1

1 INTRODUCTION
Laboratory stress-strain test is only one of the important measures of geotechnical engineering
research and practice. In addition, as it is usually very difficult to retrieve sufficiently high-
quality undisturbed samples of field geomaterial, it is often considered that the laboratory stress-
strain test is less direct (so less useful) when compared with the field-loading test. It is
particularly the case with ordinary construction projects, in which sophisticated laboratory soil
tests may be considered unwarranted, unlike a limited number of huge scale projects. Although
it is not essential, another reason why the laboratory stress-strain test has become less popular
would be that it is just painstaking and time-consuming, compared with other types of study and
investigation, such as those by theoretical and numerical analyses.
It is also true, however, that the proper characterisation of in-situ stress-strain behaviour of
geomaterial, which could result into a more rational (so safer and more cost-effective) design,
often becomes possible with a help of relevant laboratory stress-strain tests. From a more basic
point of view, the proper understanding of the stress-strain-time behaviour of geomaterials is
not possible only by field loading tests and back-analysis of full-scale behaviour, but a
comprehensive series of relevant laboratory stress-strain tests usually become necessary.
The following four topics, among others, were selected based on a consideration that their
vital importance in geotechnical engineering research and practice are not well recognised
among practicing engineers and are often ignored even in research:
1) elastic properties at very small strains as well as non-linear pre-failure stress-strain
behaviour of geomaterials;
2) inherent anisotropy in the strength and deformation characteristics of granular materials;
Impacts on Geotechnical Engineering of Several Recent Findings
from Laboratory Stress-Strain Tests on Geomaterials
F. Tatsuoka
University of Tokyo
ABSTRACT: Significant impacts on the theories and practice of geotechnical engineering of
several findings obtained from recent advanced laboratory stress-strain tests on a wide variety of
geomaterials that were performed mainly by the author and his colleagues are demonstrated and
illustrated. The laboratory stress-strain tests were performed to apply their results to theoretical
research as well as practical design. This paper discusses on: 1) elastic properties at very small
strains as well as non-linear pre-failure stress-strain behaviour of geomaterials; 2) inherent
anisotropy in the strength and deformation characteristics of granular materials; 3) strain
localization with shear banding in granular materials; and 4) viscous deformation properties of
geomaterials. The importance of knowing the limitations of using over-simplified stress-strain
models, such as isotropic linear or perfectly-plastic models, of geomaterials is emphasized. It is
attempted to show the important and essential roles of relevant laboratory stress-strain tests of
geomaterials in developing the theories and practice of geotechnical engineering. This paper is
the lecture note for the 2000 Burmister Lecture, 31st, October 2000, the Columbia University,
N.Y., U.S.A.
TATSUOUKA-2001 Burmister Lecture, Balkema
2

Figure 1.1. Preparation of inclined
specimens of sand by pluviation
through air and subsequent
moistening, followed by a sequence
of freezing and thawing (Park and
Tatsuoka 1994).
0 3 6 9 12 15
-2
0
2
4
6
8
a)

vol
-
1
relations
R -
1
relations
30
o
45
o
90
o
20
o
0
o
0
o
30
o
20
o
45
o
90
o
Ticino Sand
'
3
=0.8kgf/cm
2
OCR=1.0
(
o
) e
0.05
90 0.657
45 0.656
30 0.666
20 0.659
0 0.662
P
r
i
n
c
i
p
a
l

s
t
r
e
s
s

r
a
t
i
o
,


R
=

'
1
/

'
3
Axial strain,
1
(%)
4
0
-4
-8
-12
-16
V
o
l
u
m
e
t
r
i
c

s
t
r
a
i
n
,

v
o
l
(
%
)
0 3 6 9 12 15
-2
0
2
4
6
8
a)
vol - 1 relations
R - 1 relations
3 0
o
45
o
90
o
20
o
0
o
0
o 3 0
o
20
o
45
o
90
o Ticino Sand
'
3
=0.8kgf/cm
2
OCR=1.0
(
o
) e0.05
90 0.657
45 0.656
30 0.666
20 0.659
0 0.662
P
r
in
c
ip
a
l
s
t
r
e
s
s

r
a
t
io
,


R
=

' 1
/
' 3
Axial strain,
1
(%)
4
0
-4
-8
-12
-16
V
o
l
u
m
e
t
r
i
c

s
t
r
a
i
n
,

v
o
l
(
%
)
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.5
1.0
1.5
2.0
b)
(
o
) e
0.05
90 0.657
45 0.656
30 0.666
20 0.659
0 0.662
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,


q
=

' 1
-

'
3

(
k
g
f
/
c
m
2
)
Axial strain,
1
(LDT) (%)
0 3 6 9 12 15
-2
0
2
4
6
8
a)

vol
-
1
relations
R -
1
relations
30
o
45
o
90
o
20
o
0
o
0
o 30
o
20
o
4 5
o
90
o Ticino Sand
'3=0.8kgf/cm
2
OCR=1.0
(
o
) e
0.05
90 0.657
45 0.656
30 0.666
20 0.659
0 0.662
P
r
in
c
ip
a
l s
t
r
e
s
s
r
a
t
io
,
R
=

' 1
/
' 3
Axial strain,
1
(%)
4
0
-4
-8
-12
-16
V
o
l
u
m
e
t
r
i
c

s
t
r
a
i
n
,


v
o
l
(
%
)
0.000 0.001 0.002 0.003 0.004 0.005
0.00
0.02
0.04
0.06
0.08
0.10
0.12
c)
Axial strain,
1
(LDT) (%)
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,


q
=

' 1
-

' 3

(
k
g
f
/
c
m
2
)
(
o
) e
0.05
90 0.657
45 0.656
30 0.666
20 0.659
0 0.662
Elastic properties

Figure 1.2. Anisotropy in PSC tests on air-dried air-
pluviated Ticino sand; is the angle of
1
with respect to
bedding plane: stress-strain relationships a)
1
11%, b)
1

0.5% and c)
1
0.005% (Park and Tatsuoka 1994;
Tatsuoka and Kohata 1995).

3) strain localisation with shear banding in granular materials; and
4) time-dependent deformation properties of geomaterials.
Some representative data
illustrating these points above are
first shown. Fig. 1.1 shows a
method to prepare rectangular
prismatic specimens having the
axis of compression loading in-
clined relative to the bedding
plane direction, prepared for plane
strain compression tests (Park and
Tatsuoka 1994). The specimen
was first made by pluviating air-
dried sand particles through air,
subsequently made moist and then
frozen under a restraint against
the expansion of specimen upon
freezing. The specimen was
thawed under partial vacuum after
being set in the triaxial cell. The
specimen was made fully
saturated and isotropically
consolidated to 78 kPa. The
details of the plane strain
apparatus and the plane strain
compression procedure are
described in Shibuya et al. (1994),
Yasin et al. (1999a & b), Masuda
et al. (1999) and Yasin and
Tatsuoka (2000).
Fig. 1.2 shows a result typical
of the tests performed by Park and
3

Figure 1.3. A shear band seen at
ave
11.8 % in a PSC test on
Toyoura sand (D
50
= 0.206 mm;
3
78 kPa) (Yoshida et al.
1995: Yoshida and Tatsuoka
1997).

Shear band in air-dried Toyoura sand (made moist after the test)
in a plane strain bearing capacity test of rigid rough strip footing
B= 10 cm

Figure 1.4. A shear band network observed in the sand bed of
Toyoura sand in a bearing capacity test of strip footing; the
sand was air-dried during loading and made moist after the test
to expose this central section (Fig. 4.20) (Tatsuoka et al. 1991).
Tatsuoka (1994). The angle means the angle of the direction of the major principal stress
1

during plane strain compression relative to the bedding plane (see Fig. 1.1). It may be seen that
the stress-strain behaviour of sand is strongly affected by
the angle (i.e., inherent anisotropy), except for at very
small strains and at the residual state. It is likely that the
initial anisotropic structure is substantially damaged until
the residual condition in the shear band. At strains less
than about 0.005 %, the stress-strain relationships are
rather linear (and reversible as shown later); i.e., elastic
behaviour. The relationship between the major and minor
principal strains
1
and
3
is less sensitive to the angle than the stress-strain relation. The
reason for such a variance as above is not well understood.
On the other hand, Fig. 1.3 shows a shear band that was observed at the residual conditions
on the
2
plane through the transparent confining platen in a test at 90
o
(i.e., the
conventional plane strain compression test). It may be seen from Fig. 1.3 that the shear band
has a noticeable thickness, which means that the deformation characteristics of shear band could
have important effects on the kinematics of a failing soil mass in a boundary value problem,
such as the bearing capacity problem (as discussed later). Fig. 1.4 shows a shear band network
that was observed in a level deposit of Toyoura sand supporting a strip footing (Tatsuoka et al.
1991).
These four issues are only a part of a number of important geotechnical engineering issues
that are still not well understood, but their proper understanding is essential for the development
of geotechnical engineering theories and practice. In the following, these four topics will be
explained, trying to show their engineering implications as much as possible. It is the final aim
of the paper to show that geotechnical engineering is still young (although a number of myths
are already existing in these issues).


2 ELASTIC PROPERTIES AT VERY SMALL STRAINS AND NON-LINEAR PRE-
FAILUER STRESS-STRAIN BEHAVIOUR OF GEOMATERIALS

4
To predict ground deformations and structural
displacements at working loads;
Why is the elastic property
one of the main factors in characterising
th pre-failure deformation property ?


- Strains in the ground at working load
are relatively small.
- Deformation properties at these small strains
can be linked to the elastic properties,
although the stress-strain behaviour could be
highly non-linear at these small strains.


Figure 2.1. Several reasons why the elastic deformation
properties of geomaterials are important in many geotechnical
engineering problems.

Figure 2.2. General view of Akashi Kaikyo (Strait) Bridge and geological conditions (Tatsuoka and
Kohata 1995).
2.1 Engineering needs

The elastic deformation
property is the key reference
property for the stress-strain
behaviour of a given
geomaterial subjected to cyclic
loading (e.g. Hardin and
Drnevich 1972; Iwasaki and
Tatsuoka 1977; Tatsuoka et al.
1978, 1979a & b). It is to be
noted however that the elastic
deformation property is also an
important and essential
parameter for so-called static
geotechnical loading problems.
That is, one of the main
features of the recent
developments in the charac-
terisation of geomaterial pre-
failure deformation properties
to predict ground deformations
and structural displacements at
working loads is focusing on
the elastic deformation properties of concerned geomaterials (Fig. 2.1) (e.g. Jardine et al. 1985:
Burland 1989; Jamiolokowski et al. 1991, Atkinson and Sllfors, 1991, Tatsuoka and Shibuya
1992, Mair 1993; Tatsuoka and Kohata 1995, Jardine 1995; Hight and Higgins 1995; Tatsuoka
et al. 1995b, 1999a).
5
a)
180
160
140
120
100
80
60
10
-6
10
-5
10
-4
10
-3
10
-2
E
l
e
v
a
t
i
o
n

@
T
P
-
(
m
)
Vertical strain @

2
3
b)
Figure 2.3. a) Method used to measure the ground settlement; and b) centre-line vertical strains in
the gravel and sedimentary softrock below the piers 2P and 3P for the worlds longest suspension
bridge: Akashi Kaikyo Bridge. The average contact pressure and foundation diameters are 5.3
kgf/cm
2
and 80 m for Pier 2P; 4.8 kgf/cm
2
and 78 m for Pier 3P, respectively (Takeuchi et al. 1997).
Akashi Kaikyo Strait Bridge (Fig. 2.2) is the world longest suspension bridge. This case
would be typical of those showing the importance of elastic deformation properties of
geomaterials. Fig. 2.3 shows the center-line vertical strains in the gravel and sedimentary
softrock below the piers (3P and 2P) (Tatsuoka and Kohata 1995, Tatsuoka et al. 1999a). As
may be seen from this figure, the strains in the ground are generally small, lower than 0.5 %,
which was due to:
1) the foundations were designed allowing only a limited amount of footing displacement; and
2) the foundations were constructed on relatively stiff ground (although the ground conditions
were worst in the Honshu-Shikoku connection bridge network).
Fig. 2.4 shows the relationship between the secant Youngs modulus values
FEM
E at
different depths that were back-calculated by liner FEM from the measured ground settlement
and the known footing load, divided by the respective corresponding elastic Youngs modulus
f
E obtained from the field shear wave velocity that was measured before construction and
the measured ground vertical strain, for Piers 2P and 3P for Akashi Kaikyo Bridge (Tatsuoka
and Kohata 1995). The relationships are compared also with those from the conventional
pressure-meter tests. The following trends of behaviour may be noted:
1) the relationships evaluated from the field full-scale behaviour are highly non-linear (n.b.,
the non-linearity of the relationships is not due totally to the strain-non-linearity, but it is
also affected by the changes in the pressure level during construction, in particular by those
due to ground excavation);
6
0.0001 0.001 0.01 0.1 1
0.0
0.5
1.0
1.5
E
f
:
from V
s
before construction)
E
PMT
/E
f
(strais for
E
PMT
unreported)
o
Gravelly soil
range
for
soft rock 2
5
1
1
4
( )
3
7
2
6
3
4
5
6*
9
8
7
3P case1 for (p)
ave
= 0 5.2 kgf/cm
2
1 6 Sedimentary soft rock (P3)
7 Granite
* E
f
was estimated as 5 x E
PMT
2P case1 for (p)
ave
= 0 5.3 kgf/cm
2
1 5 Gravelly soil (Akashi)
6 9 Sedimentary soft rock



E
F
E
M
/
E
f
,


E
P
M
T
/
E
f
Measured ground vertical strain:
1
(%)
o
E
f
E
FEM


Figure 2.4. Relationship between the Youngs modulus back-calculated from the ground
settlement divided by the respective elastic Youngs modulus from the field shear wave velocity
measured before construction and the measured ground strains, compared with those from the
conventional pressure-meter tests, Akashi Kaikyo Bridge Piers 2P and 3P (Tatsuoka and Kohata
1995).


2) as the operating ground strain approaches 0.001 % (i.e., as the concerned soil layer becomes
deeper) , the
FEM
E value approaches the respective corresponding
f
E value; and
3) the Youngs modulus values
PMT
E from the conventional pre-bored pressure-meter tests
(i.e., linear interpretation of primary loading curves) are noticeably lower than those
operating in the ground. This is very likely due to large strains involved in the tests and the
effects of wall disturbance and bedding error at the wall face.
At the design stage, there was not a distinct consensus with respect to the stiffness value that
should be used to predict the instantaneous settlement of foundation among not only the
engineers in charge of this project but also geotechnical engineers in general. In the
conventional approach in rock mechanics, the tangent Youngs modulus defined at a deviator
stress that is a half of the peak strength (
50
E ) or the
PMT
E values cited above or the values
(
PLT
E ) from the conventional plate loading tests (linear interpretation of primary curve or
unload/reload cycle curve) have been often used. For such a case of foundation design as this
case, the plate loading test results are often considered to be most relevant. In addition, the
effects of strain-non-linearity and effects of pressure (or more generally, effects of stress state)
on the stiffness were not considered in a systematic way.
Fig. 2.5 compares the Youngs modulus values obtained from a set of field loading tests that
were performed at the site of Anchor 1A and laboratory stress-strain tests on samples retrieved
from the bottom of the excavation (Tatsuoka and Kohata 1995). The ground, consisting mainly
of sedimentary soft rock, was excavated to a depth of about 60 m for constructing the
anchorage. In this figure, the values of Youngs modulus (
f
E ) from the field shear wave
velocities obtained by the down-hole suspension method performed before the excavation are
also shown. It may be seen that the statically measured Youngs modulus values (i.e., those
from the conventional-type pressure-meter tests, plate loading tests and unconfined compression
7
100 1000 10000 100000
-70
-65
-60
-55
-50


D
e
p
t
h

(
m
)
Young's modulus E (kgf/cm
2
)


E
max
from CU and CD TC tests

c
' =
v
' (in situ)= 5.2 (kgf/cm
2
)
Kobe group softrock


1A-1
1A-2
1A-3
1A-4
1A-5
1A-6
E
f
(from shear wave velocity)
{ 4060 2040, 020,
range of plate pressure(kgf/cm
2
)
E
PLT
; tangent modulus
in primary loading
{
E
BHLT
; primary loading
E
50
unconfined
compression tests
(from external axial strains)

E
max
E
max
Average
(from CD TC tests;
axial strains measured with LDTs)


Figure 2.5. Youngs modulus values (in log scale) from field and laboratory tests at Anchor 1A,
Akashi Kaikyo Bridge; the diameter of plate in the PLTs= 60 cm (Tatsuoka and Kohata 1995).
tests) are substantially lower than those obtained by the dynamic method (i.e., the field shear
wave velocity measurement). It is usual to observe such a large difference as above in such
construction projects as this case. This would be one of the origins (perhaps the most important
one) for the popular but wrong notion that a given geomaterial mass has static and dynamic
Youngs modulus values (or static and dynamic elastic stiffness values) that are different
material properties (Tatsuoka and Shibuya 1991; Tatsuoka and Kohata 1995; Tatsuoka et al.
1995a-d, 1999a).
A series of CD and CU triaxial compression tests were performed on undisturbed samples
obtained by block sampling at the bottom of excavation performed for the construction of the
anchor A1 (Tatsuoka and Kohata 1995; Tatsuoka et al. 1995d; Siddiquee et al. 1994, 1995a).
Fig. 2.6 shows the relationship between the deviator stress q and the axial strain
1
typical of
those from CD TC tests on samples of soft sandstone that were isotropically reconsolidated to
the respective original field vertical stress. The axial strains were measured locally by means of
a pair of LDTs (explained later) as well as externally outside the triaxial cell (i.e., the
conventional method). It may be seen from Fig. 2.6 that the axial strains measured externally
are utterly unreliable, which is due mostly to significant effects of bedding error at the top and
bottom ends of specimen partly to the deformation of the loading piston and specimen cap (i.e.,
the system compliance). Each drained elastic Youngs modulus at the field stress state was
evaluated from the initial linear and reversible part of the respective relationship between q and
the locally measured
1
. By using the relevant Poissons ratios for drained and undrained
conditions, this drained value of elastic Youngs modulus was converted to the value under the
undrained conditions, which is relevant to the field seismic investigation. These undrained
values of Youngs modulus (
max
E ) are plotted in Fig. 2.5. It may be seen that these Youngs
modulus values (
max
E ) are very similar to those from the field shear wave velocities. This fact
indicates that in this case, the dynamically and statically measured elastic Youngs modulus
8
0.00 0.02 0.04 0.06 0.08 0.10
0.0
0.2
0.4
0.6
0.8
v
External
LDT
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
M
P
a
)
Axial strain, (%)
0 1 2 3
0
2
4
6
8
10
12
0 max
q = 9.39 MPa, E = 1520 MPa
h
'= 0.51 MPa (CD)
Sedimentary soft sandstone (Kobe Formation)
External LDT
v
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
M
P
a
)
Axial strain, (%)
0.0000 0.0005 0.0010 0.0015 0.0020
0.00
0.01
0.02
0.03
0.04
0
E = 1520 MPa
1
LDT v
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
M
P
a
)
Axial strain, ( ) (%)
Triaxial compression test
on an undisturbed sample

Figure 2.6. Typical relationship between the deviator stress and the axial strain from a CD TC test
on a specimen obtained by block sampling at the bottom of excavation for Anchor A1, Akashi
Kaikyo Bridge (Tatsuoka and Kohata 1995).

values are very similar to each other, while the effects of anisotropy, discontinuities and
inhomogeneity of the ground are not significant, if any.
The direct use of the Youngs modulus values determined by the conventional pressuremeter
tests, plate loading tests and unconfined compression tests could largely over-estimate the actual
settlement of the plate in the plate loading tests and the full-scale structure in the field as shown
below. Fig. 2.7 shows results from three plate loading tests using a 60 cm-diameter rigid plate
performed at the bottom of excavation for Anchor A1 for Akashi Kaikyo Bridge (n.b. the data
from one test from the four tests performed is excluded, because the result is extraordinary due
very likely to the effects of a large joint that was opened by ground excavation,) (Tatsuoka and
Kohata 1995; Tatsuoka et al. 1999a). In this figure, their simulations by linear FEM using the
following Youngs modulus values are also shown;
a) the average of the Youngs modulus values from unconfined compression tests (E
50
),
shown in Fig. 2.5;
b) the average of the Youngs modulus values from the pressuremeter tests (E
PMT
), shown
in Fig. 2.5; and
c) the value obtained from the back-analysis of the full-scale behaviour of Pear P3 (shown
in Fig. 2.9).
The result from a simulation by a non-linear elasto-plastic FEM using the elastic Youngs
modulus from the field shear wave velocity measurements together with on the pressure-
dependency and strain-non-linearity of stiffness evaluated by triaxial compression tests on
disturbed samples from the site is also shown in Fig. 2.7 (Siddiquee et al. 1994, 1995a). It may
be seen that the only the non-linear elasto-plastic FEM analysis reasonably simulates the results
of the plate loading tests. In particular, the concave shape of the relationship between the plate
pressure and the plate settlement can be well simulated, which can be attributed to the
following:
9

0 2 4 6 8 10
0
10
20
30
40
50
60
70
1
E
s
unloading (S-3)
1
E
t
1
D
E
=
1
0
0
0
0
k
g
f
/
c
m
2
(
f
r
o
m

f
u
l
l
-
s
c
a
l
e

f
i
e
l
d

b
e
h
a
v
i
o
u
r

o
f

3
P
)
E
P
M
T
=
3
1
2
5
k
g
f
/
c
m
2
E 5
0
=
1
7
7
7
k
g
f
/
c
m
2

(
f
r
o
m

U

t
e
s
t
s
)
linear elastic FEM
elasto-plastic FEM
Plate loading
test(D=60cm)
S-1
S-3
S-4
(
p
)
a
v
e
.
(
k
g
f
/
c
m
2
)
Settlement, S (mm)

