Material Models in Plaxis

PLAXIS-GiD
Material Models
Manual
Version 1
TABLE OF CONTENTS
TABLE OF CONTENTS
1 Introduction ................................................................................... 1-1
1.1 On the use of different models...................................................... 1-1
1.2 Limitations ............................................................................... 1-3
2 Preliminaries on material modelling.................................................. 2-1
2.1 General denitions of stress ......................................................... 2-1
2.2 General denitions of strain ......................................................... 2-4
2.3 Elastic strains ............................................................................ 2-5
2.4 Undrained effective stress analysis (effective stiffness parameters) ..... 2-7
2.5 Undrained effective stress analysis (effective strength parameters) .....2-12
2.6 Undrained effective stress analysis (undrained strength parameters) ...2-13
2.7 Undrained total stress analysis (undrained parameters) ....................2-14
2.8 The initial pre-consolidation stress in advanced models ...................2-14
2.9 On the initial stresses .................................................................2-15
3 The Mohr-Coulomb model (perfect-plasticity) ................................... 3-1
3.1 Elastic perfectly-plastic behaviour................................................. 3-1
3.2 Formulation of the Mohr-Coulomb model ...................................... 3-3
3.3 Basic parameters of the Mohr-Coulomb model................................ 3-5
3.4 Advanced parameters of the Mohr-Coulomb model.........................3-10
4 The Hardening Soil model (isotropic hardening) ................................ 4-1
4.1 Hyperbolic relationship for standard drained triaxial test ................... 4-2
4.2 Approximation of hyperbola by the Hardening Soil model ................ 4-3
4.3 Plastic volumetric strain for triaxial states of stress .......................... 4-5
4.4 Parameters of the Hardening Soil Model ........................................ 4-6
4.5 On the cap yield surface in the Hardening Soil model ......................4-11
5 The Hardening Soil model with small-strain stiffness (HSsmall) .......... 5-1
5.1 Describing small-strain stiffness with a Simple Hyperbolic Law......... 5-2
5.2 Applying the Hardin-Drnevich Relationship in the HS model ............ 5-3
5.3 Virgin (initial) loading vs. unloading/reloading ............................... 5-5
5.4 Model parameters....................................................................... 5-6
5.5 On the parameters G
0
and
0.7
...................................................... 5-7
5.6 Model initialization .................................................................... 5-9
5.7 Other differences between the HS and the HSsmall model................5-10
6 The Soft Soil model.......................................................................... 6-1
6.1 Isotropic states of stress and strain ('
1
= '
2
= '
3
)......................... 6-1
6.2 Yield function for triaxial stress state ('
2
= '
3
) ............................. 6-2
6.3 Parameters of the Soft Soil model ................................................. 6-5
7 Soft Soil Creep model (time dependent behaviour) ............................. 7-1
i
MATERIAL MODELS MANUAL
7.1 Introduction .............................................................................. 7-1
7.2 Basics of one-dimensional creep................................................... 7-2
7.3 On the variables
c
and
c
............................................................ 7-4
7.4 Differential law for 1D-creep ....................................................... 7-6
7.5 Three-dimensional-model ............................................................ 7-7
7.6 Formulation of elastic 3D-strains .................................................7-10
7.7 Review of model parameters .......................................................7-11
7.8 Validation of the 3D-model .........................................................7-14
8 The Jointed Rock model (anisotropy) ................................................ 8-1
8.1 Anisotropic elastic material stiffness matrix.................................... 8-2
8.2 Plastic behaviour in three directions .............................................. 8-4
8.3 Parameters of the Jointed Rock model ........................................... 8-7
9 User-dened soil models................................................................... 9-1
9.1 Introduction .............................................................................. 9-1
9.2 Implementation of UD models in calculations program..................... 9-1
9.3 Input of UD model parameters via user-interface ............................9-11
10 Structural behaviour ......................................................................10-1
10.1 Anchors...................................................................................10-1
10.2 Beams .....................................................................................10-1
10.3 Geogrids..................................................................................10-2
10.4 Plates ......................................................................................10-3
11 References .....................................................................................11-1
Appendix A - Symbols
Appendix B - Fortran subroutines for User-dened soil models
Appendix C - Creating a debug-le for User-dened soil models
ii PLAXIS-GiD
INTRODUCTION
1 INTRODUCTION
The mechanical behaviour of soils may be modelled at various degrees of accuracy.
Hooke's law of linear, isotropic elasticity, for example, may be thought of as the simplest
available stress-strain relationship. As it involves only two input parameters, i.e. Young's
modulus, E, and Poisson's ratio, , it is generally too crude to capture essential features
of soil and rock behaviour. For modelling massive structural elements and bedrock layers,
however, linear elasticity tends to be appropriate.
1.1 ON THE USE OF DIFFERENT MODELS
Mohr-Coulomb model (MC)
The elastic-plastic Mohr-Coulomb model involves ve input parameters, i.e. E and for
soil elasticity; and c for soil plasticity and as an angle of dilatancy. This Mohr-
Coulomb model represents a 'rst-order' approximation of soil or rock behaviour. It
is recommended to use this model for a rst analysis of the problem considered. For
each layer one estimates a constant average stiffness. Due to this constant stiffness,
computations tend to be relatively fast and one obtains a rst estimate of deformations.
Hardening Soil model (HS)
The Hardening Soil model is an advanced model for the simulation of soil behaviour.
As for the Mohr-Coulomb model, limiting states of stress are described by means of the
friction angle, , the cohesion, c, and the dilatancy angle, . However, soil stiffness is
described much more accurately by using three different input stiffnesses: the triaxial
loading stiffness, E
50
, the triaxial unloading stiffness, E
ur
, and the oedometer loading
stiffness, E
oed
. As average values for various soil types, E
ur
3E
50
and E
oed
E
50
are
suggested as default settings, but both very soft and very stiff soils tend to give other ratios
of E
oed
/ E
50
, which can be entered by the user.
In contrast to the Mohr-Coulomb model, the Hardening Soil model also accounts for
stress-dependency of stiffness moduli. This means that all stiffnesses increase with
pressure. Hence, all three input stiffnesses relate to a reference stress, usually taken as
100 kPa (1 bar).
Besides the model parameters mentioned above, initial soil conditions, such as pre-
consolidation, play an essential role in most soil deformation problems. This can be taken
into account in the initial stress generation.
Hardening Soil model with small-strain stiffness (HSsmall)
The Hardening Soil model with small-strain stiffness (HSsmall) is a modication of the
above Hardening Soil model that accounts for the increased stiffness of soils at small
strains. At low strain levels most soils exhibit a higher stiffness than at engineering strain
1-1
levels, and this stiffness varies non-linearly with strain. This behaviour is described in the
HSsmall model using an additional strain-history parameter and two additional material
parameters, i.e. G
re f
0
and
0.7
. G
0
is the small-strain shear modulus and
0.7
is the strain
level at which the shear modulus has reduced to about 70% of the small-strain shear
modulus. The advanced features of the HSsmall model are most apparent in working load
conditions. Here, the model gives more reliable displacements than the HS model. When
used in dynamic applications, the HSsmall model also introduces hysteretic material
damping.
Soft Soil model (SS)
The Soft Soil model is a Cam-Clay type model especially meant for primary compression
of near normally-consolidated clay-type soils. Although the modelling capabilities of this
model are superceded by the Hardening Soil model, the Soft Soil model is still retained in
the current version, because existing PLAXIS users might be comfortable with this model
and still like to use it in their applications.
Soft Soil Creep model (SSC)
The above Hardening Soil model is suitable for all soils, but it does not account for viscous
effects, i.e. creep and stress relaxation. In fact, all soils exhibit some creep and primary
compression is thus followed by a certain amount of secondary compression.
The latter is most dominant in soft soils, i.e. normally consolidated clays, silts and peat,
and PLAXIS thus implemented a model under the name Soft Soil Creep model. Please
note that the Soft Soil Creep model is a relatively new model that has been developed
for application to settlement problems of foundations, embankments, etc. For unloading
problems, as normally encountered in tunnelling and other excavation problems, the
Soft Soil Creep model hardly supersedes the simple Mohr-Coulomb model. As for the
Hardening Soil model, proper initial soil conditions are also essential when using the Soft
Soil Creep model. This also includes data on the pre-consolidation stress, as the model
accounts for the effect of over-consolidation.
Jointed Rock model (JR)
The Jointed Rock model is an anisotropic elastic-plastic model, especially meant to
simulate the behaviour of rock layers involving a stratication and particular fault
directions. Plasticity can only occur in a maximumof three shear directions (shear planes).
Each plane has its own strength parameters and c. The intact rock is considered to
behave fully elastic with constant stiffness properties E and . Reduced elastic properties
may be dened for the stratication direction.
Analyses with different models
It is advised to use the Mohr-Coulomb model for a relatively quick and simple rst
analysis of the problem considered.
1-2 PLAXIS-GiD
INTRODUCTION
In many cases, provided one has good data on dominant soil layers, it is recommended to
use the Hardening Soil model or the HSsmall model in an additional analysis. No doubt,
one seldomly has test results from both triaxial and oedometer tests, but good quality data
from one type of test can be supplemented by data from correlations and/or in situ testing.
Finally, a Soft Soil Creep analysis can be performed to estimate creep, i.e. secondary
compression in very soft soils. The above idea of analyzing geotechnical problems with
different soil models may seem costly, but it tends to pay off. First of all due to the fact
that the Mohr-Coulomb analysis is relatively quickly and simple, and in the second place
the procedure tends to reduce errors and inaccuracies.
1.2 LIMITATIONS
The PLAXIS code and its soil models have been developed to perform calculations
of realistic geotechnical problems. In this respect PLAXIS can be considered as
a geotechnical simulation tool. The soil models can be regarded as a qualitative
representation of soil behaviour whereas the model parameters are used to quantify
the soil characteristics. Although much care has been taken for the development of the
PLAXIS code and its soil models, the simulation of reality remains an approximation,
which implicitly involves some inevitable numerical and modelling errors. Moreover, the
accuracy at which reality is approximated depends highly on the expertise of the user
regarding the modelling of the problem, the understanding of the soil models and their
limitations, the selection of model parameters, and the ability to judge the reliability of
the computational results.
Both the soil models and the PLAXIS code are constantly being improved, so that each new
version has the character of an update. Some of the limitations in the currently available
models are listed below:
Linear Elastic model
Soil behaviour is highly non-linear and irreversible. The linear elastic model is insufcient
to capture the essential features of soil. The use of the linear elastic model may, however,
be considered to model strong massive structures in the soil or bedrock layers. Stress
states in the linear elastic model are not limited in any way, which means that the model
shows innite strength. Be careful using this model for materials that are loaded up to
their material strength.
Mohr-Coulomb model
The Mohr-Coulomb model is a rst order model that includes only a limited number
of features that soil behaviour shows in reality. Although the increase of stiffness with
depth can be taken into account, the Mohr-Coulomb model does neither include stress-
dependency nor stress-path dependency of stiffness or anisotropic stiffness. In general,
effective stress states at failure are quite well described using the Mohr-Coulomb failure
1-3
criterion with effective strength parameters ' and c'. For undrained materials, the Mohr-
Coulomb model may be used with the friction angle set to 0
and the cohesion c set to

