Beruflich Dokumente
Kultur Dokumente
Literature review
1.1 Stiffness
In order to predict the mechanical respone of soils and rocks the costitutive law should
be defined. A relationship between the stress and strain increments, through a stiffness
tensor C
1. Volumetric variations are associated with the volumetric deformation modulus K??
(or simply Bulk modulus);
Now it should be noted that the experimental data show a mechanical behavior of the
soils that is far from linear elasticity, and this fact implies on the operational plane the
need to take account of the deformation level. For this reason, we need to introduce a
G tangent modulus
d
Gt =
d
1.1 Stiffness 2
which describes the variation of the stress due to a small variation of deformation
around the current state, or a G secant modulus
Gsec =
that identifies a mean stiffness in a stress interval determined from a reference zero
as shown in Fig1.1.
It is clear that the two values of the moduli coincide in the initial phase of the test,
while as deformation increases more and more the two values become very different. In
particular, in a mechanically unstable material (like OC clay or sand in dense conditions
as shown in Fig1.2), the tangent modulus tends to decrease to the peak value, it is null
at peak strength and becomes negative in the post-peak part, thus losing of any physical
meaning. On the contrary, the secant modulus decreases monotonically as deformation
level increases and it tends to zero for infinite deformations (Lancellotta, 2012)[? ].
1. A typical plot to highlight this non-linear behavior is to represent the secant mod-
ulus as a function of deformation as shown in the figure 1.3 . As expressed by
Lancellotta (2012)[? ] and reaffirmed by Foti (2015)[? ], it is possible to observe:
a range where a linear threshold (t ) of the order of 105 exists , below which the
1.1 Stiffness 3
Figure 1.2: CID TX compression tests on sand and clay (Deangeli 2016)[? ]
Figure 1.3: Shear modulus decay as a function of shear strain (Lancellotta, 2012) [? ].
1.1 Stiffness 4
initial tangent modulus G0 or Gmax remains constant, the shear strains and volu-
muetric strains are decoupled (for this reason we dont observe interstitial over-
pressures in undrained tests), load cycles do not exhibit hysteresis and the material
has no memory of the previous deformations. This linear threshold increases with
the plasticity of the material (PI) and as the confining pressure increases.
Over t range, the behavior is irreversible. In this way, the stiffness modulus
decreases significantly as deformations become larger and larger and now every
loading cycle is characterized by hysteresis phenomena.
Therefore, when it is necessary to estimate the true behavior of the earth, an elastic
pattern would be appropriate only below the linear threshold; however, in problems
involving monotone loading conditions , an equivalent modulus can be used, only if the
latter is representative of the deformation level (Lancellotta, 2012).
by comparing with the ground response from geotechnical construction and the measure-
ment accuracy from laboratory investigation (Atkinson and Sallfors, 1991; Mair, 1993)
as shown in Fig.1.4[? ].
Some study done on ground movements induced by tunneling using a finit element
studied on London underground revealed that non-linear small strain stiffness are nec-
essary to achieve ground settlement predictions. Furthermore they highlight as the
deformation level reached by the ground under operating conditions is between 104
and 103 , excluding the contact areas at the load boundary, where plastic deformations
may occur (Atkinson and Sallfors and Burghignoli et al., 1991).
The small strain behaviour of soils may be described using a simple scheme of two
Kinematic Sub-Yield Surfaces located within a Bounding Surface. Performing triaxial
tests, assuming negligible creep and a monotonic load and representing the results in a
normalised stress space, three zone where the behaviour of soil is quite different can be
detected (Fig 1.5):
These zones are delimited by boundaries that are not fixed, but they can move.
The first surface forms the boundary of the linear elastic behaviour (Zone I) and we
will see below that it is difficult to locate or prove for most soils, the second surface marks
the limit to still recoverable zone, but non linear behaviour (Zone II). Between Zone II
aand the initial Bounding surface there is a zone called Zone III, where the behaviour
is transitional and plastic component of strains become ever larger as the stress point
approaches the Bounding Surface [? ].
1.1 Stiffness 7
1.1.2.1 Zone 1
This zone is the smallest where the stiffness can be assumed costant (Ei=Emax or
Gi=Gmax).
In reality, the perfect linear elastic behaviour doesnt exist but its a good approxi-
mation only in particular cases and only at very small strain. In addition the anisotropy
should be taken into account. As it will be discussed, many dynamic tests (resonant
column, bender element test, ..) has shown that the behaviour of naturall soils depends
also on the direction at which they are subjected.