Figure 2.7. Results from plate loading tests performed at the bottom of excavation for Anchor A1
for Akashi Kaikyo Bridge; and their simulations by linear FEM using a) the average Youngs
modulus values from unconfined compression tests (E
50
), pressuremeter tests (E
PMT
) and the back-
analysis of the full-scale behaviour of Pear P3 and by elasto-plastic FEM analysis using the
Youngs modulus from the field shear wave velocity and the pressure-dependency and the non-
linearity of stiffness from laboratory stress-strain tests (Siddiquee et al. 1994, 1995a; Tatsuoka and
Kohata 1995; Tatsuoka et al. 1999a).
a) the elastic deformation occupies a large part of the ground deformation in the plate loading
tests; and
b) the pressure-level dependency of elastic deformation characteristic of the ground is properly
modelled in the simulation.
The points a & b are also relevant to the field full-scale behaviour. Fig. 2.8, for example,
shows the relationships between the average contact pressure and the total, elastic and
irreversible settlements (S ,
e
S and
ir e
S S S ) of Pier 3P, for which the elastic part was
obtained by the FEM analysis based on the elasticity model (Tatsuoka et al. 1999a). A similar
result has also been obtained from Pier 2P (Siddiquee et al. 1994, 1995a). For the
decomposition of the footing settlement, FEM analysis based on a hypo-cross elasticity model,
which is explained later in this paper, was performed. The parameters of the model were
determined based on the initial elastic stiffness values from the field shear wave velocities
measured at the sites of Piers 2P & 3P and the pressure-dependency of the elastic deformation
properties that were obtained from laboratory cyclic and monotonic triaxial tests on undisturbed
samples retrieved from the site. The separation of the total settlement into two components
(elastic and irreversible), not into three components (elastic, plastic and viscous), is not
arbitrary. This issue is discussed in detail by Tatsuoka et al (2000).
It is to be noted from Fig. 2.8 that the tangential slope of the relationship between the average
contact pressure at the footing base and the elastic component of the settlement of the pier
10
increases with the increase in the contact pressure. This behaviour can be attributed to the fact
that the elastic Youngs modulus of the sedimentary soft sandstone increases with the increase
in the pressure, as observed in the triaxial compression tests. This issue is discussed in detail in
Tatsuoka et al. (1995a-d, 1999a) and Kohata et al. (1997).
A fluctuation seen in the relationships between the footing contact pressure and the total or
irreversible footing settlement is not due to simple measurement errors, but it is very likely that
this fluctuation was basically due to the viscous deformation property of the ground (as
discussed later). That is, the tangential slope is small when the construction rate was slow and
vice versa, and the stiffness becomes very large immediately after construction was restarted at
a normal construction speed following a long period of construction stop. It will be shown later
that a relevant elasto- viscoplastic model could explain this behaviour.
Fig. 2.9 shows the full-scale behaviour of Pear P3 until the average footing pressure became
about 95 kPa (refer to the full-range behaviour show in Fig. 2.8). The simulation of this
behaviour by the following methods are also shown:
a) linear FEM using:
i) the average Youngs modulus value from unconfined compression tests (E
50
); and
ii) that from the pressuremeter tests (E
PMT
) obtained at the site of A1; and;
b) the non-linear elasto-plastic FEM analysis using the elastic Youngs modulus from the field
shear wave velocity and the pressure-dependency and non-linearity of stiffness from
laboratory stress-strain tests on undisturbed samples obtained at the site of 3P (Tatsuoka and
Kohata 1995; Tatsuoka et al. 1999a; Siddiquee et al. 1994, 1995a).
It may be seen again that the field behaviour can be reasonably simulated only by the relevant
FEM analysis b). By comparing Figs. 2.7 and 2.9, it may be seen that for the result of a linear
isotropic analysis to fit the full-scale behaviour of Pier 3P, a Youngs modulus value that is
much larger than the value to be used to fit the plate loading test result should be used. This
difference is due to the so-called scale effect on the stiffness of ground. This scale effect in this
respect can be attributed to an initially inhomogeneous distribution of elastic stiffness and the
pressure- and strain-non-linearity of stiffness. The latte factor can be evaluated by relevant
laboratory stress-strain tests.
-5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
0
2
4
6
8
10
.
S
ir
= 0.05 (mm/day)
.
S
ir
(mm/day)
b)
Fitted to
3P
S
t
S
ir
S
e

0.05
0.00-0.05
0.05-0.10
0.10-0.15
0.15-0.20
0.20-0.25
>0.25
A
v
e
r
a
g
e

c
o
n
t
a
c
t

p
r
e
s
s
u
r
e
,

(
p
)
a
v
e


(
k
g
f
/
c
m
2
)
Settlement, S (mm)
Figure 2.8. Decomposition of the measured settlement of Pier 3P into elastic and irreversible
components, Akashi Kaikyo Bridge (Tatsuoka et al. 2001a).
11

0
2
4
6
8
10
12
0 20 40 60 80 100
Settlement, S (mm)
(
p
)
a
v
e
.

(
k
g
f
/
c
m
2
)
measured
E=10000kgf/cm
2
E
50
(from 1A site)
=1777kgf/cm
2
FEM
E
PMT
(from 3P site)
=2890kgf/cm
2

Figure 2.9. Full-scale behaviour of Pear P3 for Akashi Kaikyo Bridge; and its simulations by linear
FEM using a) the average Youngs modulus values from unconfined compression tests (E
50
), and
pressuremeter tests (E
PMT
) obtained at the site of A1; and b) elasto-plastic FEM using the Youngs
modulus from the field shear wave velocities measured at the site of 3P and the pressure-dependency
and non-linearity of stiffness from laboratory stress-strain tests (Sddiquee et al. 1994, 1995a;
Tatsuoka and Kohata 1995; Tatsuoka et al. 1999a).

Another typical example is the behaviour of the foundations for another suspension bridge,
Rainbow Bridge, located in the Tokyo area (Fig. 2.10). This bridge was completed in 1993.
The foundations were caissons that were constructed on a thick layer of sedimentary soft
mudstone of a geological age of about 1.5 million years having a compressive strength of about
2.0 - 3.0 MPa (from CD triaxial compression tests). It was considered at the initial design stage
that the foundation should be supported by pile foundations, anticipating excessive settlements
of foundations. The four foundations were constructed directly on the mudstone layers based on
the judgement made referring to the results of analysis using stiffness values from conventional
pressure-meter tests (Tatsuoka and Kohata 1995: Izumi et al. 1997). Fig. 2.11 shows the
comparison between the measured instantaneous settlement of anchorage A4 that took place
when constructing the top anchorage block having a weight of 140,000 tonf and its predictions.
As seen from Fig. 2.11, the actually observed settlement was substantially smaller than the
value predicted before the start of construction based on the stiffness values obtained from the
primary loading curves of oedometer tests on undisturbed samples. This substantially large
over-estimation of the observed value is most likely due to very large effects of bedding error at
the top and bottom surfaces of a thin specimen with a thickness of as small as 2 cm. The
settlement predicted based on the unload/reload stiffness from the oedometer tests is better, but
it still largely overestimates the observed value. In addition, there is no sound reason for the use
of the stiffness values of unload/reload cycles, unlike the field stress history, in the oedometer
tests. The prediction based on the pressuremeter stiffness is also noticeably larger than the
observed value.
12

Figure 2.10. Rainbow Bridge

- Over-estimation of the footing settlement
by the conventional methods.
- Accurate estimation when based on .
140,000 tonf
Anchorage
Observed
Non-liner FEM
using the shear
modulus from Vs
and the non-linearity
from CD TC tests
Observed
Based on m
v
from
primary loading
curves of oedometer
tests
Based on PMTs
Based on m
v
from
unload/reload curves
of oedometer tests
Figure 2.11. Comparison between the observed behaviour of Anchor A4 and the predictions by the
conventional method and the simulation by non-linear 3D FEM using stiffness values obtained from
SOA geotechnical investigation, Rainbow Bridge (Izumi et al. 1997).

To find a reason(s) for this inconsistency, after the completion of the bridge, a SOA
geotechnical investigation was carried out, consisting of a suspension-type downhole seismic
survey, undisturbed sampling by rotary core tube sampling, triaxial compression tests measuring
axial strains locally using LDTs and non-linear 3D FEM analysis based on the Youngs
modulus (or shear modulus) from the field shear wave velocity and strain-non-linearity from the
triaxial compression tests (Izumi et al. 1997). In this FEM analysis, the effects of changes in the
effective pressure on the stress-strain behaviour of the sedimentary soft mudstone were not
considered based on the triaxial test results. The result from this SOA investigation is also
shown in Fig. 2.11. It may be seen that the observed behaviour is best simulated by this non-
linear FEM analysis using the stiffness obtained from the SOA geotechnical investigation. Fig.
2.12 compares the shear modulus values, as a function of shear strain, obtained from the CD
triaxial compression tests, the conventional pressuremeter tests and the field full-scale behaviour
for some range of depth below the anchorage, which is typical of similar comparisons in this
case (Izumi et al. 1977). The following trends of behaviour may be seen from Fig. 2.12, which
were basically the same as those with the foundations for Akashi Kaikyo Bridge: that is:
1) the conventional pressuremeter stiffness values are too small, representing the stiffness

2
f s
G V
13
Full-scale behaviour
CD Triaxial
compression
tests using LDTs
Pressure-meter tests
(Linear interpretation of
primary loading curve)
Shear wave velocities
(Suspension method)
1
(%) Strain
0.1 1.0 10 0.01 0.001
0.0001
S
h
e
a
r

m
o
d
u
l
u
s
,

G

(
M
P
a
)


Figure2.12. Typical comparison between the stiffness values as a function of strain from CD triaxial
compression tests (GL 54 64 m and 64 74 m), conventional pressuremeter tests and field full-
scale behaviour, Anchor A4, Rainbow Bridge (Izumi et al. 1997).
values at strains that are considerably larger than those operated in the ground; and
2) the stiffness values from the CD TC tests are consistent with:
a) the elastic shear modulus from field shear wave velocities (defined at strains less than
about 0.0001 %);
b) those from the conventional pressuremeter tests measured at strains from a range
from about 0.5 % to about 1.0 %, and
c) those from the field full-scale behaviour at strains of about 0.01 %.

Summary: The elastic deformation characteristics that are to be referred to when predicting the
ground deformation and structural displacements in field static loading cases are often
considered to have no link to the dynamically measured deformation properties. In addition,
linear deformation analysis of filed loading tests and full-scale behaviour is stiff popular. In
many cases, the strains operated in the ground are relatively small. In such cases, prediction of
ground deformation based on the elastic stiffness obtained from field shear wave velocities,
while considering non-linearity and pressure level-dependency of stiffness at relatively small
strains, could be relevant. Laboratory stress-strain tests can contribute in many important
aspects to such geotechnical engineering issues as described above.

2.2 Modelling of small strain stiffness

A brief overview: Previously, the most common practice to obtain the elastic deformation
properties was dynamic tests such as the resonant-column (RC) tests and the wave propagation
tests (e.g. Hardin and Richart 1963; Hardin and Back 1968). Later, Hardin and Drnevich
(1972), Iwasaki et al (1978), Teachavorasinskun et and (1991a & b), Shibuya et al. (1992) and
others showed that the strain rate-dependency of the stiffness at small strains of geomaterials in
cyclic torsional shear is very low. Based on such experimental results as above, Woods (1991)
and Tatsuoka and Shibuya (1992) pointed out that it is not necessary to distinguish between
dynamically and statically measured elastic stiffness values when they are measured under
otherwise the same conditions.
In the meantime, we have become rather confident with reliable measurements of strains less
than about 0.001 % and associated small stresses in both triaxial and torsional shear tests. It has
therefore become rather popular to obtain the stiffness and damping values under cyclic loading
14

Figure 2.14. Relationships between the peak-to-peak secant shear
modulus and the single amplitude shear strain from a torsional RC
test (Lo Presti 1989, Lo Presti et al. 1993), a cyclic torsional shear
test and a cyclic triaxial test; and between the secant shear modulus
and the shear strain from a monotonic torsional shear test and a
monotonic triaxial compression test, all on an isotropically
consolidated specimen of air-pluviated Ticino sand
(Teachavorasinskun et al. 1991a & b: Tatsuoka et al. 1995b,
1999a).
1) Elastic deformation characteristics :
can be obtained only by dynamic tests ?
2) Statically and dynamically determined elastic
deformation properties are different ?
Static tests (monotonic or cyclic);
stress-strain properties from stresses and strains !
Dynamic tests (RC tests & wave-propagation tests);
stress-strain properties from dynamic responses !

Figure 2.13. Some relevant questions with respect to
the elastic deformation characteristics of geomaterials .
conditions for a full range of concerned
strain (usually from lower than 0.001 %
to around 1 %) by cyclic static loading
tests using a single specimen. It is only
recent however that it becomes possible
to evaluate confidently the elastic
deformation properties, as well as the
whole pre-peak stress-strain behaviour
and peak strength, by monotonic
loading tests using a single specimen
(e.g. Tatsuoka and Shibuya 1992;
Tatsuoka 1994, Tatsuoka et al. 1990c,
1994a & b, 1995a & b).
It seems however that it is still
popular in ordinary engineering
practice to define and treat separately
dynamic and static stiffness values of a
given mass of geomaterial. Therefore,
the questions listed in Fig.
2.13 are still relevant. In
Fig. 2.14, the shear modulus
values of Ticino sand that
was isotropically
consolidated to 49 kPa
obtained from dynamic and
static (cyclic and
monotonic) tests are
compared. It may be seen
that the shear modulus at
shear strains less than about
0.001 % is essentially the
same among the dynamic
tests (i.e., resonant-column
tests) and the static tests
(i.e., monotonic and cyclic
triaxial and torsional tests).
It has also been shown that
for various types of fine-
graded sands, the elastic
stiffness from RC tests and
static tests is essentially the
same with the values from
the bender element tests, as
summarised in Tatsuoka et
al. (1999a). The differences
seen among the shear modulus values from the different testing methods at strains exceeding the
elastic limit strain, equal to about 0.001 %, is due mostly to different stress paths and different
strain histories (i.e., with and without cyclic loading) and due only partly to different strain
rates.
Fig. 2.15 shows a summary of data showing the effects of strain rate on the very small strain
Youngs modulus E
v
defined for (
v
)
SA
of 0.001 % or less (Tatsuoka et al., 1999a &b). These
data were obtained from the following series of tests, which are mostly static tests; cyclic tests
and monotonic tests (only for kaolin), except for dynamic tests on hard rock cores, concrete and
mortor:
1) cyclic triaxial tests (U; undrained and D; drained):
a) Sagamihara soft mudstone ( ' '
v h
= 4.8 kgf/cm
2
; Tatsuoka et al. 1995a & b);
b) OAP clay ( '
v
=6.9 kgf/cm
2
and '
h
=3.4 kgf/cm
2
; Tatsuoka et al. 1995a);
15
Sandy gravel (D)
sand (U)
Hostun sand (D)
Resonant-column
Hard rock core
Ultrasonic wave
Concrete
Mortar
Sagamihara soft rock (U)
OAP clay (U)
Air-dried
Wet Chiba gravel (D)
Metramo silty sand (U)
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
10
1
10
2
10
3
10
4
10
3
10
4
10
5
10
6
Vallericca clay
.
E
v

(
k
g
f
/
c
m
2
)
Axial strain rate, d
v
/dt (%/min)
N.C. Kaolin (CU TC)
Saturated Toyoura

Figure 2.15. Summary of the effects of strain rate on the very small strain Youngs modulus E
v
defined for (
v
)
SA
of 0.001 % or less (Tatsuoka et al., 1999a,b).


c) Chiba gravel (e= 0.247, w
0
= 3.7 % and ' '
v h
= 0.2 kgf/cm
2
; Jiang et al. 1999;
Tatsuoka et al. 1999b);
d) air-pluviated Toyoura sand (e= 0.658 and ' '
v h
= 1.0 kgf/cm
2
; Tatsuoka et al. 1995a);
16
Figure 2.16. Large triaxial specimen (30
cm in dia) with local strain measurements at
the University of Tokyo (Tatsuoka et al.,
1994a).

Figure 2.17. Triaxial testing systems using a small
cylindrical specimen at the University of Tokyo
(Tatsuoka et al., 1995a, 1999a) .

Phosphor bronze
strain-gaged strip
LDT
Pseudo-hinge
Membrane
Heart of LDT
(includes electric resistance strain gages,
terminals, wiring, sealant)
Scotch tape used to fix wire
on the specimen surface
Instrument Leadwire
Membrane Surface
Figure 2.18. Local deformation transducer (Goto et al.
1991; Hoque et al. 1997).

e) air-pluviated Hostun sand (e= 0.72, '
v
= 0.8 - 2.5 kgf/cm
2
and '
h
= 0.8 kgf/cm
2
; Hoque
1996, 1997; Hoque and Tatsuoka 1998; Tatsuoka et al., 1999b); and
f) compacted Metramo silty sand ( ' '
v h
= 4.0 kgf/cm
2
; Santucci de Magistris et al.
1998, 1999, Santucci de Magistris and Tatsuoka 1999).
2) CU TC tests on NC kaolin (p
c
= 3.0 kgf/cm
2
and K
c
=0.6 - 1.0) (Tatsuoka et al. 1995a).
3) unconfined cyclic tests and
ultrasonic tests on hard rocks,
concrete and mortar (Sato et al.
1997a, b); the strain rates in the
ultra-sonic tests were evaluated
from the wave frequency and
particle velocity.
The following trends of behaviour
could be seen from Fig. 2.15:
a) With hard rock cores, concrete
and mortar specimens, effects of
strain rate on the elastic Youngs
modulus are very small in the
static tests. These test results
indicate that the effects of strain
rate are very small commonly
with these materials.
b) With concrete and mortar speci-
mens, a good agreement can be
seen between the elastic stiffness
values from the resonant column
tests and static tests at the
17
-0.0010 -0.0005 0.0000 0.0005 0.0010
-4
-3
-2
-1
0
1
2
3
4
f (Hz) E
v(s)
(MPa)
10 477.9
5 479.0
1 484.8
0.2 476.0
0.1 469.0
0.02 470.3
0.01 458.3
0.002 455.3

Chiba gravel
5
th
cycle

h
=19.6 kPa
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
k
P
a
)
Axial strain,
v
(%)
a)
-0.0010 -0.0005 0.0000 0.0005 0.0010
-4
-3
-2
-1
0
1
2
3
4


Chiba gravel
5
th
cycle

h
=19.6 kPa
f(Hz) d
v
/dt (%/min)
10 3.6x10
-1
5 1.8x10
-1
1 3.6x10
-2
0.2 7.2x10
-3
0.1 3.6x10
-3
0.02 7.2x10
-4
0.01 3.6x10
-4
0.002 7.2x10
-5
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
k
P
a
)
Axial strain,
v
(%)

b)

Figure 2.19. Relationships between the deviator stress q and the
axial strain
v
(measured with a pair of LDTs) at the fifth cycle at
different strain rates from drained cyclic triaxial tests performed at a
constant confining pressure
h
= 19.6 kPa on an isotropically
consolidated very dense specimen of mo ist well-compacted well-
graded gravel of crushed sand stone (Chiba gravel) (Jiang et a1.
1999; Tatsuoka et al. 1999b); and b) overlapped relationships of the
stress-strain curves in Fig. A).


same strain rate.
Considering that the
average material
property of a given
specimen is measured in
both types of test, these
test results indicate that
with these materials, the
measuring methods,
dynamic or static, have
no effects on the
measured stiffness
value.
c) With hard rocks and
mortar, a good
agreement can be seen
also between the elastic
stiffness values from
the-propagation tests
and the static tests.
Considering that if the
specimen is not
homogeneous, the body
wave velocity reflects
the property of stiffer
zones to a larger extent
than that of softer zones,
these tests results
indicate that these
materials are essentially
homogeneous in terms
of the wave length in
the wave propagation
tests. A similar result
has also been obtained
with cement-mixed sand
related to the ground
improvement work for
the Trans-Tokyo Bay
High project (Tatsuoka
and Shibuya 1991;
Tatsuoka et al. 1997b)
and sedimentary
softrock (Tatsuoka et al.
1997a).
d) With concrete, on the other hand, the stiffness from the wave propagation tests is noticeably
larger than that from the static tests even when compared at the same strain rate. These test
results indicate that concrete is noticeably heterogeneous in terms of the wave length in the
wave propagation tests. A similar result has been obtained with a gravel (Modoni et al.
1999, 2000). Tanaka et al. (1995) showed that the ratio of the elastic stiffness evaluated by
the wave propagation test to that by the static test increased with the increase in the particle
size when the particle size exceeded some limit. Souto et al. (1994) showed that the ratio of
the elastic stiffness of gravels, as used for road pavement, evaluated by the bender element
test to that by the resonant-column test increased with the particle diameter. Sato et al.
(1977a & b) showed that with hard rock cores, the ratio of the elastic stiffness evaluated by
the wave propagation test to that by the static test became a minimum, much less than unity,
when the density of crack was a certain intermediate value (as quoted in Fig. 7.5 of
18
-0.0008 -0.0004 0.0000 0.0004 0.0008
-0.0006
-0.0004
-0.0002
0.0000
0.0002
0.0004
0.0006


5
th
cycle

h
= 19.6 kPa
f (Hz)
10
5
1
0.2
0.1
0.02
0.01
0.002
R
a
d
i
a
l

s
t
r
a
i
n
,

h

(
%
)
Axial strain,
v
(%)
a)
-0.0008 -0.0004 0.0000 0.0004 0.0008
-0.0003
-0.0002
-0.0001
0.0000
0.0001
0.0002
0.0003


Chiba gravel
5
th
cycle

h
= 19.6 kPa
f (Hz)
10
5
1
0.2
0.1
0.02
0.01
0.002
R
a
d
i
a
l

s
t
r
a
i
n
,

h

(
%
)
Axial strain,
v
(%)
b)
Figure 2.20. a) Relationships between the radial strain

h
(measured with four pairs of proximity transducers)
and
v
(measured with a pair of LDTs) at the fifth cycle
for different strain rates (the
h
values have been shifted
arbitrarily so that the loops do not overlap) from cyclic
triaxial tests on an isotropically consolidated very dense
specimen of Chiba gravel (e
0
= 0.247 and w
0
= 3.73 %)
(Jiang et a1., 1999; Tatsuoka et al. 1999b); and b)
overlapped hysteresis loops.

Tatsuoka et al. 1999a). Sato et al.
(1997a & b) also showed that the
ratio was nearly one when the
amount of crack was negligible
(as shown in Fig. 2.15), while the
ratio tended to approach the unity
as the crack density increased
exceeding the above-mentioned
intermediate value (as it is with
fine-grained granular materials).
All these results indicate that the
wave propagation test becomes
not relevant for the purpose of
evaluating the average elastic de-
formation property when the
material becomes discontinuous
when compared with the wave
length in the wave propagation
test.
e) Also with the other types of
geomaterials, the effects of strain
rates on the elastic stiffness
evaluated by triaxial tests are
generally not significant (as
discussed more in detail below).
To evaluate the stress-strain
behaviour under both monotonic and
cyclic loading conditions of
undisturbed samples of geomaterial,
the triaxial testing method is the most
practical and relevant one. Note
however that to obtain reliable stress-
strain behaviour from strains less than
0.0001 % to that at peak by this
testing method, strains and stresses
should be measured accurately for this
full range of strain. The data from
static loading tests presented in Fig.
2.15 were obtained by local axial
strain measurements (except for the
data of reconstituted kaolin). Indeed,
the effects of bedding error in triaxial
compression tests on specimens that do not exhibit noticeable compression during consolidation
could be significant (e.g. Kim et al. 1994; Tatsuoka et al., 1995a; see also Fig. 2.6). Different
methods have been developed to locally measure axial strains (as summarised by Tatsuoka et al.
1999a,b); such as the inclinometer at the Imperial College (UK), LVDTs at several laboratories;
and LDTs (local deformation transducers at the University of Tokyo (Tatsuoka 1988; Tatsuoka
et al. 1991; Goto et al. 1991; Hoque et al. 1997; Santucci di Magistris et al., 1999). Figs. 2.16
and 2.17 show how LDTs are used in triaxial compression tests and Fig. 2.18 shows the details
of the arrangement of a LDT. The LDT is a sort of clip gauge made of a thin narrow strip of
phosphor bronze, which is pinched, after slightly bent, between two small metal pieces glued on
the lateral surface of the latex membrane of specimen. Electric-resistant strain gauges attached
to the central part of the strip form a full bridge, which detects very sensitively the compression
of the strip in the axial direction. The details of manufacturing, calibrating and setting of LDTs
are described by Goto et al. (1991), Hoque et al. (1996, 1997) and Santucci de Magistris et al.
(1999).

