was slightly different from that of the rest of the
day) long before failure developed. dataset, and its effective stress path does not join the others.
The effective stress paths exhibited a common inclination
d(z  Ł )/d(2p9) up to the final dilative stage near failure,
ANISOTROPY OF LONDON CLAY FROM 10.5 m AND suggesting that the apparent ‘dilatancy’ is predominantly
5.2 m BGL a response to the (z  Ł )-mode deformation, and that
General behaviour and anisotropy of deformation ªzŁ -mode deformation has comparatively little effect. Such
characteristics deformation characteristics are similar to those of a cross-
Gasparre et al. (2007a, 2007b) and Hight et al. (2007) anisotropically elastic material with the axis of material
describe the characterisation of the full Heathrow T5 profile. symmetry set in the z-direction. Estimates for the effective
The following sections refer to intensive HCA studies made stress-path inclinations, d(z  Ł )/d(2p9), were computed by
on samples from two levels; a later section relates these Nishimura (2006) by applying cross-anisotropic elastic theo-
results to behaviour observed at other depths in simple shear, ry and small-strain stiffness parameters obtained experimen-
triaxial compression and triaxial extension tests. tally by Gasparre (2005). The calculated inclinations were
54 NISHIMURA, MINH AND JARDINE
τzθ τzθ
Stress path

NC4505

IC4500

2αdσ IC2300
(σz  σθ)/2
IC4503

IC0000
AC6705 IC4505
AC4505
IC6705 IC2305 b  0 AC0000

AC2305
AC4507
AC9005 IC9005 IC0005
AC4510
AC0005 b  0·5 (σz  σθ)/2
AC6710
No-go conditions
AC9010 (Symes, 1983)
IC9010
b1 Undrained
Drained
b
End of reconsolidation
(a)

τzθ
AM5005
AM6005 AM3005

AM4500
)
ge

AM3003
an
-ch

b0 AM0000
AM0003

AM9005 IM9005 IM0005 AM0005

b  0·5 (σz  σθ)/2


AM4510
AM6010 Undrained
AM9010 Drained
b1 End of reconsolidation

b
(b)

Fig. 3. Undrained shear schemes in HCA stress-path test series: (a) Series AC, IC and
NC; (b) Series AM and IM (end of reconsolidation points not the same for all tests; see
Table 2)

found to be consistent, if not always quantitatively coinci- Bishop et al., 1965; Hight & Jardine, 1993) and the T5 data
dent, with the observed values. The general trend of pre- presented by Gasparre et al. (2007a) and Hight et al. (2007)
failure ‘dilative’ behaviour for extensional loading (Æd . indicate curved peak q–p9 envelopes and hence a peak q/p9
458) and ‘contractive’ behaviour for compressional loading ratio as a function of p9. Although similar trends may well
(Æd , 458) is one of the causes for the anisotropy of the apply to the HCA experiments, the authors’ programmes did
clay’s undrained shear strength, as discussed below. The not include multiple tests where non-triaxial Æ and b values
above observations applied irrespective of the particular b were maintained while the p9 levels were varied. Noting that
value applied in each test. the range of p9 values at which the HCA specimens failed is
relatively narrow (for example, 338 kPa  23% in Series
AC and 361  20% in Series IC), one approach would be to
Anisotropy in undrained shearing resistance neglect potential envelope curvature (and perhaps true effec-
Accurate representation of the observed shear strength in tive cohesion) and denote the shear strength by q/p9. In order
terms of effective stress was hindered by the fact that to avoid a biased interpretation caused by this simplification,
samples failed at different p9 values in different tests, with the data are also reported in total stress terms by plotting
the p9 values at peak strength varying systematically with peak Su against Æ. These two parameters, q/p9 and Su ,
consolidation K value, b and Æ. Earlier triaxial testing (e.g. represent the shear strength characteristics of two types of
SHEAR STRENGTH ANISOTROPY OF NATURAL LONDON CLAY 55
1·5 1·5

αpeak  0°
αpeak  0°, b  0*
1·0 1·0
90° 23°
67° 45°, 0·7
q/p

q/p
48° 46°, 1
72°, 1
90°, 1
0·5 0·5

Before rupture Before rupture


After rupture After rupture

0 0
0 0·02 0·04 0·06 0·08 0 0·02 0·04 0·06 0·08
ε1  ε3 (redefined after b change) ε1  ε3 (redefined after b change)
(a) (b)

1·5 1·5

90° (IC9005B)