c
u
(s
u
), to enable a direct control of undrained shear strength. In that case note that the
model does not automatically include the increase of shear strength with consolidation.
Hardening Soil model
Although the Hardening Soil model can be regarded as an advanced soil model, there are
a number of features of real soil behaviour the model does not include. It is a hardening
model that does not account for softening due to soil dilatancy and de-bonding effects. In
fact, it is an isotropic hardening model so that it models neither hysteretic and cyclic
loading nor cyclic mobility. Moreover, the model does not distinguish between large
stiffness at small strains and reduced stiffness at engineering strain levels. The user has
to select the stiffness parameters in accordance with the dominant strain levels in the
application. Last but not least, the use of the Hardening Soil model generally results in
longer calculation times, since the material stiffness matrix is formed and decomposed in
each calculation step.
Hardening Soil model with small-strain stiffness
As the Hardening Soil model with small-strain stiffness (HS small model) incorporates the
loading history of the soil and a strain-dependent stiffness, it can, to some extent, be used
to model cyclic loading. However, it does not incorporate a gradual softening during cyclic
loading, so is not suitable for cyclic loading problems in which softening plays a role. In
fact, just as in the Hardening Soil model, softening due to soil dilatancy and debonding
effects are not taken into account. Moreover, the HS small model does not incorporate the
accumulation of irreversible volumetric straining nor liquefaction behaviour with cyclic
loading. The use of the HS small model will generally result in calculation times that are
even longer than those of the Hardening Soil model.
Soft Soil model
The same limitations (including these in the Soft Soil Creep model) hold in the Soft Soil
model. In fact the Soft Soil model is superceded by the Hardening Soil model, but is
kept for users who are familiar with this model. The utilization of the Soft Soil model
should be limited to the situations that are dominated by compression. It is certainly not
recommended for use in excavation problems.
Soft Soil Creep model
All above limitations also hold true for the Soft Soil Creep model. In addition this model
tends to over-predict the range of elastic soil behaviour. This is especially the case for
excavation problems, including tunnelling. Care must also be taken with the generation of
initial stresses for normally consolidated soils. Although it would seemlogical to use OCR
= 1.0 for normally consolidated soils, such use would generally lead to an over-prediction
of deformations in problems where the stress level is dominated by the initial stresses.
1-4 PLAXIS-GiD
INTRODUCTION
Therefore, for such problems it is recommended to use a slightly increased OCR-value
to generate the initial stress state. In fact, in reality most soils tend to show a slightly
increased pre-consolidation stress in comparison with the initial effective stress. Before
starting an analysis with external loading it is suggested to perform a single calculation
phase with a short time interval and without loading to verify the surface settlement rate
based on common practice.
Jointed Rock model
The Jointed Rock model is a rst order anisotropic model that includes a limited number
of features that rock behaviour shows in reality. Plasticity can only occur in a maximum
of three shear directions (shear planes). Each plane has its own shear strength parameters
i
and c
i
and tensile strength t
i
. Hence, the maximum shear stress is linearly dependent
on the normal stress, and not curved as in reality. The intact rock is considered to behave
fully elastic with constant stiffness properties E and . Reduced elastic properties may be
dened for the stratication direction. Note that failure is limited to the predened shear
plane directions. It is possible that realistic potential failure mechanisms are not captured
by the model because of the elastic behaviour in any other direction than the three shear
planes.
Interfaces
Interface elements are generally modelled by means of the bilinear Mohr-Coulomb model.
When a more advanced model is used for the corresponding cluster material data set, the
interface element will only pick up the relevant data (c, , , E, ) for the Mohr-Coulomb
model, as described in Section ?? of the Reference Manual. In such cases the interface
stiffness is set equal to the elastic soil stiffness. Hence, E = E
ur
where E
ur
is stress level
dependent, following a power law with E
ur
proportional to
m
. For the Soft Soil model
and Soft Soil Creep model, the power m is equal to 1 and E
ur
is largely determined by the
swelling constant *.
Undrained behaviour
In general, care must be taken in undrained conditions, since the effective stress path that
is followed in any of the models may deviate signicantly from reality. Although PLAXIS
has an option to deal with undrained behaviour in an effective stress analysis, the use of
undrained shear strength (c
u
or s
u
) is recommended over the use of effective strength
properties (' and c') in such cases. Please note that direct input on undrained shear
strength does not automatically include the increase of shear strength with consolidation.
If, for any reason, the user decides to use effective strength properties in undrained
conditions, it is strongly recommended to check the resulting mobilised shear strength
using the corresponding option in the PLAXIS Output program.
1-5
1-6 PLAXIS-GiD
PRELIMINARIES ON MATERIAL MODELLING
2 PRELIMINARIES ON MATERIAL MODELLING
A material model is described by a set of mathematical equations that give a relationship
between stress and strain. Material models are often expressed in a form in which
innitesimal increments of stress (or 'stress rates') are related to innitesimal increments
of strain (or 'strain rates'). All material models implemented in PLAXIS are based on a
relationship between the effective stress rates, ', and the strain rates, . In the following
section it is described how stresses and strains are dened in PLAXIS. In subsequent
sections the basic stress-strain relationship is formulated and the inuence of pore
pressures in undrained materials is described. Later sections focus on initial conditions
for advanced material models.
2.1 GENERAL DEFINITIONS OF STRESS
Stress is a tensor which can be represented by a matrix in Cartesian coordinates:
=
_
xx