19
0.00001 0.00010 0.00100 0.01000 0.10000
0
200
400
600
800
1000
1200
1400
'
c
= 98.1 kPa
'
c
=196.2 kPa
'
c
=392.4 kPa
.
(a)
I
n
i
t
i
a
l

Y
o
u
n
g
'
s

m
o
d
u
l
u
s
,

E

0

(
M
P
a
)
Axial strain rate,
v
(%/min)
0.00001 0.00010 0.00100 0.01000 0.10000
0
2
4
6
8
10
.

Metramo silty sand
test MO03
undrained

a,SA
= 0.00075 %
e
0
= 0.307
'
c
= 98.1 kPa
'
c
=196.2 kPa
'
c
=392.4 kPa
(b)
I
n
i
t
i
a
l

d
a
m
p
i
n
g

r
a
t
i
o
,

h

0

(
%
)
Axial strain rate,
v
(%/min)
Figure 2.21a. Relationships between; a) the peak-peak secant Youngs modulus; and b) damping ratio
and the strain rate from cyclic undrained triaxial tests on an isotropically consolidated saturated
specimen of compacted Metramo silty sand (Santucci de Magistris et al. 1999).

Strain rate-dependency of small strain stiffness: Fig. 2.19 shows the relationships between the
deviator stress and the axial strain at very small strains obtained from a series of cyclic triaxial
tests at different loading frequencies on Chiba gravel (shown in Fig. 2.15; Jiang et al. 1999,
Tatsuoka et al. 1999b). The confining pressure was kept constant during each of the tests. It
may be seen that for the examined wide range of strain rate, the relationship between the
deviator stress and the axial is rather independent of loading frequency (i.e., independent of
strain rate). The associated relationships between the axial and lateral strains are presented in
Fig. 2.20. A similar insensitivity of stress-strain behaviour to changes in the strain rate may be
seen in these relationships.
It may be seen by carefully examining Fig. 2.15, however, that the dependency of the peak-
to-peak secant Youngs modulus of the gravel is not perfectly negligible, in particular at smaller
strain rates, while the dependency becomes smaller as the strain rate becomes larger. It is also
the case in a more clear manner with a high-compacted silty sand, which was used as the core
material for a rockfill dam (Metramo silty sand from Italy; Santucci de Magistris et al. 1999,
Santucci de Magistris and Tatsuoka 1999). The detailed result for this silty sand is presented in
Figs. 2.21a. The behaviour presented in Fig. 2.21a was simulated based on a linear three-
component rheology model by Di Benedetto and Tatsuoka (1997).
0.0000 0.0005 0.0010 0.0015
0
2
4
6
8
10
12
14
16
18
20
22
larger strain rate
start of loading
Metramo silty sand
MO03UT
3
rd
cycle
'
c
= 392.4 kPa
(a)
Axial strain rate
v
3.52x10
- 5
%/min
1.55x10
- 4
%/min
4.11x10
- 4
%/min
8.08x10
- 4
%/min
2.44x10
- 3
%/min
8.00x10
- 3
%/min
2.44x10
- 2
%/min
D
e
v
i
a
t
o
r

s
t
r
e
s
s

i
n
c
r
e
m
e
n
t
,

q

(
k
P
a
)
Axial strain increment,
v
(%)
-0.0015 -0.0010 -0.0005 0.0000
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
larger strain rate
start of unloading
(b)
Axial strain rate
v
3.52x10
-5
%/min
1.55x10
-4
%/min
4.11x10
-4
%/min
8.08x10
-4
%/min
2.44x10
-3
%/min
8.00x10
-3
%/min
2.44x10
-2
%/min
D
e
v
i
a
t
o
r

s
t
r
e
s
s

i
n
c
r
e
m
e
n
t
,

q

(
k
P
a
)
Axial strain increment,
v
(%)
Figure 2.21b. Relationships between the deviator stress q and the axial strain
v
(measured with a
pair of LDTs) at strains less than about 0.0015 % during a) loading ( q > 0) and b) unloading ( q <
0) fromcyclic undrained triaxial tests on isotropically consolidated Metramo silty sand (Santucci de
Magistris et al. 1999); in each figure, the origins have been reset to the common starting point of
loading and unloading.

v
&

v
&

20
0.00001 0.00010 0.00100 0.01000 0.10000
800
900
1000
1100
1200
1300
1400
Axial strain, 2(
v
)
sa
1.05x10
-6
2.02x10
-6
5.00x10
-6
1.47x10
-5
Metramo silty sand
MO03UT
3
rd
cycle
'
c
= 392.4 kPa
.

S
e
c
a
n
t

Y
o
u
n
g
'
s

m
o
d
u
l
u
s
,

E
s
e
c

(
M
P
a
)
Axial strain rate,
v
(%/min)
Elastic property
Quasi-elastic
property
Figure 2.22. Relationships between the secant Youngs modulus
defined at different small strains obtained from the data presented in
Fig. 2.21 (Santucci de Magistris et al. 1999).


@ @ q Elastic limiting line

Slope; E
o
at Increasing
strain rate 1 constant strain rate

@Strain rate 1 @
@ @ @ @ @ @ @ @ @ @ @ @ @ @
@ @ @ @ @ @ @ @ @ @ @

; Limit of elastic behaviour at each strain rate
0 0.001 %

Figure 2.23. Framework of stress-strain relationship at small strains (Tatsuoka and Shibuya 1992:
Tatsuoka et al. 1999a).
To examine the
dependency of the stiffness
and damping ratio on the
strain rate seen in Fig. 2.21a,
the respective stress and
strain relationships during
loading and unloading in
cyclic loading tests were
plotted separately (Fig.
2.21b). It may be seen that
the stress-strain relation
becomes more non-linear
and its strain rate-
dependency becomes more
obvious with the increase in
the strain level (defined from
the moment of reversing the
loading direction): in other
words, the stress-strain
relationship becomes more
linear with the increase in the strain rate, while the initial slope is rather insensitive to the strain
rate. Fig. 2.22 shows the secant modulus values that are defined at different strain levels from
the moment of reversing the loading direction, plotted against the average strain rate in each
loading cycle. It may be seen that the secant Youngs modulus increases with the decrease in
the strain level even at strains less than 0.001 %, while the dependency of Youngs modulus on
the strain rate decreases with the decrease in the strain level. At a strain of 0.0001 %, the
Youngs modulus is essentially independent of strain rate, which could therefore be defined as
the elastic Youngs modulus. The Young modulus values defined at a strain of 0.001 %
exhibits strain rate-dependency with a noticeable damping value (Fig.2.21), which should
therefore be called the quasi-elastic Youngs modulus (Tatsuoka et al. 1999a).
The test results shown above indicate that the framework shown in Fig. 2.23 is relevant to
the stress-strain relationships at small strains of geomaterial (Tatsuoka and Shibuya 1991;
Tatsuoka et al. 1999a). That is, as the strain rate becomes larger, the stress-strain relation
approaches an upper-bound relation. So, the length of elastic (i.e., reversible and rate-
independent) stress-strain relation becomes larger as the strain rate becomes larger, while, if the
strain rate is very low, the initial stress-strain relation could be located below the upper-bound
relation from the start of loading. This framework is consistent with the linear three-component
v
&
21

10 100 1000 5000
10
100
1000
5000
RCT
BS+DC
(
1
:
2
)
(
1
:
1
)
Range for Soft rocks and
Cement-treated soils
(BS+DC) and clays
RCT
BS+DC
Kazusa


Sedimentary soft rock
Local axial strain measurements
G
0
=
E
0
/
{
2
(
1
+

)
}

(
M
P
a
)
(

=
0
.
5

f
o
r

c
l
a
y
s

a
n
d

0
.
4
2

f
o
r

s
o
f
t
r
o
c
k
s
)
G
f
=(V
s
)
vh
2
(MPa)
Kobe

Sagara

Miura

Uraga-A

Slurry


Dry


Pleistocene clay site
Cement-treated soil
DMM


TS
BS
Tokyo
bay

Osaka
bay



OAP

RCT=rotary coring
DC=direct coring
BS=block sampling
TS=fixed-piston thin-wall sampling
Suginami



Uraga-B

Tokoname


Figure 2.24. Comparison of the shear modulus
2
f s
G V from the field downhole seismic survey
with the respective corresponding value
0
0
0
2(1 )
E
G

+
from triaxial tests on undisturbed specimens
of stiff clays, sedimentary softrocks and cement-mixed sand and clay obtained either by block or
direct sampling or by rotary core tube sampling (Tatsuoka et al. 1995a & c, 1999a).
rheology model (Di Benedetto and Tatsuoka 1997) and the non-linear three-component
rheology models described later.
Today, the pre-peak stress-strain relationships for a strain range from less than 0.001 % to
that at the peak stress state can be evaluated by means of a relevant static stress-strain test using
a single specimen. The relationships between static and dynamic experiments, between
laboratory and field techniques, and between testing and field full-scale behaviour, which have
been rather understood in a separated manner, can be better described based on the elastic
deformation characteristics, while referring to other important features; including a) effects of
strain and pressure non-linearity; b) kinematic yielding; c) effects of recent stress-time history;
d) anisotropy; e) structuration and destructuration; and e) effects of cyclic loading (Tatsuoka et
al., 1999a). This methodology is considered more relevant when the elastic modulus values
22
; independent of
; independent of
0
0
( )
m
v
v v
E E

_

,

1) E
v
/E
h
is proportional to the inherent anisotropy.
2) E
v
/E
h
increases in a non-linear fashion with .
0
0
( )
( )
v v
m
h
v
h h
E
E
E
E

0
0
( )
m
h
h h
E E

_

,

Figure 2.26. Cross-anisotropic elasticity model (Tatsuoka et al.,
1999a).
1
0 0 0
1
0 0 0
1
0 0 0
1
1
2
2
1
0 0 0 0 0
1
1
2
2
1
0 0 0 0 0
1
2
1
0 0 0 0 0
hv hv
v h h
e
v v
vh hh
e
h h
v h h
e
h
h
vh hh
e v h h
vh
vh
e
vh
hv
e
hv
hh
hh
E E E
d d
d d
E E E
d
d
E E E
d
d
G
d
d
G d
G

1

1
1
1
1
1

1
1
1
1
1
1

1
1
1

1
1
1
1
1
1
1
1
1
1
1
1 1
]
1
1
1
]
1
2
vh
hh
d

1
1
1
1
1
1
1
1
1
1
1
1
]

Figure 2.25. Compliance matrix for cross-anisotropic materials
(Tatsuoka and Kohata, 1995).
from the corresponding
field shear wave velocity
measurements and
laboratory stress-strain
tests are consistent to
each other. Fig. 2.24
compares the respective
pair of the shear modulus
2
f s
G V obtained from
the field downhole
seismic survey with the
corresponding value
obtained from triaxial
tests on undisturbed
specimens reconsolidated
to the field pressure level
(Tatsuoka et al. 1995a &
c, 1999a). The respective
value of
0
G shown in this
figure is the averaged of several data corresponding to a range of depth for which the value of
f
G was measured. Each
value of
0
G was obtained
from a Youngs modulus
value
0
E measured at
axial strains less than
about 0.001 % using a
relevant Poissons ratio
0
. In this global
comparison, the possible
effects of inherent and
stress system-induced
anisotropy and
discontinuities are
considered to be
secondary. It may be
seen that with stiff clays
of Pleistocene Era, these
two types of shear
modulus are consistent to each other. With sedimentary soft rocks, the agreement is generally
satisfactory when the values of
0
0
0
2(1 )
E
G

+
were evaluated by using undisturbed samples
obtained by block sampling or direct coring at the site. It can be seen on the other hand that for
many data points, the values of
0
G are noticeably lower than the corresponding
2
f s
G V
value. With most of these data points, samples that were obtained by the rotary core tube
sampling operated from the ground surface were used for the triaxial tests. It is very likely that
these samples were more-or-less disturbed. This issue is discussed in detail by Tatsuoka et al.
(1995c, 1999a). Taking into account the effects of sample disturbance, it could be concluded
based on the data presented in Fig. 2.24 that when the ground consists of fine-grained soils or
rocks without a noticeable amount of discontinuity, the elastic deformation property of
geomaterial in the field could be evaluated by the field shear wave velocity measurement.
/
v h

23

Proximeter for
h
see
Fig. 2b
Lateral LDT
Vertical LDT
W=23 cm
W=23 cm

h
H=57 cm
Proximeter for
v
Lubricated
9.5 cm
19 cm
9.5 cm
3.5 cm


Figure 2.27. Rectangular prismatic specimen with local strain
measurements of axial and lateral strains for triaxial tests (Jiang
et al. 1997, 1999: Hoque et al., 1996).


0.01 0.1 1
50
100
1000
2000
h h
v v
h
h
v
' (kPa)
' (kPa)
49
98
147
196
245
294
343
392
441
h
v
E



o
r

E




(
M
P
a
)
h v
' or ' (MPa)
25~108
49~216
74~323
98~431
123~539
147~647
172~755
196~862
221~970
Nerima gravel
TC test ( '=49kPa)
E ~ '
E ~ '
stress states
Isotropic

Figure 2.28. Relationships between: a) vertical Youngs modulus and vertical stress; and b) lateral
Youngs modulus and lateral Youngs modulus, at isotropic stress states and triaxial stress states
(Kohata et al. 1997).
Modelling of elastic stress-
strain behaviour (hypo-elastic
models): The definition of
elasticity for geomaterials is
not simple, because the elastic
deformation properties of
geomaterial are usually not
constant with respect to
changes in the stress state even
for a given element of geo-
material. Moreover, for a
given type of geomaterial at a
certain stress state, they are
also a function of density,
stress and strain history and so
on. So, only hypo-elasticity
models, for which elastic strain
increments are related to stress
increments through a stiffness
or compliance matrix that is a
function of instantaneous stress
state (and density, stress and
strain history and others) are
relevant. For example, Fig.
2.25 shows the compliance
matrix for a cross-anisotropic
material having the axis of
symmetry in the vertical
direction, which is relevant to, for example, a mass of geomaterial that has horizontal bedding
planes and the principal stresses working in the vertical and horizontal directions. In addition,
careful distinctions should be made between elastic, plastic and viscous properties as follows;
24

0.0 0.5 1.0 1.5 2.0 2.5
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
h
v
h v
anisotropy
Stress system-induced
Inherent anisotropy
Perfectly isotropic material
Toyoura sand
SLB sand
Ticino sand
Hime gravel
Chiba gravel
Nerima gravel
E



/
E
'/ '
0
0
( )
( )
v v
m
h
v
h h
E
E
E
E

Figure 2.30. Relationships between the ratio of vertical and horizontal elastic Youngs modulus
values and the principal stress ratio for cross-anisotropic sands and gravels (Kohata et al. 1997,
Tatsuoka et al. 1999a).
0.01 0.1 1
100
1000
m
h
0
e
v
Toyoura sand 0.49
SLB sand 0.47
Ticino sand 0.53
Hostun sand 0.47
Hime gravel 0.51
Nerima gravel 0.52
Chiba gravel 0.52
Nagoya gravel 0.57
E



/
f
(
e
)

(
=
E



/
f
(
e
)
)

(
M
P
a
)
' = ' (MPa)
Power m 0.5 for of uncemented materials,
( )
m
v v
E

Figure 2.29. Relationships between the elastic Youngs modulus
and the confining pressure at isotropic stress states of
uncemented sands and gravels (Kohata et al. 1997).

- elasticity; reversible and
time-independent deformation
properties for a given stress
history;
- plasticity; irreversible (with
energy dissipation) and time-
independent deformation
properties for a given stress
history; and
- viscosity; irreversible (with
energy dissipation) and time-
dependent deformation
properties for a given stress-
history.
One of the most primitive
hypo-elastic models for
uncemented geomaterial is the
one in which the elastic
deformation property is
isotropic and the elastic
Youngs modulus E is a
function of instantaneous
minor principal stress
3
. It is readily seen that this model is too simple when compared with
the actual elastic deformation properties of geomaterials, which is inherently anisotropic and
can also become more anisotropic at more anisotropic stress states. Hardin (1978) proposed on
the other hand that, for sands, the elastic Youngs modulus
x
x
x
E

in any particular direction

25
X is a unique function of the normal stress
x
working in direction X, independent of the
normal stresses acting in the other orthogonal directions.
The deformation characteristics developed at very small to intermediate strains of a variety of
geomaterials are now often evaluated by testing using modern laboratory stress-strain tests. A
great amount of such data as above supports the proposal of Hardin (1978) described above (e.g.
Kohata et al. 1994, 1997, Tatsuoka et al. 1999a, Jiang et al. 1997, Hoque and Tatsuoka 1998).
Based on results from such tests, a hypo-elasticity model with inherent and stress system-
induced anisotropy has been developed by extending the above proposal by Hardin (Tatsuoka et
al., 1999a, b & c). The major feature of this model is summarised in Fig. 2.26.
Fig. 2.27 shows a rectangular prismatic large triaxial specimen with local strain
measurements by means of a pair of vertical LDTs for axial strains and four pairs of lateral
LDTs for lateral strains. In this way of testing, both axial and lateral strains that are free from
effects of bedding error at the top, bottom and lateral surfaces of specimen can be evaluated.
Very small unload/reload cycles of vertical stress, with strain amplitudes of the order of
0.001 %, were applied at a constant confining pressure, while very small unload/reload cycles of
lateral stress were applied at a constant vertical stress. Such cyclic tests were performed at
various isotropic and anisotropic stress states. As the two lateral orthogonal principal stresses
are always the same, the lateral Youngs modulus
h
E was obtained as
( )
.
(1 ) /
v
h hh h h
const
E

, while assuming that the Poisson ratio

hh
in this case is equal
to the value of
vh
at the stress ratio where
v h
E E (Jiang et al. 1997). Fig. 2.28 shows typical
results for a very dense well-graded gravel consisting of crushed sandstone (Kohata et al. 1997).
It may be seen that the vertical Youngs modulus
v
E measured at isotropic and anisotropic
stress states is essentially a unique function of the vertical stress
v
, while the lateral Youngs
modulus
h
E is essentially unique function of the lateral stress
h
. Similar results have been
obtained for poor-graded sands (Hoque and Tatsuoka, 1998), for a reconstituted gravel (Jiang et
al. 1997, Balakrishnaiyer et al. 1998) and for a undisturbed gravel (Koseki et al., 1999). This
result shows that the model illustrated in Fig. 2.26 is relevant. Some data points of
v
E and
h
E
obtained near the failure state in triaxial compression deviate to values lower than the respective
value that is obtained at isotropic stress states. This is due likely to effects of damage by
shearing to the elastic deformation properties (Flora et al. 1994; Kohata et al. 1997; Tatsuoka et
al. 1999a,b; Koseki et al. 1998). Fig. 2.29 summarises the power law at isotropic stress states
for inherently cross-anisotropic geomaterials (Kohata et al. 1997). It may be seen that the
power m is around 0.5.
Referring to Fig. 2.26, we have the following relationship between the vertical and horizontal
elastic modulus values:

0
0
( )
( )
m
v v v
h h h
E E
E E

_


,
(2.1)

This equation means that the ratio
v
h
E
E
increases in a non-linear fashion with
v
h
R

. This
prediction is supported by the data (Fig. 2.30).
It is reasonable to assume that the compliance matrix for quasi-elastic strain increments for
sands and gravels is symmetric as:

vh hv
v h
E E

(2.2)

26
With m= 0.5, we can then assume the following equations for the Poissons ratios
vh
and
hv

using the value for the isotropic behaviour
0
:
26

Figure 3.1. Typical results from plane strain compression tests
of air-pluviated specimens of SLB sand (Park and Tatsuoka
1994).
0.5
0
( )
m
vh
a R = and
0.5
0
/( )
m
hv
a R = (2.3)

where a is /
v h
E E at
1 R = ; and
0
is
vh hv
=
when
1/ n
R a

=

(Tatsuoka and
Kohata 1995). Eq. 2.3 means
that the elastic Poissons ratio
is not constant either, but it
changes with the stress ratio
(as supported by the data
presented in Fig. 2.31).
Similar data are reported also
in Tatsuoka et al. (1999b). In
this respect, this model is
somewhat different from the
one proposed by Hardin and
Bradford (1989).

Summary: The elastic
Youngs modulus of
uncemented sands and
gravels can be modelled by a
hypo-elastic model representing both inherent and stress system-induced anisotropy. A model
for a cross-anistropic case with the vertical symmetric axis is shown.
Another important topic that could not be touched upon in this report is the non-linearity of
stress-strain relation due to changes in strain and stress state under cyclic loading conditions
(e.g., Iwasaki and Tatsuoka 1977, Tatsuoka et al., 1978, 1979a & b) and that under monotonic
loading conditions (e.g., Shibuya et al., 1991, Tatsuoka et al., 1999a).


3 INHERENT ANISOTROPY IN THE STRENGTH AND DEFORMATION
CHARACTERISTICS OF GRANULAR MATERIALS

Arthur,R. (UK) and Oda,M.
(Japan) are two among the
pioneers who disclosed
systematically inherently
anisotropic deformation and
strength characteristics of
granular materials (Tatsuoka
1987). Tatsuoka et al.
(1986a), Lam and Tatsuoka
(1988a & b) and Park and
Tatsuoka (1994) performed a
systematic study on this
subject. Fig. 3.1 shows
another example showing the
inherently anisotropic stress-
strain behaviour of sand (Park
and Tatsuoka 1994). Fig. 3.2
shows grain size distribution
curves of poorly graded sands
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25
0.15
0.20
.
Toyoura Sand
Expt. data

vh = (vh)R=1 ( v/h)
nv/2

vh

v
' /
h
'

Figure 2.31. Relationship between the elastic Poissons ratio
vh

and
v
h
R

= at triaxial extension and compression stress states

and the isotropic stress state and the fitting of the data by Eq. 2.3
(Hoque and Tatsuoka 1998).
27
Figure 3.3. Summary of the effects of inherent anisotropy on
the angle of internal friction for poorly graded granular
materials obtained from plane strain compression tests (see Fig.
3.2 for the gradings of the sands) (Park and Tatsuoka 1994).

Figure 3.2. Grain size distribution curves of sands
for which inherent anisotropy was evaluated by
plane strain compression tests (see Fig. 3.3) (Park
and Tatsuoka 1994).
for which the effects of inherent anisotropy
on the stress-strain properties were
evaluated by plane strain compression tests
(as shown in Fig. 3.1). Fig. 3.3 summarises
the effects of inherent anisotropy on the
angle of internal friction in plane strain
compression of air-pluviated samples of
these sands. Each angle of internal friction
has been divided by the respective value at
90
o
= for the same void ratio. It may be
seen that all these sands, having different
origins in the world, have a very similar
and marked trend of inherently anisotropic
strength. Figs. 3.4 and 3.5 show a similar,
but more marked, trend of inherent
anisotropy for undisturbed samples of sand
(Tatsuoka et al. 1990a).