αpeak  0°, b  0*
23°, 0
1·0 23° 1·0

45° 90° 45°, 0·3


q/p

q/p
(IC9005) 45°, 0
67°
90°, 0
0·5 0·5

Before rupture Before rupture


After rupture After rupture
0 0
0 0·02 0·04 0·06 0·08 0 0·02 0·04 0·06 0·08
ε1  ε3 ε1  ε3
(c) (d)

1·5

α  0°
30°
1·0
60°
q/p

50°

0·5 90°

Accidental 4-hour pause


and creep
0
0 0·02 0·04 0·06 0·08
ε1  ε3 (redefined after α change)
(e)

Fig. 4. Observed stress ratio–shear strain relationships: (a) 10.5 m bgl, Series AC, b 0.5; (b) 10.5 m bgl, Series AC, other b values
(b value changed from 1 to 0 stepwise during shear); (c) 10.5 m bgl, Series IC, b 0.5; (d) 10.5 m bgl, Series IC, other b values; (e)
5.2 m bgl, Series AM, b 0.5

idealised material: namely, a purely frictional material with some 40% as Æ increases from 08 to 458. A similar trend is
constant 9 (for any given b and Æ), and a purely cohesive followed by the 5.2 m bgl samples for this Æ range, but with
material where shear strength is independent of p9 respec- higher effective stress ratios. The slightly higher q/p9 ratios
tively. These two perspectives bracket the available possibi- developed by the shallower samples for compressional load-
lities when assessing the anisotropic shear strength trends; ing modes are compatible with the general failure envelope
the true (effective) stress-dependence of anisotropy remains curvature, shown by the London Clay at T5, reflecting
to be established. variations with depth of OCR, and possibly of lithology
The peak stress ratios, q/p9, observed for b ¼ 0.5 are (Gasparre et al., 2007a; Hight et al., 2007). For 458 , Æ ,
plotted in Figs 6(a) and 6(b) against the Æ values applying 908, the 10.5 m bgl samples exhibited peak q/p9 rising again
at failure for 10.5 m and 5.2 m bgl respectively. The shear at larger Æ values, but this pattern was not established from
strength anisotropy is clearly recognisable in each case. The the limited number of tests on 5.2 m bgl samples. The low
AC and IC series tests on 10.5 m bgl samples exhibit close shear strengths recorded for Æ ¼ 908 in Series AM and IM
agreement for 08 , Æ , 458, with peak q/p9 decreasing by may be due partly to the larger specimen size in the ICHCA
56 NISHIMURA, MINH AND JARDINE
200 200
α  0°
α  0°

Accidental 4-hour pause


100 23° 100 and creep
30°

(σz  σθ)/2: kPa 0 0

(σz  σθ)/2: kPa


45° 50°

60°
100 67° 100
90°

90°
(IC9005)
200 200
End of reconsolidation
Before rupture
90° End of α-change
After rupture
(IC9005B)
300 300
200 300 400 500 200 300 400 500
p: kPa p: kPa
(a) (b)

200 200
αpeak  0°, b  0
αpeak  0°, b  0

100 100 23°, 0


(σz  σθ)/2: kPa
(σz  σθ)/2: kPa

45°, 0·7
0 0
46°, 1

45°, 0
100 72°, 0·7 100 and 45°, 0·3

90°, 1

90°, 1
200 200

200 300 400 500 200 300 400 500


p: kPa p: kPa
(c) (d)

Fig. 5. Observed undrained effective stress paths: (a) Series AC and IC, b 0.5; (b)
Series AM, b 0.5; (c) Series AC, other b values; (d) Series IC, other b values

II and hence the inclusion of more complete discontinuity natural samples and of how these developed during shearing,
network within the specimen. A sample-size effect might showing in many cases a correlation between the original
also have led to the apparently lower shear strengths at high discontinuity patterns, the final shear mechanisms and the
Æ values of the larger specimens (Series AM) tested with b shear strength.
¼ 1, shown in Fig. 6(c), but any such effect was absent in It is noticed in Fig. 6(c) that test AM4500 gave extra-
the tests with Æ , 458 at all b values. Such a directionally ordinarily high shear strength in relation to the other data
dependent sample-size effect might result from the abundant points. The reason for this was not established by re-
sub-horizontal natural discontinuities, which were often acti- examination of the physical properties, the test specimen
vated as part of the torsional and extensional shear failure itself and the failure mode. The only possible explanation at
modes. It is worth noting that the test IC9005B (Æ ¼ 908 the moment is that the sample may have contained an
and b ¼ 0.5) developed a single spiral shear plane upon untypical local concentration of cementing and/or a lower
shearing, indicative of bifurcation in intact clay, whereas the degree of chemical alteration by pyritisation.
equivalent tests on 5.2 m bgl samples (AM9005 and Figures 6(d), 6(e) and 6(f) show the anisotropy of the
IM9005) simply failed along a natural discontinuity. undrained shear strength Su . In interpreting these results it
Nishimura (2006) and Minh (2006) made careful observa- should be borne in mind that, with brittle soils such as the
tions of the patterns of discontinuities that existed in the London clay, the peak undrained shear strengths tend to be
SHEAR STRENGTH ANISOTROPY OF NATURAL LONDON CLAY 57
2·0 2·0
10·5 m bgl (Series AC) 5·2 m bgl (Series AM)
10·5 m bgl (Series IC) 5·2 m bgl (Series IM)
1·5 1·5
Peak q/p

