Soils and Foundations 2013;53(2):315–334

The Japanese Geotechnical Society

Soils and Foundations

www.sciencedirect.com
journal homepage: www.elsevier.com/locate/sandf

Understanding sampling disturbance and behaviour of structured clays

through constitutive modelling
Gianfrancesco Rocchia,1, Giovanni Vaciagoa,n, Maurizio Fontanab,2, Monica Da Prata,1
a
Studio Geotecnico Italiano srl, via Ripamonti 89, 20139 Milano, Italy
b
Autosoft di ing. M. Fontana & C. sas, via XXV Aprile, 21047 Saronno (VA), Italy
Received 29 August 2011; received in revised form 5 January 2013; accepted 19 January 2013
Available online 29 March 2013

Abstract

The effects of sampling disturbance on structured clays with initial states above the oedometric Intrinsic Compression Line (ICLoe)
are investigated numerically by applying a modified form of the Ideal Sampling Approach (ISA) and an elasto-viscoplastic constitutive
model. The main features of the behaviour of structured clays during and after sampling reported in the technical literature can be
replicated. Notwithstanding the inevitable damage to the microstructure of the ‘‘nominally undisturbed’’ specimens, the reconsolidation
procedures commonly applied in laboratory practice are confirmed to be beneficial, even where they are not conceptually justified, as is
the case of the SHANSEP approach. While it is possible to determine the strength and compressibility of ‘‘ideal’’ specimens from
numerical interpretation of laboratory tests on ‘‘nominally undisturbed’’ specimens, the problems related to their direct use in
conventional simplified design approaches for stability and settlement predictions are highlighted.
& 2013 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Clay; Consolidation; Microstructure; Visco-plasticity; Yield; Sampling; Disturbance

1. Introduction changes of deviatoric stress during sampling and to

Disturbance due to tube sampling in saturated natural
clays is thought to derive from a combination of damage
to the microstructure (fabric and bonding) and variation in
mean effective stress compared to geostatic conditions due
to excess pore pressures induced by undrained cyclic
n
Corresponding author. Tel.: þ 39 02 5220141; fax: þ39 02 5691845.
E-mail addresses: g.rocchi@studiogeotecnico.it (G. Rocchi),
g.vaciago@studiogeotecnico.it (G. Vaciago),
autosoft@tiscalinet.it (M. Fontana),
m.daprat@studiogeotecnico.it (M. Da Prat).
1
Tel.: þ39 02 5220141; fax: þ39 02 5691845.
2
Tel/fax: þ 39 02 96248492.
Peer review under responsibility of The Japanese Geotechnical Society.
drained swelling, at equal natural water content (wo),
associated with a progressive decrease of the degree of
saturation, mainly after sample extrusion (Skempton and
Sowa, 1963; Ladd and Lambe, 1963; Kirkpatrick and
Khan, 1984; Graham and Lau, 1988; Tanaka, 2000;
Watabe and Tsuchida, 2001; Hight, 2001; Santagata and
Germaine, 2002; Tanaka and Tanaka, 2006).
These processes start during drilling and continue during
penetration and extraction of the sampler, transport to the
laboratory, storage, extrusion of the sample, specimen
preparation and assembly in the testing apparatus. They
may alter the behaviour of ‘‘nominally undisturbed’’ speci-
mens from that of ‘‘ideal’’ specimens.
Previous studies on this subject addressed several aspects:

a) the influence of sampler type and of field and laboratory

techniques on the degree of disturbance (La Rochelle

0038-0806 & 2013 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.sandf.2013.02.011
316 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334

Nomenclature r parameter of the elasto-viscoplastic constitutive

model which defines the position of the current
A clay activity state of the material with respect to the IRYL
B inner diameter of the sampler re parameter which defines the dimension of the
Bw bulk modulus of water elastic domain around the initial effective stress
cae coefficient of secondary consolidation [e-log10(T)] state of the material
cae coefficient of secondary consolidation [e-log10(T)] rn parameter of the elasto-viscoplastic constitu-
Cc * e100* e1000* (after Burland, 1990) tive model which defines the position of the
CF clay fraction current natural yield locus with respect to
e100* intrinsic void ratio at s0 v ¼ 100 kPa (Burland,
the IRYL
1990) s0 (s0 1 þ s0 3)/2
e1000* intrinsic void ratio at s0 v ¼ 1000 kPa (Burland, SD1 degree of sample disturbance according to
1990) Hong and Han (2007), suffix 24 h or EOP for
Eoe oedometric modulus
incremental loading (24 h) and end of primary
Iv void index (after Burland, 1990)
(EOP) oedometer tests, respectively
(Ko)NC earth pressure coefficient at rest for normally
T time
consolidated UPC
T90 time to 90% dissipation of excess pore pressures
k slope of the EOP isotropic swelling/recompres-
t (s0 1  s0 3)/2
sion line in the v log10(p0 ) plane or in the tmax (s0 1  s0 3)max/2
ln(v)  ln(p0 ) plane, for ‘‘nominally undis- ts thickness of the sampler wall
turbed’’ specimens tsp presumed duration of sampler penetration
ki slope of the intrinsic isotropic swelling/recom- us shear induced pore pressure
pression line in the v log10(p0 ) plane or in the v 1 þ e¼ current specific volume
ln(v)  ln(p0 ) plane vl specific volume at a reference mean effective
l parameter related to the first mechanism of the stress p0 ¼ 1 kPa on the EOP isotropic virgin
elasto-viscoplastic constitutive model compression line, inferred from interpretation
L length of the sampler of laboratory tests on ‘‘nominally undisturbed’’
LI liquidity index ¼ (wo  wP)/(wL  wP) specimens
m parameter related to the first mechanism of the vli specific volume at a reference mean effective
elasto-viscoplastic constitutive model stress p0 ¼ 1 kPa on the ICLi
M strength parameter ea axial strain
n parameter related to the second mechanism of the (ea)s imposed compression–extension axial strains in
elasto-viscoplastic constitutive model (assumed the Ideal Sampling Approach
equal to zero in this study) ea* axial strain in one dimensional compression
p0 (s0 1 þ s0 2 þ s0 3)/3 ¼ mean effective stress tests after reconsolidation to the geostatic
p0 e equivalent mean effective stress on the ICLoe effective stress
corresponding to the void ratio at the end of e_a axial strain rate
the reconsolidation phase j0 cv constant volume (critical state) angle of shear
p0 cio initial mean effective stress corresponding to the resistance
intersection of the intrinsic isotropic swelling/ l slope of the EOP isotropic virgin compression
recompression line (ICLi) with the elastic wall line in the v log10(p0 ) plane or in the
passing through the point eo, p0 o (after Rocchi ln(v) ln(p0 ) plane, for ‘‘nominally undis-
et al., 2003) turbed’’ specimens
p0 cno initial mean effective stress linked to the initial li slope of the ICLi and ICLoe in the v log10(p0 )
stress state and to the yield stress ratio plane or in the ln(v) ln(p0 ) plane
(YSR ¼ s0 vy/s0 vo; after Rocchi et al., 2003) mi viscosity parameter related to the first mechan-
p0 o mean effective stress at the initial in situ state ism of the elasto-viscoplastic constitutive model
p0 r0 residual mean effective stress after tube sampling mn viscosity parameter related to the second
and extrusion of the sample at zero storage time mechanism of the elasto-viscoplastic constitu-
and no loss of saturation tive model
p0 r1 residual mean effective stress after tube sam- y Lode angle
pling, transportation, storage and extrusion s0 vmax maximum vertical effective stress applied in the
with no loss of saturation SHANSEP procedure
p0 r2 residual mean effective stress after tube sam- s0 hmax maximum horizontal effective stress applied in
pling, transportation, storage extrusion and the SHANSEP procedure
partial loss of saturation s0 vy vertical effective yield stress in oedometer tests
q (s0 1  s0 3)¼ deviatoric stress above which significant irrecoverable time-
qo deviatoric stress at the initial in situ state dependent strains occur in structured clays
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
et al., 1981; Atkinson et al., 1992; Tanaka and Tanaka,
1999; Tanaka and Tanaka, 2006, Horng et al., 2010)
which depends on the characteristics of the material,
especially its sensitivity, permeability and mesostructure
(fissures, laminations, etc.);
b) the reconsolidation procedures which attempt to remedy
the damage from sample disturbance and to recover in
the laboratory the behaviour of the ‘‘ideal’’ specimen
(Bjerrum, 1973; Ladd and Foott, 1974);
c) the effects of sample disturbance on the results of
laboratory tests, including comparative studies of the
disturbance induced by different sampling techniques
and laboratory and/or numerical simulations of the
sampling processes (Kirkpatrick and Khan, 1984;
Kirkpatrick et al., 1986; Baligh et al., 1987; Graham
and Lau, 1988; Budhu and Wu, 1992; Hight et al., 1992;
Clayton et al., 1992; Georgiannou and Hight, 1994;
Shogaki, 1996; Hong and Onizuka, 1998; Siddique
et al., 1999; Tanaka and Tanaka, 1999; Santagata and
Germaine, 2002; Noda et al., 2005a; Santagata et al.,
2006; Tanaka and Tanaka, 2006; Hong and Han, 2007);
d) the empirical, theoretical and/or numerical procedures
to correct the results obtained in laboratory tests on
‘‘nominally undisturbed’’ specimens to estimate the
behaviour of ‘‘ideal’’ specimens (Schmertmann, 1953;
Nagaraj et al., 1990; Nagaraj et al., 2003; Noda et al.,
2005a; Hong and Han, 2007; Takaine et al., 2010).
In homogeneous specimens it is helpful to address these
issues by considering the state and behaviour of natural
materials relative to those of reconstituted materials at
comparable effective stresses and pore water chemistry,
where the effects of microstructure have been removed
(Leroueil and Vaughan, 1990; Burland, 1990; Shibuya
et al., 1999; Cotecchia and Chandler, 2000; Asaoka
et al., 2000a; Nagaraj and Miura, 2001; Rocchi et al.,
2003). At any (initial or current) point in space and time
the state of the material is defined by its void ratio (e) and
vertical (s0 v) and horizontal (s0 h) effective stresses. The
void ratio is usefully expressed in terms of void index IV=
(e100* e)/(e100*  e1000*) (see Notation), as proposed by
Burland (1990), who showed that, once normalized in this
manner, the virgin oedometric compression curves of
reconstituted clays of different plasticity (liquid limit)
overlap and are referred to as ‘‘Intrinsic Compression
Line’’ (ICLoe).
In particular, it is useful to compare in the (Iv) (s0 v)
plane the oedometric curves of ‘‘nominally undisturbed’’
specimens with the ICLoe and the natural (initial) void
index (Ivo) at the geostatic vertical effective stress (s0 vo)
with the corresponding value (Iv2) on the ICLoe (Fig. 1).
Mechanically overconsolidated, unbonded materials
have initial states and oedometric curves below the ICLoe;
the oedometric curves tend to converge to and overlap
with the ICLoe at high applied pressures; these materials
have stable microstructures, resulting in limited strain rate
dependence of their behaviour.
317
Initial states and oedometric curves located above the
ICLoe indicate metastable microstructures, whereby varia-
tions in applied stress beyond a threshold value (signifi-
cantly lower than the conventional preconsolidation
pressure s0 vp) cause irreversible activation of destruction
processes and viscoplastic deformations (Nagaraj et al.,
1990; Burland, 1990; Rocchi et al., 2003).
In these ‘‘structured’’ materials, definitions of the degree
of sample disturbance based solely on the residual mean
effective stress (p0 r) in the specimen (Ladd and Lambe,
1963; Skempton and Sowa, 1963; Okumura, 1971; Tanaka
et al., 1996) or on the axial (ea*) or volumetric (ev*) strain,
both equal to (eo  e1)/(1 þ eo), where eo and e1 are
respectively the void ratio in situ and after reconsolidation
to the geostatic effective stress (Andresen and Kolstad,
1979; Lunne et al., 1997), do not have general validity
(Tanaka, 2000; Tanaka and Tanaka, 2006; Hong and Han,
2007). For materials with different initial states (expressed
in terms of Ivo  Iv2), similar values of p0 r or ea* may
indicate different degrees of disturbance. The degree of
sample disturbance may be defined more rationally using
the intrinsic compression concept as SD1=(Ivo  Iv1)/
(Ivo  Iv2)—see Fig. 1, as proposed, for example, by Hong
and Han (2007) with different notation. Even though it is a
derived parameter, SD1 provides a more robust indication and
an unambiguous definition of sample disturbance with a clear
physical meaning and values always between 0% and 100%.
Several aspects of sample disturbance in structured clays
with initial states above the ICLoe are investigated with the
aid of numerical simulations of laboratory tests on the
Upper Pisa Clay (UPC), where the effects of sample
disturbance are modelled by a modified version of the
‘‘Ideal Sampling Approach’’ (ISA) procedure (Baligh et al.,
1987; Santagata and Germaine, 2002), which is considered
to improve other procedures such as that adopted by Noda
et al. (2005a).
This clay is well characterized (Rampello et al., 1996;
Mesri et al., 1997; Rampello and Callisto, 1998; Callisto
and Calabresi, 1998; Callisto et al., 2002; Lo Presti et al.,
Fig. 1. Degree of sample disturbance in structured materials with initial
states above the ICLoe in terms of void indexes, after Hong and Han
(2007).
318 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
2003) on the basis of laboratory tests carried out on
‘‘nominally undisturbed’’ specimens taken by standard
thin-walled Shelby samplers and the Laval sampler, speci-
fically developed to minimize disturbance in soft sensitive
clays (La Rochelle et al., 1981). For comparison, reference
is also made to experimental data for other clays with
similar initial states relative to the ICLoe.
Whilst many advanced elasto-plastic time-independent con-
stitutive models exist which account for destruction processes
(Asaoka et al., 2000b; Kavvadas and Amorosi, 2000; Rouainia
and Muir Wood, 2000), an elasto-viscoplastic constitutive
model is used here because destruction processes and asso-
ciated irreversible strains require time for their development,
independently from hydrodynamic phoenomena.
Among the many such models available in the literature
(e.g.: Kimoto and Oka, 2005; Hinchberger and Qu, 2009;
Yin et al., 2011; Nakai et al., 2011), the model developed
by Rocchi et al. (2003, 2006) has been used here because it
includes a) irreversible activation of viscoplastic strains
once a threshold is reached, even if stress is subsequently
reduced below the threshold; b) termination of viscoplastic
strains on reaching an intrinsically stable condition; c) rate
of viscoplastic strains dependent on damage.
Two main issues are addressed here:
 the laboratory behaviour of ‘‘ideal’’ specimens, together
with a critical appraisal of some procedures proposed in
the technical literature to remedy sample disturbance;
 the field behaviour compared to the laboratory beha-
viour of the ‘‘ideal’’ specimen. This distinction is
common when dealing with undrained strengths, less
so when dealing with compressibility. The use of
compression curves of ‘‘ideal’’ specimens to represent
‘‘field behaviour’’, combined with simple constitutive
models and/or simplified procedures (i.e. CAM CLAY
model, Terzaghi’s consolidation theory, uniqueness of the
End Of Primary (EOP) void ratio-effective stress relation-
ship), as often done in standard practice, may lead to
predictions of settlements and excess pore pressures
significantly different from those observed in practice.
2. Geotechnical characteristics of UPC
The soil profile in Pisa is characterized by three geological
units, Horizons A (upper deposits), B (clayey deposits) and C
(lower sandy deposits) (Rampello et al., 1996). The ground
level and the water table are respectively at þ 1/þ 2 m and
þ 0 m m.s.l.
The materials considered here are the plastic clays of
layers B1 and B3, between  7.5 m and  18 m m.s.l. (UPC,
or ‘‘Pancone’’), whose typical characteristics are summar-
ized in Table 1, based on extensive ground investigations.
The UPC formed approximately 7000 years ago in a
marine environment (Berardi et al.,1991; Mesri et al.,
1997); it has open (metastable) microstructures, as indi-
cated by SEM photographs and by initial states and
oedometric curves above the ICLoe in the Iv  s0 v plane
(Fig. 2). As a consequence of destruction processes, at high
stress the oedometric curves converge towards the ICLoe;
in the conventional interpretation of oedometer tests, high
values (4 1–1.5%) of the coefficient of secondary con-
solidation (cae) are observed between the initial (geostatic)
vertical effective stress (s0 vo) and the conventional precon-
solidation pressure (s0 vp).
The initial states, the oedometric curves and the ICLoe in
Fig. 2 are normalized as proposed by Burland (1990), based
on the measured values of particle density (Gs ¼ 2.65–2.75)
and natural water content (wo) and assuming full saturation
in situ (degree of saturation St ¼ 1). The specimen taken using
the Shelby sampler has ea*¼ 8–9% and a degree of sample
disturbance (SD1)24 h ffi35%; the specimen taken using the
Laval sampler has ea*¼ 2–3% and (SD1)24 h ffi 10%.
The permeability coefficient (k) at the initial in situ state
has been evaluated using the empirical relationships proposed
by Tavenas et al. (1983a,b) and Feng (1991), based on direct
measurements from laboratory tests on several natural clays;
in structured clays this approach overcomes the limitations of
indirect evaluations based on Terzaghi’s consolidation theory
(Tavenas et al., 1983a).
Conventional interpretation of oedometer tests would
indicate an apparent overconsolidation ratio (OCR*)¼ s0 vp/
s0 vo ffi 1.4. Considering the geological history of the site,
whereby it seems unlikely that the UPC has ever sustained
greater vertical stress than at present, the initial effective stress
condition has been determined by assuming (Ko)NC ¼ s0 ho/
s0 vo ¼ 0.6, compatible with a constant volume (critical state)
angle of shear resistance j0 cv ¼ 241. Corrections for mechanical
overconsolidation (Schmidt, 1966) or for delayed consolida-
tion (Mesri and Hayat, 1993) do not seem applicable where
the apparent overconsolidation is due to microstructure only
and delayed consolidation is prevented by bonding.
3. Constitutive model and parameters
The elasto-viscoplastic constitutive model described by
Rocchi et al. (2003) considers two yield loci and how they
change in response to viscoplastic strain related to desc-
truction processes (DPs) in saturated structured clays
whose initial states are located above the ICLoe. The
intrinsic yield locus (IRYL) represents structurally stable
conditions where DPs may be considered negligible; it is
defined by the modified CAM CLAY model and intrinsic
parameters determined by tests on reconstituted speci-
mens. The initial yield locus (INYL) has the same shape
as and lies outside the IRYL, reflecting the presence of
metastable microstructures (bonding); it is equivalent to
the superloading surface of the Asaoka et al. (2000b)
model. Following the general principles of viscoplasticity
(Perzyna, 1963), viscoplastic strains (flow rules) and rela-
ted changes of the yield loci (hardening rules) may be
governed by two mechanisms, both activated when the
current effective stress locus first reaches the INYL. The
two mechanisms consider the overstress of the current
effective stress locus (equivalent to the subloading surface
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334 319