xy

xz
yx

yy

yz
zx

zy

zz
_
_
(2.1)
In the standard deformation theory, the stress tensor is symmetric such that
xy
=
yx
,
yz
=
zy
and
zx
=
xz
. In this situation, stresses are often written in vector notation, which
involve only six different components:
=
_
xx

yy

zz

xy

yz

zx
_
T
(2.2)
According to Terzaghi's principle, stresses in the soil are divided into effective stresses,
', and pore pressures,
w
:
= ' +
w
(2.3)
y
z
x
xx
xy
xz
yy
yx
yz
zz
zx
zy
Figure 2.1 General three-dimensional coordinate system and sign convention for stresses
2-1
Pore pressures are generally provided by water in the pores. Water is considered not to
sustain any shear stresses. As a result, effective shear stresses are equal to total shear
stresses. Positive normal stress components are considered to represent tension, whereas
negative normal stress components indicate pressure (or compression). Moreover, water
is considered to be fully isotropic, so all pore pressure components are equal. Hence, pore
pressure can be represented by a single value, p
w
:
w
=
_
p
w
p
w
p
w
0 0 0
_
T
(2.4)
Material models for soil and rock are generally expressed as a relationship between
innitesimal increments of effective stress and innitesimal increments of strain. In such
a relationship, innitesimal increments of effective stress are represented by stress rates
(with a dot above the stress symbol):
' =
_
'
xx
'
yy
'
zz

xy

yz

zx
_
T
(2.5)
It is often useful to apply principal stresses rather than Cartesian stress components
when formulating material models. Principal stresses are the stresses in such a coordinate
system direction that all shear stress components are zero. Principal stresses are, in fact,
the eigenvalues of the stress tensor. Principal effective stresses can be determined in the
following way:
det
_
' ' I
_
= 0 (2.6)
where I is the identity matrix. This equation gives three solutions for ', i.e. the principal
effective stresses ('
1
, '
2
, '
3
). In PLAXIS the principal effective stresses are arranged in
algebraic order:
'
1
'
2
'
3
(2.7)
Hence, '
1
is the largest compressive principal stress and '
3
is the smallest compressive
principal stress. In this manual, models are often presented with reference to the principal
stress space, as indicated in Figure 2.2.
In addition to principal stresses it is also useful to dene invariants of stress, which are
stress measures that are independent of the orientation of the coordinate system. Two
useful stress invariants are:
p' =
1
3
('
xx
+'
yy
+'
zz
) (2.8a)
q =
_
1
2
_
('
xx
'
yy
)
2
+('
yy
'
zz
)
2
+('
zz
'
xx
)
2
+6
_
2
xy
+
2
yz
+
2
zx
_
_
(2.8b)
2-2 PLAXIS-GiD
where p' is the isotropic effective stress, or mean effective stress, and q is the equivalent
shear stress. The equivalent shear stress, q, has the important property that it reduces to
q =|'
1
'
3
| for triaxial stress states with '
2
= '
3
.
-
-
-
- = - = - '
1
'
1
'
2
'
2
'
3
'
3
Figure 2.2 Principal stress space
Principal effective stresses can be written in terms of the invariants:
'
1
= p' +
2
3
qsin
_

2
3
_
(2.9a)
'
2
= p' +
2
3
qsin() (2.9b)
'
3
= p' +
2
3
qsin
_
+
2
3
_
(2.9c)
in which T is referred to as Lode 's angle (a third invariant), which is dened as:
=
1
3
arcsin
_
27
2
J
3
q
3
_
(2.10)
with
J
3
= ('
xx
p')('
yy
p')('
zz
p') ('
xx
p')
2
yz
('
yy
p')
2
zx
('
zz
p')
2
xy
+2
xy
yz
zx
(2.11)
2-3
2.2 GENERAL DEFINITIONS OF STRAIN
Strain is a tensor which can be represented by a matrix with Cartesian coordinates as:
=
_
xx

xy

xz
yx

yy

yz
zx

zy

zz
_
_
(2.12)
Strains are the derivatives of the displacement components, i.e.
i j
= (u
i
/ j +u
j
/i),
where i and j are either x, y or z. According to the small deformation theory, only the
sum of complementing Cartesian shear strain components
i j
and
ji
result in shear stress.
This sum is denoted as the shear strain . Hence, instead of
xy
,
yx
,
yz
,
zy
,
zx
and
xz
the
shear strain components
xy
,
yz
and
zx
are used respectively. Under the above conditions,
strains are often written in vector notation, which involve only six different components:
=
_
xx

yy

zz

xy

yz

zx
_
T
(2.13)
xx
=
u
x
x
(2.14a)
yy
=
u
y
y
(2.14b)
zz
=
u
z
z
(2.14c)
xy
=
xy
+
yx
=
u
x
y
+
u
y
x
(2.14d)
yz
=
yz
+
zy
=
u
y
z
+
u
z
y
(2.14e)
zx
=
zx
+
xz
=
u
z
x
+
u
x
z
(2.14f)
Similarly as for stresses, positive normal strain components refer to extension, whereas
negative normal strain components indicate compression.
In the formulation of material models, where innitesimal increments of strain are
considered, these increments are represented by strain rates (with a dot above the strain
symbol).
=
_

xx

yy

zz

xy

yz

zx
_
T
(2.15)
In analogy to the invariants of stress, it is also useful to dene invariants of strain. A strain
invariant that is often used is the volumetric strain,
, which is dened as the sum of all

2-4 PLAXIS-GiD
normal strain components:
v
=
xx
+
yy
+
zz
=
1
+
2
+
3
(2.16)
The volumetric strain is dened as negative for compaction and as positive for dilatancy.
For elastoplastic models, as used in PLAXIS, strains are decomposed into elastic and
plastic components:
=
e
+
p
(2.17)
Throughout this manual, the superscript e will be used to denote elastic strains and the
superscript p will be used to denote plastic strains.
2.3 ELASTIC STRAINS
Material models for soil and rock are generally expressed as a relationship between
innitesimal increments of effective stress ('effective stress rates') and innitesimal
increments of strain ('strain rates'). This relationship may be expressed in the form:
' = M (2.18)
where M is a material stiffness matrix. Note that in this type of approach, pore-pressures
are explicitly excluded from the stress-strain relationship.
Figure 2.3 Parameters tab for the Linear Elastic model
The simplest material model in PLAXIS is based on Hooke's law for isotropic linear elastic
2-5
behaviour. This model is available under the name Linear Elastic model, but it is also the
basis of other models. Hooke's law can be given by the equation:
_
_
'
xx
'
yy
'
zz
'
xy
'
yz
'
zx
_
_
=
E'
(12')(1+')
_
_
1' ' ' 0 0 0
' 1' ' 0 0 0
' ' 1' 0 0 0
0 0 0
1
2
' 0 0
0 0 0 0
1
2
' 0
0 0 0 0 0
1
2
'
_
_
_
_