Peak q/p
1·0 1·0

0·5 0·5 Series AC and IC, b  0·5


5·2 m bgl (12·5 mOD): CAU, b  0·5
Performed in ICRCHCA
0 0
0 30 60 90 0 30 60 90
α: degrees α: degrees
(a) (b)
2·0 300
10·5 m bgl

Undrained shear strength, Su: kPa


b  0 (Series IC)
1·5 b  0 (Series AC)
b  1 (Series AC)
AM4500 200
Peak q/p

1·0

100
5·2 m bgl
0·5
b  0 (Series AM)
b  0·3 (Series AM) 10·5 m bgl (Series AC)
b  1 (Series AM) 10·5 m bgl (Series IC)
0 0
0 30 60 90 0 30 60 90
α: degrees α: degrees
(c) (d)

300 300
5·2 m bgl (12·5 mOD): CAU, b  0·5 10·5 m bgl
Undrained shear strength, Su: kPa

Undrained shear strength, Su: kPa

b  0 (Series IC)
Performed in ICRCHCA
b  0 (Series AC)
200 b  1 (Series AC)
200
AM4500

100 Series AC and IC, b  0·5 100


5·2 m bgl
b  0 (Series AM)
5·2 m bgl (Series AM)
b  0·3 (Series AM)
5·2 m bgl (Series IM)
b  1 (Series AM)
0 0
0 30 60 90 0 30 60 90
α: degrees α: degrees
(e) (f)

Fig. 6. Anisotropy of peak shear strength: (a) b 0.5, 10.5 m bgl; (b) b 0.5, 5.2 m bgl; (c) Other b values; (d) b 0.5,
10.5 m bgl; (e) b 0.5, 5.2 m bgl; (f) Other b values

reached shortly after engaging their failure envelopes with- for b ¼ 1 show little variation in peak q/p9 or Su in the case
out showing extended periods of dilative behaviour. Hence where Æ . 458.
Su depends on the effective stress state under which the
envelope is first engaged, and is not uniquely related to
water content or void ratio (e.g. Jardine et al., 2004). There- Influence of reconsolidation conditions on shear strength
fore a soil with a fixed failure envelope and a fixed water anisotropy
content could exhibit a variety of Su values, depending on its Jardine et al. (1997) and Zdravkovic & Jardine (2001)
initial effective stress states and the effective stress paths demonstrated that the shear strength anisotropy of low-OCR
followed towards failure. Considering this fact, the anisotro- granular soils depends strongly on their consolidation his-
pically reconsolidated test results (i.e. Series AC and AM) tory, especially when this involved rotation of the 1 axis. It
are the most relevant for estimating the in situ undrained was important to assess whether similar features applied to
shear strength anisotropy. the London Clay. Figs 7(a) and 7(b) show the rupture (i.e.
As noted above, extensional loading with Æ . 458 resulted onset of shear strain localisation) and failure envelopes
in a more dilative pore water pressure response, which obtained from test Series IC, AC and NC with b ¼ 0.5
affects the Su anisotropy and leads to a significantly different plotted in (z  Ł )/2p9–zŁ /p9 space. Such a space de-
pattern from the q/p9 trend. It is interesting that all the data scribed by tensorial stresses offers a simpler and clearer
58 NISHIMURA, MINH AND JARDINE
Series IC Final points after cally with Æ, giving Su,Ƽ90 /Su,Ƽ0 ¼ 0.3–0.6. However, the

τzθ/p
Series AC reconsolidation upper data curve for the London Clay, obtained from Series
0·5
Series NC AC, follows the opposite trend, with the Su,Ƽ90 /Su,Ƽ0 ratio
being as high as 1.5. Whereas this upper bound is consid-
ered to reflect the influences of both the micro- and macro-
NC fabric, the lower data curve obtained from Series AM may
reflect more the discontinuous (highly fissured) macrofabric.
Heavy overconsolidation (i.e. OCR . 9) appears to change
AC IC the microstructural anisotropy from that seen at low OCR,
0 while discontinuities developed (most probably during or
0·5 0 0·5
(σz  σθ)/2p after the overconsolidation) modify the anisotropy en masse
(a) further. A further investigation is currently being undertaken
by testing reconstituted London Clay samples at OCRs
45° comparable to those of the natural samples.
τzθ/p