Table 1
Geotechnical characteristics of the UPC samples – layers B1 and B3 (data after Studio Geotecnico Italiano project files – Shelby sample – and Rampello
et al., 1996 – Laval sample).

Sample and depth gta wLa IPa wo a Sta eo1a, b

eo2a, c
Aa, d s0 voa s0 vpa Cc1a, e
Cc2a, f
ka j0 cva
(m m.s.l.) (kN/m3) (%) (%) (%) ( ) () ( ) () (kPa) (kPa) ( ) ( ) (m/s) (1)

Shelby (–17.5) 15.56 70 47 62.5 0.97 1.70 1.65 0.6 145 180 0.54 0.433 3  10–10 24
Laval (–16.0) 15.80 70 40 63.0 0.97 1.78 1.73 0.6 135 190 0.95 0.456 5  10–10 24

Notes:
a
see Notation.
b
as reported in the laboratory certificates, corresponding to St ffi97%.
c
imposing St ¼100%.
d
assumed.
e
immediately beyond s0 vp.
f
¼ Cc*, intrinsic compression index (Burland, 1990).

of the Asaoka et al., 2000b model) compared with (1) the 2.0
Profile of initial states assumed for
IRYL and (2) the INYL. The flow rule of the first 1.5 settlement calculations of Fig. 25
1.0
mechanism accounts for progressive damage of bonding Laval Sample
0.5
through the damage factor Rmax, defined as the maximum

Iv ( - )
Shelby Sample
0.0
overstress relative to the INYL ever reached by the current -0.5
effective stress locus. For simplicity, the second mechanism -1.0 Initial state ICLoe
has been deactivated in the analyses presented here. -1.5
Laval sample
Shelby sample
To improve the match with the experimental data for a -2.0
10 100 1000 10000
broader range of clays than originally considered and to
σ'v (kPa)
make the original model applicable to more general loading
conditions, the following changes have been implemented Fig. 2. UPC – IL24h oedometric curves and initial states of ‘‘nominally
(see Notation and Rocchi et al., 2006): undisturbed’’ samples recovered by Laval and Shelby samplers (See Table 1).

 The flow rule of the first mechanism has been modified  Considering the curvature of the oedometric (ICLoe)
as follows: and isotropic (ICLi) intrinsic compression lines in the
v ln(p0 ) plane, the model has been optimized by
ðr1Þl ðRmax Þm ½1expð1rÞ 
Fi ¼ ð1Þ rearranging their formulation so that they are defined
mi in the ln(v) ln(p0 ) plane, where they are nearly linear.
where Fi ¼ 0 for (r 1) o 0 and/or rn ¼ 1.
 The M ¼ constant condition has been replaced by the More details regarding the geotechnical framework, the
M ¼ variable condition, by ensuring both its dependence mathematical formulation and the parameters of the
on the Lode angle y (Grammatikopoulou et al., 2007) model, including those which define the initial conditions
and a constant critical state angle j0 cv (Parry, 1960), as represented by the initial state (given by s0 vo, s0 ho and eo),
follows: the INYL and the IRYL (given by p0 cno and p0 cio respec-
pffiffiffiffiffiffiffiffiffiffiffiffi 0 tively), can be found in the original paper.
3 3 þ h2 sinjcv The numerical simulations of laboratory tests on UPC
M¼ ð2Þ
3 þ hsinj0cv have been performed using the parameters in Table 2.
where Based on the data published by Bjerrum (1967), the true
0 0
2s s s
0
yield vertical effective stress in oedometric conditions
h¼ 2 0 1 0 3 (s0 vy), which defines the INYL, has been assumed equal
s1 s3
to 1.15s0 vo ( o s0 vp).
 To simulate destruction processes along unloading The parameters mi, m and l have been backfigured from
stress paths, the first viscoplastic mechanism is irrever- the oedometric curve of the high quality Laval specimen;
sibly activated when the current state of the material the effects of disturbance has been incorporated in the
moves outside an initial elastic domain defined by the back analysis, assuming (ea)s ¼ 0.5%, which is compatible
following equation: both with ISA and with (SD1)24 h ffi 10%.
qq 2  0 2 The coefficient of permeability (k) has been updated in
o 0 p re
þ ðp0 po Þ ¼ cno ð3Þ the analyses on the basis of the current void ratio (e) and
M 2
the relationship proposed by Tavenas et al. (1983a).
where re, p0 o and qo define the dimensions and the centre As will be shown later, with the parameters backfigured
of the initial elastic domain, assumed centred on the from the oedometric curve it was possible to replicate quite
initial effective stress state and tangential to the INYL. well also the undrained shear response observed in triaxial
320 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
Table 2
Parameters of the elasto-viscoplastic constitutive model for simulations of laboratory tests on the UPC samples – layers B1 and B3.
wL a IP a eoa lia,b,c vlia,b p0 cno/p0 cio
(%) (%) () ( ) ( ) ()
70 40 1.73 0.095 3.394 27
Notes:
a
see Notation and Rocchi et al. (2003).
b
referred to the ICLi.
c
in the ln(v)–ln(p0 ) plane.
tests, both in compression and in extension, without any
adjustment.
4. Procedure for numerical simulation
The simulation of sampling and laboratory tests on
‘‘nominally undisturbed’’ specimens has been carried out
according to the following steps:
a) Application of initial conditions. Whilst in principle the
estimate of the initial conditions could be refined using
back analyses of laboratory tests (Takaine et al., 2010),
the work presented here is based on a unique (best
estimate) definition of initial conditions.
b) Simulation of sampling. The penetration of the tube
sampler imposes a cycle of compression–extension axial
strain 7 (ea)s in undrained triaxial conditions and
presumed duration 60 s. Based on the ISA approach
(Baligh et al., 1987; Santagata and Germaine, 2002)
(ea)s=0.387ts/B, where ts is the wall thickness and B is
the inner diameter of the sampler. For JPN, Laval and
ELE100 samplers with B/ts=50 to 60 (Shibuya et al.,
1999; Tanaka, 2000), (ea)s ffi0.6% to 0.8%; for Shelby
samplers as detailed in ASTM D1587-94 (B/ts=20 to 40),
(ea)s ffi 1% to 2%. Values of (ea)s up to 10% need to be
considered to reflect higher degrees of disturbance that
may arise from more severe strains caused by different
tube and tip geometry (La Rochelle et al., 1981; Clayton
et al., 1998; Clayton and Siddique, 1999) and operator
effects, including drilling before sampling. Higher values
did not result in further variation of the residual mean
effective pressure p0 r0 reached at the end of step (d1) for
zero storage time and no loss of saturation.
c) Simulation of transport and storage, by introducing a
delay of one month in which the strains attained at the
end of step (b) are maintained and no drainage is
allowed; negligible and 1 year duration have also been
examined for comparison.
d) Simulation of the extrusion from the sampler, where:
(d1) The deviatoric stress at the end of step (b) or (c)
is removed in undrained conditions; the residual
mean effective pressure p0 r at this stage depends on
storage time; hereafter p0 r0 and p0 r1 are for zero and
non zero storage time respectively.
(d2) Samples are often observed to have a degree of
a
na ka Bwa j0 cva kia,c mia ma la
( ) (m/s) (kPa) (1) () (days) ( ) ( )
0.25 5.0  10–10 2  106 24 0.018 1.91  1015 11 9
saturation St less than 1, even when the shortest
possible time is allowed between extrusion and
testing, presumably because the samples can sustain
only part of the suction associated with p0 r0 or p0 r1.
Since the constitutive model used in the analyses
does not address partially saturated soils, this is
modelled by allowing the sample to swell in drained
conditions at constant natural water content (wo)
along the swelling/recompression line to reach a
void ratio compatible with the degree of saturation
observed in the laboratory (typically St ffi 0.97 to
0.98 in UPC). At the end of this stage p0 r=p0 r2.
(d3) possible additional damage to the microstruc-
ture is allowed by artificially increasing the damage
factor Rmax beyond the value reached in step (b);
this allows simulation of remoulding.
e) Simulation of various laboratory tests and test proce-
dures, as detailed below.
Incremental loading (IL24h and ILEOP) oedometer tests
have been simulated with initial specimen height Ho=2 cm
and drainage at both boundaries. The load increments are
applied every 24 h in the first case and on reaching residual
excess pore pressures equal to 5% of the applied pressure
increment Ds0 v in the second case.
Unconfined compression (UC) and unconsolidated
undrained triaxial tests (TX-UU) have been simulated by
imposing an axial strain rate ėa equal to 1%/h, as in the
experimental study carried out by Santagata and Germaine
(2002). The simulations have been repeated at different
strain rates, from 100%/h to 0.1%/h, to cover the full
range of strain rates encountered in US and Japanese
laboratory testing practice and in the field (Ladd and
Foott, 1974; Germaine and Ladd, 1988; Ladd, 1991;
Tsuchida, 2000; De Groot, 2001). In the simulation of TX-
UU tests, step (d2) is ignored, assuming that standard
laboratory procedures are effective in ensuring saturation of
the specimen at constant water content before loading.
Anisotropically consolidated undrained triaxial tests in
compression and extension (TX-CAU-C; TX-CAU-E) have
been simulated by adapting different procedures as applied
in practice for the consolidation phase. The simulations
have been carried out for the two limiting cases, i.e.
reconsolidation to the initial vertical and horizontal stres-
ses s0 vo, s0 ho (Bjerrum, 1973), and the SHANSEP approach
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334 321