xx

yy

zz

xy

yz

zx
_
_
(2.19)
The elastic material stiffness matrix is often denoted as D
e
. Two parameters are used in
this model, the effective Young's modulus, E', and the effective Poisson's ratio, '. In the
remaining part of this manual effective parameters are denoted without dash ('), unless a
different meaning is explicitly stated. The symbols E and are sometimes used in this
manual in combination with the subscript ur to emphasize that the parameter is explicitly
meant for unloading and reloading. A stiffness modulus may also be indicated with the
subscript re f to emphasize that it refers to a particular reference level (y
re f
)(see next page).
According to Hooke's law, the relationship between Young's modulus E and other stiffness
moduli, such as the shear modulus G, the bulk modulus K, and the oedometer modulus
E
oed
, is given by:
G =
E
2(1+)
(2.20a)
K =
E
3(12)
(2.20b)
E
oed
=
(1)E
(12)(1+)
(2.20c)
During the input of material parameters for the Linear Elastic model or the Mohr-Coulomb
model the values of G and E
oed
are presented as auxiliary parameters (alternatives),
calculated from Eq. (2.20). Note that the alternatives are inuenced by the input values
of E and . Entering a particular value for one of the alternatives G or E
oed
results in a
change of the E modulus.
It is possible for the Linear Elastic model and the Mohr-Coulomb model to specify a
stiffness that varies linearly with depth. This can be done by entering the Advanced
parameters window using the Advanced button, as shown in Figure 2.3. Here one may
enter a value for E
increment
which is the increment of stiffness per unit of depth, as indicated
in Figure 2.4.
2-6 PLAXIS-GiD
Together with the input of E
increment
the input of y
re f
becomes relevant. Above y
re f
the
stiffness is equal to E
re f
. Below the stiffness is given by:
E(y) = E
re f
+
_
y
re f
y
_
E
increment
y < y
re f
(2.21)
Figure 2.4 Advanced Parameters window
The Linear Elastic model is usually inappropriate to model the highly non-linear behaviour
of soil, but it is of interest to simulate structural behaviour, such as thick concrete walls
or plates, for which strength properties are usually very high compared with those of soil.
For these applications, the Linear Elastic model will often be selected together with Non-
porous type of material behaviour in order to exclude pore pressures from these structural
elements.
2.4 UNDRAINED EFFECTIVE STRESS ANALYSIS (EFFECTIVE STIFFNESS
PARAMETERS)
In PLAXIS it is possible to specify undrained behaviour in an effective stress analysis using
effective model parameters. This is achieved by identifying the type of material behaviour
(Material type) of a soil layer as Undrained. In this section, it is explained how PLAXIS
deals with this special option.
The presence of pore pressures in a soil body, usually caused by water, contributes to the
total stress level. According to Terzaghi's principle, total stresses can be divided into
effective stresses ' and pore pressures
w
(see also Eq. 2.3). However, water is supposed
not to sustain any shear stress, and therefore the effective shear stresses are equal to the
total shear stresses:
xx
= '
xx
+ p
w
(2.22a)
yy
= '
yy
+ p
w
(2.22b)
2-7
zz
= '
zz
+ p
w
(2.22c)
xy
= '
xy
(2.22d)
yz
= '
yz
(2.22e)
zx
= '
zx
(2.22f)
Note that, similar to the total and the effective stress components, p
w
is considered
negative for pressure.
A further distinction is made between steady state pore stress, p
steady
, and excess pore
stress, p
excess
:
p
w
= p
steady
+ p
excess
(2.23)
Steady state pore pressures are considered to be input data, i.e. generated on the basis
of phreatic levels. This generation of steady state pore pressures is discussed in Section
4.3.6 of the Reference Manual. Excess pore pressures are generated during plastic
calculations for the case of undrained material behaviour or during a consolidation
analysis. Undrained material behaviour and the corresponding calculation of excess pore
pressures are described below.
Since the time derivative of the steady state component equals zero, it follows:
p
w
= p
excess
(2.24)
Hooke's law can be inverted to obtain:
_
_

e
xx

e
yy

e
zz

e
xy

e
yz

e
zx
_
_
=
1
E'
_
_
1 ' ' 0 0 0
' 1 ' 0 0 0
' ' 1 0 0 0
0 0 0 2+2' 0 0
0 0 0 0 2+2' 0
0 0 0 0 0 2+2'
_
_
_
_
'
xx
'
yy
'
zz
'
xy
'
yz
'
zx
_
_
(2.25)
2-8 PLAXIS-GiD
Substituting Eq. (2.22) gives:
_
_

e
xx

e
yy

e
zz

e
xy

e
yz

e
zx
_
_
=
1
E'
_
_
1 ' ' 0 0 0
' 1 ' 0 0 0
' ' 1 0 0 0
0 0 0 2+2' 0 0
0 0 0 0 2+2' 0
0 0 0 0 0 2+2'
_
_
_
_

xx
p
w

yy
p
w

zz
p
w

xy

yz

zx
_
_
(2.26)
Considering slightly compressible water, the rate of excess pore pressure is written as:
p
w
=
Kw
n
_

e
xx
+
e
yy
+
e
zz
_
(2.27)
in which K
w
is the bulk modulus of the water and n is the soil porosity.
The inverted form of Hooke's law may be written in terms of the total stress rates and the
undrained parameters E
u
and
u
:
_
_

e
xx

e
yy

e
zz

e
xy

e
yz

e
zx
_
_
=
1
E
u
_
_
1
u

u
0 0 0
u
1
u
0 0 0
u

u
1 0 0 0
0 0 0 2+2
u
0 0
0 0 0 0 2+2
u
0
0 0 0 0 0 2+2
u
_
_
_
_
'
xx
'
yy
'
zz

xy

yz

zx
_
_
(2.28)
where:
E
u
= 2G(1+
u
)
u
=
' +(1+')
1+2(1+')
(2.29)
=
1
3n
K
w
K'
K' =
E'
3(12')
(2.30)
Hence, the special option for undrained behaviour in PLAXIS is such that the effective
parameters G and are transferred into undrained parameters E
u
and
u
according to
Eqs. (2.29) and (2.30). Note that the index u is used to indicate auxiliary parameters for
undrained soil. Hence, E
u
and
u
should not be confused with E
ur
and
ur
as used to
denote unloading / reloading.
Fully incompressible behaviour is obtained for
u
= 0.5. However, taking
u
= 0.5 leads to
singularity of the stiffness matrix. In fact, water is not fully incompressible, but a realistic
2-9
bulk modulus for water is very large. In order to avoid numerical problems caused by
an extremely low compressibility,
u
is taken as 0.495, which makes the undrained soil
body slightly compressible. In order to ensure realistic computational results, the bulk
modulus of the water must be high compared with the bulk modulus of the soil skeleton,
i.e. K
w
>> n K'. This condition is sufciently ensured by requiring ' 0.35. Users will
get a warning as soon as larger Poisson's ratios are used in combination with undrained
material behaviour.
Consequently, for undrained material behaviour a bulk modulus for water is automatically
added to the stiffness matrix. The value of the bulk modulus is given by:
Kw
n
=
3(
u
')
(12
u
)(1+')
K' = 300
0.495'
1+'
K' > 30K' (2.31)
at least for ' 0.35. In retrospect it is worth mentioning here a review about the
Skempton B-parameter.
Skempton B-parameter
When the Material type (type of material behaviour) is set to Undrained, PLAXIS
automatically assumes an implicit undrained bulk modulus, K
u
, for the soil as a whole
(soil skeleton + water) and distinguishes between total stresses, effective stresses and
excess pore pressures (see Undrained behaviour):
Total stress: p = K
u
Effective stress: p' = (1B)p = K'
Excess pore pressure: p

w
= Bp =
K
w
n

Note that effective stiffness parameters should be entered in the material data set, i.e.
E' and ' and not E
u
and
u
, or the respective stiffness parameters in advanced models.
The undrained bulk modulus is automatically calculated by PLAXIS using Hooke's law of
elasticity:
K
u
=
2G(1+
u
)
3(12
u
)
where G =
E'
2(1+')
and
u
= 0.495 (when using the Standard setting)
or
u
=
3' +B(12')
3B(12')
(when using the Manual setting)
A particular value of the undrained Poisson's ratio,
u
, implies a corresponding reference
bulk stiffness of the pore uid, K
w,re f
/ n:
2-10 PLAXIS-GiD
K
w,re f
n
= K
u
K' where K' =
E'
3(12')
This value of K
w,re f
/ n is generally much smaller than the real bulk stiffness of pure water,
K
0
w
(= 210
6
kN/m
2
).
If the value of Skempton's B-parameter is unknown, but the degree of saturation, S, and
the porosity, n, are known instead, the bulk stiffness of the pore uid can be estimated
from:
K
w
n
=
K
0
w
K
air
SK
air
+(1S)K
0
w
1
n
where K' =
E'
3(12')
and K
air
= 200 kN/m
2
for air under atmospheric pressure. The value of Skempton's B-
parameter can now be calculated from the ratio of the bulk stiffnesses of the soil skeleton
and the pore uid:
B =
1
1+
nK'
K
w
The rate of excess pore pressure is calculated from the (small) volumetric strain rate,
according to:
p
w
=
Kw
n