Fig. 3.6 compares the angles of
internal friction, as a function
of the angle , obtained from
a series of triaxial compression
tests, plane strain compression
tests and torsional simple shear
tests performed on air-
pluviated specimens of
Toyoura sand (Tatsuoka et al.
1988, 1990a). Each angle of
internal friction has been
divided by the respective value
from the plane strain
compression test at 90
o
= for
the same void ratio. Note that
the data presented in Fig. 3.6
include those from triaxial
compression tests, plane strain
compression test and torsional
simple shear tests in which the
3
value at failure was very
low (such as 10 kPa or less). In such tests, very precise measurement of the effective confining
pressure and relevant stress correction for the effect of membrane force is essential to obtain
accurate results (Tatsuoka et al. 1986a & b, 1988). It can be seen from Fig. 3.6 that the strength
(and also deformation) properties obtained from different testing methods can be linked to each
other only when taking into account the effects of inherent anisotropy, among other parameters
(Tatsuoka et al. 1996c; Tatsuoka 1988; Tatsuoka et al. 1988; Pradhan et al., 1988a & b). A
more comprehensive analysis in this respect is made in Lam and Tatsuoka (1988a & b).

Summary: It has been shown above that commonly with different types of granular materials,
the pre-peak deformation properties and peak strength could be markedly anisotropic. Then, for
numerical analysis and design of boundary value stability problems in geotechnical engineering:
e.g., earth pressure, slope stability, bearing capacity of footing and so on, the following
important questions arise:
1) What is the meaning of using an isotropic value of in analysis and design related to the
28
Figure 3.5. Effects of inherent anisotropy of undisturbed
samples of resedimented Shirasu (volcanic pumice) (see
Fig. 3.4) (Tatsuoka et al. 1990a).

Figure 3.6. Comparison of as a function of , from a series of triaxial compression (TC)
tests, plane strain compression (PSC) tests and torsional simple shear (TSS) tests of air-pluviated
specimens of Toyoura sand (Tatsuoka et al. 1988, 1990a); the details of the torsional shear testing
method are described in Tatsuoka et al. (1986c) and Pradhan et al. (1988a).


Figure 3.4. Effects of inherent anisotropy of undisturbed
samples of resedimented Shirasu (volcanic pumice) (Tatsuoka et
al. 1990a).

failure of sand and gravel ?
2) What is the meaning of the classical limit equilibrium stability analysis assuming
isotropic perfectly-plastic
properties of soil and
gravel ?
In fact, the stress-strain
behaviour of soil is over-
simplified in the classical soil
mechanics. The factor of
anisotropic strength and
deformation characteristics is
only one of several essential
factors that are ignored in such
classical theories, which
includes:
a) effects of pressure level on

b) different definitions of
between compression tests
and simple and direct shear
tests; and
c) progressive failure associated
with shear banding, thereby
associated with effects of
particle sizes.
These factors, or most of them,
are often unduly ignored even in
recent research, as discussed by
Tatsuoka et al. (1989a & b, 1992,
1994c; Tatsuoka and Huang
1991) and even in some modern
numerical analysis by FEM.
The consideration of the effects
of these factors is equally
important when analysing the
29

Figure 4.1. Relationships between the stress and the strain
(averaged for the whole specimen size) from a series of
special plane strain compression tests on different granular
materials having a wide range of particle size (see Fig. 4.2:
Yoshida et al. 1995: Yoshida and Tatsuoka 1997).
failure of reinforced soil (Huang and Tatsuoka 1990; Huang et al. 1994; Huang and Tatsuoka
1994).


4 STRAIN LOCALISATION WITH SHEAR BANDING IN GRANULAR MATERIALS

4.1 Strain localisation in plane strain compression tests

Fig. 4.1 shows results from a
series of special plane strain
compression tests performed at
3
= 78 kPa and 392 kPa, with
lubricated top and bottom ends,
on a number of different types
of granular material having a
wide range of particle size (Fig.
4.2) (Yoshida et al. 1995;
Yoshida and Tatsuoka 1997). A
deformable grid, which was
made of latex rubber, had been
printed on one
2
plane of the
specimen membrane (Fig. 1.3).
Pictures of deformed grid-
printed
2
plane were taken
through the transparent
confining platen and the triaxial
cell at a number of loading
stages in each plane strain
compression test. Figs. 4.3 and
4.4 shows three typical shear
strain fields immediately before
and after the peak stress state
and immediately after the start
of residual state. These figures
were constructed from observed
displacement fields on the
2

plane, as shown in Fig. 1.3. It
may be seen from Figs. 4.3 and
4.4 that the strain has already
been noticeably localised into
some zones before the peak
stress state. In the post-peak
resume, only one of these shear
zones seen at the peak stress
state developed into a distinct
shear band while the remaining
part of specimen was unloaded
with negative local shear strain increments.
From such displacement field seen on the
2
plane as seen in Fig. 1.3, shear deformation and
volume change of a shear band at each loading stage, defined as shown in Fig. 4.5, were
obtained. Fig. 4.6 shows typical relationships between the shear stress level, defined by Eq. 4.1,
and the shear deformation of shear band in the post-peak regime for the different types of
30
Figure 4.2. Grain size distributions of granular
materials used in the plane strain compression tests
to observe strain localisation (see Fig. 4.1)
(Yoshida et al. 1995: Yoshida and Tatsuoka 1997).

Figure 4.3. Shear strain fields constructed from the observed displacement field on the
2
plane:
A) immediately before the peak stress state;
B) immediately before the peak stress state; and
C) immediately after the start of residual condition
(SLB sand;
3
= 78.4 kPa; see Fig. 4.1a for the locations of A, B and C) (Yoshida et al. 1995).
granular material (i.e., the shear defor-
mation is defined as zero at the peak stress
state).

1 3 1 3
1 3 1 3
( / ) ( / )
( / ) ( / )
peak
n
peak residual
R

(4.1)

R
n
= 1.0 means the peak stress state and R
n
=
0.0 means the residual stress state. The
corresponding relationships between the
shear stress level and the volume change of
shear band are reported in Yoshida et al.
(1995). It may be seen from Fig. 4.6 that
the post-peak stress-shear deformation
relationship is markedly different for
different grain sizes: i.e., the shear
displacement increment *
s
u that takes place as the stress state changes from the peak state to
the residual state increases with the increase in the particle size. This feature is summarised in
Fig. 4.7 (Yoshida and Tatsuoka 1997). The value of *
s
u is slightly smaller when
3
= 392 kPa
than when
3
= 78 kPa. The reason for the above is not known. It is also to be noted that the
value of *
s
u is not proportional to the mean diameter
50
D , but the ratio
50
*/
s
u D decreases
noticeably with the increase in
50
D . It seems that this is due to that particle properties other
than
50
D change with the changes in
50
D .
A similar result has been obtained from triaxial compression tests and plane strain
compression tests on sedimentary soft rock (Tatsuoka and Kim 1995; Hayano et al. 1999).
These results indicate that the post-peak deformation properties of geomaterials are controlled
by a characteristic scale that is specific to each geomaterial type under each specific stress
D (mm)
31

Figure 4.5. Definition of shear
deformation and volume change
of shear band.


Figure 4.6. Relationships between the shear stress level and the shear
displacement increment from the peak stress state for different
particle sizes at
3
= 78 kPa (Yoshida and Tatsuoka 1997).

Figure 4.4. Shear strain fields constructed from the observed displacement field on the
2
plane:
A) immediately before the peak stress state;
B) immediately before the peak stress state; and
C) immediately after the start of residual condition
(Karlsrule sand;
3
= 392 kPa; see Fig. 4.1b for the locations of A, B and C) (Yoshida et al. 1995).
conditions (and others). It seems that for granular materials, the mean diameter
50
D is the most
important parameter representing this characteristic scale. This fact indicates that the prototype
soil mass (in a large scale) cannot be scaled down into a smaller model by using the same type
of soil as the prototype. This point is discussed in the next section.

32

Figure 4.8. Bearing capacity on sand of a strip footing in plane strain.
4.2 Implications of shear banding in the issue of the bearing capacity of strip footing on sand
and particle size effects

Most of the classical theories for the bearing capacity of strip footing on sand assume that the
sand is an isotropic perfectly plastic material having a constant angle of internal friction. For
the upper bound analysis and the limit equilibrium analysis, it is further assumed that the
thickness of failure planes is zero, or independent of particle size at best. It has already been
shown in the above however that these assumptions overly simplify the actual behaviour of real
soils. In this section, the bearing capacity of strip footing on sand, as defined in Fig. 4.8, will be
discussed to demonstrate some essential limitations of the classical stability analysis and the
importance of particle size effects. In the following, only the mechanism of the bearing
capacity of a rigid strip footing with a rough and smooth footing base placed on the surface of a
homogeneous level sand layer subjected to vertical central load (i.e., the basic case) will be


Figure 4.7. Relationship between the shear displacement increment that is needed for the stress state
changes from the peak to residual states and the mean particle size (Yoshida and Tatsuoka 1997).
33

Figure 4.9. Summary of strength anisotropy in plane strain
compression tests of air-pluviated Toyoura sand (see Fi. 3.6)
(Tatsuoka et al. 1986a, 1991).
Figure 4.10. Summary of pressure level-dependency in
plane strain compression tests of air-pluviated Toyoura sand
(Tatsuoka et al. 1986a, 1991).
examined.
Toyoura sand was used in the
model tests that are explained
below. The strength
characteristics of Toyoura sand
are first summarised.
1) Fig. 4.9 summarises the
inherent anisotropy of
strength (i.e. the angle of
internal friction) obtained
from a series of plane strain
compression tests (Tatsuoka
et al. 1986a; see Fig. 3.6).
2) Fig. 4.10 shows the pressure
level-dependency of
1 3
1 3
max
arcsin

=

+

at
90
o
= from the plane strain
compression tests (Tatsuoka
et al. 1986a).
3) Fig. 4.11a compares the
relationships between the
friction angles and the void
ratio obtained from the
following different tests:
a) at 90
o
= from plane
strain compression tests,
which is the largest value
with respect to the angle
(i.e. the conventional
plane strain compression
tests);
b) at 23
o
= from plane
strain compression tests,
which is the smallest value with respect to the angle ;
c) at 90
o
= from triaxial compression tests, which is noticeably smaller than the
corresponding value from the plane strain compression test at 90
o
= (i.e. the
conventional triaxial compression tests); and
d) the simple shear angle of friction defined as
max
arctan
hv
SS
v

=


from torsional simple
shear tests using hollow cylindrical specimens in which the bedding plane is horizontal
(Pradhan et al. 1988a & b).
In the torsional simple shear tests, as the magnitudes and directions of the three principal
stresses were continuously measured. In these tests, the directions of the principal stresses
rotate before reaching the residual state. Fig. 4.11b compares the friction angles,
1 3
1 3
max
arcsin

=

+

and
max
arctan
hv
SS
v

=


, from the torsional simple shear tests. It
may be seen that the friction angle
1 3
1 3
max
arcsin

=

+

is consistently larger than the
34
Figure 4.11a. Comparison of the relationships
between the friction angles and the void ratio
from different tests for air-pluviated Toyoura sand
(Tatsuoka et al. 1991; Siddiquee et al. 1999).


Figure 4.11b. Comparison of the values of
and
SS
for the same torsional simple
shear test data for air-pluviated Toyoura
sand (Tatsuoka et al. 1991; Siddiquee et al.
1999).

simple shear angle
max
arctan
hv
SS
v

=


(Pradhan et al. 1989, Pradhan and Tatsuoka 1989,
Tatsuoka et al. 1991). This is due to the fact that the horizontal plane, along which the
normal strain is always zero, is not the plane of maximum stress obliquity due to the fact
that the dilatancy angle is significantly smaller that the angle of internal friction (e.g.
Pradhan et al. 1989a & b: Tatsuoka et al. 1988). It is to be noted that
SS
values obtained
by conventional direct shear tests could be subjected to some large experimental errors,
which could mask the relationship shown in Figs. 4.11a & b (Qui et al. 2000).
For these data, all the specimens were prepared by the same method (pluviation of air-dried
particles through air). The same preparation method was also used to prepare the sand bed for
the model tests that are described below. It may be seen from Fig. 4.9 4.11 that the range of
strength for the same void ratio is very large among these different test methods performed
under otherwise the same testing conditions.
Fig. 4.12 summarises the representative classical bearing capacity theories for the coefficient
for sand weight N

(Tani 1986; Tatsuoka et al. 1991). Then, only from the fact that the friction
angle is not unique for the same mass of sand, it is understood that it is very hard to predict the
realistic bearing capacity of a footing in sand based on any of these classical bearing capacity
theories. Even based on the same assumptions (i.e., isotropic perfectly plasticity with a constant
and zero shear band thickness), the N

values by these classical theories differ from each

other due to different assumptions with respect to the distribution of the friction angle on the
footing base or the failure mechanism of the active zone immediately below the footing.
Although the variation in the N

value among the classical theories is not small with a range of

about two times, the difference between the classical theories and the actual value is much
larger, as shown below.

35

Figure 4.12. Summary of relationships between the factor of soil
weight N

and by the representative classical bearing

capacity theories (Tani 1986; Tatsuoka et al. 1991).

Fig. 4.13 compares the
relationships between the value
of N

value and the friction

angle according to the
classical bearing capacity
theories, which are denoted by
a band in the figure, with
experimental results for air-
pluviated Toyoura sand
(Tatsuoka et al. 1991). The
experimental results were
obtained from a comprehensive
series of element tests, as
described above, and plane
strain model tests using
different sizes of strip footing
under gravitational acceleration
(1g). All the experimental
results were obtained with air-
pluviated Toyoura sand and
corrected to the same void ratio
(e= 0.66). Therefore, these
data points should collapse into
a single point if the sand were
an isotropic perfectlyplastic
material having a single value
of . The data scatter very
largely, however, due to the
following reasons:
1) In the model tests (1g), the N

value decreases with the increase in the physical footing size

(
0
B ). This behaviour has been called the scale effect.
2) For a single value of N

for the respective model test condition with a single value of

0
B ,
different values of and
SS
from different types of shear tests are plotted. These friction
angles were obtained at the respective
3
value equal to one tenth of the evaluated average
footing pressure. So, these values decrease with the increase in
0
B .
It is seen from Fig. 4.13 that if we substitute the value of from plane strain compression
tests at 90
o
= into a classical bearing capacity theory, the actual N

value is largely over-

estimated to a degree that cannot be covered by the global safety factor that is used in usual
practice, such two to three. For example, ( ; 90 ) 49
o o
PSC = = for
0
B = 50 cm is substituted
into the classical bearing capacity theories, we obtain N

800, which is about eight times as

large as the measured value (about 100). This extremely large overestimation is due to the fact
that not only the effects of strength anisotropy but also the effects of progressive failure of sand
bed, which is explained in detail below, are not considered in the above-mentioned procedure to
obtain the N

value. That is, the consideration of the pressure level-dependency of is not

sufficient to explain the discrepancy between the classical theories and the real behaviour of
sand.
To validate the above-mentioned notion, results obtained from a comprehensive series of
model bearing capacity tests of a strip footing on sand are shown below.

36

Figure 4.13. Comparison between the classical bearing capacity theories and the experimental results
for air-pluviated Toyoura sand with a void ratio of 0.66 with N

values from FEM analysis plotted

against ( ; 90 )
o
PSC = (Tatsuoka et al. 1991).


Small- scale 1g model tests (Tani 1986): Fig. 4.14 shows the sand box in which a sand bed of
Toyoura sand was made by the air-pluviation technique. The side walls were lubricated by
using a thin latex membrane smeared with a controlled thickness (0.05 mm) of a selected
silicone grease. A very low friction angle with this configuration has been confirmed (Tatsuoka
et al. 1984; Goto et al. 1993). The footing load was measured by using eleven load cells, each
measuring normal and shear loads. These local load cells were arranged at the central third of
footing (Fig. 4.15); so the footing load measured in this way is therefore essentially free from
the effects of side wall friction, if any. A grid had been printed on the surface of the membrane,
and displacements at the nodes of the grid were obtained from pictures of the grid that were
taken at several loading stages during each test. Strain fields in the sand bed were then obtained
from these observations. In some tests, the sand bed, which was air-dried during each loading
test, was made wet after each loading test had been over so that stable vertical faces of the sand
bed could be excavated and exposed without a support (as shown in Fig. 4.17). It was
confirmed that the deformation of the membrane in contact with the side wall, which was seen
through the transparent side wall, be essentially the same as that seen in the exposed central
section of the sand bed (Tani 1986). The sand box was made very stiff so that the plane strain
conditions could be satisfied. The bottom face of footing was made rough in most series of
tests, while it was lubricated in one series of tests. The loading of footing was made by
displacement control. Similar configurations were taken in the other series of model tests (i.e.
large scale 1g tests and centrifuge tests that are described later in this paper).
The relationships between the normalised footing pressure and the normalised footing
settlement obtained from two typical 1g tests using a rough footing with a width
0
B of 10 cm
are shown in Fig. 4.16. Fig. 4.17 shows the central section of the sand bed in the test in which
37

Plane strain bearing capacity tests
air-dried Toyoura sand (B
0
= 10 cm in 1 g)
Sand box: 40 cm wide,
183 cm long and
49 cm (sand depth )
Lubricated


Figure 4.14. Small scale plane strain sand box (Tani 1986).
B
0
= 10 cm
1/3 of footing length = 40/3 cm
Eleven two-component load cells (normal and shear stresses)

Figure 4.15. View from the bottom of the strip footing having a
width of 10 cm (Tani 1986).
the loading was terminated
when the footing load became
nearly the peak state (see Fig.
4.16). Thin layers of black-
dyed Toyoura sand seen in the
picture had been placed
around the central section of
the sand bed when it was
prepared. It may be seen
from Figs. 4.16 and 4.17 that
the footing settlement was
about 5 % of
0
B at the peak
footing load, and by this
moment, shear bands had
developed for some length
from the edges of the footing.
It may also be seen that the
developed shear bands were
only a part of the full
potential shear bands which
are assumed to have
developed at the moment of peak footing load in the classical bearing capacity theories. Fig.
4.18 shows the shear strain field corresponding to Fig. 4.17. It can also be seen from this figure
that the strain in the sand bed are extremely non-uniform, so it is the case along the potential
failure planes. This fact indicates a highly progressive nature of the failure of the ground.
Fig. 4.19 shows the central section of sand bed that was exposed in another test in which
loading was continued until the footing settlement became as large as about 70 % of
0
B . It may
be seen that the full potential shear bands, reaching the ground surface, had developed only far
38


Figure 4.17. Picture of the exposed central section of the sand bed in the test in which loading was
stopped around the peak footing load (the footing width= 10 cm; the horizontal back colour strips are
black-dyed Toyoura sand placed only around the central section of the sand bed) (see Fig. 4.16)
(Tatsuoka et al. 1991).
after the peak footing load had
been attained. Fig. 4.20 is the
zoom-up of a local zone below
the footing. It can be seen from
Fig. 4.20 that the failure planes
are not planes without a
thickness, but they are bands
having an intrinsic thickness,
and the shear deformation is not
uniform along the shear bands.
It is practically impossible to
evaluate the local stress states
inside the model sand bed in
such model tests as those in this
series of tests. Therefore, the
stress field was estimated as
follows.
1) The relationship between the
mobilised angle of friction
1 3
1 3
arcsin
mob

=

+

and
the shear strain
1 3
=
for different angles of
were constructed from the results of the plane strain compression tests of Toyoura sand
explained before (Fig. 4.21) (Tatsuoka et al. 1991). The shear strains shown in this figure
are local values that were averaged for a 1 cm-wide band including a shear band. A value
of 1 cm was selected to be equal to the spacing between the lines of the grid printed on the
Figure 4.16. Typical relationships between normalised
footing pressure and normalised footing settlement from
two typical 1g model tests with
0
10 B cm = (Tatsuoka et
al. 1991) .
Fig. 4.17
39
membrane in the model
tests. Note that the
decreasing rate of stress
level in the post-peak
regime is very low in
these stress-strain curves
when compared with
those seen in the usual
relationships between
the stress and the strain
that is averaged for the
whole of a specimen. In
this sense, relationships
between the stress and
the strain that is
averaged for the whole
of a specimen are not
objective in the sense
that the post-peak stress-
strain curves are
controlled by the ratio of
the particle size to the
specimen size.
2) It was assumed that the
direction of
1
be the
same as the direction of
1
measured at each
point in the sand bed. This was based on the consideration that the most part of the strains
that takes place by the moment when the footing load becomes the peak value are inelastic
and the principal directions of inelastic strain increment are close to the instantaneous
principal directions of stress, as validated with Toyoura sand by Pradhan et al. (1988a & b,
1989) and Pradhan and Tatsuoka (1989). .

Figure 4.18. Local shear strain (%) contours in the zone below the
footing around the peak footing load state (the grid spacing is 1 cm)
(Tatsuoka et al. 1991).

Figure 4.19. Picture of the exposed central section of the sand bed in which loading was
continued to S/B
0
of about 0.7, far after the moment when the peak footing load was attained; a
shear band has developed up to the surface of the sand bed in each side of the footing (Tatsuoka et
al. 1991).

40
- The shear displacement along shear band is not uniform,
indicating the progressive failure of the ground !
- The shear band has an intrinsic thickness !


Figure 4.20. Zoom up of the central zone of Fig. 4.19.

Figure 4.21. Relationships between the mobilised angle of friction
1 3
1 3
arcsin
mob

=

+

and
the shear strain
1 3
= for different angles of from plane strain compression tests of
Toyoura; the shear strains are local values averaged for a 1 cm-wide band including a shear
bands (Fig. 1.3) (Tatsuoka et al. 1991).
3) The value of
1 3
1 3
arcsin
mob

=

+

at each point in the sand bed was then obtained by
41
substituting the measured local shear strain into the relationship between the mobilised
angle of friction and local shear strain presented in Fig. 4.21.
Fig. 4.22a shows the distribution of local shear strain at the moment of the peak footing load
in the second test (e= 0.722) shown in Fig. 4.16. Fig. 4.22b shows the corresponding
distribution of the mobilised angle of friction
1 3
1 3
arcsin
mob

=

+

estimated by the method
described above. It can be seen that the failure of the ground is not simultaneous at all in the
sense that at any moment of loading, the peak local strength is never mobilised simultaneously

a)
b)
c)
Figure 4.22. a) Shear strain field at the peak footing load state; b) the corresponding field of
mobilised angle of friction (in degree); and c) stress state below footing at the peak footing load (see
Fig. b) (B
0
= 10 cm; e= 0.66; Tatsuoka et al. 1991).