0.50
(Ladd and Foott, 1974), where the specimen is consoli-
dated to vertical and horizontal effective stresses equal to Initial state

s0 vmax ¼ (1.5–2)  s0 vp and s0 hmax ¼ (Ko)NC  s0 vmax and subse- 0.25

end of sampling

quently unloaded in oedometric conditions to s0 vc ¼ (1.5– end of extrusion (d1)

2)  s0 vp/OCR*, where s0 vp is the conventional preconsoli-

t/σ'vo (-)
0.00
dation pressure determined by interpretation of the
oedometric curve; (Ko)NC ¼ (1  sin j0 cv) is the earth pres-
sure coefficient at rest for normally consolidated UPC; -0.25

OCR* ¼ s0 vp/s0 vo is the apparent overconsolidation ratio.

In the simulations predetermined values of s0 vo and s0 ho -0.50
or s0 vmax and s0 hmax have been applied to the specimen -10 -8 -6 -4 -2 0 2 4 6 8 10

(Lunne et al., 1997; De Groot, 2001); subsequent unload- εa (%)

ing in the SHANSEP approach has been carried out in 0.50

oedometric conditions (zero radial strains); the reconsoli-
dation phase has been carried out in drained conditions end of sampling
0.25
at ėa of the order of 0.1%/h; during undrained loading ėa
has been imposed equal to 1%/h. Initial state

t/σ'vo (-)
As recommended for example, by Tatsuoka et al. (1997), 0.00

the effects of setting time at the end of the reconsolidation

phase have been explored. -0.25
end of extrusion (d1)

In UC, TX-UU and TX-CAU tests the undrained shear (εa)s=10%

(εa)s=5%
(εa)s=2.5%
strength has been conventionally taken as one half of the (εa)s=1.0%
-0.50
maximum difference between the maximum and minimum 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
principal stresses (tmax). The simulations refer to uniformly s'/σ'vo (-)
strained, homogeneous specimens, without pre-existing
0.80
discontinuities (natural or induced by sampling, Tsuchida, end of sampling

2000). The results do not account for the localization of plastic

strains that may occur as a consequence of the dimensions and 0.60
end of extrusion (d1)
shape of the specimen and of the use of rigid frictional platens
us/σ'vo (-)

(see Boudali, 1995; Callisto and Calabresi, 1998). 0.40

5. Results
0.20

5.1. Sampling, transport, storage and extrusion Initial state

The results of the numerical simulations of the sampling
and extrusion phases (b) þ (d1), for different values of (ea)s,
are shown in Fig. 3 in terms of stress–strain curves, stress-
paths and shear induced pore pressures. These numerical
simulations are similar to the laboratory tests by Santagata
and Germaine (2002), (Fig. 6 therein) on samples of
Resedimented Boston Blue Clay (RBBC) prepared in the
laboratory inducing flocculation to reproduce the beha-
viour of natural clays, including significant strain rate
dependence. Considering the differences in materials and
strain rates (60% to 600%/h in the numerical simulations;
1%/h in the laboratory tests), the numerical simulations
replicate quite well the main features of the experimental
work, including low axial strains at failure and significant
post-failure strain-softening in compression, dependence of
the strength in extension on the strains reached in com-
pression, substantial reduction of the mean effective
pressures p0 and development of significant shear induced
pore pressures as a consequence of destruction processes
and associated visco-plastic strains. Similar trends have
been observed by Hight (2001) in experiments on Both-
kennar clays. For all values of (ea)s, the majority of the
0.00
-10 -8 -6 -4 -2 0 2 4 6 8 10
εa (%)
Fig. 3. UPC – Numerical simulations of the sampling and extrusion
phases (b)þ(d1), ėa ¼60% to 600%/h: (a) stress–strain curves, (b) stress
paths and (c) shear induced pore pressures.
reduction in p0 occurs during the compression part of the
strain cycle. Preventing extension of the sample during
sampling does not significantly alter the results of the
numerical simulations and is not considered in this study.
Fig. 4 shows the numerically simulated values of
(p0 o  p0 r)/p0 o versus imposed (ea)s caused by sampling and
undrained extrusion phases (b) þ (d1) only (p0 r ¼ p0 r0) or by
the complete processes, including storage and swelling
phases (b) þ (c) þ (d1) þ (d2) (p0 r ¼ p0 r2), for different trans-
portation and storage times and a degree of swelling in
step (d2) resulting in St ¼ 0.97 as is common for UPC
(Table 1); the results compare well with the laboratory
simulations on RBBC and Bothkennar clay, phases (b) þ
(d1) only, reported by Santagata and Germaine (2002) and
Hight (2001), also shown in the figure.
322 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334

1.00 2.00
p'r2
0.80 1.75
(p'o-p'r)/p'o (-)

0.60 1.50
Numerical simulations on UPC
phases (b)+(c)+(d1)+(d2) - storage time 365 days
phases (b)+(c)+(d1)+(d2) - storage time 30 days

e (-)
0.40 1.25
phases (b)+(c)+(d1)+(d2) - storage time 0 days
p'r0 phases (b)+(d1)
0.20 Serie7 simulations
Laboratory
Bothkennar clay - phases (b)+(d1) [Hight, 2001] 1.00
RBBC - phases (b)+(d1) [Santagata&Germaine, 2002]
0.00
0 1 2 3 4 5 6 7 8 9 10 0.75

(εa)s (%)
0.50
1 10 100 1000 10000
0
Fig. 4. UPC – Numerical simulations of the decrease in p r due to σ'v (kPa)
sampling disturbance and comparison with experimental data for RBBC
and Bothkennar clay. 2.00

1.75
As seen by comparing the curves of p0 r0 and p0 r2 for
storage time 0 days in Fig. 4, the effect on p0 r of swelling 1.50
and desaturation is progressively less significant as (ea)s

e (-)
increases; this may explain the difference observed by 1.25

Santagata et al. (2006) between the results of laboratory 1.00

tests carried out by considering phases (b) þ (d1) only or
phases (b) þ (d1) þ (d2). 0.75

For (ea)s ¼ 1–1.5%, representative of good quality sam-

0.50
pling, p0 r2 values are roughly equal to (1/5–1/8)s0 vo, i.e. of 1 10 100 1000 10000

the same order of magnitude as those reported by Tanaka σ'v (kPa)

and Tanaka (1999) for Ariake and Bothkennar clays.

Fig. 5. UPC – Numerical oedometer tests and comparison with experi-
The effect of storage – phase (c) – on p0 r2 values is mental data: (a) IL24h and (b) ILEOP.
significant for high quality samples having high values of p0 r2
(close to 1/3s0 vo) on immediate extrusion and measurement in
the field (Hight et al., 1992) and becomes progressively less 8.00

significant at decreasing sample quality. 7.00

Significant decay with time of mean effective stress has 6.00
been reported by Kirkpatrick and Khan, 1984, Kirkpatrick
5.00
et al. (1986), Graham et al. (1987) and Graham and Lau
εa* (%)

(1988) in clay (kaolin and illite) materials manufactured 4.00

from consolidated slurries, sampled with minimum dis- 3.00

turbance, extruded and stored in the laboratory. In the
2.00
authors0 opinion the observed behaviour may be caused by
1.00
both undrained creep (relaxation) (Tanaka et al., 2006),
activated by destruction processes in metastable micro- 0.00
0 1 2 3 4 5 6 7 8 9 10
structures, and the inability of the clays to maintain the
(εa)s (%)
suction induced by removal of ‘‘in situ’’ total stress; even
though the tested materials have been reconstituted in the Fig. 6. UPC – Numerical oedometer tests: ea* (ea)s relationships and
laboratory, their initial states appear to be located well comparison with experimental data for RBBC and Bothkennar clay.
above the ICLoe determined by means of the Burland
(1990) empirical procedure. represent high [(SD1)24 h ffi 10%] and poor [(SD1)24 h ffi35%]
A conventional transport and storage time of 30 days is quality samples respectively.
considered in all the simulations presented below. For a given set of (ea)s and Rmax a greater value of ea*
(and therefore SD1) is predicted by the elasto-viscoplastic
5.2. Oedometer tests model in IL24 h compared to ILEOP tests due to the different
duration of the test.
Fig. 5 shows the calculated IL24 h and ILEOP oedometric As SD1 increases, the compression curve is more rounded,
curves for ‘‘nominally undisturbed’’ specimens with different becoming progressively steeper at s0 v os0 vp and flatter at
values of (ea)s, the corresponding (or imposed) damage factor s0 v 4s0 vp, in line with the general trend reported in the
Rmax at the end of phase (d) and the resulting degree of technical literature, first highlighted by Schmertmann (1953).
disturbance SD1, together with the initial state (eo; s0 vo), the The damage induced by loading progressively prevails
ICLoe and the experimental IL24 h curves for two samples over that induced by sampling disturbance, the behaviour
obtained with Laval and Shelby samplers, which here of all specimens converging as applied pressures increase.
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334 323

300 0.0

250

200 -0.5
σ'vp (kPa)

150
-1.0

ΔH (mm)
100

50
-1.5
0
0 5 10 15 20 25 30 35 40
-2.0
SD1 (%)