67° 0·5 23°

NC INFLUENCE OF b ON PEAK SHEAR STRENGTH


The influence of b on the shape of the failure surface
projected onto the  plane has been identified as an impor-
90° AC IC α  0° tant but often neglected factor in geotechnical analysis (e.g.
0 Potts, 2003). This feature for the London Clay was investi-
0·5 0 0·5
(σz  σθ)/2p gated by plotting the peak q/p9 points on the  plane in Fig.
(b) 9 for groups of tests with common Æ ranges. Although the
axes are normalised by p9 here, it should be noted that the
Fig. 7. Influence of reconsolidation regimes on development of pore water pressure development was predominantly a func-
shear strength: (a) rupture; (b) peak shear strength tion of Æd , and was affected much less by b. As a result, p9
at failure fell within a narrow range for a given Æpeak value.
For example, the failure p9 range for the six data points
view of the distortion and kinematics of the failure envelope from Series IC and AC in Fig. 9(a) was 303–343 kPa. Tests
than the q/p9–Æ plot. Distortion of the rupture envelope with Æ ¼ 45–488 were performed covering the entire range
towards the reconsolidation effective stress points is clearly of b, from 0 to 1. As discussed above, the peak shear
observed. However, the peak strength surface was less strength for torsional shear may not be significantly affected
affected, with only two data points at Æ ¼ 678 and 908, by the K values applied during triaxial reconsolidation. No
suggesting the possibility that reconsolidation affected the distinction is therefore made here between the data from
peak shear strength surface. Series AC and IC. Also shown as references in Fig. 9(a) are
the Mohr–Coulomb and Prager–Drucker (corresponding to
(s ij s ij )1=2 /p9 ¼ constant, where s ij ¼ ij   ij  kk /3) envel-
opes, fitted to pass through the b ¼ 0 test data. The former
Influence of overconsolidation and natural soil structure was thought to apply best to the natural London Clay by
It is of interest to compare the anisotropy of the natural Bishop (1966), whereas the latter is preferred in some
London Clay with the more familiar patterns known for low- critical state soil model variants. The Mohr–Coulomb failure
OCR reconstituted soils. Fig. 8 shows the Su anisotropy of line for 9 ¼ 23.58 and c9 ¼ 0, fitted at b ¼ 0, gives a well-
the natural London Clay shown earlier for b ¼ 0.5, normal- defined lower bound of shear strength for Æ ¼ 45–488,
ised by Su at Æ ¼ 08 and set against a range of results from although some data points lie well above this line. Mohr–
previously published tests on low-OCR (OCR ¼ 1–4) K0 - Coulomb failure lines drawn with c9 ¼ 0 and passing
reconstituted soils. They include clay, silt and clayey and through either the b ¼ 0 or 1 data points are presented in
silty sands (Whittle et al., 1994; Menkiti, 1995; Zdravkovic, Figs 9(b) to 9(e) for other Æ ranges. Although the datasets
1996). Low-OCR soils generally exhibit Su falling monotoni- are less complete, the Mohr–Coulomb lines again provide
lower bounds of shear strength, whereas b ¼ 0.5 conditions
tended to result in development of higher 9. It is difficult,
2·0
however, to propose specific curved envelopes that encapsu-
Natural London Clay late the potential influence of b based on these limited data.
10·5 m bgl (Series AC: 70 mm dia.)
5·2 m bgl (Series AM: 100 mm dia.)
In the absence of further data, an anisotropic Mohr–
1·5
Coulomb criterion calibrated to data obtained at b ¼ 0 or 1
may be regarded as a reasonably safe option, although not
Su/Su,α0

the best fit.