v
(2.32)
The types of elements used in PLAXIS are sufciently adequate to avoid mesh locking
effects for nearly incompressible materials.
This special option to model undrained material behaviour on the basis of effective model
parameters is available for all material models in the PLAXIS program. This enables
undrained calculations to be executed with effective input parameters, with explicit
distinction between effective stresses and (excess) pore pressures.
Such an analysis requires effective soil parameters and is therefore highly convenient when
such parameters are available. For soft soil projects, accurate data on effective parameters
may not always be available. Instead, in situ tests and laboratory tests may have been
performed to obtain undrained soil parameters. In such situations measured undrained
Young's moduli can be easily converted into effective Young's moduli by:
E' =
2(1+')
3
E
u
(2.33)
For advanced models there is no such direct conversion possible. In that case it is
recommended to estimate the required effective stiffness parameter from the measured
undrained stiffness parameter, then perform a simple undrained test to check the resulting
undrained stiffness and adapt the effective stiffness if needed.
2-11
2.5 UNDRAINED EFFECTIVE STRESS ANALYSIS (EFFECTIVE STRENGTH
PARAMETERS)
In principle, undrained effective stress analysis as described in Section 2.4 can be used in
combination with effective strength parameters ' and c' to model the material's undrained
shear strength. In this case, the development of the pore pressure plays a crucial role
in providing the right effective stress path that leads to failure at a realistic value of
undrained shear strength (c
u
or s
u
). However, note that most soil models are not capable
of providing the right effective stress path in undrained loading. As a result, they will
produce the wrong undrained shear strength if the material strength has been specied on
the basis of effective strength parameters. Another problem is that for undrained materials
effective strength parameters are usually not available from soil investigation data. In
order to overcome these problems, some models allow for a direct input of undrained
shear strength. This approach is described in Section 2.6.
If the user wants to model the material strength of undrained materials using the effective
strength parameters ' and c', this can be done in PLAXIS in the same way as for
drained materials. However, in this case the Material type must be set to Undrained. As
a result, PLAXIS will automatically add the stiffness of water to the stiffness matrix (see
Section 2.4) in order to distinguish between effective stresses and (excess) pore pressures
(= effective stress analysis). The advantage of using effective strength parameters in
undrained loading conditions is that after consolidation a qualitatively increased shear
strength is obtained, although this increased shear strength could also be quantitatively
wrong, for the same reason as explained before.
Shear strength
increase
Shear strength:
(1) MC model
reality (2)
(3) consolidation
(4) MC model
MC failure line
model
realistic
c', '
q
p'
c
u
Figure 2.5 Illustration of stress paths; reality vs. Mohr-Coulomb model
Figure 2.5 illustrates an example using the Mohr-Coulomb model. When the Material type
2-12 PLAXIS-GiD
is set to Undrained, the model will follow an effective stress path where the mean effective
stress, p', remains constant all the way up to the failure (1). It is known that especially soft
soils, like normally consolidated clays and peat, will follow an effective stress path in
undrained loading where p' reduces signicantly (2). As a result, the maximum deviatoric
stress that can be reached in the model is over-estimated in the Mohr-Coulomb model. In
other words, the mobilized shear strength in the model supersedes the available undrained
shear strength. If, at some stress state, the soil is consolidated, the mean effective stress
will increase (3). Upon further undrained loading with the Mohr-Coulomb model, the
observed shear strength will be increased (4) compared to the previous shear strength, but
this increased shear strength may again be unrealistic, especially for soft soils.
On the other hand, advanced models do include, to some extent, the reduction of mean
effective stress in undrained loading, but even when using advanced models it is generally
advised to check the mobilised shear strength in the Output program against the available
(undrained) shear strength when this approach is followed.
Note that whenever the Material type parameter is set to Undrained, effective values must
be entered for the stiffness parameters (Young's modulus E and Poisson ratio in case of
the Mohr-Coulomb model or the respective stiffness parameters in the advanced models.)
2.6 UNDRAINEDEFFECTIVESTRESS ANALYSIS (UNDRAINEDSTRENGTH
PARAMETERS)
For undrained soil layers with a known undrained shear strength prole, PLAXIS offers
for some models the possibility of an undrained effective stress analysis, as described in
Section 2.4 (Material type = Undrained), with direct input of the undrained shear strength,
i.e. setting the friction angle to zero and the cohesion equal to the undrained shear strength
( =
u
= 0
; c = c
u
). Also in this case, distinction is made between pore pressures and
effective stresses. Although the pore pressures and effective stress path may not be fully
correct, the resulting undrained shear strength is not affected, since it is directly specied
as input parameter.
The option to perform an undrained effective stress analysis with undrained strength
properties is only available for the Mohr-Coulomb model, the Hardening Soil model and
the HS small model. Since most soils show an increasing shear strength with depth, it is
possible to specify the increase per unit of depth in PLAXIS in the Advanced Parameters
window.
Note that if the Hardening Soil model or the HS small model is used with = 0
, the
stiffness moduli in the model are no longer stress-dependent and the model exhibits
no compression hardening, although the model retains its separate unloading-reloading
modulus and shear hardening.
Further note that whenever the Material type parameter is set to Undrained, effective
values must be entered for the stiffness parameters (Young's modulus E and Poisson
ratio in case of the Mohr-Coulomb model or the respective stiffness parameters in the
2-13
advanced models.)
2.7 UNDRAINED TOTAL STRESS ANALYSIS (UNDRAINED PARAMETERS)
If, for any reason, it is desired not to use the Undrained option in PLAXIS to perform
an undrained effective stress analysis, one may simulate undrained behaviour using a
total stress analysis with all parameters specied as undrained. In that case, stiffness is
modelled using an undrained Young's modulus E
u
and an undrained Poisson ratio
u
, and
strength is modelled using an undrained shear strength c
u
(s
u
) and =
u
= 0
. Typically,
for the undrained Poisson ratio a value close to 0.5 is selected (between 0.495 and 0.499).
A value of 0.5 exactly is not possible, since this would lead to singularity of the stiffness
matrix.
In PLAXIS it is possible to perform a total stress analysis with undrained parameters if
the Mohr-Coulomb model, the Hardening Soil model or the HS small model is used. In
this case, one should select Non-porous as the Material type (and not Undrained, since
that would also add water stiffness, as described in Section 2.4, which would still give
excess pore pressures!). The disadvantage of the undrained total stress analysis is that
no distinction is made between effective stresses and pore pressures. Hence, all output
referring to effective stresses should now be interpreted as total stresses and all pore
pressures are equal to zero.
Note that in graphical output of stresses the stresses in Non-porous clusters are not plotted.
If one does want graphical output of stresses one should select Drained instead of Non-
porous for the type of material behaviour and make sure that no pore pressures are
generated in these clusters.
Also note that a direct input of undrained shear strength does not automatically give the
increase of shear strength with consolidation.
This type of approach is not possible when using the Soft Soil model or the Soft Soil Creep
model. If the Hardening Soil model or HS small model is used in a total stress analysis
using undrained parameters, i.e. =
u
= 0
, the stiffness moduli in the model are no

longer stress-dependent and the model exhibits no compression hardening, although the
model retains its separate unloading-reloading modulus and shear hardening.
2.8 THE INITIAL PRE-CONSOLIDATION STRESS IN ADVANCED MODELS
When using advanced models in PLAXIS an initial pre-consolidation stress has to be
determined. In the engineering practice it is common to use a vertical pre-consolidation
stress,
p
, but PLAXIS needs an equivalent isotropic pre-consolidation stress, p
eq
p
to
determine the initial position of a cap-type yield surface. If a material is over-consolidated,
information is required about the Over-Consolidation Ratio (OCR), i.e. the ratio of the
greatest effective vertical stress previously reached,
p
(see Figure 2.6), and the in-situ
2-14 PLAXIS-GiD
effective vertical stress, '
0
yy
.
OCR =

p
'
0
yy
(2.34)
=
'
0
yy
'
0
yy
p
OCR
a. Using OCR
'
0
yy

p
POP
b. Using POP
Figure 2.6 Illustration of vertical pre-consolidation stress in relation to the in-situ vertical
stress
It is also possible to specify the initial stress state using the Pre-Overburden Pressure
(POP) as an alternative to prescribing the over-consolidation ratio. The Pre-Overburden
Pressure is dened by:
POP =|
p
'
0
yy
| (2.35)
These two ways of specifying the vertical pre-consolidation stress are illustrated in Figure
2.6.
The pre-consolidation stress
p
is used to compute p
eq
p
which determines the initial
position of a cap-type yield surface in the advanced soil models. The calculation of p
eq
p
is
based on the stress state:
'
1
=
p
and: '
2
= '
3
= K
nc
0