Fig. 7. UPC — Numerical oedometer tests: s0 vp–SD1 relationships. -2.5

0 50 100 150 200 250 300 350 400

Fig. 6 shows the different ea*–(ea)s relationships from the T½ (s½)

numerical simulation of different types of oedometer tests pffiffiffiffi

Fig. 8. UPC – Numerical IL24h oedometer tests: DH– T curves.
(IL24 h and ILEOP) where the additional destruction described
as phase (d3) is not considered. They compare well with the 2.00
experimental results for RBBC (Santagata and Germaine,
2002) and Bothkennar clay (Hight, 2001), considering that 1.75

they refer to different clays and that the experimental results 1.50
refer to specimens subjected to phases (b)þ (d1) only, resulting

e (-)
in lower ea* than normally observed in practice, as shown by 1.25

Santagata et al. (2006). The results of the simulation also

support the Santagata et al. (2006) conclusion that, even in the
most favourable case represented by (ea)s lower than 1%, for
this type of material samples are unlikely to meet requirements
such as ea* less than 1%. Significant creep strains may occur
even in good quality specimens recompressed to s0 vo, as
experimentally observed for example by Tanaka (2005a); such
creep strains are not necessarily experienced in the field by
relatively old, undisturbed clayey deposits.
Fig. 7 shows relationships between conventional pre-
consolidation pressures s0 vp and SD1 for the numerically
simulated oedometer tests on samples not subjected to the
additional destruction described as phase (d3); s0 vp values
have been determined according to both Casagrande’s (1936)
method and the bi-logarithmic method (Butterfield, 1979;
Onitsuka et al., 1995; Nagaraj et al., 2003; Hong and Han,
2007), with no substantial difference between s0 vp obtained
from IL24 h and ILEOP tests for both methods.
The commonly accepted assumption that disturbance
always leads to an underestimation of s0 vp is also not
supported by the experimental work of Rampello et al.
(1996) on UPC and Santagata and Germaine (2002) on
RBBC; in their Fig. 11, the latter propose a possible
explanation for this. As observed in Fig. 5, s0 vp is found to
increase with increasing disturbance (Fig. 7), unless addi-
tional destruction beside that caused by (ea)s occurs, as in
phase (d3). As also pointed out by Tanaka and Tanaka
(1999), s0 vp is not a suitable indicator of sample quality.
While in Fig. 5 the first part of the ILEOP e  log10(s0 v)
curve is above the IL24 h e  log10(s0 v) curve, the opposite
applies at s0 v higher than 600 kPa, where the time neces-
sary for complete excess pore pressure dissipation
(Du o 0.05Ds0 v) becomes longer than 24 h.
Fig. 8 shows the DH (T)0.5 curves for the simulated IL24 h
test on a sample which has undergone (ea)s ¼ 2.5%. Serious
difficulties are encountered in trying to determine ‘‘conven-
tional’’ consolidation parameters, such as coefficients of
1.00
0.75
0.50
10 100 1000 10000
σ'v (kPa)
Fig. 9. UPC – Numerical oedometer tests: ILEOP curves of the ‘‘ideal’’
specimen compared with those of ‘‘nominally undisturbed’’ specimens
corrected for sampling disturbance.
primary consolidation cv, oedometric moduli Eoe and perme-
ability coefficients k. For example, the Taylor (1948) method
would not predict correctly the ‘‘true’’ time to 90% dissipa-
tion of excess pore pressures (T90) determined from the pore
pressures calculated by the model, shown in Fig. 8, especially
for pressures above s0 vp. This confirms the well known
limitations of Terzaghi’s theory when applied to structured
clays, as highlighted for example by Tavenas et al. (1983a);
the same difficulties arise for the method proposed by
Casagrande and Fadum (1944) and those reported in
Nagaraj and Miura (2001). In oedometer tests it is therefore
highly advisable to measure excess pore pressures directly.
The simulated ILEOP e log10(s0 v) curve of ‘‘ideal’’
specimens characterized by (ea)s ¼ 0%, Rmax ¼ 1 and
(SD1)EOP ¼ 0, is compared in Fig. 9 with the e-log10 (s0 v)
curves resulting from the simulated ILEOP and IL24 h
curves for the nominally undisturbed specimens shown in
Fig. 5, corrected for sampling disturbance according to the
empirical procedures suggested by Schmertmann (1953) or
Nagaraj et al. (2003). For samples not subjected to the
additional destruction described in phase (d3), only minor
differences are observed between the corrected curves
obtained by the two empirical procedures for both ILEOP
and IL24 h tests; both procedures tend to overcorrect, with
increasing error as sample disturbance increases.
324 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334

2.00 0.40

0.35
1.75
0.30
1.50
0.25

t /σ'vo (-)
e (-)

1.25 0.20

0.15
1.00
0.10

0.75
0.05

0.50 0.00
20 0.0 0.2 0.4 0.6 0.8 1.0
Eoe (MPa)

15 s' /σ'vo (-)

10
0.40
5
0 0.35
1.0E-08
0.30
k (m/s)

1.0E-09
0.25
1.0E-10

t /σ'vo (-)
1.0E-11 0.20
10
0.15
cv (m2/y)

5 0.10

0.05
0
1 10 100 1000 10000
0.00
σ'v (kPa) 0 2 4 6 8 10 12 14
εa (%)
Fig. 10. UPC – Numerical ILEOP oedometer tests: Eoe s0 v, k s0 v and
cv  s0 v relationships. Fig. 11. UPC – Numerical UC tests: (a) effective stress paths; (b) stress–
strain curves (normalization to s0 vo)—test terminated on reaching s0 3 ¼ 0.
Fig. 10 shows the calculated cv, Eoe and k for the
simulated ILEOP tests on specimens characterized by
(SD1)EOP equal to 0 (‘‘ideal’’ sample), 3.5% (high quality
sample) and 13% (good to fair quality samples). The 0.40

dramatic reduction of the values of cv and Eoe observed in 0.35

a
high quality samples at pressures around s0 vp rapidly 0.30
disappears as the degree of disturbance increases, as high-
0.25
lighted for example by Jamiolkowski et al. (1983). However,
t / σ'vo (-)

the influence of the degree of disturbance on the values of cv 0.20

and Eoe from ILEOP oedometer tests is significant only in a 0.15

narrow range of applied pressures, say between s0 vo and 0.10

1.1s0 vp; in field conditions, where strain rates and therefore
0.05
s0 vp are lower (Leroueil et al., 1985), this range is even
narrower, it being limited to the true yield stress s0 vy. Such a 0.00
0.0 0.2 0.4 0.6 0.8 1.0
limited range is seldom significant for practical applications, s' /σ'vo (-)
especially on consideration of possible disturbance to the
0.40
deposit by construction activities (ground improvement b
works, pile installation). 0.35

0.30

5.3. UC and TX-UU tests 0.25

t /σ'vo (-)

0.20
Figs. 11 and 12 show the numerically simulated stress–
0.15
strain curves and effective stress paths of UC and TX-UU
0,10
tests carried out at ėa=1%/h on ‘‘nominally undisturbed’’
specimens characterized by different degrees of distur- 0.05

bance, compared with the simulation of an undrained 0.00

0 2 4 6 8 10 12 14
triaxial compression test on an ‘‘ideal’’ specimen consoli-
εa (%)
dated to the initial in situ effective stresses. As experimen-
tally observed by Tanaka and Tanaka (1999) on Ariake, Fig. 12. UPC – Numerical TX-UU tests: (a) effective stress paths;
Bothkennar and Drammen clays, Hight (2001) on (b) stress–strain curves (normalization to s0 vo).
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334 325

1.40 0.50
Shear Strength Ratio (=1 at 1%/min)

0.40
1.20
0.30
1.00 0.20

0.10
0.80

t /σ'vo (-)
0.00
0.60
-0.10

0.40 -0.20

-0.30
0.20
-0.40

0.00 -0.50
0.0001 0.001 0.01 0.1 1 10 0.0 0.2 0.4 0.6 0.8 1.0

εa (%/min) s' /σ'vo (-)

0.50
Fig. 13. UPC – Numerical TX-UU tests: undrained shear strength – axial
strain rate relationship. 0.40

0.30

Bothkennar clay, Santagata and Germaine (2002) on 0.20

RBBC clay and Horng et al. (2011) on Takuhoku clay, 0.10

t / σ'vo (-)
increasing disturbance results in: (a) a decrease of the 0.00

undrained shear strength as a consequence of damage to -0.10

the microstructure and the associated reduction of mean -0.20

effective stress (Okumura, 1971; Tanaka, 2000; Tsuchida, -0.30

2000; Watabe and Tsuchida, 2001; Horng et al., 2011); (b) -0.40

an increase in the strain at failure and a corresponding -0.50

reduction of stiffness; (c) a loss of strain softening; (d) a
change from contractive to dilatant behaviour; (e) an
increase in the maximum obliquity (t/s0 at peak strength).
As commonly observed, TX-UU tests on ‘‘nominally
undisturbed’’ specimens result in higher strengths than UC
tests, the difference becoming less significant with increasing
disturbance.
The dependence of the undrained shear strength on ėa
determined by the numerical simulation of TX-UU tests
on specimens characterized by (SD1)EOP ¼ 13% (Fig. 13) is
similar to that reported by Tsuchida (2000) based on
laboratory tests on several natural clays.
For ėa ¼ 1%/min (typical laboratory practice), the
undrained shear strength from numerically simulated
TX-UU tests on high to good quality samples, i.e.
(SD1)EOP ffi 3.5–13%, is in the range of 40  60 kPa, which
compares with the values obtained experimentally by
Rampello et al. (1996) and Rampello and Callisto (1998).
5.4. TX-CAU tests—Bjerrum reconsolidation procedure
For the same ‘‘nominally undisturbed’’ specimens, Fig. 14
shows the numerically simulated stress-strain curves and the
stress-paths for TX-CAU-C and TX-CAU-E tests carried out
at ėa ¼ 1%/h on specimens reconsolidated to the initial in situ
stresses following the Bjerrum (1973) reconsolidation proce-
dure; the results are compared with those obtained from
benchmark tests on ‘‘ideal’’ specimens. Relatively low levels of
disturbance – (SD1)EOP ¼ 0%, 3.5% and 13% – result in
increasing strains to failure, together with a reduction of
stiffness and strength; some regain in strength and stiffness
occurs at intermediate levels of disturbance – (SD1)EOP ¼ 21%
and 27%, which is lost again at higher levels of disturbance –
0 2 4 6 8 10 12 14
εa (%)
Fig. 14. UPC – Numerical TX-CAU tests – Bjerrum (1973) reconsolida-
tion procedure: (a) effective stress paths and (b) stress–strain curves
(normalization to s0 vc ¼s0 vo).
(SD1)EOP ¼ 89%; for all degrees of disturbance, while it fails to
recover completely the degree of strain softening of the ‘‘ideal’’
specimen, which is almost completely lost at (SD1)EOP ¼ 89%,
the Bjerrum (1973) reconsolidation procedure results in a
significant regain of strength compared to UC and TX-
UU tests.
As highlighted by Hight (2001), the difference in strength
and in overall behaviour observed in samples with different
degrees of disturbance depends on the opposite and concur-
rent effects of the amount of irreversible damage to the
microstructure of the clay and of the changes in void ratio
during the reconsolidation phase; in the constitutive model,
these processes are accounted for by the hardening and the
flow rules and by continuously updating the specific volume v
and the terms r and Rmax. The irreversible damage in the
specimens can be inferred by normalizing the results of each
test to the equivalent mean effective pressure p0 e on the ICLoe
corresponding to the void ratio of each specimen at the end of
the reconsolidation phase, shown in Fig. 15. When compared
with Fig. 14, the normalized results shown in Fig. 16 clearly
highlight that the reconsolidation procedures partly hides but
does not remedy the significant changes which occur in the
damaged specimens, which have both less pronounced strain
softening and considerably lower normalized strengths.
Allowing the ‘‘nominally undisturbed’’ specimen with
(SD1)EOP ¼ 3.5% to age (drained creep) at the end of the
326 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334

2.00 0.50

1.75
0.40
1.50
e (-)

1.25
0.30

t/σ'vo (-)
1.00

0.75 0.20

0.50
1 10 100 1000 10000 0.10
p' (kPa)

0.00
Fig. 15. UPC – Numerical TX-CAU tests – Bjerrum (1973) reconsolida- 0.0 0.2 0.4 0.6 0.8 1.0
tion procedure: current states at the end of the reconsolidation phase in s'/σ'vo (-)
the e log10(p0 ) plane.
0.50

10

8 0.40

6
0.30

t/σ'vo (-)
4

2
t / p'e (-)

0.20
0

-2
0.10
-4

-6
0.00
-8 0 2 4 6 8 10 12 14
εa (%)
-10
0 2 4 6 8 10 12 14 16 18 20
s'/ p'e (-) Fig. 17. UPC – Numerical TX-CAU tests – Bjerrum (1973) reconsolida-
tion procedure plus different ageing times: (a) effective stress paths and (b)
10 stress-strain curves (normalization to s0 vc ¼s0 vo).
8

6 1.4
4 1.2
1.0
2
t / p'e (-)

0.8
0 0.6
-2 0.4
q / p'o (-)

0.2
-4
0.0
-6 -0.2
-8 -0.4
-0.6
-10
0 2 4 6 8 10 12 14 -0.8
εa (%) -1.0
-1.2
Fig. 16. UPC – Numerical TX-CAU tests – Bjerrum (1973) reconsolida- -1.4
-5 -4 -3 -2 -1 0 1 2 3 4 5
tion procedure: (a) effective stress paths and (b) stress–strain curves
εa (%)
(normalization to p0 e)..