1·0 Finally, the influences of Æ and b on the peak shear
strength, q/p9, are summarised in Figs 10(a) and 10(b) for
10.5 m and 5.2 m bgl samples respectively. As discussed
0·5 later, the shear strength exhibited in test AM4500 was
Typical Range for low-OCR soils exceptionally high, and this outlying data point is excluded
(b  0·5 or plane strain) from Fig. 10(b). Despite the detailed differences between the
0 two horizons discussed above, the overall shapes are gener-
0 30 60 90
α: degrees
ally similar. In the range explored, the highest peak value of
q/p9 is available for Æ ¼ 0 with 0 , b , 0.5, and the lowest
Fig. 8. Undrained shear strength anisotropy of low-OCR K0 - is encountered for Æ ¼ 458 with b ¼ 0.5 or Æ ¼ 908 and
reconstituted soils and natural London Clay b ¼ 1.
SHEAR STRENGTH ANISOTROPY OF NATURAL LONDON CLAY 59
AM4500
q/p Isotropically reconsolidated
Anisotropically reconsolidated
1·5
σ1/p

1·0

(sijsij)1/2/p  0·53
0·5
1 .0

0.5
b
Mohr-Coulomb
0
(φ  23·5°, c  0) 22
.5
α: 45
de
gre 67
es .5
90
Lode's angle, θ (a)
Series AC 30°

Series IC 30° q/p Isotropically reconsolidated
Series AM Anisotropically reconsolidated
Series IM 1·5
σ2/p σ3/p

(a)
1·0
σ1/p σ1/p

0·5
1 .0

0.5
b
φ  32° φ  26°
0
22
.5
α: 45
de
gre 67
σ2/p σ3/p σ3/p es .5
σ2/p
(b) (c) 90
σ1/p σ1/p (b)

Fig. 10. Influence of Æ and b on peak shear strength: (a) 10.5 m


bgl (7.2 mOD); (b) 5.2 m bgl (12.5 mOD)
φ  28°
φ  27°
φ  23°
φ  20° comparison of the shear strength in these three different
shear modes is therefore subject to mixed effects of Æ and b.
Nevertheless, compilation of such test data provides useful
σ2/p σ3/p σ2/p σ3/p benchmarks for the possible variations in anisotropy through-
(d) (e) out the stratum. The water contents and index properties of
the samples tested are included in the overview given in the
Fig. 9. Peak shear strength points plotted on -plane for five Æ companion paper (Hight et al., 2007), and are omitted here
ranges: (a) Æ 45–488; (b) Æ 08; (c) Æ 23–308; (d) Æ 60– for reasons of space.
728; (e) Æ 908
The effective stress paths observed during simple shear
and triaxial extension tests are shown in Fig. 11. Generally,
p9 changed little in the simple shear tests before engaging
PEAK SHEAR STRENGTH ANISOTROPY FOR 6–35 m the final failure surface and experiencing accelerated p9
BGL increases, whereas the pre-failure effective stress paths were
The intensive investigation of the shear strength aniso- consistently ‘dilative’ (p9 increasing) in the triaxial extension
tropy developed at 5.2 m bgl and 10.5 m bgl was extended tests, features that are consistent with the small-strain stiff-
through sets of triaxial compression, triaxial extension and ness anisotropy, as described earlier. The points of maximum
HCA simple shear tests conducted between 6 and 35 m bgl. q/p9 are plotted in Fig. 12, set against the envelopes for
Although the (Æ, b) combinations in triaxial compression triaxial compression originally proposed by Hight & Jardine
and extension are fixed to (08, 0) and (908, 1) respectively, (1993) for high-quality rotary cores of the London Clay,
neither variable can be pre-specified in strain-controlled mainly from central London and partly from Suffolk, slightly
simple shear tests. In the present cases of the London Clay modified here according to Gasparre (2005). The potential
simple shear tests, the peak shear stress q was reached at curvature of the failure envelopes mentioned earlier is
508 , Æ ,658 and 0.5 , b , 0.7 (Nishimura, 2006). The evident here, along with variations due to changes in depth
60 NISHIMURA, MINH AND JARDINE
200 Undrained shear strength, Su: kPa
Before rupture 0 200 400 600 0 200 400 600
0
After rupture Triaxial compression, Gravel
τzθ: kPa

20
100 mm
100 Simple shear C

Depth below ground level: m


6–35 m bgl 10
Upper bound 10
for TC

Elevation: mOD
0
0 200 400 600 20 B2
p: kPa 0
0 200 400 600 Lower
0
Triaxial extension 30 B1 and
(σzσr)/2: kPa

(100 mm diameter) upper 10


6–30 m bgl bounds
100
40 for TE&SS
A3
Triaxial extension, 20
Lower bound 100 mm
200 50 for TC Triaxial extension,
A2 70 mm
Fig. 11. Effective stress paths observed for HCA simple shear Triaxial extension,
and triaxial extension tests (only data for 100 mm diameter 38 mm
samples are shown) Simple shear,
70 mm