Pre-peak stress state
Post-peak stress state
Near-peak stress state
42

Figure 4.23. Comparison between the stress characteristics
solutions assuming the perfect plasticity and the measured
bearing capacity (Tatsuoka et al. 1991).
along the potential failure planes.
At the moment of peak footing
load, the following trends of be-
haviour could be seen from Fig.
4.22b (see also Fig. 4.22c):
1) Some part adjacent to the
footing edges is already in the
post-peak regime.
2) The peak and near peak stress
states are attained in only a
limited zone below the footing.
3) A small zone immediately
below the footing base and the
large remaining part outside the
footing width are in the pre-
peak regime.
The significantly progressive
nature of ground failure described
above was further confirmed by
calculating the footing load by the
stress characteristics method based
on the anisotropic peak strength
which is a function of the values of
and
3
at each point in the sand
bed while using the friction angles
at the footing base measured at the
moment of peak footing load (Fig.
4.23). In this figure, the
relationships between the measured
value of N

and the void ratio for

both a rough footing and a smooth
footing with a lubricated footing
base are shown. These measured
bearing capacity values are compared with the respective theoretical values obtained as above
for the rough and smooth footings. It may be seen that even when the strength anisotropy is
considered, the use of peak strength along the potential failure planes results into a significant
over-estimation of the measured bearing capacity.

Large- scale 1g model tests: The above-mentioned fact was further confirmed by performing
similar model tests but in a scale that was larger by a factor of five than the above-mentioned
series of model tests (Morimoto 1990; Tatsuoka et al. 1991; Siddiquee et al. 1999, 2001). Figs.
4.24, 4.25, 4.26 and 4.27 show the test configurations of the large model tests. All the test
conditions (i.e., the shape of the sand bed, the relative size of the sand bed to the footing size,
the plane strain conditions with the lubrication of side wall, the model sand, the air-pluviation
method to prepare the sand bed, the use of eleven two-component load cells at the central third
of footing, the displacement-control loading, the observation of the deformation of the latex
rubber membrane used for the lubrication to obtain the deformation of sand bed and so on,
except for the size of model) were made the same with those for the small-scale model tests
described above.
Fig. 4.28 shows the results from two representative 1g tests using a rough footing with a
physical width
0
B equal to 50 cm (n.b., the results from a centrifuge test shown in this figure
are discussed later in this paper). Fig. 4.29 shows a picture taken when the footing load became
nearly the peak in one of these two tests. Fig. 4.30 shows a typical shear strain field that was
constructed based on such a picture. It may be seen that the strain field is extremely non-
43
Lubrication of the side wall
using a thin latex membrane
smeared with silicone grease
Pluviation of air-dried
Toyoura sand
Through air

Figure 4.25. Preparation of a lubrication layer on the side walls and preparation of sand bed by
pluviating air-dried sand particle through air from a slit of a moving hopper (Morimoto 1990;
Tatsuoka et al. 1991).

uniform, and the degree of non-uniformity is much larger than the one that was observed in the
corresponding small-scale model test (Fig. 4.18). This was due to the fact that the ratio of the
thickness of shear band, which was essentially independent of the scale of model test, to the
footing size was smaller by a factor of five compared with that in the small scale model tests.
This means that the degree of progressive failure (i.e., the degree of non-simultaneous
mobilisation of local peak strength) becomes larger with the increase in the ratio of footing size
to particle size; i.e., the bearing capacity factor N

becomes lower with the increase in the

footing width
0
B when the footing is placed on the same type of sand. Such a decrease in the
N

values with the increase in the

0
B value has been called the scale effect.
Rough footing
(0.5 m wide & 2 m long)
Three local load cells on each 1/3 of footing
Eleven local load cells on central 1/3 of footing
Large pit (2m wide, 7 m long and 4 m deep

Figure 4.24. Large-scale model tests; a) footing (the footing base shown in this picture); and b)
general view (a loading reaction frame set above the footing) (Morimoto 1990; Tatsuoka et al. 1991).
44
Fig. 4.31 shows the picture taken nearly at the end of loading, at a footing settlement ratio of
about 16 %. It may be seen that a wedge has developed below the footing, but the full potential
failure planes have not developed at all by this loading stage.

Small- scale centrifuge model tests: It is shown below that the scale effect on the
0
/ N S B :
relationship as well as the values of N

and
0
/ S B at the peak footing load consists of the
following two components (Tatsuoka et al. 1991; Siddiquee et al. 1999):
1) Pressure-level effect, which is due purely to the effect of pressure changes on the stress-
strain behaviour of sand. This effect can be purely observed in centrifuge tests changing the
pressure level (or vertical acceleration level) keeping the same size of footing and using the
same sand type.
2) Pparticle size effect, which is due purely to the effect of the ratio of sand particle size
relative to footing size. This effect can be observed by comparing the bearing capacity
characteristics between different tests using the same sand for the same equivalent footing
width
o
B B n = , but for different physical footing widths
o
B (n is the acceleration level; n=
1.0 means the gravitational acceleration). This is indeed a comparison of bearing capacity
characteristics behaviour between the 1g tests and the centrifuge tests shown in Fig. 4.28
and Fig. 4.32. That is, Figs. 4.28 and 4.32, respectively, compare the
0
/ N S B :
relationships from a set of corresponding 1g and centrifuge tests for the same (or very
similar) equivalent footing width
o
B B n = , but for different physical footing widths
o
B ;
Fig. 4.28 is for the case of 50 B cm = , while Fig. 4.32 is for the case of 21 23 B cm = : . It
may be seen from these figures that, as the
0
/ S B becomes larger, the prototype behaviour
(i.e., the 1g model tests in this case) is less satisfactorily simulated by a centrifuge test using
the same sand as the prototype but the behaviour is more affected by the particle size effect:
i.e.,
(1) the initial load-settlement curve at small footing settlements in the centrifuge test is


Fig. 4.26 Setting of footing on the
surface of prepared sand bed (Morimoto
1990; Tatsuoka et al. 1991).


Figure 4.27. Side view of the model seen
through the transparent side wall and the
lubrication layer (Morimoto 1990; Tatsuoka
et al. 1991).
45
similar to the corresponding prototype
behaviour;
(2) but the peak footing load and the settlement at
the peak footing load in the centrifuge test are
noticeably larger than those of the prototype
footing (i.e., the 1 g test in this case)..
With respect to the pressure effect, when the pressure level (i.e., the acceleration level) is
changed in centrifuge tests using the same size of footing and the same type of sand, we would
not observe any change in the
0
/ N S B : relationship as well as the values of N

and
0
/ S B at
the peak footing load if the sand has the following properties:
a) the relationship among the stress ratio
1 3
/ , the shear strain and the volumetric
strain
vol
, together with the angle of internal friction , during the shearing process of
the sand used in the model tests is independent of pressure level; and

Figure 4.28. Relationships between normalised footing
load and normalised footing pressure from two 1g tests
with
0
50 B cm = and the corresponding centrifuge test
(Morimoto 1990; Tatsuoka et al. 1991).

Figure 4.29. Sideview of the model
around the peak footing load state (see
Fig.4.28). (Tatsuoka et al. 1991).


Figure 4.30. Shear strain field around the near peak footing load state (grid spacing= 1 cm;
see Fig. 4.28) (Morimoto 1990; Tatsuoka et al. 1991).
46


Figure 4.31. Side view of the model around at
the end of loading (see Fig. 4.28) (Morimoto 1990;
Tatsuoka et al. 1991).
b) the relationship between the
logarithm of the mean pressure
' p and
vol
during the
compression process of the sand
used in the model tests is
independent of pressure level.
If the actual properties of sand were as
above and when the behaviour of a
prototype footing before the effects of
shear banding become significant is to be
evaluated, small scale 1g model tests
using the prototype are more than
sufficient to predict the prototype
behaviour. However, the actual sand does
not have such properties as described
above, and therefore, as the pressure level
increases, the initial slope of
0
/ N S B :
relation decreases,
0
/ S B at the peak
footing load increases and the value of
N

decreases.
It is also possible to observe the particle
size effect by comparing the bearing
capacity characteristics between different
tests either in 1g or in centrifuge tests
using the same footing size on different
model sand beds made of different sands
having different particle sizes but having
the same global pre-peak stress-strain
properties and peak strength and also the same post-peak stress-strain properties within the
shear band (so different stress ratio-shear deformation relationships in the post-peak regime of
shear band). This latter type of comparison is practically very difficult. Instead, some
approximated method was attempted as shown later in this paper (Tatsuoka et al. 1997b).
Fig. 4.33 summarises the N

values from the physical model
tests, which are plotted versus
the equivalent footing size,
defined as
n
B B n = , and the
corresponding results from a
series of FEM analysis
(explained below). It may be
seen that the N

values from the

1g model tests exhibits a large
scale effect, which consists of
the pressure level effect and the
particle size effect.

4.3 FEM simulation
Realistic results can be obtained
by FEM analysis only when
taking into account properly the
actual complicated deformation
and strength characteristics of
Figure. 4.32. Comparison of bearing capacity between
1g and centrifuge tests for the same sand Toyoura sand
and a very similar equivalent footing width
0
B B n = of
21 23 cm (Morimoto 1990; Tatsuoka et al. 1991).
47
geomaterial (i.e., sand in this case) as described above. The constitutive modelling of
48
Toyoura sand that is described in Tatsuoka et
al. (1993) is used in the FEM simulation
described below. In this modelling, the
observed non-linear stress-strain behaviour
together with the observed dilatancy
characteristics are formulated, while the flow
characteristics are not associated with the
peak frictional angle. For the FEM analysis
of the model bearing capacity tests described
above (Tatsuoka et al. 1991: Siddiquee et al.
1999), the simulation of strain localisation
into a shear band(s) is the most difficult part.
The following assumptions were adopted
(Siddiquee et al. 1995b):
1) each FEM element has a specific peak
strength that is independent of boundary
conditions and stress and strain histories;
2) in each FEM element, when the peak
strength is reached, strain localisation starts taking place into a single shear band having a
thickness that is specific to a given type of sand (i.e. Toyoura sand in the present case),
while the relationship between the stress ratio and the shear deformation is specific to a
given type of sand; and
3) the average stress-average strain relationship in each FEM element is the same as that of a
plane strain specimen having the same size as the FEM element.
1 10 100 1000
0
100
200
300
400
500
Particle size effect
Scale effect
Pressure level effect
1g test results
Centrifuge test results
(B
0
=3 cm)
Void ratio, e=0.66
N

=(2q/B)
max
1g simulation
Centrifuge simulation


B=n.B
0
(cm) in Log
B
e
a
r
i
n
g

c
a
p
a
c
i
t
y

f
a
c
t
o
r
,

N

Scale effects
+
= pressure level effects* + particle size effects
+ by 1g tests
for different footing sizes
and the same sand type
*by centrifuge tests under
different n values
for the same footing size
and the same sand type

Figure 4.33. Comparison of the N

values from the model tests as a function of B (Siddiquee et al.

1999):
n
B B n = , and the corresponding relation from FEM analysis (Tatsuoka et al. 1991).

Figure 4.34. Initial and deformed FEM
meshes for
0
50 B cm = in 1g (Siddiquee et al.
1999).

49
In this way, for a given
type of sand, the post-
peak stress-strain
relationship of each FEM
element becomes
dependent on the mesh
size, while the result of
FEM analysis can become
mesh size-independent.
In the same way, the post-
peak stress-strain relation
in each FEM element
becomes dependent on
the particle size, and
therefore, the effects of
particle size can be
simulated by the FEM
analysis. A more detailed
description of the FEM
analysis is given in
Siddiquee et al. (1995b;
1999, 2001), Kotake et al.
(1999) and Peng et al.
(2000).
Fig. 4.34 shows the
mesh used in the FEM analyses, and Fig. 4.35 compares the results from the following FEM
analyses using the mesh shown in Fig, 4.34 with the corresponding experimental results for
0
B = 50 cm in 1 g:
a) the value of from the plane strain compression test at 90
o
= performed at pressure
levels lower than the critical value was used (Fig. 4.10): the value is essential constant
with the respect to the changes in the pressure at this low pressure level and the value at
this low pressure level is the highest value for a given mass of sand.
b) For the above value of , the pressure-dependency of , shown in Fig. 4.10, was
considered.
c) In addition to the above, the factor of inherent anisotropy of the deformation and strength
characteristics of the test sand, as shown in Fig. 4.9, was considered.
d) In addition to the above, the post-peak strain softening was considered in such a way that
the post-peak stress-strain relationship in each element did not depend on the mesh size,
while it was the same with the relationship between the average stress and the average strain
for the whole of a specimen (20 cm-high, 16 cm-long and 8 cm-wide in the present case);
and:
e) In addition to the case c), the strain localisation into shear bands having a specific thickness
and a specific relationship between the stress level and the shear deformation was
considered.
It may be seen Fig. 4.35 that the peak footing pressure by analysis a) is consistent with the
classical solution obtained for the same strength characteristics. It may be seen however that the
peak footing load is attained at a very large (so unrealistic) footing settlement in this FEM
analysis. This is because a very large settlement of footing is necessary to mobilise the peak
strength fully along the potential failure planes when using the realistic pre-failure deformation
characteristics of sand. It may also be seen that only the solution of analysis e) is realistic. The
N

values from these analyses are plotted and compared with the experimental results in Fig.
4.13. These N

values by the solution of analysis e) are plotted against the equivalent footing
width
0
B B n = in Fig. 4.33. It may be seen from these figures that only analysis e) simulates
very well the observed scale effect, pressure-level effect and particle size effect.
Figure 4.35. Comparison among FEM simulations based on different
assumptions of sand stress-strain properties and experimental results
(
0
50 B cm = in 1g) (Tatsuoka et al. 1991).
50

Figure 4.36. Relationships between the shear stress
level
n
R (E. 4.1) and the post-peak shear deformation
of a shear band for three types of granular materials
having different particle sizes (see Fig. 4.2) (Tatsuoka
et al. 1997c; Siddiquee et al. 1999).

Small-scale 1g model tests using
different types of sands having
different particle diameters: As the last
series of physical model test, 1g and
centrifuge tests were performed using a
coarser sand (Silver Leighton Buzzard
sand) and a small-diameter gravel
(Hime gravel) (Tatsuoka et al. 1997c).
Fig. 4.36 shows the relationships
between the shear stress level
n
R and
the post-peak relationships between the
shear stress level and the shear
deformation of shear band of the three
granular materials (Toyoura sand, SLB
sand and Hime gravel), obtained by
Yoshida and Tatsuoka (1997).
Fig. 4.37 summarises the
relationships between the
experimentally obtained N

value and
the corresponding respective value of
( 90 )
o
PSC at = for the three
types of granular materials. A set of
N

values for each type of granular

material was obtained
from 1g tests using
different footing sizes (see
Fig. 4.13 for Toyoura
sand). When comparing
these experimental
relationships with the
relationships between the
N

and from the

classical bearing capacity
theories shown in Fig.
4.12, it can be seen that as
the particle size increases
(i.e., as the shear
deformation increment
that is necessary for the
stress state to change from
the peak to residual states
increases), the N

value
increases, approaching the
relationships by the classi-
cal bearing capacity theo-
ries. Fig. 4.38 summa-
rised the relationships
between the following two
quantities for the three
types of granular materi-
als:

Figure 4.37. Comparison of the relationships between the N

and
from the representative classical bearing capacity theories with the
measured relationships between N

from 1g tests using different

footing sizes and ( 90 )
o
PSC at = for three types of granular
materials (Tatsuoka et al. 1997c).
51
1) the ratio of the
respective
experimentally obtained
N

value to the
corresponding ( )
theory
N

value obtained by
substituting the
corresponding value of
( 90 )
o
PSC at =
into the isotropic
perfectly plastic solution
by Meyerhof (1951); and
2) the logarithm of the ratio
of the mean diameter
50
D to the physical
footing width
0
B .
The experimental data
plotted in Fig. 4.38 were
obtained from all the 1g and
centrifuge tests reported in
this paper. The variation in
the value of /( )
theory
N N


for the same
50 0
/ D B value
seen in the data is due
basically to the pressure
level effect. Such a general
trend as that the
/( )
theory
N N

value increases with the increase in the
50 0
/ D B value can be clearly seen. It may
also be noted that the values of /( )
theory
N N

of the data from the 1g tests on Hime gravel
noticeably exceed the unity. This fact is apparently not consistent with the results of Toyoura
sand; i.e., with Toyoura sand, this value should be always smaller than the unity, as the factors
of strength anisotropy and the progressive failure explained in the above should control the ratio
/( )
theory
N N

. This result of Hime gravel can be explained by the fact that in these model tests,
the potential shear band would have been very thick, compared with the footing width.
Therefore, a failure mechanism consisting of distinct shear bands (i.e., an active wedge and so
on) did not develop below the footing. In such a case, the analysis assuming that the material is
continuous is not relevant even when strain localisation with shear banding is taken into
account.
The effects of strain localisation could be taken into account in the limit equilibrium analysis
although it reflects only partially the actual strain localisation phenomenon. Koseki et al.
(1997) and Tatsuoka et al. (1998) showed such a method in modifying the dynamic earth
pressure theory by Mononobe and Okabe, while Leshchinsky (2001) in the stability analysis of
reinforced soil structures.

Summary: It has been shown how progressive the failure of ground could be in the sense that
the local peak strength is not mobilised simultaneously along the full potential failure planes. It
was also shown that the scale effect observed in the bearing capacity of a footing on an unbound
granular material is due not only to the effect of confining pressure on the peak friction angle
and deformation characteristics (i.e., the pressure level effect), but also the effect of the particle
size relative to the footing size (i.e., the particle size effect). It was also shown that the failure
of a mass of dense granular material can be numerically simulated reasonably by the FEM (and

Figure 4.38. Relationship between /( )
theory
N N

and log. of
50 0
/ D B value for three types of granular materials having different
particle sizes; these ( )
theory
N

values were obtained by substituting

the respective corresponding value of ( 90 )
o
PSC at = into the
isotropic perfectly plastic solution by Meyerhof (1951) (Tatsuoka
et al. 1997c).
52
others) only when the pressure-level dependency of and the deformation characteristics, the
inherent anisotropy in the strength and deformation characteristics, the deformation
characteristics of shear band (as a function of particle size), among other parameters, are taken
into account. All the results of the physical model tests and the FEM analysis of the
experimental data indicate that geotechnical engineers should be very careful when using
classical bearing capacity theories in engineering practice.
In ordinary engineering practice, a low value of , such 30 - 35 degrees, is used, which is
rather equivalent to the residual angle of friction. When the design N

is obtained by
substituting this value into the classical bearing capacity theories, a very conservative result
would be obtained, as seen from Fig. 4.13. The use of such a design value of N

as above
would be too conservative (i.e. not economical), in particular when a foundation structure is
constructed on a large-particle granular material and/or on a granular material that has a peak
strength that is much large than the residual strength (such as very dense well-graded gravels).
That is, the design could not be consistent among different types of granular materials that are
compacted to different relative densities.


5 TIME-DEPENDENT DEFORMATION PROPERTIES OF GEOMATERIALS

5.1 Introduction

The last topic of this paper is also one of the oldest and most classical topics of geotechnical
engineering, but it seems that this topic is still a new and challenging topic one. I discussed on
this issue in my three recent keynote lectures in Hamburg in 1997 (Tatsuoka et al. 1999a),
Napoli in 1998 (Tatsuoka et al. 2000) and Torino in 1999 (Tatsuoka et al. 2001a). It was shown
in these three lecture notes that this issue is still full of many topics that are important in both
geotechnical engineering practice and research but only poorly understood.

0 250 500 750 1000 1250 1500 1750 2000
-70
-60
-50
-40
-30
-20
-10
0
10
3P
The 1995 Hyogo-ken Nambu
earthquake


S
e
t
t
l
e
m
e
n
t
,

S

(
m
m
)
Elasped time (days)
0
2
4
6
8
10
b)
End of tower construction
26th Jan. 1990

A
p
p
l
i
e
d

p
r
e
s
s
u
r
e
,

(
p
)
a
v
e

(
k
g
f
/
c
m
2
)




Figure 5.1. Time histories of the applied average footing pressure and the settlement of Pier 3 of
Akashi Kaikyo Oh-hashi bridge (Tatsuoka et al. 2001a).
52
5.2 Engineering needs

It is often required to predict the long-term residual deformation of ground and displacements of
a completed structure subjected to sustained static loads and dynamic load, such as traffic load.
Fig. 5.1 shows the time history of the settlement of Pier 3 of the Akashi Kaikyo Oh-hashi
Bridge, which was open to service in 1998 (n.b., similar time histories of Pier 2 are reported in
Tatsuoka et al. (2001a). Fig. 2.8 shows the relationship between the footing settlement and the
footing average pressure constructed using the data presented in Fig. 5.1. The prediction of the
settlement during construction and the post-construction residual settlement of this and other
footings was of the important geotechnical engineering issues with this bridge. It may be seen
from Fig. 5.1 that the construction speed was not constant. Perhaps for this reason, the tangent
modulus of the relationship between the footing pressure and the settlement shown in Fig. 2.8 is
not smooth. For example, there were three relatively long periods where the footing load was
kept nearly constant (i.e., the creep stages) at an intermediate construction stage where the
footing pressure was about a half of the final value and before and after the construction of the
tower. During these periods, the footing settlement increased despite a nearly constant footing
pressure. When the construction was restarted at a normal construction speed following the
respective creep stage, the tangent modulus of the footing pressure and settlement relationship
was relatively high. And generally, the tangent modulus was larger when the construction speed
was larger. This behaviour was, at least partly, due to the viscous deformation property of the
ground.
It may also be seen from Fig. 5.1 that the time period before the opening of the bridge to
service is quite long. In contrast, the time period used for the loading stage until reaching the
creep loading stage in laboratory creep tests that are performed to predict such a residual
settlement of foundation as described above is substantially shorter. Although this difference
should be properly considered when predicting the field behaviour based on results from such
laboratory creep tests (Tatsuoka et al. 2001a), it is usually not the case.
Summarising the above, it is often required to predict:
1) load-deformation behaviour at different rates of construction or loading;
2) creep deformation and stress relaxation during a period following continuous construction or
loading at different rates (as the case described above);
3) load-deformation behaviour after construction or loading is restarted at a certain rate
following a long period of intermission; and
4) creep (residual) deformation and stress relaxation at unloaded conditions (e.g. Uchimura et al.
1996).
For such a prediction as above, the characterisation of the time-dependent (viscous) stress-
strain behaviour of geomaterial is essential, in particular the following aspects:
a) effects of constant strain rate on the stress-strain behaviour, including those on the peak
strength;
b) changes in the stress-strain behaviour when the strain rate is suddenly or gradually increased
or decreased from a certain value to another,
c) creep deformation and stress relaxation;
d) stress-strain behaviour when loading is resumed at a constant strain rate after a stage of
creep or strain relaxation; and
e) time-dependent stress-strain behaviour in the course of unloading and reloading,
or more generally the stress-strain-time behaviour for a arbitrarily general stress history,
including cyclic loading. A number of different constitutive models have been proposed to
simulate the behaviour described above.
Described below is one type of constitutive modelling, which has been developed recently by
the author and his colleagues. This modelling procedure is still on a long way towards the final
goal: i.e.. the development of a three-dimensional model which can simulate and predict the
time-dependent behaviour (ageing effects and loading rate effects) of geomaterials in general
subjected to arbitrary loading histories.