Fig. 18. UPC – Numerical TX-CAU tests – Bjerrum (1973) reconsolida-

reconsolidation phase (as recommended by Tatsuoka et al., tion procedure plus 40 h ageing and comparison with corresponding
1997) results in a significant increase in both stiffness and laboratory test results (normalization to p0 c ¼p0 o).
undrained strength (Fig. 17), even beyond those of the
benchmark test of the ‘‘ideal’’ specimen, as a result of the reconsolidated according to the Bjerrum (1973) procedure
progressively lower void ratio at which undrained loading and allowed to ‘‘age’’ for 40 h under drained conditions before
takes place. undrained loading (Fig. 18); the simulations have been carried
The results of the simulations described above are out with axial strain rates equal to 1%/h (as common practice)
compatible with the experimental results published by and 0.1%/h, since Callisto et al. (2002) do not specify the axial
Tanaka and Tanaka (1999), Hight (2001) and Horng strain rate used for the laboratory tests.
et al. (2011) for many structured clays. The model predicts the experimentally observed undrained
Once normalized to the initial mean effective pressure (p0 o), post-peak strain softening without invoking localization,
the results of the simulations are also in agreement with the suggested by Callisto et al. (2002) as a possible cause.
experimental results by Callisto et al. (2002) for TX-CAU-C Due to the dependence of M on the Lode angle y, the
and TX-CAU-E tests on high quality UPC samples simulations yield an undrained strength in triaxial extension
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334 327

(tmax E) which is approximately 75% of that in triaxial 2.00

compression (tmax C), in line with experimental results for 1.75

the UPC and as reported for other high plasticity clays 1.50
(Jamiolkowski et al., 1985; Ladd, 1991; Tsuchida, 2000). At

e (-)
1.25
large strains the ratio tmax  E/tmax C reduces to 0.85.
1.00
Consideration is being given to modifying the constitutive
0.75
model to address evolving inherent anisotropic clay beha-
viour, as suggested for example by Yin et al. (2011), to 0.50
1 10 100 1000 10000
improve the prediction of Fig. 18 and to reflect experimental p' (kPa)
observations of differences between compression and exten-
sion behaviour also in plane strain conditions (as found by Fig. 20. UPC – Numerical TX-CAU tests – SHANSEP reconsolidation
Vaid and Campanella, 1974 for sensitive Haney clays). procedure: current states at the end of the reconsolidation phase in the
elog10(p0 ) plane.

5.5. TX-CAU tests—SHANSEP approach 10

8
Fig. 19 shows the numerically simulated stress-strain curves 6
and the stress-paths for TX-CAU-C and TX-CAU-E tests 4
carried out at ėa ¼ 1%/h on ‘‘nominally undisturbed’’ speci- 2

t / p'e (-)
mens prepared following the SHANSEP reconsolidation 0
procedure, with s0 vmax ¼ 2s0 vp and OCR*¼ 1.4, together with -2
the results of benchmark tests on ‘‘ideal’’ specimens, all results -4
being normalized to the vertical effective consolidation -6
pressure s0 vc at the start of undrained loading. -8
For (SD1)EOP o27%, the shape of the stress–strain curves -10
0 2 4 6 8 10 12 14 16 18 20
and the effective stress paths of the damaged specimens,
s' / p'e (-)
normalized to s0 vc, are similar to the behaviour of the ‘‘ideal’’
specimen; all these specimens reach the same state before 10

undrained loading, irrespective of the degree of disturbance; 8

6
0.50
4
0.40
2
t / p'e (-)

0.30
0
0.20
-2
0.10
t /σ'vc (-)

-4
0.00
-6
-0.10
-8
-0.20
-10
-0.30 0 2 4 6 8 10 12 14

-0.40 εa (%)

-0.50
0.0 0.2 0.4 0.6 0.8 1.0 Fig. 21. UPC – Numerical TX-CAU tests - SHANSEP reconsolidation
s' /σ'vc (-) procedure: (a) effective stress paths and (b) stress–strain curves (normal-
ization to p0 e).
0.50

0.40

0.30
this is because they are consolidated to a pressure where the
0.20
oedometric curves for the different degrees of disturbance
0.10
converge, Fig. 5.
t /σ'vc (-)

0.00
At (SD1)EOP ¼ 89% the degree of strain softening of the
-0.10

-0.20
‘‘ideal’’ specimen is nearly lost, even though, in spite of the
-0.30
induced damage, the SHANSEP reconsolidation proce-
-0.40
dure results in a significant regain of strength compared to
-0.50
UC and TX-UU tests.
0 2 4 6 8 10 12 14 The irreversible damage caused by the SHANSEP
εa (%)
procedure can be inferred by normalizing the results of
Fig. 19. UPC – Numerical TX-CAU tests – SHANSEP reconsolidation
each test to the p0 e on the ICLoe corresponding to the void
procedure: (a) effective stress paths; (b) stress–strain curves (normal- ratio of each specimen at which undrained loading takes
ization to s0 vc). place (Fig. 20), as shown in Fig. 21.
328 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334

Similar observations have been made by Burland (1990) 700

600
and Horng et al. (2011) on the basis of experimental results
500
on Troll Field (upper clay) and Takuhoku clay respectively.

tmax (kPa)
400
As highlighted by De Groot (2001), strictly speaking, the 300
inherent assumptions (normalized behaviour, governed by 200
mechanical overconsolidation) make the method inap- 100
propriate in structured clays like UPC. Ladd and Foott 0
(1974) noted that in structured materials the SHANSEP 0.2 0.4 0.6 0.8 1 1.2
ln(v)
technique destroys important aspects of the microstruc-
ture, feeling however that, in spite of this, the procedure Fig. 23. UPC – Numerical TX-CAU-C tests – Undrained shear strength
provides reasonable estimates of the in situ strength and vs specific volume for high quality and remoulded specimens.
reasonable results in engineering projects involving a
variety of clays of low to moderate sensitivity, especially 1.5

when compared with UC and TX-UU tests. 1.3

up
In spite of the greater damage induced in the specimens pe
rb
ou
1.0 nd
by the SHANSEP procedure, due to the greater void ratio (M
itch
ell,
19
reductions, the normalized undrained shear strength from 0.8 76
)
low
the SHANSEP and the Bjerrum (1973) approaches do not er

LI (-)
bou
0.5 nd
(M
differ significantly. itch
ell,
197
0.3 6)
In apparent contrast with reports by Shibuya et al. (1999),
Tsuchida (2000) and De Groot (2001), the normalized 0.0

undrained shear strength obtained by applying the SHANSEP -0.3

procedure is over 25% higher than that obtained by applying
the Bjerrum (1973) reconsolidation procedure. This difference
would reduce if the strength obtained by the two procedures
were normalized by the mean effective consolidation pressure
p0 c reached before the start of undrained loading, rather than
s0 vc, cancelling the effect of higher Ko values induced by the
SHANSEP procedure. If the normalized undrained strength is
divided by (OCR)0.85, it becomes less than that of the ‘‘ideal’’
specimen and in line with that obtained with the Bjerrum
reconsolidation procedure. Comparable results are obtained if
the SHANSEP procedure is modified omitting the unloading
phase so that s0 vc ¼ s0 vmax.
The relative effectiveness of the two procedures depends
on the specific clay under examination and the details of
the experimental programme.
5.6. Persistence of microstructure
For two typical levels of disturbance, i.e. (SD1)EOP ¼ 3.5%
(high quality specimen) and (SD1)EOP ¼ 89% (very poor
700
600
500
tmax (kPa)
-0.5
1 10 100 1000
tmax (kPa)
Fig. 24. UPC – Numerical TX-CAU-C tests – Undrained shear strength
vs Liquidity Index for high quality and remoulded specimens.
quality specimen, representative of remoulded conditions),
Fig. 22 shows the undrained shear strength in compression
found by simulations of anisotropically consolidated
(s0 hc ¼ 0.6s0 vc) tests on UPC at different values of s0 vc. In line
with the experimental results reported by Hong et al. (2006)
and Hong et al. (2007) for the Ariake and Fukuoka clays and
by Graham and Li (1985) for the Winnipeg clay, whilst the
undrained shear strength of the remoulded specimens increa-
ses linearly with s0 vc, for high quality specimens this is true
only for consolidation stresses above a threshold value (s0 vc)tr,
below which the specimens have proportionally higher
strengths.
At equal specific volume (Fig. 23) or Liquidity Index (LI)
(Fig. 24), the high quality specimen has higher undrained shear
strength than the remoulded specimen, thanks to some
microstructure being preserved in the high quality specimen,
as demonstrated also by the behaviour in the oedometer tests.
At equal s0 vc 4(s0 vc)tr, the undrained shear strength of the

400
remoulded specimen is higher than that of the high quality
300
200
specimen, due to different void ratios (or specific volumes)
100
reached by the specimens at the end of the consolidation
0 phase; for example, at s0 vc ¼ 2  s0 vo ¼ 270 kPa the void ratio of
0 500 1000 1500 2000 2500 the remoulded and of the high quality specimens are approxi-
σ'vc (kPa)
mately equal to 1.1 and 1.5 respectively (Fig. 23).
Fig. 22. UPC – Numerical TX-CAU-C tests – Undrained shear strength
Similar considerations are suggested by experimental results
vs vertical effective consolidation pressure for high quality and remoulded on Ariake clays (Hong et al., 2007, Figs. 3, 6 and 7 therein);
specimens. one may be misled into not recognizing the different undrained
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
shear strengths of high quality and remoulded specimens by
the lack of tests on remoulded specimens at high specific
volumes.
The undrained shear strengths found by the simulations
on remoulded specimens (Fig. 24) compare well with both
the undrained shear strengths reported for remoulded clays
by Mitchell (1976) and Leroueil et al. (1983) and the
‘‘critical state’’ undrained strength predicted for normally
consolidated materials using the modified CAM-CLAY
model combined with the intrinsic deformability (li ; nli ; ki)
and effective strength parameters of the UPC (Table 2).
The persistence of microstructure in well sampled struc-
tured clays, as reflected by higher undrained shear strengths
than for specimens reconstituted at equal void ratio, is also
shown by the tests on Saint Jean Vienney clays discussed by
Saihi et al. (2002). Their experimental evidence and the
results of simulations presented here indicate that in triaxial
tests on well sampled structured clays the large strain
(ea ffi 10–15%) undrained shear strength is often still affected
by residual microstructural effects and may not represent
remoulded (critical state) conditions. Similar conclusions are
drawn by Burland (1990) on the basis of experimental results
on Troll Field upper clay.
6. Laboratory versus field behaviour of structured clays
Sampling and testing techniques influence the results of
laboratory tests on structured clays. Advanced constitutive
modelling and careful back-analysis of laboratory tests can be
used to determine the behaviour of ‘‘ideal’’ specimens, also
validating to some extent the procedures proposed in the past
to achieve this. However, the laboratory behaviour of ‘‘ideal’’
specimens does not necessarily reflect the field behaviour of
structured clays in engineering applications.
It is widely recognized that in stability analyses involving
structured clays it is necessary to consider several peculiar
aspects of their behaviour, such as inherent and stress
induced anisotropy, strain softening with associated pro-
gressive failure mechanisms, as well as strain/stress rate
effects (see for example Bjerrum, 1972; Ladd and Foott,
1974; Ladd et al., 1977; Track et al., 1980; Chapuis, 1982;
Graham et al., 1983; Jamiolkowski et al., 1985; Mesri,
1989; Ladd, 1991; Tsuchida, 2000).
For the simplest case of single stage construction, in
routine stability analyses, typically still based on limit
equilibrium methods, this is normally done by applying
one of the following empirical approaches to determine the
‘‘operational’’ undrained shear strength (su,op) using:
 empirical correction factors together with well defined
sampling and laboratory testing procedures; for example,
the Japanese ‘‘qu method’’ (Nakase, 1967; Tsuchida, 2000)
relies on undrained shear strengths from UC tests carried
out at an axial strain rate of 60%/h on ‘‘nominally
undisturbed’’ specimens taken by JPN piston samplers;
 (a) reconsolidation procedures to remedy the effects of
sample disturbance, (b) different stress paths to account
329
for inherent and stress induced anisotropy and strain
compatibility and (c) correcting the laboratory results to
account for strain/stress rate effects (Ladd, 1991).
Strictly speaking, empirical correction factors should be
obtained from back analyses of full scale tests to failure,
appropriately designed to reproduce situations comparable
to those foreseen for the proposed construction (La
Rochelle et al., 1974; Crawford et al., 1995). Published
empirical correction factors should be used with care,
recognizing that they may have been derived with reference
to specific sampling and testing techniques. As highlighted
by Hight (2001), there are inherent dangers in improving
sampling and testing methods if the results are used in
conjunction with procedures calibrated on the basis of soil
data from other experimental methods.
Even where the same sampling and testing methods are
used, differences in soil layering and microstructure of the
clays or in the geometry and construction details (including
time schedule) of the works may generate a large scatter in the
data used to determine the empirical correlations and possibly
invalidate their applicability to the specific case of interest.
While supporting the merits of the second approach, this
study shows that for good quality samples of UPC the
Japanese ‘‘qu method’’ could yield comparable operational
undrained shear strengths. For specimens characterized by
s0 vo ¼ 135 kPa and (SD1)EOP of the order of 3.5%–13%, the
operational undrained shear strength can be determined as:
 qu/2 ffi 33 kPa (for ėa ¼ 60%/h, Japanse method), based
on qu/2 ffi25 kPa (Figs. 11, ėa ¼ 1%/h) and a strain rate
correction factor equal to 1.3 to 1.35 (Fig. 13).
 su,op ¼ 0.5  (suC þ suE) ¼ 35 kPa, based on j0 cv ¼ 241
(Table 1), suC ¼ tmax  C cos j0 cv, suE ¼ tmax  E cos j0 cv
(Ladd, 1991) and the results of TX-CAU-C and TX-
CAU-E tests (Fig. 14) carried out according to the
Bjerrum (1973) reconsolidation procedure at strain rates
(e_a ¼ 0.5–1%/h) generally assumed to be comparable
with field conditions (Nakase, 1967; Ladd and Foott,
1974; Tsuchida, 2000).
Substantially consistent results have been reported by
Tsuchida (2000) for the Tokyo and Osaka bay structured
clays when tested by the two approaches described above,
irrespective of the reconsolidation procedure adopted
(Bjerrum or SHANSEP).
These values of operational undrained strength, which refer
to incipient failure, are much higher than the undrained shear
strength predicted by the model at equal void ratio (or
Liquidity Index) for remoulded conditions, where the micro-
structure effects are fully removed (for UPC,o5 kPa,
i.e.o0.04s0 vp, at s0 vc ¼ s0 vo), as may be relevant to landslide
run-out or to the smear zone around driven piles during and
immediately after installation (Fig. 24).
Whilst it is widely recognized that in conventional limit
equilibrium stability analyses the undrained shear strength
from high quality ‘‘nominally undisturbed’’ specimens
needs to be reduced to reflect the true field behaviour of
330 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334