Fig. 13. Undrained shear strength profiles of three different


600 shear modes; triaxial compression data from Gasparre (2005).
m
0m

40 Triaxial ext. (100 mm dia.) The T5 lithological stratigraphy described by Hight et al. (2007)
– after Gasparre (2005)
27 is shown in the central column between the two Su profiles
–6
50

400 m the other shear modes. Considering that the simple shear
24
16– tests were performed with 70 mm diameter hollow speci-
q: kPa

mens, whereas the rest were typically performed with


100 mm diameter solid specimens, allowing for any sample-
200 m size effect would further accentuate the anisotropic hierarchy
12
8– Triaxial ext. (100 mm dia.) in Su values.
1) Triaxial ext. (71/38 mm HC) Despite the apparent increases in shear strength anisotropy
(b 
17° Triaxial ext. (38 mm dia.)
φ  Simple shear (71/38 mm HC)
with depth, the overall hierarchy, (triaxial compression) .
(simple shear) . or  (triaxial extension), is fully compa-
0
0 200 400 600 tible with the finding from the HCA investigation in Fig. 10.
p: kPa

Fig. 12. Comparison of triaxial extension and simple shear CONCLUSIONS


strength with triaxial compression envelopes given by Hight & The present study was undertaken to investigate the mech-
Jardine (1993)
anical anisotropy of the natural London Clay through a
comprehensive programme of laboratory tests performed in
HCAs and stress-path triaxial cells. This paper focused only
and structure. Multiple en-echelon upper-bound envelopes on the peak shear strength; a companion paper (Gasparre et
were employed to account for changes in dilatancy and an al., 2007b) discusses the anisotropy found at smaller strains.
apparently increasing degree of cementation in the deeper The directional dependence of q/p9 or Su was evaluated at
samples. Hight et al. (2003) and Gasparre (2005) found that two fixed depths (5.2 m and 10.5 m below ground level) by
this set of envelopes broadly applied to triaxial compression multiple tests involving fixed values of the intermediate
strength data for T5 samples. The fissure shear strength line principal stress ratio b. Although the patterns for these two
(9 ¼ 178 and c9 ¼ 0) given also by Hight & Jardine (1993) measures differ because anisotropic deformation character-
provides a lower bound for all test types. The deepest istics lead to direction-dependent development of pore water
samples from 35 m bgl, tested with p9 . 400 kPa, show the pressure up to failure, both q/p9 and Su typically exhibited
largest gap between the data points and the 27–40 m bgl their maxima at either Æ ¼ 08 or 908 and minima at around
triaxial compression envelope, indicating that the anisotropy Æ ¼ 458 in tests with b ¼ 0.5. The effects of anisotropy led
appears to grow with depth. This is particularly the case to maximum and minimum q/p9 values that could differ by
with the simple shear, which gave strength closer to the 40%. This feature contrasts strongly with the behaviour of
triaxial compression upper bound in shallower horizons. This low-OCR K0 -reconstituted soils, for which shear strength
finding may be related to the SEM observation made by generally decreases monotonically with Æ from 08 to 908.
Gasparre (2005) that the deeper clay had more uniform The tests with larger samples suggested the possibility that
particle orientation. As discussed in the companion paper the shear strength can be equally very low against exten-
(Gasparre et al., 2007b), the small-strain stiffness also sional loading (i.e. Æ . 458), if samples contain an intense
exhibited stronger anisotropy in the deeper samples. discontinuity network. For all the combinations of Æ and b,
The undrained shear strength profiles obtained with the excluding the ‘no-go’ combinations in HCA testing, the
three shear modes are shown in Fig. 13. The triaxial com- lowest peak shear strength was found at (Æ, b) ¼ (458, 0.5)
pression data are quoted from Gasparre (2005). Reflecting and (908, 1), with another possible minimum at (908, 0.5).
the above discussion, the shear strengths in the three differ- Profiles of triaxial extension, compression and simple shear
ent shear modes are not significantly different for shallow tests on rotary cores from 6 m to 35 m bgl indicated the
samples, but the upper bound of the triaxial compression same hierarchy of shear strengths and also suggested an