p
(2.36)
Where K
nc
0
is the K0
-value associated with normally consolidated states of stress, which is
an input parameter for the advanced soil models. For the Hardening Soil model the default
parameter settings is such that we have the Jaky formula K
nc
0
1sin.
2.9 ON THE INITIAL STRESSES
In overconsolidated soils the coefcient of lateral earth pressure is larger than for normally
consolidated soils. This effect is automatically taken into account for advanced soil
models when generating the initial stresses using the K
0
-procedure. The procedure that
is followed here is described below.
2-15
-
-
-
-
1
1-
-
'
yy
p
'
0
yy
'
xx
'
0
xx
K
nc
0
ur
ur
Figure 2.7 Overconsolidated stress state obtained from primary loading and subsequent
unloading
Consider a one-dimensional compression test, preloaded to '
yy
=
p
and subsequently
unloaded to '
yy
= '
0
yy
. During unloading the sample behaves elastically and the
incremental stress ratio is, according to Hooke's law, given by (see Figure 2.7):
'
xx
'
yy
=
K
nc
0

p
'
0
xx
p
'
0
yy
=
K
nc
0
OCR'
0
yy
'
0
xx
(OCR1)'
0
yy
=

ur
1
ur
(2.37)
where K
nc
0
is the stress ratio in the normally consolidated state. Hence, the stress ratio of
the overconsolidated soil sample is given by:
'
0
xx
'
0
yy
= K
nc
0
OCR

ur
1
ur
(OCR1) (2.38)
The use of a small Poisson's ratio, as discussed previously, will lead to a relatively large
ratio of lateral stress and vertical stress, as generally observed in overconsolidated soils.
Note that Eq. (2.38) is only valid in the elastic domain, because the formula was derived
from Hooke's law of elasticity. If a soil sample is unloaded by a large amount, resulting in
a high degree of over-consolidation, the stress ratio will be limited by the Mohr-Coulomb
failure condition.
2-16 PLAXIS-GiD
THE MOHR-COULOMB MODEL (PERFECT-PLASTICITY)
3 THE MOHR-COULOMB MODEL (PERFECT-PLASTICITY)
Plasticity is associated with the development of irreversible strains. In order to evaluate
whether or not plasticity occurs in a calculation, a yield function, f , is introduced as a
function of stress and strain. Plastic yielding is related with the condition f = 0. This
condition can often be presented as a surface in principal stress space. A perfectly-plastic
model is a constitutive model with a xed yield surface, i.e. a yield surface that is fully
dened by model parameters and not affected by (plastic) straining. For stress states
represented by points within the yield surface, the behaviour is purely elastic and all strains
are reversible.
3.1 ELASTIC PERFECTLY-PLASTIC BEHAVIOUR
The basic principle of elastoplasticity is that strains and strain rates are decomposed into
an elastic part and a plastic part:
=
e
+
p
=
e
+
p
(3.1)
Hooke's law is used to relate the stress rates to the elastic strain rates. Substitution of Eq.
(3.1) into Hooke's law Eq. (2.19) leads to:
' = D
e

e
= D
e
(
p
) (3.2)
According to the classical theory of plasticity (Hill, 1950), plastic strain rates are
proportional to the derivative of the yield function with respect to the stresses. This means
that the plastic strain rates can be represented as vectors perpendicular to the yield surface.
This classical form of the theory is referred to as associated plasticity. However, for Mohr-
Coulomb type yield functions, the theory of associated plasticity overestimates dilatancy.
Therefore, in addition to the yield function, a plastic potential function g is introduced.
The case g = f is denoted as non-associated plasticity. In general, the plastic strain rates
are written as:

p
=
g
'
(3.3)
in which is the plastic multiplier. For purely elastic behaviour is zero, whereas in the
case of plastic behaviour is positive:
= 0 for: f < 0 or :
f
'
T
D
e
0 (Elasticity) (3.4a)
> 0 for: f = 0 and:
f
'
T
D
e
> 0 (Plasticity) (3.4b)
3-1
'
Figure 3.1 Basic idea of an elastic perfectly plastic model

These equations may be used to obtain the following relationship between the effective
stress rates and strain rates for elastoplasticity (Smith & Grifth, 1982; Vermeer & Borst,
1984):
' =
_
D
e
d
D
e
g
'
f
T
'
D
e
_
(3.5a)
where:
d =
f
'
T
D
e
g
'
(3.5b)
The parameter is used as a switch. If the material behaviour is elastic, as dened by Eq.
(3.4a), the value of is equal to zero, whilst for plasticity, as dened by Eq. (3.4b), the
value of is equal to unity.
The above theory of plasticity is restricted to smooth yield surfaces and does not cover
a multi surface yield contour as present in the Mohr-Coulomb model. For such a yield
surface the theory of plasticity has been extended by Koiter (1960) and others to account
for ow vertices involving two or more plastic potential functions:

p
=
1
g
1
'
+
2
g
2
'
+... (3.6)
Similarly, several quasi independent yield functions ( f
1
, f
2
, ...) are used to determine the
magnitude of the multipliers (
1
,
2
, ...).
3-2 PLAXIS-GiD
3.2 FORMULATION OF THE MOHR-COULOMB MODEL
The Mohr-Coulomb yield condition is an extension of Coulomb's friction law to general
states of stress. In fact, this condition ensures that Coulomb's friction law is obeyed in any
plane within a material element.
The full Mohr-Coulomb yield condition consists of six yield functions when formulated
in terms of principal stresses (see for instance Smith & Grifth,1982):
f
1a
=
1
2
('
2
'
3
) +
1
2
('
2
+'
3
)sin c cos 0 (3.7a)
f
1b
=
1
2
('
3
'
2
) +
1
2
('
3
+'
2
)sin c cos 0 (3.7b)
f
2a
=
1
2
('
3
'
1
) +
1
2
('
3
+'
1
)sin c cos 0 (3.7c)
f
2b
=
1
2
('
1
'
3
) +
1
2
('
1
+'
3
)sin c cos 0 (3.7d)
f
3a
=
1
2
('
1
'
2
) +
1
2
('
1
+'
2
)sin c cos 0 (3.7e)
f
3b
=
1
2
('
2
'
1
) +
1
2
('
2
+'
1
)sin c cos 0 (3.7f)
The two plastic model parameters appearing in the yield functions are the well-known
friction angle and the cohesion c. The condition f
i
= 0 for all yield functions together
(where f
i
is used to denote each individual yield function) represent a hexagonal cone in
principal stress space as shown in Figure 3.2.
-
1
-
2
-
3
Figure 3.2 The Mohr-Coulomb yield surface in principal stress space (c = 0)
3-3
In addition to the yield functions, six plastic potential functions are dened for the Mohr-
Coulomb model:
g
1a
=
1
2
('
2
'
3
) +
1
2
('
2
+'
3
)sin (3.8a)
g
1b
=
1
2
('
3
'
2
) +
1
2
('
3
+'
2
)sin (3.8b)
g
2a
=
1
2
('
3
'
1
) +
1
2
('
3
+'
1
)sin (3.8c)
g
2b
=
1
2
('
1
'
3
) +
1
2
('
1
+'
3
)sin (3.8d)
g
3a
=
1
2
('
1
'
2
) +
1
2
('
1
+'
2
)sin (3.8e)
g
3b
=
1
2
('
2
'
1
) +
1
2
('
2
+'
1
)sin (3.8f)
The plastic potential functions contain a third plasticity parameter, the dilatancy angle
. This parameter is required to model positive plastic volumetric strain increments
(dilatancy) as actually observed for dense soils. A discussion of all of the model
parameters used in the Mohr-Coulomb model is given in the next section.
When implementing the Mohr-Coulomb model for general stress states, special treatment
is required for the intersection of two yield surfaces. Some programs use a smooth
transition from one yield surface to another, i.e. the rounding-off of the corners (see for
example Smith & Grifth,1982). In PLAXIS, however, the exact form of the full Mohr-
Coulomb model is implemented, using a sharp transition fromone yield surface to another.
For a detailed description of the corner treatment the reader is referred to the literature
(Koiter, 1960; van Langen & Vermeer, 1990).
For c > 0, the standard Mohr-Coulomb criterion allows for tension. In fact, allowable
tensile stresses increase with cohesion. In reality, soil can sustain none or only very small
tensile stresses. This behaviour can be included in a PLAXIS analysis by specifying a
tension cut-off. In this case, Mohr circles with positive principal stresses are not allowed.
The tension cut-off introduces three additional yield functions, dened as:
f
4
= '
1
t
0 (3.9a)
f
5
= '
2
t
0 (3.9b)
f
6
= '
3
t
0 (3.9c)
When this tension cut-off procedure is used, the allowable tensile stress,
t
, is, by default,
3-4 PLAXIS-GiD
taken equal to zero, but this value can be changed by the user. For these three yield
functions an associated ow rule is adopted.
For stress states within the yield surface, the behaviour is elastic and obeys Hooke's law
for isotropic linear elasticity, as discussed in Section 2.3. Hence, besides the plasticity
parameters c, , and , input is required on the elastic Young's modulus E and Poisson's
ratio .
3.3 BASIC PARAMETERS OF THE MOHR-COULOMB MODEL
The Mohr-Coulomb model requires a total of ve parameters, which are generally familiar
to most geotechnical engineers and which can be obtained frombasic tests on soil samples.
These parameters with their standard units are listed below:
E : Young's modulus [kN/m
2
]
: Poisson's ratio [-]
: Friction angle [
]
c : Cohesion [kN/m
2
]
: Dilatancy angle [
]
Figure 3.3 Parameter tab sheet for Mohr-Coulomb model
3-5
Instead of using the Young's modulus as a stiffness parameter, alternative stiffness
parameters can be entered. These parameters with their standard units are listed below:
G
re f
: Shear modulus [kN/m
2
]
E
oed
: Oedometer modulus [kN/m
2
]
Young's modulus (E)
PLAXIS uses the Young's modulus as the basic stiffness modulus in the elastic model and
the Mohr-Coulomb model, but some alternative stiffness moduli are displayed as well.
A stiffness modulus has the dimension of stress. The values of the stiffness parameter
adopted in a calculation require special attention as many geomaterials show a non-linear
behaviour from the very beginning of loading. In soil mechanics the initial slope (tangent
modulus) is usually indicated as E
0
and the secant modulus at 50% strength is denoted as
E
50
(see Figure 3.4). For materials with a large linear elastic range it is realistic to use E
0
,
but for loading of soils one generally uses E
50
. Considering unloading problems, as in the
case of tunnelling and excavations, one needs E
ur
instead of E
50
.
For soils, both the unloading modulus, E
ur
, and the rst loading modulus, E
50
, tend to
increase with the conning pressure. Hence, deep soil layers tend to have greater stiffness
than shallow layers. Moreover, the observed stiffness depends on the stress path that
is followed. The stiffness is much higher for unloading and reloading than for primary
loading. Also, the observed soil stiffness in terms of a Young's modulus may be lower for
(drained) compression than for shearing. Hence, when using a constant stiffness modulus
to represent soil behaviour one should choose a value that is consistent with the stress
level and the stress path development. Note that some stress-dependency of soil behaviour
is taken into account in the advanced models in PLAXIS which are described in Chapter
4 and 5. For the Mohr-Coulomb model, PLAXIS offers a special option for the input of
a stiffness increasing with depth (see Section 3.4). Note that for material data sets where
the type of material behaviour is set to undrained, the Young's modulus has the meaning
of an effective Young's modulus, whereas the Undrained setting takes care of the low
compressibility.
Poisson's ratio ()
Standard drained triaxial tests may yield a signicant rate of volume decrease at the very
beginning of axial loading and, consequently, a low initial value of Poisson's ratio (
0
).
For some cases, such as particular unloading problems, it may be realistic to use such a
low initial value, but in general when using the Mohr-Coulomb model the use of a higher
value is recommended.
The selection of a Poisson's ratio is particularly simple when the elastic model or Mohr-
Coulomb model is used for gravity loading. For this type of loading PLAXIS should give
realistic ratios of K
0
=
h
/
.
As both models will give the well-known ratio of
h
/
= / (1) for one-dimensional

compression it is easy to select a Poisson's ratio that gives a realistic value of K
0
. Hence,
3-6 PLAXIS-GiD
1 1
strain -
1
E
0
E
50
|
1
3
|
Figure 3.4 Denition of E
0
and E
50
for standard drained triaxial test results
is evaluated by matching K
0
. This subject is treated more extensively in Section 4.1.6
of the Reference Manual, which deals with initial stress generation. In many cases one
will obtain values in the range between 0.3 and 0.4. In general, such values can also
be used for loading conditions other than one-dimensional compression. For unloading
conditions, however, it is more appropriate to use values in the range between 0.15 and
0.25. Please note that in this way it is not possible to create K
0
values larger than 1, as
may be observed in highly overconsolidated stress states.
Further note that for material data sets where the type of material behaviour is set to
Undrained, the Poisson's ratio has the meaning of an effective Poisson's ratio, whereas the
Undrained setting takes care of the low compressibility. To ensure that the soil skeleton
is much more compressible than the pore water, the effective Poisson's ratio should be
smaller than 0.35 for undrained behaviour.
Cohesion (c)
The cohesive strength has the dimension of stress. In the Mohr-Coulomb model, the
cohesion parameter may be used to model the effective cohesion c' of the soil, in
combination with a realistic effective friction angle ' (see Figure 3.5a). This may not
only be done for drained soil behaviour, but also if the type of material behaviour is
set to Undrained, as in both cases PLAXIS will perform an effective stress analysis.
Alternatively, the cohesion parameter may be used to model the undrained shear strength
c
u
(or s
u
) of the soil, in combination with =
u
= 0 (see Figure 3.5b).
The disadvantage of using effective strength parameters c' and ' in combination with the
material type being set to Undrained is that the undrained shear strength as obtained from
the model may deviate from the undrained shear strength in reality because of differences
in the actual stress path being followed. In this respect, advanced soil models generally
perform better than the Mohr-Coulomb model, but in all cases it is recommended to
compare the resulting stress state in all calculation phases with the present shear strength
in reality (|
1
3
| 2c
u
).
3-7
On the other hand, the advantage of using effective strength parameters is that the
change in shear strength with consolidation is obtained automatically, although it is still
recommended to check the resulting stress state after consolidation.
The advantage of using the cohesion parameter to model undrained shear strength (in
combination with = 0) is that the user has direct control over the shear strength,
independent of the actual stress state and stress path followed. Please note that this option
may not be appropriate when using advanced soil models.
PLAXIS can handle cohesionless sands (c = 0), but some options will not perform well.
To avoid complications, non-experienced users are advised to enter at least a small value
(use c > 0.2 kPa). Please note that a positive value for the cohesion will lead to a tensile
strength, which may be unrealistic for soils. The Tension cut-off option may be used to
reduce the tensile strength. See ?? in the Reference Manualfor more details.
PLAXIS offers a special option for the input of layers in which the cohesion increases with
depth (see Section 3.4).
Friction angle ()
The friction angle (phi) is entered in degrees. In general the friction angle is used to
model the effective friction of the soil, in combination with an effective cohesion c' (Figure
3.5a). This may not only be done for drained soil behaviour, but also if the type of material
behaviour is set to Undrained, since in both cases PLAXIS will perform an effective stress
analysis. Alternatively, the soil strength is modelled by setting the cohesion parameter
equal to the undrained shear strength of the soil, in combination with = 0 (Figure 3.5b).
3
shear
stress
stress
normal
c'
'
0
a. Using effective strength parameters
3
shear
stress
stress
normal
c = c
u
= 0
b. Using undrained strength parameters
Figure 3.5 Stress circles at yield; one touches Coulomb's envelope.
High friction angles, as sometimes obtained for dense sands, will substantially increase
plastic computational effort. The computing time increases more or less exponentially
with the friction angle. Hence, high friction angles should be avoided when performing
preliminary computations for a particular project. Computing time tends to become
large when friction angles in excess of 35 degrees are used. The friction angle largely
determines the shear strength as shown in Figure 3.5 by means of Mohr's stress circles.
A more general representation of the yield criterion is shown in Figure 3.6. The Mohr-
Coulomb failure criterion proves to be better for describing soil strength for general stress
3-8 PLAXIS-GiD
3
Figure 3.6 Failure surface in principal stress space for cohesionless soil
states than the Drucker-Prager approximation.
Dilatancy angle ()
The dilatancy angle, (psi), is specied in degrees. Apart from heavily over-consolidated
layers, clay soils tend to show little dilatancy ( 0). The dilatancy of sand depends
on both the density and on the friction angle. In general the dilatancy angle of soils is
much smaller than the friction angle. For quartz sands the order of magnitude is
30
. For -values of less than 30
, however, the angle of dilatancy is mostly zero. A