the soil, irrespective of the nature of the clay, settlement  the modified CAM-CLAY elasto-plastic constitutive
analyses are often based on: model, consistent with the principle of the uniqueness
of the EOP compression curves, assuming l¼ 0.177;
a) laboratory compression curves from high quality ‘‘nom- nl ¼ 4.457; k ¼ 0.053, OCR ¼ 1.3–1.4, determined from
inally undisturbed’’ specimens, corrected for disturbance EOP curves of ‘‘ideal’’ specimens.
effects;
b) the principle of the uniqueness of the EOP void ratio-
effective stress relationship (Ladd et al., 1977; Mesri In both cases the initial void ratios are represented by a
and Choi, 1985a; Jamiolkowski et al., 1985); line parallel to the ICLoe, passing through eo ¼ 1.73
c) the assumption that during ‘‘primary’’ consolidation the (corresponding to Ivo ¼ 1.2 for LL ¼ 70%) and s0 vo ¼ 135
‘‘conventional’’ preconsolidation pressure s0 vp repre- kPa (Fig. 2); furthermore, the model parameters fit the
sents the boundary between the occurrence of small same oedometric compression curve representative of the
(elastic) versus large (elasto-plastic) strains; behaviour of an ‘‘ideal’’ specimen characterized by eo ¼
d) the occurrence of additional settlements beyond those 1.73 and s0 vo ¼ 135 kPa (Fig. 9). The elasto-plastic para-
developed at the EOP condition, evaluated according to meters of the two models are not directly comparable.
the popular cae/Cc=constant criterion (Mesri and In the elasto-viscoplastic model, li refers to the intrinsic
Godlewski, 1977; Mesri, 1987). parameters of reconstituted material. The value of l of the
elasto-plastic model is necessarily higher because it refers
As highlighted, for example, by Chang (1981), Kabbaj to the compressibility of structured material and it needs to
et al. (1988), Leroueil et al. (1988), Yin et al. (1994), Kim incorporate the viscous component of compressibility,
and Leroueil (2001), Zhu and Yin (2001), Rocchi et al. albeit in a gross and inappropriate manner.
(2003), Mimura and Jang (2005), Tanaka (2005a, 2005b), The calculated settlement (dv) at the top of the UPC
Rocchi et al. (2006, 2007), Inagaki et al. (2010), Tashiro layer and the excess pore pressure in the middle of the layer
et al. (2011), Mirjalili et al. (2012), for structured clays this (Dum) are shown in Fig. 25 as a function of time for the
approach may lead to erroneous predictions of both two constitutive models in terms of dv/Ho  log10(T/H2d)
settlement and excess pore pressure dissipation rates. and Dum/D0 sv  log10(T/H2d), where Ho is the initial thick-
This view is further supported by numerical calculations of ness of the layer and Hd is the length of the drainage path;
settlements carried out for a hypothetic soil profile consisting the corresponding numerically simulated oedometer tests
of a 16 m thick layer of UPC overlain by 2 m of sand and on an ‘‘ideal’’ specimen sampled in the middle of the layer
resting on a rigid, pervious base, with the groundwater table are also shown for comparison.
at 2 m depth. This profile is progressively loaded in 90 days in Unlike ‘‘theory A’’ (Ladd et al., 1977) and predictions
1D (oedometric) conditions by a pressures Ds0 v ¼ 92.5 kPa. based on the modified CAM CLAY model (Fig. 25b), the
The initial conditions and stress changes in the UPC are results of the elasto-viscoplastic model do not follow the
summarized in Table 3. The boundary value problem has H2d rule (Fig. 25a).
been solved by FEM analyses. As a consequence of the destruction processes activated
Two sets of calculations have been carried out, using: in the metastable microstructure and of the low perme-
ability of the materials, excess pore pressures in the UPC
 the elasto-viscoplastic constitutive model developed by layer may even exceed the applied pressure Ds0 v.
Rocchi et al. (2003), modified as detailed earlier, and the This is consistent with the results published by Tanaka
associated parameters in Table 2; (2005b) for interconnected oedometer tests on structured
Pleistocene Osaka clays and with monitoring results from
many projects on structured clays (Chang, 1981; Hansbo
Table 3 et al., 1983; Mesri and Choi, 1985b; Kabbaj et al., 1988;
Hypothetic UPC profile for settlement calculations. Leroueil et al., 1988; Indraratna et al., 1994; Kim and
Leroueil, 2001; JGS, 2005; Noda et al., 2005b), which show
za,b s0 vo s0 vpc s0 vf d s0 vf/ s0 vp eoe ka
high rates of settlement associated with high excess pore
(m) (kPa) (kPa) (kPa) (–) (–) (m/s)
pressures in contrast with predictions based on conven-
2.5 39.9 53.9 132.4 2.45 2.07 1.2  10  9 tional elasto-plastic models which do not simulate destruc-
6.0 60.2 81.3 152.7 2.34 1.95 8.9  10  10 tion processes.
10.0 83.4 112.6 175.9 1.56 1.86 7.0  10  10
As implicit in the model, Fig. 25b shows that the modified
14.0 106.6 143.9 199.1 1.38 1.79 5.9  10  10
18.0 129.8 175.2 222.3 1.27 1.74 5.2  10  10 CAM-CLAY model follows the H2d rule; however the
modified CAM CLAY model significantly under-predicts
Notes: both settlements and excess pore pressures compared to the
a
See notation. best estimate of field behaviour provided by the elasto-
b
depth b.g.l.
c 0
s vp=(1.3 1.4)  s0 vo.
viscoplastic model. The settlement underprediction is not
d 0
s vf=s0 vo+ns0 v. remedied by consideration of ‘‘secondary consolidation’’
e
determined from Fig. 2. starting after dissipation of excess pore pressures.
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334 331

0 0.0
reasonable accuracy from the ‘‘back analysis’’ of tests
2
on ‘‘nominally undisturbed’’ specimens;
4 0.3
 empirical corrections typically applied to oedometer test
6
results (Schmertmann, 1953 or Nagaraj et al., 2003)

Δum/Δσ'v (-)
δv / Ho (%)

8 0.6

10
tend to overcorrect both s0 vp and Cc, with increasing
12 0.9
error as sampling disturbance increases;
14
 there is no possibility of recovering the damage to the
16 1.2 microstructure induced by sampling; however, the
18 reduction of void ratio induced by the reconsolidation
20 1.5 techniques proposed by Bjerrum (1973) and by Ladd
1 10 100 1000 10000 100000
and Foott (1974) (and subsequent setting time, where
T / Hd2 (h / m2)
carried out as suggested by Tatsuoka et al., 1997), tends
0 0.0 to compensate the effects of damage, typically resulting
2 in a substantial regain of the strength lost due to
4 0.3 sampling disturbance;
6
 there are inherent dangers in applying conventional
Δum/Δσ'v (-)
δv / Ho (%)

8 0.6
simplified procedures for stability and settlement pre-
10
dictions, such as are often described in codes of practice
12 0.9
and manuals, particularly if reference is made to test
14
16 1.2
results obtained by improved methodologies (technolo-
18
gical, experimental and/or interpretative, as those
20 1.5 described in this paper) aimed to determine the response
1 10 100 1000 10000 100000 of ‘‘ideal’’ specimens.
T / Hd2 (h / m2)