strength below 10 m bgl locates substantially above those for increase in the degree of anisotropy with depth. The top 5 m
SHEAR STRENGTH ANISOTROPY OF NATURAL LONDON CLAY 61
layer of the London Clay stratum appeared to be much less Burland, J. B. (1990). On the compressibility and shear strength of
anisotropic than the layers below, particularly when ex- natural clays. Géotechnique 40, No. 3, 329–378.
pressed in terms of Su . Burland, J. B., Rampello, V. N., Georgiannou, V. N. & Calabresi, G.
The influence of b on the shear strength was less well (1996). A laboratory study of the strength of four stiff clays.
defined. For a given Æ value, the Mohr–Coulomb failure Géotechnique 46, No. 3, 491–514.
Chandler, R. J. & Apted, J. P. (1988). The effect of weathering on the
criterion fitted at either b ¼ 0 or 1 seems to provide a strength of London Clay. Q. J. Engng Geol. 21, No. 1, 59–68.
reasonable lower-bound shear strength envelope, although Gasparre, A. (2005). Advanced laboratory characterization of Lon-
higher strengths were generally mobilised at intermediate don Clay. PhD thesis, Imperial College London (downloadable
values of b. from www.imperial.ac.uk/geotechnics).
Gasparre, A., Nishimura, S., Coop, M. R. & Jardine, R. J. (2007a).
The influence of structure on the behaviour of London Clay.
ACKNOWLEDGEMENTS Géotechnique 57, No. 1, 19–31.
The authors express their gratitude to their former collea- Gasparre, A., Nishimura, S., Minh, N. A., Coop, M. R. & Jardine,
gue Dr Akihiro Takahashi (now of PWRI, Japan) and the R. J. (2007b). The stiffness of natural London Clay. Géotechni-
que 57, No. 1, 33–47.
Imperial College laboratory technicians, Steve Ackerley,
Hight, D. W., Gens, A. & Symes, M. J. P. R. (1983). The develop-
Graham Keefe and Alan Bolsher. Dr Apollonia Gasparre is ment of a new hollow cylinder apparatus for investigating the
thanked for providing her triaxial test data and Cedric effects of principal stress rotation in soils. Géotechnique 33, No.
Allenou for his help with block sampling at Heathrow T5. 4, 355–384.
The research was funded by EPSRC, British Airports Hight, D. W. & Jardine, R. J. (1993). Small-strain stiffness and
Authority and London Underground Ltd. The second author strength characteristics of hard London tertiary clays. In Geo-
was also funded by the Vietnamese Government Overseas technical engineering of hard soils and soft rocks (eds A.
Scholarship Programme Project 322. Anagnostopoulos, F. Schlosser, N. Kalteziotis & R. Frank), Vol.
1, pp. 533–552. Rotterdam: Balkema.
Hight, D. W., McMillan, F., Powell, J. J. M., Jardine, R. J. and
Allenou, C. P. (2003). Some characteristics of London Clay. In
Characterisation of engineering properties of natural soils (ed.
NOTATION T. S. Tan), pp. 851–908. Rotterdam: Balkema.
B Skempton’s pore water pressure coefficient Hight, D. W., Gasparre, A., Nishimura, S., Minh, N. A., Coop, M.
b intermediate principal stress ratio R. & Jardine, R. J. (2007). Characteristics of the London Clay
c9 effective cohesion from the Terminal 5 site at Heathrow Airport. Géotechnique 57,
Gs specific gravity No. 1, 3–18.
GzŁ shear modulus Jardine, R. J. (1996). Development of the new HCA, internal report.
K coefficient of earth pressure London: Imperial College.
K0 coefficient of earth pressure at rest Jardine, R. J. & Smith, P. R. (1991). Evaluating design parameters
p mean normal stress for multi-stage construction. Proc. Int. Conf. on Geotechnical
p9 mean normal effective stress Engineering for Coastal Development, Geo-coast ’91, Yokosuka,
q deviatoric stress 1, 197–202.
Su undrained shear strength Jardine, R. J., Zdravkovic, L. & Porovic, E. (1997). Anisotropic
w natural water content consolidation including principal stress axis rotation: experi-
Æ angle between major principal stress axis and vertical ments, results and practical implications. Proc. 14th Int. Conf.
Æd  angle between major principal stress increment axis Soil Mech. Found. Engng, Hamburg, 4, 2165–2168.
and vertical Jardine, R. J., Gens, A., Hight, D. W. & Coop, M. R. (2004).
Æpeak value of Æ at maximum q/p9 Developments in understanding soil behaviour. Advances in
ª unit weight Geotechnical Engineering, Proceedings of the Skempton Confer-
ªzŁ shear strain in polar coordinate system ence, Vol. 1, pp. 103–206. London: Thomas Telford.
 ij Kronecker delta King, C. (1981). The stratigraphy of the London Clay and asso-