It can be stated that Zone 1 exists and varies in size and shape according to:
OCR;
Number of cycles;
Drained tests show a linear trend and greater stiffness than Undrained tests. The
stiffness of the drained test was constant for the entire test before brittle failure.
In this case Zone II and Zone III doesnt exist but they are engulfed by Zone I.
In Undrained conditions, Zone I could not exist. CIU tests showed not only a
softer response, but also stiffness was a little bit reduced. Furthermore, sometimes
in CIU condition is not possible to identify a linear trend. It means Zone I could
not exist in undrained condition for cemented soils.
If Zone I exists, its much smaller. Less cemented soils have shown that Zone I is
smaller than cemented soils (as expected) and also the irrecoverable strains (Zone
III) start at an early stage (Tatsuoka et al., 1990).
Resonant column results on two different clays demonstrate that Zone I exists even if
small and initial moduli are generally overestimated due to two main reasons:
1.1 Stiffness 8
The strain rate allowed in terms of time is quite different than the standard
tests;
In addiction the mechanical response of soils depends also on the stress directions
at which the specimen is subjected.
Linear elastic;
The Linear Elastic model is used to describe materials which respond as follows:
3. the material returns to its original shape when the loads are removed, and the
unloading path is the same as the loading path ( elastic );
This model well represents the engineering materials up to their elastic limit.
So doing all these assumptions, the Isotropic Linear Elastic material is described
using only 2 constants (E and ) that are independent and the stress-strain law can be
written as:
xx = 1
[
E xx
(yy + zz )
yy = 1
[
E yy
(xx + zz )
zz = 1
[
E zz
(xx + yy )
1.1 Stiffness 9
xy = 1+
E
xy
xz = 1+
E
xz
yz = 1+
E
yz
where
This is known as Hookes Law. In addition other Elastic moduli can be retrieved from
these two constants, depending on the cases to be analyzed:
Its important to recall that elastic parameters should be differentiated according to the
conditions at which tests are performed. Undrained analysis are studied in terms of
total stress, instead in drained tests effective stress must be used.
In the case of undrained analysis, if the material is saturated, the kinematic constraint
of no volume variation requires that the Poissons coefficient is equal to 0.5. In these
conditions, the volumetric deformation module assumes infinite value, and therefore
the deformations in terms of volume are zero, whatever the entity of the applied load.
In drained conditions the Poissons coefficient results experimentally always positive
between 0 and 0,5.
Its important to underline that the shear modulus, due to the fact that is indipen-
dent from interstitial pressure, it has the same value in both situations (undrained and
drained) [? ]. Thats why many tests are performed in order to retrieve the G moduli.
1.2 Static and Dynamic elastic moduli 10
For this reason, when it is necessary to distinguish the cases from which the param-
eters are derived, we will refer to the following nomenclature:
The values of the shear modulus to be used in the case of very small variations in the
tension state can be determined by measuring the wave velocity propagation in situ,
or in the lab on undisturbed specimens. The values of the tangential modulus thus
1.2 Static and Dynamic elastic moduli 11
Figure 1.6:
obtained are used in dynamic problems, and are mainly used to obtain an upper limit
of the modulus value.
Due to the fact that the shear modulus G must be the same in drained and undrained
conditions, the difference between the shear wave velocity through a saturated soil and
on the same dry specimen must be very small if the porosity index (e) and the effective
stress are the same in both situations.
Instead the Young modulus in undrained conditions, at very small stress variation,
could be evaluated on specimen in the lab measuring the longitudinal wave velocity (...).
The P-wave velocity in water, it means on a fully saturated sample, is more or less
equal to 1500 m/s and naturally is much bigger than a dry sample. So the P-wave
velocity is more dependent on the fluid phase compressibility and less dependent on
the solid skeleton compressibility; it means the wave velocity is independent from the
effective stress.
So knowing the wave velocity on the porous media which is a value easily obtainable
by in situ tests, does not provide useful indications on the stiffness of the solid skeleton
of saturated soils [? ].
1.2 Static and Dynamic elastic moduli 12
Many tests have shown that the stess-strain modulus for undrained loading becomes
greater after the first cycle of loading. This happens cause part of apparent strains in
the initial loading is due to closure of cracks on the sample. This errors are largely
greater in unconfined compression tests and lead to underestimate the moduli than in
standard triaxial test. Tests on London clay showed that the stress-strain modulus of
the second loading cycle was 1.5 times greater than the one obtained from the first cycle
[? ].