5.3 Experimental issue
53

It has been considered that effects of bedding errors at the top and bottom ends of specimen be
negligible in triaxial creep tests, considering that they could be significant only when the
effective axial stress increases (Tatsuoka et al. 1999b, 2000; Hayano et al. 2001). This is not the
case at least with sedimentary soft rock, however, as shown in Fig. 5.2. In this figure,
significant effects of bedding error can be noted as differences in the axial strain between;
a) the external measurement from the axial displacement of the specimen cap detected by
means of a proximity transducer (or gap sensor), denoted as AGS (cap), and
b) the local measurement by means of a pair of LDTs, denoted as CLDT, and two pairs of
proximity transducers, denoted as BGS (local).
It may be noted that the effects of bedding errors increase not only during monotonic loading
stages but also during creep stages. This behaviour could be attributed to extra time-dependent
deformations of a thin disturbed zone that should have been formed at the specimen ends during
specimen preparation. Large differences in the axial strain between the external gauge (denoted
as @EXT) and the proximity transducer (denoted as AGS (cap)) is due to the deformation of the
triaxial apparatus (i.e., the system compliance). This result indicates that the use of local axial
strain gauge is imperative in such triaxial creep tests.

5.4 Constitutive modelling-1 (Isotach type modelling)

One of the relevant frameworks for constitutive modelling for the present purpose is the general
three component model (Di Benedetto et al. 2001a). A simplified version that is used herein is
shown in Fig. 5.3. The heart of the model is as follows:
1) Strains are first defined in terms of increments, and each strain rate & is decomposed into
elastic and inelastic (or irreversible) components,
e
& and
ir
& . The irreversible strain
increment
ir
& cannot be decomposed linearly into plastic and viscous components (Tatsuoka
et al. 2000).
2) Stress is decomposed into the inviscid and viscous components,
f
and
v
.
kcs
External di al gauge
Load cell
Cap
Gap sensors
Target
Pedest al
Specimen
Loading pi ston
High-pressure cell
Gap sensor(proximeter)
hi nge
0.0 0.1 0.2 0.3 0.4 0.5 0.6
0
1
2
3
4
5
6
C
B
A
@
(a)
Creep
Creep
CLDT
BGS(local)
AGS(cap)
@Ext.


D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
M
P
a
)
Axial strain,
v
(%)

Figure 5.2. Drained triaxial creep test on sedimentary soft mudstone; each creep period is three days
(Tatsuoka et al. 1999b,2000; Hayano et al. 2001); the details of the testing method is described in
Hayano et al. (1977).
54
According to the conventional isotach
model, the stress is a unique function
of instantaneous total strain and its rate
t

& . Tatsuoka et al. (1999d, 2000)

showed the limitations of this type of
model: for example, this model cannot
simulate the stress relaxation process.
This is because when the condition of
stress relaxation, 0 & , is given to a
geomaterial element, the stress
becomes suddenly and discontinuously
the value corresponding to 0 & , which
is not realistic.
A great deal of experimental results
indicate that, with geomaterials, the stress
is a unique function of instantaneous
irreversible strain
ir
and its rate
ir
& (Tatsuoka et al. 2000). Tracking this line and following
the framework of the three-component model described in Fig. 5.3, Tatsuoka et al. (1999d, 2000,
2001b) proposed the so-called New Isotach model (Fig. 5.4). According to the New Isotach
model, we obtain different stress-strain relationships for monotonic loading at different constant
strain rates. The creep deformation is a process where
ir
& continuously decreases towards zero
at a constant . Similarly, the stress relaxation is the process where
ir
& continuously decreases
towards zero at a constant
e ir
+ with negative
e
d and positive
ir
d . The elastic strain
increment is obtained by a relevant hypo-elastic model, which is described in Chapter 2, for
which the tangent stiffness is a function of instantaneous stress. Fig. 5.5 is a typical example of
this strain decomposition for a sedimentary soft rock (Tatsuoka et al. 2000; Hayano et al.
20001). Note that the irreversible strain can be obtained only after such strain decomposition.
The other important features of this model, as summarised in Fig. 5.4 are:
1)
f
is a function of the instantaneous value of
ir
;
Non-linear
inviscid component; @ @ @ @ @ @
@ @ @ @ @ @ @ @ @ @

Hypo-elastic
component:


&

@ @ @ @ @ @ @ @ @ @ @

@
Non-linear viscous component;
@
@ @ @ @ @ @ @ @ @ @ @



&
e

&
vp
f

Hypo-elastic
model
ir
&
Figure 5.3. Framework used in the development of
models described in this paper (Di Benedetto et al.
2001a & b; Tatsuoka et al. 2001b).

0
10 & &



0
& &



0
/10 & &




Lower bound at
0
i r
&
:
( )
f ir


Creep
0

@New Isotach Model
i the simplest three-component model j
( ) ( )
,
f ir v ir ir
+ &

( ) ( ) , ( )
v ir ir f ir ir
v
g & &

The stress is always a unique
function of instantaneous values
of and .
Different stress-strain relations
develop by loading at different
strain rate; and corresponding to
the above, creep and stress
relaxation take place.
( ) [1 exp{1 ( 1) }] ( 0)
ir
ir m
v ir
r
g

+
&
&
&
f

Non-linear
inviscid component;
@ @ @ @ @ @
@ @ @ @ @ @ @ @ @ @

Hypo-elastic
component:


&


@ @ @ @ @ @ @ @ @ @ @

@
Non-linear viscous component;
@
@ @ @ @ @ @ @ @ @ @ @



&
e

&
vp
ir

ir
&

Figure 5.4. New Isotach model (Tatsuoka et al. 1999d, 2000, 2001b).
55
2)
v
is always
proportional to
f
, which
is one of the specific
features of this model; and
3) ( )
ir
v
g & is the viscous
function, which is a highly
non-linear function of
instantaneous value of
ir
&
(this function is explained
in Fig. 5.13).
Note also that the elastic
strain, which is obtained by
integrating elastic strain
increments for a closed
loop of stress path, could
be irreversible (Puzrin and
Tatsuoka 1998). On the
other hand, the relationship
between and
ir
, which
is obtained by integrating
ir
& , may form a closed
loop. In this sense,
reversibility and
irrversibility of strain is
0.00 0.05 0.10 0.15 0.20 0.25
0
20
40
60
(a)
C: Drained creep
for one day
C
C
C
C

0
/100

0
/100

0
/100
Irreversible
Elastic
Total
Axial strain,
v
(%)
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
k
g
f
/
c
m
2
)
Sedimentary soft rock (mudstone; Kazusa group)


Figure 5.5. Separation of total axial strains into elastic and
irreversible parts for the stress-strain relation from a drained TC test
on sedimentary soft rock (
0
& = 0.01 %/min) (Tatsuoka et al. 2000;
Hayano et al. 2001).
0.00 0.05 0.10 0.15 0.20 0.25
0
10
20
30
40
50
.
.
.
.
.
.
.
.
.
()
0
/100
()
0
/100
()
0
/100
()
0
()
0
()
0
()
0
()
0
()
0
(e)
C
Measured
Calculated
C: Drained creep
for one day
C
C
C
C
()
0
Total axial strain,
v
(%)
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
k
g
f
/
c
m
2
)
0.00 0.02 0.04 0.06 0.08 0.10
0
10
20
30
40
50
(f)
n=1
Calculated q-(
ir
)
static
Calculated q-
i r
C: Drained creep
for one day
C
C
C
C
C

0
/100

0
/100
Calculated irreversible axial strain,
v
ir
(%)
D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q

(
k
g
f
/
c
m
2
)
Simulation by
the New Isotach model
Viscous effects
Figure 5.6. a) Simulation of the test result presented in Fig. 5.5 by the New Isotach model; and b)
decomposition of the stress component in the simulation (Tatsuoka et al. 2000; Hayano et al. 2001).

a)
b)
56
defined only with
respect to strain
increments.
Fig. 5.6 shows the
simulation of the test
result presented in Fig.
5.5 by the New Isotach
model (Tatsuoka et al.
2000; Hayano et al.
2001). According to
this model, the
relationship between
f
and
ir
is obtained
by monotonic loading
at an infinitively slow
strain rate, which will
therefore be called the
reference curve. In this
simulation, the
reference curve was
obtained so that the
entire test result can be
simulated by using the
same viscous function
( )
ir
v
g & . It may be
seen that this model
can simulate well the
entire stress-strain
behaviour, including
the creep and
unloading behaviour.
In particular, the
decrease in the axial
strain with time at the
creep stage in the
course of global
unloading (i.e., the
creep recovery
phenomenon) is well
simulated.
The model is also
capable to simulate
the effects of step
increase and decrease
in the constant strain
rate as well as the
behaviour during and immediately two creep stages (Fig. 5.7).
Fig. 5.8 shows the simulation by the New Isotach Model of the relationship between the
average footing pressure and the footing settlement of Pier 3P, Akashi Kaikyo Bridge, shown in
Fig. 2.8 (also refer to Fig. 5.1). In this considerably simplified simulation, the supporting
ground was treated as one soil element, and the parameters of the model were determined to
obtain the best fit. Despite the above, it is seen that the overall effects of loading rate are well
captured: i.e., the fluctuation of the relation is not due to simple measurement errors, but it is
0.0 0.3 0.6 0.9 1.2 1.5
0
1
2
3
4
5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

0
/100

0
/100

0
/100

0
/100

0
/10

0
/10

0
/10

0
/10

0
Measured
(a)
Silt-sandstone
'
c
=1.29MPa
0
=0.01%/min
C: Drained creep
Simulated

0
C
C
Total axial strain,
v
t
(%)


D
e
v
i
a
t
o
r

s
t
r
e
s
s
,

q
=

'
v
-

'
h

(
M
P
a
)
Sedimentary soft rock (mudstone; Kazusa group)
Figure 5.7. Simulation by the New Isotach model of the stress-strain
behaviour in CD triaxial compression test on sedimentary soft rock
(Tatsuoka et al. 2000; Hayano et al. 2001).

0 10 20 30 40 50
0
2
4
6
8
10
b-1)
Load control
3P

Field data
Simulation
A
v
e
r
a
g
e

c
o
n
t
a
c
t

p
r
e
s
s
u
r
e
,

(
p
)
a
v
e

(
k
g
f
/
c
m
2
)
Irreversible settlement, S
ir
(mm)
Simulation of the foundation behaviour by the New Isotach model:
(the ground is treated as a single element)

Figure 5.8. Approximated simulation by the New Isotach Model of the
relationship between the average footing pressure and the footing
settlement of Pier 3P, Akashi Kaikyo Bridge; the ground is treated as
one element (Tatsuoka et al. 2001a).


57
due to changes of loading rate. More discussion on this simulation is given in Tatsuoka et al.
(2001a). Of course, the final goal of this study is to simulate the observed behaviour by relevant
FEM analysis.
Results from PSC tests - 1
Very small difference
among the behaviour
at constant strain
rates differing by a
factor up to 500.
The stress value changes
when the strain rate is
stepwise changed by a
factor of 100, and the
stress change decays with
strain.
Apparent
contradiction I
Test name e
0.05
d
v
/ dt (%/min)
HOS01 0.6146 variable
H302C 0.6153 10
0
H303C 0.6162
0
/10
H304C 0.6149
0
/10
H305C 0.6160 10
0
H306C 0.6155
0
/10
H307C 0.6164
0
/50
0 1 2 3 4 5 6 7 8 9
3.0
3.5
4.0
4.5
5.0
5.5
6.0
.
.
Fig. 2
(graph 11)
.
.
.
.
.
.
.
.
.
.
.
Saturated Hostun sand
(Batch A)

0
= 0.0125 %/min
Test HOS01
c1-e1
0
/10
e1-f1 10
0

f1-h1
0
/10
h1-i1 10
0
i1-k1
0
/10
k1-l1 10
0
d1, g1, j1 5 times small
cyclic loading
H307C
0
/50
H304C
0
/10
H305C 10
0
H306C
0
/10
H303C
0
/10
H302C 10
0
l1
k1
j1
i1
g1
h1
f1
e1
d1
c1
Shear strain, =
v
-
h
(%)
S
t
r
e
s
s

r
a
t
i
o
,

R

=

' v
/

'
h
Figure 5.10a. Plane strain compression tests on Hostun sand by monotonic loading at different
constant strain rates and loading with step changes in the constant strain rates (Matsushita et al.
1999; Tatsuoka et al. 1999b; Di Benedetto et al. 2001a).

Plane Strain Compression Tests on Sand
'
v

PSC

' ' 3.0
v h
R

(not to scale)
Initial state
0
'
h

(kPa)
29 392
2
0
c
m
8
c
m
1
6
c
m

Figure 5.9. Plane strain compression test conditions to study the viscous properties of sands (Di
Benedetto et al. 2001b and Tatsuoka et al. 2001b); the details of the plane strain compression
testing method are described in Yasin et al. (1999a & b) and Yasin and Tatsuoka (2000).


58
5.4 Constitutive modelling-2 (TESRA model)

TESRA stands for temporary effects of strain rate and strain acceleration. The reason why
this mew model was to be developed is explained below. Fig. 5.9 shows the PSC (plane strain
compression test) procedure by which the viscous properties of sand were investigated
(Matsushita et al. 1999; Tatsuoka et al. 2001a; Di Benedetto et al. 2001a). Fig. 5.10a shows
results from a series of PSC tests on six saturated specimens of air-pluviated Hostun sand (from
France), in which monotonic loading at different constant strain rates were applied after
anisotropic consolidation. In the other test, the axial strain was changed stepwise several times
during otherwise monotonic loading at a constant strain rate. The following trends of behaviour
can be seen:
1) It may be seen from Fig. 5.10a that the overall stress-strain curves from the monotonic
loading tests at different constant strain rates are nearly the same (n.b., rigorously, the initial
part is different for different strain rates).
2) Despite the above, the stress-strain curve from the other test exhibits very stiff behaviour
immediately after the strain rate increases stepwise. Then, after having exhibited clear
yielding, the stress-strain curve tends to rejoin the stress-strain curve that would have been
obtained if the loading had continued at the strain rate before a step change.
3) The behaviour that is opposite to the above takes place after the strain rate decreases
stepwise. These two behaviours apparently contract each other.
Note that such noticeable viscous effects were observed also in similar PSC tests on air-dried
specimens and in triaxial compression tests on saturated and air-dried sand (Matsushita et al.
1999; Di Benedetto et al. 2001a).
Fig. 5.10b shows results from another PSC test on Hostun sand, in which creep and
relaxation stages are included during otherwise monotonic loading, and the test result is
compared with those from the PSC tests performed at different constant strain rates, presented
in Fig. 5.10a (Matsushita et al. 1999; Tatsuoka et al. 20001a; Di Benedetto et al. 2001a). It may
be seen Fig. 5.10b that the specimen exhibits noticeable creep deformation and stress relaxation,
Noticeable creep
deformation and stress
relaxation take place.
0 1 2 3 4 5 6 7 8 9
3.0
3.5
4.0
4.5
5.0
5.5
6.0
.

0
= 0.0125 %/min.
.
.
.
.
.
.
.
.
.
.
.
.
Fig. 3
(graph 8)
Test HOSB1
c2-d2 10
0
j2 -k2 creep
d2-e2 creep k2-m2 10
0
e2-g2 10
0
m2-n2 relaxation
g2-h2 creep n2-o2 10
0
h2-i2 accidental pressure drop, o2-p2 creep
followed by relaxation stage p2-q2
0
/10
i2 -j2
0
/10
H307C

0
/50
q2
p2
H304C
0
/10
H305C 10
0
H306C
0
/10
H303C
0
/10
H302C 10
0
o2
n2
m2
l2
k2 j2
i2
g2
h2
f2
e2
d2
c2
Shear strain, =
v
-
h
(%)
S
t
r
e
s
s

r
a
t
i
o
,

R

=

'
v
/

' h
Very small difference
among the behaviour
at constant strain
rates differing by a
factor up to 500.
Apparent
contradiction I
Results from PSC tests - 2


Figure 5.10b. Plane strain compression tests on Hostun sand by monotonic loading at different
constant strain rates and loading with creep and relaxation stages (Matsushita et al. 1999; Tatsuoka et
al. 1999b; Di Benedetto et al. 2001a).

59
which is again apparently inconsistent with the behaviour during loading at different constant
strain rates.
To simulate the above peculiar behaviour, a new model was developed as follows. Fig. 5.11
shows the procedure at the preparation stage for the above. That is, the viscous stress
v
of the
TESRA (temporary effect of
strain rate and acceleration) model @ P

( )
f ir v
+



1 1
1
( , ) ( )
( )
( )
{ ( )}
( )
( )
ir ir
ir
ir i r
ir
ir
v f ir
v
v f ir
v
ir f ir
ir f v
v ir ir ir
g
d d g
g
g d

1 1
] ]
_ _ _
+


, , ,

&
&
& &&
&
& &



( 0)
v



f

Non-linear
inviscid component; @ @ @ @ @ @
@ @ @ @ @ @ @ @ @ @
Hypo-elastic
component:


&

@ @ @ @ @ @ @ @ @ @ @
@
Non-linear viscous component;
@
@ @ @ @ @ @ @ @ @ @ @



&
e

&
vp
the New Isotach Model:


Figure 5.11. Reforming of the New Isotach model into an integral form as a preparation for the
development of the TESRA model; is the value of
ir
at a certain moment before the current
state (Di Benedetto et al. 2001a).
1
1
( )
( )
( )
{ ( )}
( )
ir
ir
ir
ir
f
ir
ir ir
f v
i r
v f ir
v
v ir ir
g d
g
d
g
d

_

1
]
_
+

, ,

,

&
&
&
& &
&
&

( 0)
v

b
0
10
ir
& &

0
i r
v
i r
ir
d
d

,
&
&


v
d

0
ir
& &

0
ir
v
ir
ir
d
d

, &
Parallel a
v

{ ( ) ( )}
v f ir ir
v
d d g &

f

f f
d +
0

1
ir
&

ir
a
&

e ir
d d d +
Effect of irreversible
strain rate
Effect of irreversible
strain acceleration
The New Isotach Model:

Figure 5.12. Illustration of the effects of irreversible strain rate and acceleration with the New
Isotach Model (Di Benedetto et al. 2001a).
60
New Isotach Model is expressed in the integral form. That is, the increment of
v
consists of
its derivative with respect to
ir
times irreversible strain increment
ir
d , showing the
effect of irreversible strain rate, and its derivative with respect to
ir
& times irreversible strain
rate increment
ir
d& , showing the effect of irreversible strain acceleration (see Fig. 5.12). Note
that the viscous function ( )
ir
v
g & should be defined so that its derivative with respect to
ir
& is
always smooth (Fig. 5.13).
In the TESRA modelling, the effects of irreversible strain rate and acceleration that take place
at a certain loading stage where
ir
are considered to be temporary and decay with the
increase in the strain difference
ir
until the current state where
ir ir
(Fig. 5.14). This
property is expressed by the decay function ( )
ir
decay
g , which is explained in Fig. 5.15. The
introduction of the decay function is the heart of the TESRA model. The decay function
represents such a property of sand as that the sand tends to gradually forget what happens in the
past due to gradual changes in the structure (i.e., rearrangement of relative locations of particles)
with the increase in the irreversible strain (not with time). Fig. 5.16 illustrates how the model
behaves.
Fig. 5.17 shows a typical simulation by the TESRA model of the PSC test result of Hostun
sand in which the strain rate was changed stepwise several times. Note that the specimen was
air-dried, so the observed viscous effect was not due to the partial and delayed drainage of pore
water. It may be seen that the model simulates very well the test result. Note that the details of
the behaviour after the strain rate is changed stepwise are simulated surprisingly well.
Fig. 5.18 shows the simulation of the result of another PSC test, in which the strain rate was
increased and decreased at a constant rate respectively for some strain range and this sequence
was repeated two times. One creep stage was included between the above mentioned two
( ) [1 exp{1 ( 1) }] ( 0)
ir
ir m
v ir
r
g

+
&
&
&
Viscosity function:
This function can be determined experimentally@I
1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0.01
1
1.0
.
.
1.00
0.99
1.10
Fig. 15b
b
Hostun sand
Toyoura sand


Irreversible strain rate,
ir
(%/sec)
1
+

g
v
(

ir
)
0.000 0.002 0.004 0.006 0.008 0.010
0.00
0.02
0.04
0.06
0.08
0.10
Fig. 15a .
.
Hostun sand
Toyoura sand


Irreversible strain rate,
ir
(%/sec)
g
v
(

i
r )
The
conditions
to be
satisfied:
1)
0.0 ( )
ir
v
g &
for any value of
ir
&
between

and

;
2)
( 0) 0
ir
v
g &
; and
( )
( 0)
ir
ir v
ir
g

]
&
&
&
=
ir
r
m

&
(a finite positive value).

Figure 5.13. Structure of the viscous function (Di Benedetto et al. 2001a).
61
sequences. It
may be seen that
the model is able
to simulate very
well the whole
stress-strain
behaviour,
including the
behaviour during
the strain rate
was changed
gradually, the
creep behaviour
and the post-
creep behaviour
during loading at
a constant strain
rate. It is one of
the characteristic
features of the
TESRA model
that the stress-strain state can be located below the reference curve when the negative effects of
negative strain acceleration on the
v
value in the recent strain history become dominant.
Note that for the simulation shown above, the model parameters were determined so that the
test results were best fit. However, the same parameters are used for all the tests performed
under otherwise the same test conditions (i.e., the PSC tests referred in Figs. 5.17 and 5.18). So,
it can be concluded that the TESRA model captures the basic characteristic feature of the
viscous property of sand.
Fig. 5.19a shows the results from two special PSC tests on air-pluviated Toyoura sand
(Tatsuoka et al. 2001b). The two specimens were loaded first at constant axial strain rates that
TESRA (temporary effect of
strain rate and acceleration) model 2 @

( )
f ir v
+

v

1 1
( , ) ( )
{ } ( ) ( )
ir ir
ir
ir ir
f i ir
deca
r
v y
v
g d g d





1 1
]

]


&

( 0; 0)
v
or


Decay function:
( )
1
( )
ir
ir
decay
g r





f

Non-linear
inviscid component;
@ @ @ @ @ @
@ @ @ @ @ @ @ @ @ @

Hypo-elastic
component:

&


@ @ @ @ @ @ @ @ @ @ @

@
Non-linear viscous component;
@
@ @ @ @ @ @ @ @ @ @ @



&
e

&
vp
the same as the New
Isotach model

Figure 5.14. Introduction of the decay function for the TESRA model; r
1
is a constant lower than
unity (Di Benedetto et al. 2001a).
Sand gradually forgets
the viscous effects
that took place in the past
at with subsequent
irreversible straining.
v
d

Current state
Event of
( )
v
d

1
]

v

Strain difference

i r


f

1
ir

i r


ir

0
( )
ir
ir
decay
g r





ir

1.0

r1

( )
ir
decay
g
for
( )
( )
v
d

i r


i r

1.0 0
( )
i r
decay
g
( )
1
( )
ir
ir
decay
g r



Decay function:

Fig. 5.15 Explanation of the decay function (Di Benedetto et al. 2001a).
62
were different by a factor of 100 as
v
& = 0.25 and 0.0025 %/min. The stress-strain relationships
of the two tests gradually collapse into a single relationship, as simulated by the TESRA model.
In both the tests, creep tests were then performed at two stages, each stage lasting for 24 hours.
The following important trends of behaviour may be seen from Fig. 5.19a:
1) Despite nearly the same stress and strain states at the start of creep in the two tests, the time
history of creep strain is significantly different between the two tests. This fact should be
explained by the fact that despite the same stress and nearly the same strain at the start of
each creep stage in the two tests, the initial creep axial strain rate at the start of creep is very
different (by a factor of about 100).
2) Immediately after loading is restarted at constant but different
v
& values following each
creep stage, the difference in the stress-strain relations between the two tests becomes large
again, which is to a larger extent than it is immediately after the start of loading. As loading
continues at a constant
v
& , after having exhibited clear yielding, the difference becomes
gradually smaller again.
These facts above indicate that, even when limiting to the monotonic loading case, the stress-
strain state is not a unique function of instantaneous irreversible strain rate, but it is also
significantly affected by the recent strain history. The new isotach model is, and perhaps most
of the existing elasto-viscoplastic models are, not able to properly simulate these behaviours.
The TESRA model is able to simulate this behaviour, as shown above and also below.
Fig. 5.19b compares the measured and simulated time histories of
v
and
v
R :
relationships. It may be seen that the TESRA model simulates the measured behaviour well, in
particular the following aspects:
1) the creep strain is larger in the test in which
v
& at the start of creep is larger, despite that the
stress-strain state at the start of each creep stage is nearly the same in the two tests; and
2) the stress range in which the stiffness is very high that appears immediately after loading was
restarted following each creep stage is larger when loading is restarted at a higher
v
& , while
Stress is a specific function of
instantaneous and and
strain history.
The stress value could be the
same for the different
instantaneous and , while
creep deformation and stress
relaxation can take place.
The viscous stress could be
either positive, zero or negative
depending on the strain history.