Fig. 25. Comparison of numerically simulated field behaviour for the

hypothetic UPC soil profile loaded in 1D conditions to 92.5 kPa and
numerically simulated oedometer tests on an ideal specimen sampled at
mid-layer—normalized values.
The relative merits of predictions based on the two
models have been discussed by Rocchi et al. (2006) for two
well documented case histories in the Osaka Bay.
The correct prediction of excess pore pressure distribu-
tion within the clay deposits and its evolution with time is
also significant for evaluating the stability in ‘‘staged
construction’’ problems (Ladd, 1991).
7. Conclusions
Whilst in situ tests are useful to evaluate the spatial
variability of layering, only the determination of the initial
states relative to the ICLoe and comparative mechanical
laboratory tests on natural and reconstituted specimens
can provide the information necessary for the geotechnical
characterization of clay deposits and for the selection of
appropriate constitutive models.
In structured clays such as UPC, whose initial state is
above the ICLoe:
 the use of an elasto-viscoplastic constitutive model
which simulates the destruction processes taking place
in the microstructure, together with the application of a
modified form of the ISA approach, provides some
insight into the key effects of sampling disturbance on
the results of laboratory tests;
 the compressibility and the strength of ‘‘ideal’’ (truly
undisturbed) specimens can be determined with
Standard practice developed on the basis of a combina-
tion of simple computational methods and the use of
‘‘operational’’ strength and/or deformability parameters,
corrected for sampling disturbance, needs to be revisited.
The large scatter observed in popular empirical relation-
ships, such as those which use corrected field vane
undrained shear strength for embankment stability ana-
lyses or empirical factors applied to undrained shear index
tests such as torvane, UC and TX-UU tests for the
evaluation of pile shaft capacity, could arise from plotting
in the same graph experiences characterized by different
microstructural characteristics (sensitivity of the clay), site
investigation techniques and construction details.
The apparent conflict between settlement and excess
pore pressures dissipation rates often observed in many
projects might be solved by recognizing the inapplicability,
for these materials, of both the principle of the uniqueness
of the EOP void ratio-effective stress relationship and
Terzaghi0 s consolidation theory, rather than by casting
doubt on the reliability of the piezometric data or, in the
case of vertical drains, by placing excessive emphasis on
other aspects such as well resistance, smear effects, loss of
efficiency with time.
Significant progress in the field of geotechnical investi-
gation, soil characterization, selection of soil constitutive
models and assessment of relevant soil parameters through
back analysis of laboratory tests, including techniques
dealing with sampling disturbance, combined with current
software and hardware capabilities, make it now easier,
more reliable and less time consuming to use more
advanced design methods than those routinely adopted
so far. The increasing need to carry out ‘‘performance
based design’’, which places greater emphasis on the
332 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
prediction of displacements rather than relying only on
safety factors, pushes the profession in this direction.
Acknowledgements
The authors are well aware that the work presented here
could not have been conceived but for the efforts of so
many who have come before. For each citation made many
more were necessarily omitted; we are equally beholden to
all. We are particularly grateful to Prof. M. Jamiolkowski
for fostering the cultural and scientific environment at
Studio Geotecnico Italiano which made this work possible.
References
Andresen, A., Kolstad, P., 1979. The NGI 54 sampler for undisturbed
sampling of clays and representative sampling of coarser materials. In:
Proceedings of the International Symposium on Soil Sampling,
Singapore, I.S.S.M.F.E., pp. 13–21.
Asaoka, A., Nakano, M., Noda, T., 2000a. Superloading yield surface concept
for highly structured soil behaviour. Soils and Foundations 40 (2), 99–110.
Asaoka, A., Nakano, M., Noda, T., Kaneda, K., 2000b. Delayed
compression/consolidation of natural clay due to degradation of soil
structure. Soils and Foundations 40 (3), 75–85.
Atkinson, J.H., Allman, M.A., Boese, R.J., 1992. Influence of laboratory
sample preparation procedures on the strength and stiffness of intact
Bothkennar soil recovered using the Laval sampler. Géotechnique 42
(2), 349–354.
Baligh, M.M., Azzouz, A.S., Chin, C.T., 1987. Disturbance due to ideal
tube sampling. ASCE Journal of Geotechnical Engineering 113 (7),
739–757.
Berardi, G., Caroti, L., Giunta, M., Jamiolkowski, M., Lancellotta, R. ,
1991. Mechanical properties of upper Pisa clay. In: Proceedings of the
10th European Conference on Soil Mechanics and Foundation
Engineering, Florence, 7, pp. 7–10.
Bjerrum, L., 1967. Engineering geology of Norwegian normally consoli-
dated marine clays as related to settlements of buildings. Géotechnique
17 (2), 81–118.
Bjerrum, L., 1972. Embankments on soft ground. In: Proceedings of the
ASCE Specialty Conference on Earth and Earth Supported Struc-
tures, Purdue University, West Lafayette, 2, pp. 1–54.
Bjerrum, L., 1973. Problems of soil mechanics and construction on soft
clays. In: Proceedings of the 8th International Conference on Soil
Mechanics and Foundation Engineering, Moscow, 3, pp. 111–159.
Boudali, M., 1995. Comportement tridimensionnel et visqueux des argiles
naturelles. Ph.D. thesis, Laval University, Québec.
Budhu, M., Wu, C.S., 1992. Numerical analysis of sampling disturbances
in claly soils. International Journal for Numerical and Analytical
Methods in Geomechanics 16, 467–492.
Burland, J.B., 1990. On the compressibility and shear strength of natural
clays. Géotechnique 40 (3), 329–378.
Butterfield, R., 1979. A natural compression law for soils. Géotechnique
29 (4), 469–480.
Callisto, L., Calabresi, G., 1998. Mechanical behaviour of a natural soft
clay. Géotechnique 48 (4), 495–513.
Callisto, L., Gajo, A., Muir Wood, D., 2002. Simulation of triaxial and
true triaxial tests on natural and reconstituted Pisa clay. Géotechnique
52 (9), 649–666.
Casagrande, A., (1936): The determination of the pre-consolidation load
and its practical significance. In: Proceedings of the 1st International
Conference on Soil Mechanics and Foundation Engineering, Harvard
University, Cambridge, Massachussets, 3, D-34, pp. 60–64.
Casagrande, A., Fadum, R.E., 1944. Reply to discussion on ‘‘Application
of soil mechanics in designing building foundations’’. Transactions of
ASCE, paper no. 2213, pp. 383–490.
Chang, Y.C.E., 1981. Long term consolidation beneath the test fills at Väsby,
Sweden. Swedish Geotechnical Institute, Report no. 13, Linköping.
Chapuis, R.P., 1982. New stability method for embankments on clay
foundations. Canadian Geotechnical Journal 19 (1), 44–48.
Clayton, C.R.I., Hight, D.W., Hopper, R.J., 1992. Progressive destruc-
turing of Bothkennar clay: implication for sampling and reconsolida-
tion procedures. Géotechnique 42 (2), 219–239.
Clayton, C.R.I., Siddique, A., Hopper, R.J., 1998. Effects of sampler
design on tube sampling disturbance – numerical and analytical
investigations. Géotechnique 48 (6), 847–867.
Clayton, C.R.I., Siddique, A., 1999. Tube sampling disturbance –
forgotten truths and new perspectives. Geotechnical Engineering,
ICE 137 (3), 127–135.
Cotecchia, F., Chandler, R.J., 2000. A general framework for the
mechanical behaviour of clays. Géotechnique 50 (4), 431–447.
Crawford, C.B., Fannin, R.J., Kern, C.B., 1995. Embankment failures at
Vernon, British Columbia. Canadian Geotechnical Journal 32 (2), 271–284.
De Groot, D.J., 2001. Laboratory measurement and interpretation of soft
clay mechanical behavior. In: Proceedings of the Symposium on Soil
Behavior and Soft Ground Construction, Geotechnical Special Pub-
lication no. 119, Germaine, J.T., Sheahan, T.C., Whitman, R.V.
(Eds.), ASCE-GEO INSTITUTE, pp. 167–200.
Feng, T.W., 1991. Compressibility and permeability of natural soft clays
and surcharging to reduce settlements. Ph.D. Thesis, University of
Illinois at Urbana-Champaign.
Georgiannou, V.N., Hight, D.W., 1994. The effect of centerline tube
sampling strains on the undrained behavior of two stiff overconsoli-
dated clays. ASTM Geotechnical Testing Journal 17 (4), 475–482.
Germaine, J.T., Ladd, C.C., 1988. State-of-the-Art: triaxial testing of
saturated cohesive soils. Advanced Triaxial Testing of Soil and Rock,
ASTM STP 977, 421–459.
Graham, J., Crooks, J.H.A., Bell, A.L., 1983. Time effects on the stress–
strain behaviour of natural soft clays. Géotechnique 33 (3), 327–340.
Graham, J., Li, E.C.C., 1985. Comparison of natural and remolded plastic
clay. ASCE Journal of Geotechnical Engineering 111 (7), 865–881.
Graham, J., Kwok, C.K., Ambrosie, R.W., 1987. Stress release,
undrained storage and reconsolidation in simulated underwater clay.
Canadian Geotechnical Journal 24 (1), 279–288.
Graham, J., Lau, S.L.K., 1988. Influence of stress-release disturbance,
storage and reconsolidation procedures on the shear behaviour of
reconstituted underwater clay. Géotechnique 38 (2), 279–300.
Grammatikopoulou, A., Zdravkovic, L., Potts, D.M., 2007. The effect of
the yield and plastic potential deviatoric surfaces on the failure height
of an embankment. Géotechnique 57 (10), 795–806.
Hansbo, S., Jamiolkowski, M., Kok, M., 1983. Consolidation by vertical
drains. Géotechnique 31 (1), 45–66.
Hight, D.W., 2001. Sampling effects in soft clay: an update on Ladd and
Lambe (1963). In: Proceedings of the Symposium on Soil Behavior
and Soft Ground Construction, Geotechnical Special Publication no.
119, Germaine, J.T., Sheahan, T.C., Whitman, R.V. (Eds.), ASCE-
GEO INSTITUTE, pp. 86–121.
Hight, D.W., Boese, R., Butcher, A.P., Clayton, C.R.I., Smith, P.R.,
1992. Disturbance of the Bothkennar clay prior to laboratory testing.
Géotechnique 42 (2), 199–217.
Hinchberger, S.D., Qu, G., 2009. Viscoplastic constitutive approach for rate-
sensitive structured clays. Canadian Geotechnical Journal 46 (6), 609–626.
Hong, Z., Onizuka, K., 1998. A method of correcting yield stress and
compression index of Ariake clays for sample disturbance. Soils and
Foundations 38 (2), 211–222.
Hong, Z., Liu, S., Shen, S., Negami, T., 2006. Comparison in undrained shear
strength between undisturbed and remoulded Ariake clays. ASCE Journal
of Geotechnical and Geoenvironmental Engineering 132 (2), 272–275.
Hong, Z., Han, J.A., 2007. Evaluation of sample quality of sensitive clay
using intrinsic compression concept. ASCE Journal of Geotechnical
and Geoenvironmental Engineering 133 (1), 83–90.
Hong, Z., Shen, S., Deng, Y., Negami, T., 2007. Loss of soil structure for
natural sedimentary clays. Geotechnical Engineering, ICE 160 (GE3),
153–159.
 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
Horng, V., Tanaka, H., Obara, T., 2010. Effects of sampling tube
geometry on soft clayey sample quality evaluated by nondestructive
methods. Soils and Foundations 50 (1), 93–107.
Horng, V., Tanaka, H., Hirabayashi, H., Tomita, R., 2011. Sample
disturbance effects on undrained shear strengths—study from Taku-
hoku site, Sapporo. Soils and Foundations 51 (2), 203–213.
Inagaki, M., Nakano, M., Noda, T., Tashiro, M., Asaoka, A., 2010.
Proposal of a simple method for assessing the susceptibility
of naturally deposited clay grounds to large long-term settlement due
to embankment loading. Soils and Foundations 50 (1), 109–122.
Indraratna, B., Balasubramaniam, A.S., Ratnayake, P., 1994. Perfor-
mance of embankment stabilized with vertical drains on soft clay.
ASCE Journal of Geotechnical Engineering 120 (2), 257–273.
Jamiolkowski, M., Lancellotta, R., Wolski, W., 1983. Precompression and
speeding up consolidation. In: Proceedings of the 8th European
Conference on Soil Mechanics and Foundation Engineering, Helsinki,
3, pp. 1201–1226.
Jamiolkowski, M., Ladd, C.C., Germaine, J.T., Lancellotta, R., 1985. New
developments in field and laboratory testing of soils. In: Proceedings of
the 11th European Conference on Soil Mechanics and Foundation
Engineering, San Francisco, 1, pp. 57–153.
JGS, 2005. In: Proceedings of the Symposium on Geotechnical Aspects of
Kansai Intern. Airport. KIA Land Development Co., Ltd., Japanese
Geotechnical Society, pp. 1–96.
Kabbaj, M., Tavenas, F., Leroueil, S., 1988. In situ and laboratory stress–
strain relationships. Géotechnique 38 (1), 83–100.
Kavvadas, M., Amorosi, A., 2000. A constitutive model for structured
soils. Géotechnique 50 (3), 263–273.