On the other hand, if the load/unload cycles becomes more and more beyond the
elastic field the specimen begins to degrade and the resistance modulus decreases as
shown in Fig 1.7.
In addition Foti (2015) argued that:
if the number of cycles are the same, cyclic degradation increases as the shear
strains increases;
This is a fundamental parameter for understanding, from a general point of view, when
the static domain start to become dynamic.
Strain rate refers to the rate of strain (i.e. change in axial strain per unit time) that
is applied during undrained shear. The rate of srain can have very large effects on the
stress-strain modulus.
As studied and tested by Richardson and Whitman (1963) where they compared
data from dynamic tests (a test in which the sample is failed in 1msec) with those from
static tests (one in which a specimen is failed in 10 min), Whitman (1964) concluded
that the dynamic modulus is 1.5 to 2 times greater than static modulus [? ].
Rock strength is estimated from two common laboratory techniques: uniaxial compres-
sive strength tests and triaxial or confined compressive strength tests. Uniaxial com-
pressive strength tests are used to determine the ultimate strength of a rock, that is, the
maximum value of stress attained before failure.
TheYoung Modulus E;
ThePoisson ratio.
The dynamically determined elastic modulus (Edyn) is generally higher than that
statically determined, and both methods provide high divergent results for low elasticity
modulus rocks (Ide 1936).
Some authors (Ciccotti and Mulargia 2004) consider that the static test is a dynamic
test at a very low frequency, and they highlight the nonlinear elastic response to different
associated frequencies (f). In addition it has been observed that there is no significant
dependence on frequency below the seismic regime, confirming that seismic methods
produce accurate estimates of elastic modulus for geophysical modeling [? ].
Indeed, elastic constants can be measured using three different standard techniques
whom cover nine orders of magnitude of frequency: The main methods used to measure
them are (M. Ciccotti et al., 2004 [? ]:
1. the static method, which is based on the measurement of the deformation induced
in a material by the application of a known force, the uniaxial static compression
test (~ 10-3 Hz);
2. the forced oscillation method, which consists of measuring the ratio between force
and deformation in an intermediate range of frequency, through a device called
Dual Cantilever (da 10-2 a 200 Hz);
3. the dynamic method, which implies the measurement of the ultrasonic longitudinal
and transversal velocities (at 75 kHz and at 1 MHz).
As a result of inelastic effects, these values do not necessarily coincide (M. Ciccotti and
F. Mulargia, 2003 [? ] . In spite of this, the lack of available data forces one to use them
indifferently, albeit with persisting doubts about the reliability of such an operation.
Eissa and Kazi (1988) [? ] did several laboratory measurements, where the elastic
moduli are divided into static and dynamic moduli. Their proposed empirical relations
show that values of static moduli appear in generally from 5% up to 10% lower than
those of dynamic moduli; in some cases, due to the fact that the scatter among the data
is very huge, static measurements can differ from dynamics of an order of magnitude [?
].
rocks, sandstones of low to medium porosity, compact limestone, chalks, etc. exhibit
linear behaviour. The isotropic elastic parameters of the rock are then easily determined
using the stress-strain curves obtained at costant pore pressure (drained test) [? ].
.
1.2.7.2 Empirical
media laterally exterior to the reservoir (sideburden). However, an often used assump-
tion of the mechanical response of the reservoir is the assumption of uni-axial strain
during deformation. This simplification allows an easier characterization of rock proper-
ties using data from uni-axial lab measurement . As the fluids are withdrawn from the
reservoir, it is assumed to only compact one-dimensional in the vertical direction. This
assumption is valid if the vertical extent of the reservoir is small compared to its lateral
extent [[? ]].
= u
where is a correction factor that measures the effectiveness of the pore pressure
in counteracting the total applied load. The value of , which varies between 0 and 1,
depends on the pore geometry and the physical properties of the constituents of the solid
system. In the extreme cases when is equal to 0, the pore pressure has no effect on the
behavior of the rock, and when =1 the pore pressure is 100% effective in counteracting
the applied load.
The coefficient in the regime of linear elasticity of isotropic rocks is equal to the
so called Biot coefficient:
K
=1 Ks
where K is the drained bulk modulusof the solid skeleton and K_s is the bulk
modulus of the solid phase.
show the absence of lateral deformations only along the center of the baricentric axis at
which the longitudinal elastic modulus is reduced to the edometric.
1. Porosity;
2. Reservoir thickness;
5. Reservoir depth.