0
10
ir
& &
Reference relation:

( )
f ir


0
ir
& &

0
/10
ir
& &
@

Creep

0

Non-linear
inviscid component;
@ @ @ @ @ @
@ @ @ @ @ @ @ @ @ @

Hypo-elastic
component:

&


@ @ @ @ @ @ @ @ @ @ @

@
Non-linear viscous component;
@
@ @ @ @ @ @ @ @ @ @ @



&
e

&
vp
TESRA (temporary effect of
strain rate and acceleration)
model @ 3
ir
&
ir

ir

ir
&


Figure 5.16. Behaviour of TESRA model (Di Benedetto et al. 2001a).
63
the stress-strain curves in the two tests become gradually similar as loading continue at a
constant
v
& after having exhibited clear yielding.
It is likely that only the TESRA model, and other models having the similar basic structure
including the viscous evanescent model (Di Benedetto et al. 2001b), can simulate these aspects.
In engineering practice, three or five-component models have often been used to predict the
residual settlement of a footing. A large amount of laboratory creep tests were performed to
determine the model parameters. However, this method has the following serious drawbacks:
1) The model is not adequate because of the linear property of the three components.
2) In the analysis, it is often assumed that the creep starts after a sudden instantaneous loading.
Furthermore, it is often assumed that all the ground deformation before the start of creep
phase or the instant settlement of footing is elastic. Corresponding to the above, specified
creep load is applied to the specimen in a rather sudden manner in usual laboratory creep
tests.
It is readily seen that the residual settlement of a footing should be predicted by a model that
can take into account the effects of recent loading history, such as the New Isotach and TESRA
models.
Another important issue is the simulation of the time-dependent behaviour during unloading
process. An example of a test result and its simulation dealing with this issue is shown in Fig.
5.20. An air-dried specimen of Toyoura sand was used in this PSC test for the easiness of
testing lasting for such a long period, while based on the fact that air-dried and saturated
specimens of Toyoura sand exhibited essentially the same stress-strain behaviour in drained TC
tests (Tatsuoka et al., 1986b). The specimen was isotropically consolidated to ' '
v h
= 392
kPa prior to the start of drained PSC loading. The primary PSC loading was made at an axial
strain rate of 0.125 %/min. Two full unload/reload cycles were applied between R= 1.0 and 4.0
or 5.0. The shear stress was decreased at a constant stress rate, q& = - 98 kPa/hour (i.e., R
&
= -
0.25/hour= - 0.0042/min). Reloading was made by strain control as the primary loading. By
such stress unloading, it becomes possible to closely observe the development of positive
irreversible axial strain (and shear strain) for some shear stress range immediately after the start
PSC test on sand 3
0 1 2 3 4 5 6 7
3.0
3.5
4.0
4.5
5.0
5.5
6.0
.
Fig. 12b
(grpah3)
Test Hsd03
Experiment
Simulation
Reference curve
(in terms of total strain)
= 0.25; m=0.04;

r
ir
=10
-6
(%/sec); and
r
1
= 0.1 (for strain
difference in %)


S
t
r
e
s
s

r
a
t
i
o
,

R
=

v
'
/

h
'
Shear strain, (%)
3.5 4.0 4.5 5.0 5.5
4.6
4.7
4.8
4.9
5.0
5.1
5.2
5.3
5.4
5.5
5.6
Fig. 12d
(graph14)
Test Hsd03
Experiment
Simulation
Reference curve
(in terms of total strain)

S
t
r
e
s
s

r
a
t
i
o
,

R
=

v
'/

h
'
Shear strain, (%)


Figure 5.17. Simulation of the behaviour of air-dried Hostun sand in plane strain compression test
by the TESRA model (Di Benedetto et al. 2001a).
64
of decreasing the shear stress. In Fig. 5.20, this behaviour can be seen during the stress
unloading process starting from point U. The neutral condition is then reached, where
ir
v
&
becomes zero and a switching from positive to negative
ir
v
& values takes place. Such behaviour
as described above is often observed in load (or pressure)-controlled plate loading and
pressuremeter tests. It is usually very difficult to evaluate the elastic property by analysing load
(pressure)-displacement curves observed immediately after the start of decreasing load or
pressure (Tatsuoka et al. 2001a). Creep tests, each lasting for four hours, were performed at
every increment of q= 196 or 392 kPa (i.e., every increment of R of 0.5 or 1.0) in the course of
primary loading, unloading and reloading.
The following trends of behaviour can be seen from Fig. 5.20:
1) For some stress range immediately below point U, where the shear stress was decreasing, the
total axial and shear strains (and the irreversible axial and shear strains) are still increasing.
2) As the shear stress further decreases, the neutral state, where
ir
v
& = 0, is reached; then,
unloading with negative
ir
v
& values starts. However, the exact location of the neutral state is
not obvious.
3) At the creep stages where the shear stress is lower than that at the neutral state, the sign of
creep axial strain (and shear strain) is negative, while the absolute value of negative creep
strains becomes larger at a lower shear stress (i.e., the phenomenon called creep recovery).
Tatsuoka et al. (2001a) discussed on this phenomenon based on data from a number of field
loading tests and laboratory stress-strain tests on soft clay, sand and gravel and sedimentary
soft rock, while showing the importance of this behaviour in many geotechnical engineering
issues.
4) The sign of axial strain increments (and shear strain increments) at the creep stages during
reloading becomes positive again as it is during primary loading, and the amount of creep
strain increases as the shear stress increases.
0.0 0.5 1.0 1.5 2.0 2.5 3.0
3.0
3.5
4.0
4.5
5.0
5.5
6.0
.
Test Combi1 (Toyoura sand)
=0.25; m=0.05;

r
ir
= 10
-6
(%/sec); and
r
1
= 0.1 (for strain difference in %)
Fig. 16 b-1
(graph7)
Experiment
Reference curve
(in terms of total strain)
Simulation


Vertical (axial) strain,
v
(%)
S
t
r
e
s
s

r
a
t
i
o
,

R
=

v
'
/

h
'
PSC test on sand 4
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
4.7
4.8
4.9
5.0
5.1
5.2
5.3
Fig. 16 b-3
graph14
Experiment
Reference curve
Simulation


Vertical (axial) strain,
v
(%)
S
t
r
e
s
s

r
a
t
i
o
,

R
=

v
'/

h
'
0 10000 20000 30000 40000 50000 60000
1.25
1.30
1.35
1.40
1.45
1.50
1.55
1.60
1.65
Test Combi1
Start of creep stage
Fig. 16e
(graph 11)
Simulation
Experiment


Elapsed time (sec)
V
e
r
t
i
c
a
l

s
t
r
a
i
n
,

v

(
%
)
Figure 5.18. Simulation of the behaviour of Hostun sand in plane strain compression test by the
TESRA model (Di Benedetto et al. 2001a).
65
It can be seen from Fig. 5.20 that the TESRA model simulates very well all the details of the
stress-strain-time behaviour from loading phase towards unloading phase. It is particularly
important that the following aspects are simulated very well:
1) the increase in the positive total and irreversible axial strains continues for some stress range
immediately after the start of decreasing the shear stress at a constant negative rate;
PSC test on sand 5
0.0 0.5 1.0 1.5 2.0 2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
Creep
A,B
Low strain rate ie
i
=0.740 j
High strain rate
(higher by a factor of 100)
ie
i
=0. 742 j


Axial strain,

( )

Different creep strain rates;

despite having started from nearly the same stress and strain state !
0.0 0.5 1.0 1.5 2.0 2.5 3.0
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
Experiment (thin curve)
Low strain rate
High strain rate

Axialstrain,
v
(%)
S
t
r
e
s
s

r
a
t
i
o
,

R
=

v
/

h

Figure 5.19. Simulation of the behaviour of Hostun sand in plane strain compression test by the
TESRA model (Tatsuoka et al. 2001b).
a)
b)
66
2) creep recovery takes place at the creep stages in the course of unloading below the neutral
state, and the amount of creep recovery increases as the shear stress at the creep stage
decreases.

Summary: It has been demonstrated that the issue of time-dependent stress-strain behaviour
of geomaterials (ageing effects and loading rate effects) is one of the fresh topics in
geotechnical engineering, requiring developments of new constitutive models.
Careful and systematic laboratory tests are required to understand the time-dependent
deformation properties of geomaterial, and a comprehensive series of experimental study on
various types of geomaterials are necessary to understand the general framework of this issue.
It was shown above that erratic results could be obtained from conventional creep tests on stiff
geomaterials measuring axial strains externally and local axial strain measurement is imperative
for such tests. It was also shown that one of the relevant testing methods to validate a
constitutive model developed to simulate the time-dependent deformation properties of
geomaterial and to obtain the parameters of a model include:
1) stepwise or gradual changes in the strain rate; and
2) switching between strain- and stress- control tests; between constant rate loading and creep
and stress relaxation tests.
Constitutive models are required to simulate this behaviour for various stress histories,
including unloading. It was shown that three-component models are relevant for this purpose.
It was also shown that the viscous stress-strain behaviour of some types of geomaterials (such as
sedimentary softrocks) can be simulated by the New Isotach model (without a decay in the
viscous stress), which is an extension of the conventional three-component rheology framework.
One new model (the TESRA model), with a decay in the viscous stress, was introduced herein
to simulate the peculiar viscous behaviour of sand.
PSC test on sand 6
Increase in the strain after the stress has started decreasing !
1.0 1.2 1.4
1
2
3
4
Fig. 18b
U
Test Ulcrp3


Experiment
Simulation
S
t
r
e
s
s

r
a
t
i
o
,

R
=

' v
/

'
h
Vertical (axial) strain,
v
(%)
8.5 9.0 9.5 10.0 10.5 11.0
1.30
1.35
1.40
1.45
Start of creep stage
V
e
r
t
i
c
a
l

(
a
x
i
a
l
)

s
t
r
a
i
n
,

v

(
%
)
Fig. 19b
U
Test Ulcrp3


Experiment
Simulation
by the TESRA model
Elapsed time (hour)
35 36 37 38 39 40 41
1.00
1.05
1.10
1.15
1.20
Start of creep stage
V
e
r
t
i
c
a
l

(
a
x
i
a
l
)

s
t
r
a
i
n
,

v

(
%
)
Fig. 19c
Test Ulcrp3


Experiment
Simulation by the TESRA model
Elapsed time (hour)
Figure 5.20. Simulation of the behaviour during stress-controlled unloading and the creep in the
course of unloading (Tatsuoka et al. 2001b).
67
More study is necessary to generalise these models to apply to more general stress paths with
and without cyclic loading. In particular, more experimental study is necessary on viscous
effects on the yield locus and flow rule, as discussed by Tatsuoka and Ishihara (1973, 1974a &
b) for sand, and the hardening function, as discussed by Tatsuoka et al. (2001c) for sand.


6 CONCLUDING REMARKS

Impacts on the theories and practice of geotechnical engineering of several findings obtained
from recent advanced laboratory stress-strain tests on geomaterials have been demonstrated.
The topics discussed in this lecture note are only some of similarly important issues in
geotechnical engineering. The main objective of this lecture is to demonstrate that geotechnical
engineering is still very young.
It seems that laboratory stress-strain tests have recently become less popular. It is perhaps
because:
1) it is very difficult to retrieve high-quality undisturbed samples in many occasions;
2) laboratory stress-strain tests are often considered to be less direct (so less useful) than field
loading tests for design purposes, while laboratory stress-strain tests are just painstaking and
time-consuming.
In many cases, however, the following is also true:
1) More proper characterisation of the stress-strain property of geomaterials becomes possible
with a help of relevant laboratory stress-strain tests, which could result into more rational (so
safer and more cost-effective) design.
2) Proper understanding of the stress-strain-time behaviour of geomaterials, which is also
essential for rational (safer and more cost-effective) design, is not possible only by field
loading tests and back-analysis of full-scale behaviour. Relevant laboratory stress-strain
tests can play an essential role for this purpose.