Kim, Y.T., Leroueil, S., 2001. Modeling the viscoplastic behaviour of clays
during consolidation: application to Berthierville clay in both laboratory
and field conditions. Canadian Geotechnical Journal 38 (3), 484–497.
Kimoto, S., Oka, F., 2005. An elasto-viscoplastic model for clay
considering destructuralization and consolidation analysis of unstable
behavior. Soils and Foundations 45 (2), 29–42.
Kirkpatrick, W.M., Khan, A.J., 1984. The reaction of clays to sampling
stress relief. Géotechnique 34 (1), 29–42.
Kirkpatrick, W.M., Khan, A.J., Mirza, A.A., 1986. The effects of stress
relief on some overconsolidated clays. Géotechnique 36 (4), 511–525.
Ladd, C.C., 1991. Stability evaluation during staged construction. ASCE
Journal of Geotechnical Engineering 117 (4), 540–615.
Ladd, C.C., Lambe, R.W., 1963. The strength of undisturbed clay
determined from undrained tests. In: Proceedings of the ASTM-NRC
Symposium on Shear Testing of Soils, Ottawa, STP 361, pp. 342–371.
Ladd, C.C., Foott, R., 1974. New design procedure for stability of soft
clays. ASCE Journal of Geotechnical Engineering 100 (7), 763–786.
Ladd, C.C., Foott, R., Ishihara, K., Schlosser, F., Poulos, H.G., 1977.
Stress-deformation and strength characteristics. In: Proceedings of the
9th International European Conference on Soil Mechanics and
Foundation Engineering, Tokyo, 4, 421–494.
La Rochelle, P., Trak, B., Tavenas, F., Roy, M., 1974. Failure of a test
embankment on a sensitive Champlain clay deposit. Canadian
Geotechnical Journal 11 (1), 142–164.
La Rochelle, P., Sarrailh, J., Tavenas, F., Roy, M., Leroueil, S., 1981.
Causes of sampling disturbance and design of a new sampler.
Canadian Geotechnical Journal 18 (1), 52–66.
Leroueil, S., Tavenas, F., Le Bihan, J.P., 1983. Propriétés caractéristiques
des argiles de l0 est du Canada. Canadian Geotechnical Journal 20 (4),
681–705.
Leroueil, S., Kabbaj, M., Tavenas, F., Bouchard, R., 1985. Stress–strain–
strain rate relation for the compressibility of sensitive natural clays.
Géotechnique 35 (2), 159–180.
Leroueil, S., Kabbaj, M., Tavenas, F., 1988. Study of the validity of
a s0 v, ev , e_ v model in in situ conditions. Soils and Foundations 40 (3),
13–25.
Leroueil, S., Vaughan, P.R., 1990. The general and congruent effects of
structure in natural soils and weak rocks. Géotechnique 40 (3), 467–488.
Lo Presti, D.C.F., Jamiolkowski, M., Pepe, M., 2003. Geotechnical
characterization of the subsoil of the Pisa Tower. In: Proceedings of
333
the International Workshop on Characterization and Engineering
Properties of Natural Soils, Singapore, Tan, T.S., Phoon, K.K.,
Hight, D.W., Leroueil S. (Eds.), Swets & Zeitlinger, 2, pp. 909–946.
Lunne, T., Berre, T., Strandvik, S., 1997. Sample disturbance in soft low
plastic Norwegian clay. In: Almeida, M. (Ed.), Recent Developments
in Soil and Pavement Mechanics. Balkema, Rotterdam, pp. 81–102.
Mesri, G., 1987. The fourth law of soil mechanics: the law of compres-
sibility. In: Proceedings of the International Symposium on Geotech-
nical Engineering of Soft Soils, Mexico City, 2, pp. 179–187.
Mesri, G., 1989. A re-evaluation of su(mob) ¼ 0.22s0 p using laboratory
shear tests. Canadian Geotechnical Journal 26 (1), 162–164.
Mesri, G., Godlewski, P.M., 1977. Time- and stress-compressibility inter-
relationship. ASCE Journal of Geotechnical Engineering 103 (5),
417–430.
Mesri, G., Choi, Y.K. (1985a): The uniqueness of EOP void ratio—
effective stress relationship. In: Proceedings of the 11th International
European Conference on Soil Mechanics and Foundation Engineer-
ing, San Francisco, 2, pp. 587–590.
Mesri, G., Choi, Y.K., 1985b. Settlement analysis of embankments
on soft clays. ASCE Journal of Geotechnical Engineering 111 (4),
441–464.
Mesri, G., Hayat, T.M., 1993. The coefficient of earth pressure at rest.
Canadian Geotechnical Journal 30 (4), 647–666.
Mesri, G., Shahien, M., Hedien, J.E., 1997. Geotechnical characteristics
and compression of Pisa clay. In: Proceedings of the 14th Interna-
tional European Conference on Soil Mechanics and Foundation
Engineering, Hamburg, 1, pp. 373–376.
Mimura, M., Jang, W.Y., 2005. Long-term settlement of the Pleistocene
deposits due to construction of KIA. In: Proceedings of the of the
Symposium on Geotechnical Aspects of Kansai International Airport,
pp. 77–85.
Mirjalili, M., Kimoto, S., Oka, F., Hattori, T., 2012. Long-term
consolidation analysis of a large-scale embankment construction on
soft clay deposits using an elasto-viscoplastic model. Soils and
Foundations 52 (1), 18–37.
Mitchell, J.K., 1976. Fundamentals of soil behaviour. John Wiley & Sons,
New York.
Nagaraj, T.S., Murthy, B.R.S., Vatsala, A., Joshi, R.C., 1990. Analysis of
compressibility of sensitive soils. ASCE Journal of Geotechnical
Engineering 116 (1), 105–118.
Nagaraj, T.S., Miura, N., 2001. Soft clay behaviour—analysis and
Assessment. Balkema.
Nagaraj, T.S., Miura, N., Chung, S.G., Pasad, K.N., 2003. Analysis and
assessment of sampling disturbance of soft sensitive clays. Géotechni-
que 53 (7), 679–683.
Nakai, T., Shahin, H.M., Kikumoto, M., Kyokawa, H., Zhang, F., Farias,
M.M., 2011. A simple and unified three-dimensional model to describe
various characteristics of soils. Soils and Foundations 51 (6), 1149–1168.
Nakase, A., 1967. The f ¼0 analysis of stability and unconfined
compression strength. Soils and Foundations 7 (2), 33–50.
Noda, T., Yamada, S., Asaoka, A., 2005a. Elasto-plastic behaviour of
naturally deposited clay during/after sampling. Soils and Foundations
45 (1), 51–64.
Noda, T., Asaoka, A, Nakano, M., Yamada, E., Tashiro, M., 2005b.
Progressive consolidation settlement of naturally deposited clayey
soil under embankment loading. Soils and Foundations 45 (5),
39–51.
Okumura, T. (1971): The variation of mechanical properties of clay
samples depending on its degree of disturbance. In: Proceedings of the
4th Asian Regional Conference on Soil Mechanics and Foundation
Engineering, Bangkok, 1, pp. 78–81.
Onitsuka, K., Hong, K., Hara, Y., Yoshitake, S., 1995. Interpretation of
oedometer tests data for natural clays. Soils and Foundations 35 (3),
61–70.
Parry, R.H.G., 1960. Triaxial compression and extension tests on
remoulded saturated clay. Géotechnique 10 (4), 166–180.
Perzyna, P., 1963. The constitutive equations for rate sensitive plastic
materials. Quarterly of Applied Mathematics 20, 321–332.
334 G. Rocchi et al. / Soils and Foundations 53 (2013) 315–334
Rampello, S., Callisto, L., Viggiani, G.M.B., 1996. The Leaning Tower of
Pisa—soil parameters for the numerical modelling of the Tower
resulting from the most recent investigations. Research Contract on
‘‘Modellazione numerica del sottosuolo della Torre di Pisa’’ between
the University of Rome La Sapienza and the Consorzio Progetto
Torre di Pisa (in Italian).
Rampello, S., Callisto, L., 1998. A study on the subsoil of the Tower of
Pisa based on results from standard and high-quality samples.
Canadian Geotechnical Journal 35 (6), 1074–1092.
Rocchi, G., Fontana, M., Da Prat, M., 2003. Modelling of natural soft
clay destruction processes using viscoplasticity theory. Géotechnique
53 (8), 729–745.
Rocchi, G., Vaciago, G., Fontana, M., Plebani, F., 2006. Enhanced
prediction of settlement in structured clays with examples from Osaka
Bay. Geomechanics and Geoengineering 1 (3), 217–237.
Rocchi, G., Vaciago, G., Fontana, M., Plebani, F., 2007. Further insight
into the behaviour of Pleistocene Osaka clays at KIA Phase 1 island.
Geomechanics and Geoengineering 2 (3), 159–173.
Rouainia, M., Muir Wood, D., 2000. A kinematic hardening constitutive
model for natural clays with loss of structure. Géotechnique 50 (2),
152–164.
Saihi, F., Leroueil, S., La Rochelle, P., French, I., 2002. Behaviour of the
stiff and sensitive Saint-Jean-Vianney clay in intact, destructured and
remoulded conditions. Canadian Geotechnical Journal 39 (5),
1075–1087.
Santagata, M.C., Germaine, J.T., 2002. Sampling disturbance effects in
normally consolidated clays. J. Geot. Geoenv. Eng. 128 (12),
997–1006 ASCE.
Santagata, M.C., Sinfield, J.V., Germaine, J.T., 2006. Laboratory simula-
tion of field sampling: comparison with ideal sampling and field data.
ASCE Journal of Geotechnical and Geoenvironmental Engineering
132 (3), 351–362.
Schmidt, B., 1966. Discussion on ‘‘Earth pressures at rest related to stress
history by Brooker, E.W., Ireland, H.O., C.G.J. 2, no. 1, 1-15, 1965’’.
Canadian Geotechnical Journal, 3(4), pp. 239–242.
Schmertmann, J.H., 1953. Estimation the true consolidation behavior of clay
from laboratory test results. Transactiona of ASCE 118 (311), 1–25.
Shibuya, S., Mitachi, T., Tanaka, H., Kawaguchi, T., Lee, I., 1999.
Measurement and application of quasi-elastic properties in geotechni-
cal site characterisation. In: Proceedings of the 11th Asian Regional
Conference on Soil Mechanics and Foundation Engineering, Seoul, 2,
pp. 639–710.
Shogaki, T., 1996. A method for correcting consolidation parameters for
sample disturbance using volumetric strain. Soils and Foundations 36
(3), 123–131.
Siddique, A., Clayton, C.R.I., Hopper, R.J., 1999. The effects of varying
centerline tube sampling disturbance on the behaviour of reconstituted
clay. ASTM Geotechnical Testing Journal 22 (3), 245–256.
Skempton, A.W., Sowa, V.A., 1963. The behaviour of saturated clays
during sampling and testing. Géotechnique 13 (4), 269–290.
Takaine, T., Tashiro, M., Shiina, T., Noda, T., Asaoka, A., 2010.
Predictive simulation of deformation of peat-calcareous soil layered
ground due to multistage test embankment loading. Soils and
Foundations 50 (2), 245–260.
Tanaka, H., 2000. Sample quality of cohesive soils: lessons from three sites,
Ariake, Bothkennar and Drammen. Soils and Foundations 40 (4), 57–74.
Tanaka, H., 2005a. Consolidation behaviour of natural soils around pc
value—long term consolidation test. Soils and Foundations 45 (3), 83–95.
Tanaka, H., 2005b. Consolidation behaviour of natural soils around pc
value—interconnected oedometer test. Soils and Foundations 45 (3),
97–105.
Tanaka, H., Sharma, P., Tsuchida, T., Tanaka, M., 1996. Comparative
study on sample quality using several types of samplers. Soils and
Foundations 36 (2), 57–68.
Tanaka, H., Tanaka, M., 1999. Key factors governing sample quality. In:
Tsuchida, T., Nakase, A. (Eds.), Characterization of Soft Marine
Clays. Balkema, Rotterdam.
Tanaka, H., Tanaka, M., 2006. Main factors governing residual effective
stress for cohesive soils sampled by tube sampling. Soils and Founda-
tions 46 (2), 209–219.
Tanaka, H., Udaka, K., Nosaka, T., 2006. Strain rate dependency of
cohesive soils in consolidation settlement. Soils and Foundations 46
(3), 315–322.
Tashiro, M., Noda, T., Inagaki, M., Nakano, M., Asaoka, A., 2011.
Prediction of settlement in natural deposited clay ground with risk of
large residual settlement due to embankment loading. Soils and
Foundations 51 (1), 133–149.
Tatsuoka, F., Jardine, R.J., Lo Presti, D., Di Benedetto, H., Kodaka, T.,
1997. Characterising the pre-failure deformation properties of geoma-
terials. In: Proceedings of the 14th Asian Regional Conference on Soil
Mechanics and Foundation Engineering, Hamburg, 4, pp. 2129–2164.
Tavenas, F., Leblond, P., Jean, P., Leroueil, S., 1983a. The permeability
of natural soft clays. Part I: method of laboratory measurement.
Canadian Geotecnical Journal 20 (4), 629–644.
Tavenas, F., Jean, P., Leblond, P., Leroueil, S., 1983b. The permeability
of natural soft clays. Part II: permeability characteristics. Canadian
Geotecnical Journal 20 (4), 645–660.
Taylor, D.W., 1948. Fundamentals of Soil Mechanics. John Wiley, New York.
Track, B., La Rochelle, P., Tavenas, F., Leroueil, S., Roy, M., 1980. A
new approach to the stability analysis of embankments on sensitive
clays. Canadian Geotecnical Journal 17 (4), 526–544.
Tsuchida, T., 2000. Evaluation of undrained shear strength of soft clay
with consideration of sample quality. Soils and Foundations 40 (3),
29–42.
Vaid, Y.P., Campanella, R.G., 1974. Triaxial and plane strain behavior of
natural clay. ASCE Journal of Geotechnial Engineering 100 (3), 207–224.
Watabe, Y., Tsuchida, T., 2001. Influence of stress release on sample
quality of pleistocene clay collected from large depth in Osaka Bay.
Soils and Foundations 41 (4), 17–24.
Yin, J.-H., Graham, J., Clark, J.I., Gao, L., 1994. Modelling unantici-
pated porewater pressures in soft clays. Canadian Geotecnical Journal
31 (5), 773–778.
Yin, Z.Y., Karstunen, M., Chang, C.S., Koskinen, M., Lojander, M., 2011.
Modeling time-dependent behavior of soft sensitive clay. Journal of
Geotechnical and Geoenvironmental Engineering 137 (11), 1103–1113.
Zhu, J.G., Yin, J.-H., 2001. Deformation and porewater pressure
responses of elastic viscoplastic soil. ASCE Journal of Engineering
Mechanics 127 (9), 899–908.