1 , 3 principal strains ciated deposits. Tertiary Research Special Paper, No. 6.
Ł Lode’s angle Kovacevic, N., Hight, D. W. & Potts, D. M. (2004). Temporary
1 , 2 , 3 principal stresses slope stability in London Clay. Advances in Geotechnical En-
ij stress tensor gineering, Proceedings of the Skempton Conference, Vol. 2, pp.
z , r , Ł normal stresses in cylindrical coordinate system 842–855. London: Thomas Telford.
zŁ shear stress in cylindrical coordinate system Menkiti, C. O. (1995). Behaviour of clayey-sand, with particular
9 angle of shearing resistance reference to principal stress rotation. PhD thesis, Imperial
College, University of London.
Minh, N. A. (2006). An investigation of the anisotropic stress–
strain–strength characteristics of an Eocene clay. PhD thesis,
REFERENCES Imperial College London (downloadable from www.imperial.ac.
Agarwal, K. B. (1968). The influence of size and orientation of uk/geotechnics).
sample on the undrained strength of London Clay. PhD thesis, Molenkamp, F. (1998). Principle of axial shear apparatus. Géotech-
Imperial College, University of London. nique 48, No. 3, 427–431.
Bishop, A. W. (1966). The strength of soils as engineering materi- Nishimura, S. (2006). Laboratory study on anisotropy of natural
als. Géotechnique 16, No. 2, 91–130. London Clay. PhD thesis, Imperial College, University of
Bishop, A. W., Webb, D. L. & Lewin, P. I. (1965). Undisturbed London (downloadable from www.imperial.ac.uk/geotechnics).
samples of London Clay from the Ashford Common shaft: Porovic, E. (1995). Investigation of soil behaviour using a resonant-
strength-effective stress relationships. Géotechnique 15, No. 1, column torsional shear hollow-cylinder apparatus. PhD thesis,
1–31. Imperial College, University of London.
Bishop, A. W. & Little, A. L. (1967). The influence of the size and Potts, D. M. (2003). Numerical analysis: a virtual dream or practical
orientation of the sample on the apparent strength of the London reality? Géotechnique 53, No. 6, 535–573.
Clay at Maldon, Essex. Proceedings of the geotechnical confer- Potts, D. M., Kovacevic, N. & Vaughan, P. R. (1997). Delayed
ence, Oslo, Vol. 1, pp. 89–96. collapse of cut slopes in stiff clay. Géotechnique 47, No. 5,
Bjerrum, L. (1973). Problems of soil mechanics and construction 953–982.
on soft clays and structurally unstable soils (collapsible, expan- Pradhan, T. B. S., Tatsuoka, F. & Horii, N. (1988). Simple shear
sive and others). Proc. 8th Int. Conf. Soil Mech. Found. Engng, testing on sand in a torsional shear apparatus. Soils Found. 28,
Moscow 3, 111–159. No. 2, 95–112.
62 NISHIMURA, MINH AND JARDINE
Rolo, R. (2003). The anisotropic stress–strain–strength behaviour Ward, W. H., Marsland, A. & Samuels, S. G. (1965). Properties of
of brittle sediments. PhD thesis, Imperial College, University of the London clay at the Ashford Common shaft: in-situ and
London (downloadable from www.imperial.ac.uk/geotechnics). undrained strength tests. Géotechnique 15, No. 4, 321–344.
Saada, A. S. (1970). Testing of anisotropic clay soils. J. Soil Mech. Whittle, A. J., DeGroot, D. J., Ladd, C. C. & Seah, T.-H. (1994).
Found. Div. ASCE 96, No. SM5, 1847–1852. Model prediction of anisotropic behavior of Boston Blue Clay.
Saada, A. S. & Townsend, F. C. (1981). State-of-the-art: Laboratory J. Geotech. Engng Division, ASCE 120, No. 1, 199–224.
strength testing of soils. In Laboratory shear strength of soil, Zdravkovic, L. (1996). The stress–strain–strength anisotropy of a
ASTM STP 740, pp. 7–77. granular medium under general stress conditions. PhD thesis,
Shibuya, S. & Hight, D. W. (1987). On the stress path in simple Imperial College, University of London.
shear. Géotechnique 37, No. 4, 511–515. Zdravkovic, L. & Jardine, R. J. (2001). The effect on anisotropy of
Skempton, A. W., Schuster, R. L. & Petley, D. J. (1969). Joints and rotating the principal stress during consolidation. Géotechnique
fissures in the London Clay at Wraysbury and Edgware. Géo- 51, No. 1, 69–83.
technique 19, No. 2, 205–217. Zdravkovic, L., Potts, D. M. & Hight, D. W. (2002). The effect of
Symes, M. J. P. R. (1983). Rotation of principal stresses in sand. strength anisotropy on the behaviour of embankments on soft
PhD thesis, Imperial College, University of London. ground. Géotechnique 52, No. 6, 447–457.

S-ar putea să vă placă și