Acknowledgements: The materials that were referred to in this lecture note were obtained by
many previous and present colleagues of the author at the Institute of Industrial Science,
University of Tokyo, where the author spent nearly twenty years, and the Department of Civil
Engineering, University of Tokyo, where the author is presently working. The co-operation and
help from these colleagues are deeply acknowledged.
7 REFERENCES
Atkinson,J.H. and Sllfors,G. (1991): Experimental determination of soil properties, Proc. 10th EC on
SMFE, Firenze 3, 915-956.
Atkinson,J.H. and Sllfors,G. (1991): Experimental determination of soil properties, Proc. 10th EC on
SMFE, Firenze 3, 915-956.
Balakrishnaiyer,K., Ahn Dan,L.Q., Tatsuoka,F., Koseki,J. and Modoni,G. (1998): Deformation
characteristics at small strain levels of dense gravel, The Geotechnics of Hard Soils Soft Rocks, Proc.
of Second Int. Conf. on Hard Soils and Soft Rocks (Evamgelista & Picarelli eds.), Balkema, 1, 423-43.
Burland,J.B. (1989): Small is beautiful -the stiffness of soils at small strains, Canadian Geotechnical
Journal 26, 499-516.
Di Benedetto,H. and Tatsuoka,F. (1997): Small strain behaviour of geomaterials: Modelling of strain rate
effects, Soils and Foundations, 37-2, 127-138.
Di Benedetto,H., Tatsuoka,F. and Ishihara,M. (2001a): Time -dependent deformation characteristics of
sand and their constitutive modelling, Soils and Foundations (submitted).
Di Benedetto,H,, Sauzeat,C. and Geoffroy,H. (2001b), Modelling viscous effects and behaviour in the
small strain domains, Proc. 2nd Int. Symp. on Prefailure Deformation Characteristics of
Geomaterials, IS Torino (Jamiolkowski et al., eds.), Vol.2 (to be published).
Flora,A., Jiang,G.L., Kohata,Y. and Tatsuoka,F. (1994): Small strain behaviour of a gravel along some
triaxial stress paths, Proc. of Int. Symposium Pre-Failure Deformation of Geomaterials (Shibuya et al.,
eds.), Balkema, 1, 279-285.
68
Goto.S, Tatsuoka,F., Shibuya,S., Kim,Y.-S., and Sato,T.(1991): A simple gauge for local small strain
measurements in the laboratory, Soils and Foundations, 31-1, 169-180.
Goto,S., Park,C. -S., Tatsuoka,F., and Molenkamp,F. (1993): Quality of the lubrication layer used in
element tests on granular materials, Soils and Foundations, 33-2, 47-59.
Hardin,B.O. and Richart,F.E.Jr. (1963): Elastic wave velocities in granular soils, J. ASCE 89-SM1, 33-
65.
Hardin,B.O. and Black,W.L. (1968): Vibration modulus of normally consolidated clay, J. SMF Div.,
ASCE 95-SM6, 1531-1537.
Hardin,B.O. and Drnevich,V.P. (1972): Shear modulus and damping in soils: design equations and
curves, Jour. of SMF Div., ASCE, 98-SM7, 667-692.
Hardin,B.O. (1978): The nature of stress-strain behavior for soils, Proc. Geotech. Div. Specialty Conf. on
Earthquake Eng. and Soil Dynamics, ASCE, Pasadena, I, 3-90.
Hardin,B.O. and Bladford,G.E. (1989): Elasticity of articulate materials, J. of Geotech. Engrg., ASCE,
115-6, 788-805.
Hayano,K., Sato,T. and Tatsuoka,F. (1997), Deformation characteristics of a sedimentary softrock from
triaxial compression tests rectangular prism specimens, Gotechnique, Vol.47, No.3, Symposium In
Print, pp.439-449.
Hayano,K., Maeshiro,T., Tatsuoka,F., Sato,T., Wang,L. and Kodaka,T. (1999): Shear banding in a
sedimentary soft mudstone subjected to plane strain compression, Geotechnical Testing Journal,
ASTM, 22-1, 67-79.
Hayano,K., Matsumoto,M., Tatsuoka,F. and Koseki,J. (2001): Evaluation of time-dependent deformation
property of sedimentary soft rock and its constitutive modeling, Soils and Foundations, Vol.41, No.2.
Hight,D.W. and Higgins,K.G. (1995): An approach to the prediction of ground movements in engineering
practice: Background and application, Proc. of Int. Symposium Pre-Failure Deformation of
Geomaterials, IS Hokkaido 94 (Shibuya et al., eds.), Balkema, 2, 909-945.
Hoque,E., Tatsuoka,F. and Sato,T. (1996): Measuring anisotropic elastic properties of sand using a large
triaxial specimen, Geotechnical Testing Journal, ASTM, 19-4, 411-420.
Hoque,E., Sato,T. and Tatsuoka,F. (1997): Performance evaluation of LDTs for the use in triaxial tests,
Geotechnical Testing Journal, ASTM, 20-2, 149-167.
Hoque,E., and Tatsuoka,F. (1998): Anisotropy in the elastic deformation of granular materials, Soils and
Foundations, 38-1, 163-179.
Huang,C-C. and Tatsuoka,F. (1990): Bearing capacity of reinforced horizontal sandy ground, Geotextiles
and Geomembranes, 9-1, 51-82.
Huang,C. -C., Tatsuoka,F., and Sato,Y. (1994), Failure mechanisms of reinforced sand slopes loaded
with a footing, Soils and Foundations, 34-2, 27-40.
Huang,C. -C. and Tatsuoka,F. (1994), St ability analysis for footings on reinforced sand slopes, Soils and
Foundations, 34-3, 21-37.
Iwasaki,T., and Tatsuoka,F. (1977), Effects of grain size and grading on dynamic shear moduli of
sands, Soils and Foundations, 17-3, 19-35.
Iwasaki,T., Tatsuoka,F. and Takagi,Y. (1978), Shear moduli of sands under cyclic torsional loading,
Soils and Foundations, 18-1, 39-56.
Izumi,K., Ogihara,M. and Kameya,H. (1997): Displacements of bridge foundations on sedimentary
softrock; a case study on small strain stiffness, Gotechnique 47-3, 619-632.
Jamiolkowski,M., Leroueil,S., and Lo Presti,D.C.F. (1991): Design parameters from theory to practice,
Theme Lecture, Proc. Geo-Coast '91, Yokohama, 2, 877-917.
Jardine,R.J., Fourie,A.B., Maswoswse,J. and Burland,J.B. (1985): Field and laboratory measurements of
soil stiffness, Proc. 11th ICSMFE, San Francisco, 2, 511-514.
Jardine,R.J. (1995): One perspective of the pre-failure deformation characteristics of some geomaterials,
Proc. of Int. Symposium Pre-Failure Deformation of Geomaterials, IS Hokkaido 94 (Shibuya et al.,
eds.), Balkema, 2, 855-885.
Jiang,G.L., Tatsuoka,F., Flora,A. and Koseki,J. (1997), Inherent and stress state-induced anisotropy in
very small strain stiffness of a sandy gravel, Gotechnique, 47-3, Symposium In Print, 509-521.
Jiang,G.L., Kohata,Y., Tateyama,M. and Tatsuoka,F. (1999): Small strain deformation characteristics of
dense gravel, Proc. Second Int. Conf. on Pre-Failure Deformation Characteristics of Geomaterials, IS
Torino 99 (Jamiolkowski et al., eds.), Balkema, 1, 291-298.
Kim,Y.-S., Tatsuoka,F. and Ochi,K. (1994), Deformation characteristics at small strains of sedimentary
soft rocks by triaxial compression tests, Gotechnique, 44-3, 461-478.
Kohata,Y., Tatsuoka,F., Dong,J., Teachavorasinskun,S. and Mizumoto,K. (1994), Stress states affecting
elastic deformation moduli of geomaterials, Proc. of Int. Symposium Pre-Failure Deformation of
Geomaterials, IS Hokkaido 94 (Shibuya et al., eds.), Balkema, 1, 3-9.
69
Kohata,Y., Tatsuoka,F., Wang,L., Jiang,G.L., Hoque,E. and Kodaka,T. (1997), Modelling the non-
linear deformation properties of stiff geomaterials, Gotechnique, 47-3, Symposium In Print, 563-
580.
Koseki,J., Tatsuoka,F., Munaf,Y., Tateyama,M. and Kojima,K. (1997), A modified procedure to
evaluate active earth pressure at high seismic loads, Soils and Foundations, Special Issue, 209-216.
Koseki,J., Hamaya,S., Tatsuoka,F. and Maeshiro,N. (1998): Elastoplastic deformation characteristics of
Toyoura sand during liquefaction, Proc. of ASCE Specialty Conference on Geotechnical Engineering
and Soil Dynamics III, Geotechnical Special Publication No. 75, 1, 385-397.
Koseki,J, Balakrishnaier,K. and Tatsuoka,F. (1999): Large scale triaxial tests on elastic properties of
undisturbed gravel containing fines, Proc. Second Int. Conf. on Pre-Failure Deformation
Characteristics of Geomaterials, IS Torino 99 (Jamiolkowski et al., eds.), Balkema, 1, 299-304.
Kotake,N., Tatsuoka,F., Tanaka,T., Siddiquee,M.S.A., and Yamauchi,H. (1999): An insight into the
failure of reinforced sand in plane strain compression by FEM simulation, Soils and Foundations, 39-
5, 103-130.
Lam,W-K. and Tatsuoka,F. (1988), Triaxial compressive and extension strength of sand affected by
strength anisotropy and sample slenderness, ASTM STP No.977, 655-666.
Lam,W-K. and Tatsuoka,F. (1988), Effect of initial anisotropic fabric and
2
on strength and
deformation characteristics of sand, Soils and Foundations, 28-1, 89-106.
Leshchinsky,D. (2001): Design dilemma: Use of peak to residual strength of soil, Geotextiles and
Geomembranes 19-2, 111-125.
Lo Presti,D.C.F. (1989): Propriet Dinamiche dei Terreni, XIV Conferenza Geotechnica di Torino,
Dipartimento di Ingegneria Strutturale, Poltecnico di Torino, 62.
Lo Presti,D.C.F., Pallara,O., Lancellotta,R., Armandi,M. and Maniscalco,R. (1993): Monotonic and
cyclic loading behaviour of two sands at small strains, GTJ 16-4, 409-424.
Mair,R.J. (1993): Developments in geotechnical engineering research: application to tunnel and deep
excavation; Unwin Memorial Lecture 1992, Proc. ICE 93, 27-41.
Masuda,T., Tatsuoka,F., Yamada,S. and Sato,T. (1999): Stress-strain behaviour of sand in plane strain
compression, extension and cyclic loading tests, Soils and Foundations, 39-5, 31-45.
Matsushita,M., Tatsuoka,F., Koseki,J., Cazacliu,B., Di Benedetto,H. and Yasin,S.J.M. (1999): Time
effects on the pre-peak deformation properties of sands, Proc. Second Int. Conf. on Pre-Failure
Deformation Characteristics of Geomaterials, IS Torino 99 (Jamiolkowski et al., eds.), Balkema, 1,
681-689.
Meyerhof,G.G. (1951): The ultimate bearing capacity of foundations, Gotechnique, 2, 301-332.
Modoni,G., Flora,A., Anh Dan,L.Q., Mancuso,C., Koseki,J., Balakrishnaier,K. and Tatsuoka,F. (1999): A
simple experimental procedure for the complete characterization of small strain stiffness of gravels,
Proc. Second Int. Conf. on Pre-Failure Deformation Characteristics of Geomaterials, IS Torino 99
(Jamiolkowski et al., eds.), Balkema, 1, 123-130.
Modoni,G., Flora,A., Mancuso,C., Viggiani,C. and Tatsuoka,F. (2000): Evaluation of gravel stiffness by
pulse wave transmission tests, Geotechnical Testing Journal, ASTM, 23-4, 506-521.
Morimoto, T. (1990); Scale effect of bearing capacity of footing on sand, Master Thesis, the University
of Tokyo (in Japanese).
Park,C.-S. and Tatsuoka,F. (1994), Anisotropic strength and deformations of sands in plane strain
compression, Proc. of the 13
th
Int. Conf. on S.M.F.F., New Delhi, 1, 1-4.
Peng,F.L., Kotake,N., Tatsuoka,F., Hirakawa,D., and Tanaka,T. (2000): Plane strain compression
behaviour of geogrid-reinforced sand and its numerical analysis, Soils and Foundations, 40-3, 55-74.
Pradhan,T.B.S., Tatsuoka,F. and Horii,N. (1988a), Simple shear testing on sand in a torsional shear
apparatus, Soils and Foundations, 28-2, 95-112.
Pradhan,T.B.S., Tatsuoka,F. and Horii,N. (1988b), Strength and deformation characteristics of sand in
torsional simple shear, Soils Foundations, 28-3, 131-148.
Pradhan,T.B.S., Tatsuoka,F. and Sato,Y. (1989), Experimental stress-dilatancy relations of sand
subjected to cyclic loading, Soils and Foundations, 29-1, 45-64.
Pradhan,T.B.S. and Tatsuoka,F. (1989), On stress-dilatancy equations of sand subjected to cyclic
loading, Soils and Foundations, 29-1, pp.65-81.
Puzrin,A. and Tatsuoka,F. (1998), Elastic energy potential for uncemented granular materials, Soils and
Foundations, 38-4 , 267-274.
Qiu,J.-Y., Tatsuoka,F. and Uchimura,T. (2000): Constant pressure and constant volume direct shear tests
on reinforced sand, Soils and Foundations, 40-4, 1-17.
Sato,M., Ueda.M, Hasebe,N. and Kondo,H. (1997a): Comparison among wave velocities from seismic
observation waves and field tests for hard rock mass, J. of Japanese Society of Civil Engineers (JSCE),
V-38, 75-87 (in Japanese).
Sato,M., Ueda.M, Hasebe,N. and Umehara,H. (1997b): Dynamic elastic modulus of dam concrete
earthquake motion, J. of JSCE, X-35, 43-55.
70
Santucci de Magistris,F., Sato,T., Koseki,J. and Tatsuoka,F. (1998): Effects of strain rate and ageing on
small strain behaviour of a compact silty sand, The Geotechnics of Hard Soils Soft Rocks, Proc. of
Second Int. Conf. on Hard Soils and Soft Rocks (Evamgelista & Picarelli eds.), Balkema, 1, 843-851.
Santucci de Magistris,F., Koseki,J., Amaya,M., Hamaya,S., Sato,T. and Tatsuoka,F. (1999), A triaxial
testing system to evaluate stress-strain behaviour of soils for wide range of strain and strain rate,
Geotechnical Testing Journal, ASTM, 22-1, 44-60.
Santucci de Magistris,F. and Tatsuoka,F. (1999): Time effects on the stress-strain behaviour of Metramo
silty sand, Proc. Second Int. Conf. on Pre-Failure Deformation Characteristics of Geomaterials, IS
Torino 99 (Jamiolkowski et al., eds.), Balkema, 1, 491-555.
Shibuya,S., Tatsuoka,F., Abe,F., Kim,Y. -S., Park,C-S. and Mukabi,J.N. (1991), A new look at stress-
strain relations of soils and soft rocks, Proc. of the 9
th
Asian Regional Conf. on SMFE, Bangkok, 1,
63-66.
Shibuya,S., Tatsuoka,F., Teachavorasinskun,S., Kong,X.J ., Abe,F., Kim,Y.S. and Park,C.-S. (1992),
Elastic Deformation Properties of Geomaterials, Soils and Foundations, 32-3, 26-46.
Shibuya,S., Park,C.-S., Tatsuoka,F., .Abe,F., Teachavorasinskun,S., Kohata,Y. and Sato,T. (1994), The
significance of local lateral-strain measurement of soil specimens for a wide range of strain, Soils and
Foundations, 34-2, 95-105.
Siddiquee,M.S.A., Tatsuoka,F., Hoque,E., Tsubouchi,T., Yoshida,O., Yamamoto,S. and Tanaka,T.
(1994), FEM simulation of footing settlement for stiff geomaterials, Proc. of Int. Symposium Pre-
Failure Deformation of Geomaterials (Shibuya et al., eds.), Balkema, 1, 531-537.
Siddiquee,M.S.A., Tatsuoka,F., Kohata,Y., Yoshida,O. and Yamamoto,Y. and Tanaka,T. (1995a),
Settlement of a Pier Foundation for Akashi- Kaikyo Bridge and its numerical analysis, Proc. Int.
Workshop on Rock Foundation of Large-Scale Structures, Tokyo, Balkema, 413-420.
Siddiquee,M.S.A., Tanaka,T. and Tatsuoka,F. (1995b), Tracing the equilibrium path by dynamic
relaxation in materially non-linear problems, Int. Jour. For Numerical and Analytical Methods in
Geomechanics, 19-11, 749-767.
Siddiquee,M.S.A., Tanaka,T., Tatsuoka,F., Tani,K. and Morimoto,T. (1999), FEM simulation of scale
effect in bearing capacity of strip footing on sand, Soils and Foundations, 39-4, 91-109.
Siddiquee,M.S.A. and Tatsuoka,F. (2001): Modeling time-dependent stress-strain behavior of stiff
geomaterials and its Applications, Proc. of 10th International Conference on Computer Methods and
Advances in Geomechanics (IACMAG), Tucson, Arizona on January 7-12.
Siddiquee, M. S. A., Tatsuoka, F., Tanaka, T. , Tani, K., Yoshida, K. and Morimoto, T. (2001); Model
tests and FEM simulation of some factors affecting the bearing capacity of footing on sand, Soils and
Foundations, 41-2.
Souto,A., Hartikainen,J. And Ozudogru,K. (1994): Measurement of dynamic parameters of road
pavement materials by the bender element and resonant column tests, Gotechnique, 44-3, 519-526.
Takeuchi,T., Tanaka,T., Yamamoto,S. and Inoue,A. (1997): Akashi Kaikyo Bridge-a case history,
Tsuchi-to-kiso, Jour. of JGS 45-9, 43-48 (in Japanese).
Tanaka,Y., Kokusho,T., Okamoto,T. and Kudo,K. (1995): Evaluation of initial shear modulus of gravelly
soil by laboratory test and PS-logging, IS Tokyo 95, 1, 101-106.
Tani, K. (1986): Mechanism of Bearing Capacity of Shallow Foundation on Sand, Master of Engineering
tthesis, the University of Tokyo (in Japanese).
Tatsuoka,F. and Ishihara,K. (1973), Stress path and dilatancy performance of sand, Proc. of 10
th
Int.
Conf. on SMFE, Moscow, 1/64, 419-424.
Tatsuoka,F. and Ishihara,K. (1974a), Yielding of sand in triaxial compression, Soils and Foundations
(Journal of Japanese Society of Soil Mechanics and Foundation Engineering), 14-2, 51-65.
Tatsuoka,F. and Ishihara,K. (1974b), Drained Deformation of sand under cyclic stresses reversing
direction, Soils and Foundations, 14-3, 51-65.
Tatsuoka,F., Iwasaki,T., and Takagi,Y. (1978), Hysteretic damping of sands and its relation to shear
modulus, Soils and Foundations, 18-2, 25-40.
Tatsuoka,F., Iwasaki,T., Yoshida,S., Fukushima,S. and Sudo,H. (1979a), Shear modulus and damping by
drained tests of clean and specimens reconstituted by various methods, Soils and Foundations, 19-1,
39-54.
Tatsuoka,F., Iwasaki,T., Fukushima,S. and Sudo,H. (1979b), Stress conditions and stress histories
affecting shear modulus and damping of sand under cyclic loading, Soils and Foundations, 19-2, 29-
43.
Tatsuoka,F., Molenkamp,F., Torii,T., and Hino,T. (1984): Behavior of lubrication layers of platens in
element tests, Soils and Foundations, 24-1, 113-128.
Tatsuoka,F., Sakamoto,M., Kawamura,T. and Fukushima,S. (1986a), Strength and deformation
characteristics of sand in plane strain compression at extremely low pressures, Soils and Foundations,
26-1, 65-84.
71
Tatsuoka,F., Goto,S. and Sakamoto,M. (1986b), Effects of some factors on strength and deformation
characteristics of sand at low pressures, Soils and Foundations, 26-1, 105-114.
Tatsuoka,F., Sonoda,S., Hara,K., Fukushima,S., and Pradhan,T. B.S. (1986c), Failure and deformation of
sand in torsional shear, Soils and Foundations, 26-4, 79-97.
Tatsuoka,F. (1987), Discussion on the paper by Bolton, Gotechnique, 37-2, 219-226.
Tatsuoka,F. (1988), Some recent developments in triaxial testing system for cohesionless soils, ASTM
STP No.977, 7-67.
Tatsuoka,F., Pradhan,Tej B.S. and Horii,N. (1988), Discussion on the paper by Jewell and Wroth,
Gotechnique, 38-1, 148-153.
Tatsuoka,F., Tani,K., Okahara,M., Morimoto,T., Tatsuta,M., Takagi,S. and Mori,H. (1989a), Discussion
on the Paper by Hettler and Gudehus, Soils and Foundations, 29-4, 146-154.
Tatsuoka,F., Huang,C-C., Morimoto,T. and Okahara,M. (1989b), Discussion on the Paper by Graham et
al., Canadian Geotechnical Journal, 26-4, 748-755.
Tatsuoka,F., Nakamura,S., Huang,C-C., and Tani,K. (1990a), Strength anisotropy and shear band
direction in plane strain tests on sand, Soils and Foundations, 30-1, 35-54.
Tatsuoka,F., Shibuya,S., Goto,S., Sato,T. and Kong,X.J. (1990b), Discussion on the Paper by Clayton et
al., Geotechnical Testing Journal, 13-1, March, 63-67.
Tatsuoka,F., Shibuya,S., Teachavorasinskun,S. and Park,C-S. (1990c), Discussion on the Paper by
Bolton and Wilson, Gotechnique, 40-4, 659-663.
Tatsuoka,F., Okahara,M., Tanaka,T., Tani,K., Morimoto,T. and Siddiquee,M.S.A. (1991), Progressive
failure and particle size effect in bearing capacity of a footing on sand, Proc. ASCE Geotech.
Engineering Congress, 1991, Boulder, ASCE GSP, 27, 788-802.
Tatsuoka,F. and Huang,C.-C. (1991), Discussion of Bearing capacity of foundations in slopes by
Shields et al., Journal of Geotechnical Engineering, ASCE, 117-12, 1970-1975.
Tatsuoka,F. and Shibuya,S. (1991), Deformation characteristics of soils and rocks from field and
laboratory tests, Keynote Lecture for Session No.1, Proc. of the 9
th
Asian Regional Conf. on SMFE,
Bangkok, 1, 101-170.
Tatsuoka,F., Siddiquee,M.S.A., Tanaka,T. and Okahara,M. (1992), A new aspect of a very old issue:
Bearing capacity of footing on sand, Panel Dis cussion, Proc. of the 9
th
Asian Regional Conf. on
SMFE, 2, 358-359.
Tatsuoka,F., Siddiquee,M.S.A., Park,C. -S., Sakamoto,M. and Abe,F.(1993), Modeling stress-strain
relations of sand, Soils and Foundations, 33-2, 60-81.
Tatsuoka,F., Sato,T., Park,C.-S., Kim,Y. -S., Mukabi,J.N. and Kohata,Y. (1994a), Measurements of
elastic properties of geomaterials in laboratory compression tests, Geotechnical Testing Journal C
ASTM, 17-1, 80-94.
Tatsuoka,F., Teachavorasinskun,S., Dong,J., Kohata,Y. and Sato,T. (1994b), Importance of measuring
local strains in cyclic triaxial tests on granular materials, Proc. of ASTM Symposium Dynamic
Geotechnical TestingU, ASTM, STP 1213, 288-302.
Tatsuoka,F., Siddiquee,M.S.A. and Tanaka,T. (1994c), Link among design, model tests, theories and
sand properties in bearing capacity of footing on sand, Panel Discussion, Proc. of the 13
th
Int. Conf.
on S.M.F.E., New Delhi, 5, 87-88.
Tatsuoka,F. (1994), Measurement of static deformation moduli in dynamic tests, Panel Discussion on
Deformation of soils and displacements of structures, Panel Discussion, Proc. of the 10
th
European
Conf. on S.M.F.E., Florence, 4, 1219-1226.
Tatsuoka,F and Kim,Y.-S. (1995): Deformation of shear zone in sedimentary soft rock observed in
triaxial compression, Localisation and Bifurcation Theory for Soils and Rocks (Chambon et al., eds.),
Balkema, 181-187.
Tatsuoka,F. and Kohata,Y. (1995), Stiffness of hard soils and soft rocks in engineering applications,
Keynote Lecture, Proc. of Int. Symposium Pre-Failure Deformation of Geomaterials (Shibuya et al.,
eds.), Balkema, 2, 947-1063.
Tatsuoka,F., Lo Presti,D.C.F. and Kohata,Y. (1995a), Deformation characteristics of soils and soft rocks
under monotonic and cyclic loads and their relationships, SOA Report, Proc. of the Third Int. Conf.
on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, St Louis (Prakash
eds.), 2, 851-879.
Tatsuoka,F., Kohata,Y., Ochi,K. and Tsubouchi,T. (1995b), Stiffness of soft rocks in Tokyo
metropolitan area - from laboratory tests to full-scale behaviour, Keynote Lecture, Proc. Int.
Workshop on Rock Foundation of Large-Scale Structures, Tokyo, Balkema, 3-17.
Tatsuoka,F., Kohata,Y., Tsubouchi,T., Murata,K., Ochi,K. and Wang,L. (1995c), Sample disturbance in
rotary core tube sampling of softrock, Conf. on Advances in Site Investigation Practice, Institution of
Civil Engineers, London, 281-292.
72
Tatsuoka,F., Kohata,Y., Tsubouchi,T. and Ochi,K. (1995d), Stiffness of sedimentary soft rocks
evaluated by triaxial compression tests, Proc. of the 8
th
Int. Congress on Rock Mechanics, Tokyo, 3,
1201-1204.
Tatsuoka,F., Ochi,K., Tsubouchi,T., Kohata,Y. and Wang,L. (1997a): Sagamihara experimental
underground excavations in sedimentary softrock Geotech. Eng., Proc. ICE, 125, Oct., 206-223.
Tatsuoka,F., Uchida,K., Imai,K., Ouchi.T. and Kohata,Y. (1997b), Properties of cement-treated soils in
Trans-Tokyo Bay Highway project, Ground Improvement, Thomas Telford, 1-1, 37-58.
Tatsuoka,F., Goto,S., Tanaka,T., Tani,K., and Kimura,Y. (1997c), Particle size effects on particle size
effects on bearing capacity of footing on granular material, Proc. Int. Conf. on Deformation and
Progressive Failure in Geomechanics, IS Nagoya 97 (Asaoka, Adachi and Oka eds.), Pergamon
Press, 133-138.
Tatsuoka,F., Koseki,J., Tateyama,M., Munaf,Y. and Horii,N. (1998), Seismic Stability Against High
Seismic Loads of Geosynthetic-Reinforced Soil Retaining Structures, Keynote Lecture, Proc. 6
th
Int.
Conf. on Geosynthetics, Atlanta, 1, 103-142.
Tatsuoka,F., Jardine,R.J., Lo Presti,D., Di Benedetto,H. and Kodaka,T. (1999a), Characterising the Pre-
Failure Deformation Properties of Geomaterials, Theme Lecture for the Plenary Session No.1, Proc.
of XIV IC on SMFE, Hamburg, September 1997, 4, 2129-2164.
Tatsuoka,F., Modoni,G., Jiang,G.L., Anh Dan,L.Q., Flora,A., Matsushita,M., and Koseki,J. (1999b):
Stress-Strain Behaviour at Small Strains of Unbound Granular Materials and its Laboratory Tests,
Keynote Lecture, Proc. of Workshop on Modelling and Advanced testing for Unbound Granular
Materials, January 21 and 22, 1999, Lisboa (Correia eds.), Balkema, 17-61.
Tatsuoka,F., Correia,A.G., Ishihara,M. and Uchimura,T. (1999c): Non-linear Resilient Behaviour of
Unbound Granular Materials Predicted by the Cross-Anisotropic Hypo-Quasi-Elasticity Model, Proc.
of Workshop on Modelling and Advanced testing for Unbound Granular Materials, January 21 and
22, 1999, Lisboa (Correia eds.), Balkema, 197-204.
Tatsuoka,F., Santucci de Magistris,F. and Momoya,M. and Maruyama,N. (1999d): Isotach behaviour of
geomaterials and its modelling, Proc. Second Int. Conf. on Pre-Failure Deformation Characteristics of
Geomaterials, IS Torino 99 (Jamiolkowski et al., eds.), Balkema, 1, 491-499.
Tatsuoka,F., Santucci de Magistris,F., Hayano,K., Momoya,Y. and Koseki,J. (2000): Some new aspects
of time effects on the stress-strain behaviour of stiff geomaterials, Keynote Lecture, The Geotechnics
of Hard Soils Soft Rocks, Proc. of Second Int. Conf. on Hard Soils and Soft Rocks, Napoli, 1998
(Evamgelista and Picarelli eds.), Balkema, 2, 1285-1371.
Tatsuoka,F., Uchimura,T., Hayano,K., Di Benedetto,H., Koseki,J. and Siddiquee,M.S.A. (2001a); Time-
dependent deformation characteristics of stiff geomaterials in engineering practice, the Theme Lecture,
Proc. of the Second International Conference on Pre-failure Deformation Characteristics of
Geomaterials, Torino, 1999, Balkema (Jamiolkowski et al., eds.), 2 (to appear).
Tatsuoka,F., Ishihara,M., Di Benedetto,H. and Kuwano,R., (2001b): Time-dependent deformation
characteristics of geomaterials and their simulation, Soils and Foundations (submitted).
Tatsuoka,F., Masuda,T. and Siddiquee,M.S.A. (2001c): Modelling the stress-strain behaviour of sand in
cyclic plane strain loading, Soils and Foundations (submitted).
Teachavorasinskun,S., Shibuya,S., Tatsuoka,F., Kato,H. and Horii,N. (1991a), Stiffness and damping of
sands in torsion shear, Proc. Second Int. Conf. on Recent Advances in Geotech. Earthquake Engnrg.
and Soil Dynamics, March, St. Louis, I, 103-110.
Teachavorasinskun,S., Shibuya,S. and Tatsuoka,F. (1991b), Stiffness of sands in monotonic and cyclic
torsional simple shear, Proc. ASCE Geotechnical Engineering Congress, Boulder, Geotechnical
Special Publication, Vol.27, 863-878.
Uchimura,T., Tatsuoka,F., Sato,T., Tateyama,M. and Tamura,Y. (1996), Performance of preloaded and
prestressed geosynthetic-reinforced soil, Proc. Int. Symposium Earth Reinforcement, Fukuoka (Ochiai
et al., eds.), Balkema, 1, 537-542.
Woods,R.D. (1991), Field and laboratory determination of soil properties at low and high strains, Proc.
Second. Int. Conf. On Recent Advances in Geotechnical Earthquake Engineering and Soil Dymanics,
St. Lous, MO, 2, 1727-1741.
Yasin,S.J.M., Umetsu,K., Tatsuoka,F., Arthur,J.R.F. and Dunstan,T. (1999a), Plane strain strength and
deformation of sands affected by batch variations in two different types of apparatus, Geotechnical
Testing Journal, ASTM, 22-1, 80-1000.
Yasin,S.J.M. and Tatsuoka,F. (1999b): Stress history-dependency of sand deformation in plane strain,
Proc. Second Int. Conf. on Pre-Failure Deformation Characteristics of Geomaterials, IS Torino 99
(Jamiolkowski et al., eds.), Balkema, 1, 703-711.
Yasin,S.J.M. and Tatsuoka,F. (2000): Stress history-dependent deformation characteristics of dense sand
in plane strain; Soils and Foundations, 40-2, 77-98.
73
Yoshida,T., Tatsuoka,F., Siddiquee,M.S.A. and Kamegai,Y. (1995): Shear banding in sands observed in
plane strain compression, Localisation and Bifurcation Theory for Soils and Rocks (Chambon et al.,
eds.), Balkema, 165-179.
Yoshida,T. and Tatsuoka,F. (1997): Deformation property of shear band in sand subjected to plane strain
compression and its relation to particle characteristics, Proc. 14
th
ICSMFE, Hamburg, 1, 237-240.