small negative value for is only realistic for extremely loose sands. In the Hardening
Soil model or Hardening Soil model with small-strain stiffness the end of dilatancy, as
generally observed when the soil reaches the critical state, can be modelled using the
Dilatancy cut-off. For further information about the link between the friction angle and
dilatancy, see Bolton (1986).
A positive dilatancy angle implies that in drained conditions the soil will continue to dilate
as long as shear deformation occurs. This is clearly unrealistic, as most soils will reach
a critical state at some point and further shear deformation will occur without volume
changes. In undrained conditions a positive dilatancy angle, combined with the restriction
on volume changes, leads to a generation of tensile pore pressures. In an undrained
effective stress analysis therefore the strength of the soil may be overestimated.
When the soil strength is modelled as c =c
u
(s
u
) and =0, the dilatancy angle must be set
to zero. Great care must be taken when using a positive value of dilatancy in combination
with material type set to Undrained. In that case the model will show unlimited soil
strength due to suction.
3-9
Shear modulus (G)
The shear modulus, G, has the dimension of stress. According to Hooke's law, the
relationship between Young's modulus E and the shear modulus is given by (see Eq.
2.20a):
G =
E
2(1+)
(3.10)
Entering a particular value for one of the alternatives G or E
oed
results in a change of the
E modulus.
Oedometer modulus (E
oed
)
The oedometer modulus, E
oed
, has the dimension of stress. According to Hooke's law,
the relationship between Young's modulus E and the shear modulus is given by (see Eq.
2.20c):
E
oed
=
(1)E
(12)(1+)
(3.11)
Entering a particular value for one of the alternatives G or E
oed
results in a change of the
E modulus.
3.4 ADVANCED PARAMETERS OF THE MOHR-COULOMB MODEL
When using the Mohr-Coulomb model, the Advanced subtree can be expanded (see Figure
3.7). The advanced features comprise the increase of stiffness and cohesive strength with
depth and the use of a tension cut-off. In fact, the latter option is used by default, but it
may be deactivated here, if desired.
Increase of stiffness (E
increment
)
In real soils, the stiffness depends signicantly on the stress level, which means that
the stiffness generally increases with depth. When using the Mohr-Coulomb model, the
stiffness is a constant value. In order to account for the increase of the stiffness with depth
the E
increment
-value may be used, which is the increase of the Youngs modulus per unit
of depth (expressed in the unit of stress per unit depth). At the level given by the y
re f
parameter, and all levels above this level, the stiffness is equal to the reference Youngs
modulus, E
re f
, as entered in the Parameters tab sheet. The actual value of Youngs
modulus in the stress points is obtained from the reference value and E
increment
. Note that
during calculations a stiffness increasing with depth does not change as a function of the
stress state.
3-10 PLAXIS-GiD
Figure 3.7 Advanced Parameters Mohr-Coulomb window
Increase of cohesion (c
increment
)
PLAXIS offers an advanced option for the input of clay layers in which the cohesion
increases with depth. In order to account for the increase of the cohesion with depth the
c
increment
-value may be used, which is the increase of cohesion per unit of depth (expressed
in the unit of stress per unit depth). At the level given by the y
re f
parameter, and all levels
above this level, the cohesion is equal to the (reference) cohesion, c
re f
, as entered in the
Parameters tab sheet. The actual value of cohesion in the stress points is obtained from
the reference value and c
increment
.
Tension cut-off
In some practical problems an area with tensile stresses may develop. According to the
Coulomb envelope shown in Figure ?? this is allowed when the shear stress (radius
of Mohr circle) is sufciently small. However, the soil surface near a trench in clay
sometimes shows tensile cracks. This indicates that soil may also fail in tension instead
of in shear. Such behaviour can be included in a PLAXIS analysis by selecting the tension
cut-off. In this case Mohr circles with positive principal stresses are not allowed. When
selecting the tension cut-off the allowable tensile strength may be entered. For the Mohr-
Coulomb model and the Hardening Soil model the tension cut-off is, by default, selected
with a tensile strength of zero.
3-11
3-12 PLAXIS-GiD
THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)
4 THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)
In contrast to an elastic perfectly-plastic model, the yield surface of a hardening plasticity
model is not xed in principal stress space, but it can expand due to plastic straining.
Distinction can be made between two main types of hardening, namely shear hardening
and compression hardening. Shear hardening is used to model irreversible strains due to
primary deviatoric loading. Compression hardening is used to model irreversible plastic
strains due to primary compression in oedometer loading and isotropic loading. Both
types of hardening are contained in the present model.
The Hardening Soil model is an advanced model for simulating the behaviour of different
types of soil, both soft soils and stiff soils, Schanz (1998). When subjected to primary
deviatoric loading, soil shows a decreasing stiffness and simultaneously irreversible
plastic strains develop. In the special case of a drained triaxial test, the observed
relationship between the axial strain and the deviatoric stress can be well approximated by
a hyperbola. Such a relationship was rst formulated by Kondner (1963) and later used in
the well-known hyperbolic model (Duncan & Chang, 1970). The Hardening Soil model,
however, supersedes the hyperbolic model by far: Firstly by using the theory of plasticity
rather than the theory of elasticity, secondly by including soil dilatancy and thirdly by
introducing a yield cap. Some basic characteristics of the model are:
Stress dependent stiffness according to a power law. Input parameter m
Plastic straining due to primary deviatoric loading. Input parameter E
re f
50
Plastic straining due to primary compression. Input parameter E
re f
oed
Elastic unloading / reloading. Input parameters E
re f
ur
,
ur
Failure according to the Mohr-Coulomb model. Parameters c, and
A basic feature of the present Hardening Soil model is the stress dependency of soil
stiffness. For oedometer conditions of stress and strain, the model implies for example
the relationship E
oed
= E
re f
oed
_
/
p
re f
_
m
. In the special case of soft soils it is realistic to
use m = 1. In such situations there is also a simple relationship between the modied
compression index
, as used in the PLAXIS Soft Soil Creep model and the oedometer
loading modulus (see also Section 7.7).
E
re f
oed
=
p
re f
=

(1+e
0
)
where p
re f
is a reference pressure. Here we consider a tangent oedometer modulus at
a particular reference pressure p
re f
. Hence, the primary loading stiffness relates to the
modied compression index
or to the standard Cam-Clay compression index .

Similarly, the unloading-reloading modulus relates to the modied swelling index
or
to the standard Cam-Clay swelling index . There is the approximate relationship:
4-1
E
re f
ur

2p
re f
=

(1+e
0
)
Again, this relationship applies in combination with the input value m = 1.
4.1 HYPERBOLIC RELATIONSHIP FOR STANDARD DRAINED TRIAXIAL
TEST
A basic idea for the formulation of the Hardening Soil model is the hyperbolic relationship
between the vertical strain,
1
, and the deviatoric stress, q, in primary triaxial loading.
Here standard drained triaxial tests tend to yield curves that can be described by:
1
=
1
E
i
q
1q/
q
a
for: q < q
f
(4.1)
Where q
a
is the asymptotic value of the shear strength and E
i
the initial stiffness. E
i
is
related to E
50
by:
E
i
=
2E
50
2R
f
(4.2)
This relationship is plotted in Figure 4.1. The parameter E
50
is the conning stress
dependent stiffness modulus for primary loading and is given by the equation:
E
50
= E
re f
50
_
ccos '
3
sin
ccos +
p
re f
sin
_
m
(4.3)
where E
re f
50
is a reference stiffness modulus corresponding to the reference conning
pressure p
re f
. In PLAXIS, a default setting p
re f
= 100 stress units is used. The actual
stiffness depends on the minor principal stress, '
3
, which is the conning pressure in
a triaxial test. Please note that '
3
is negative for compression. The amount of stress
dependency is given by the power m. In order to simulate a logarithmic compression
behaviour, as observed for soft clays, the power should be taken equal to 1.0. Janbu (1963)
reports values of m around 0.5 for Norwegian sands and silts, whilst von Soos (1990)
reports various different values in the range 0.5 < m < 1.0.
The ultimate deviatoric stress, q
f
, and the quantity q
a
in Eq. (4.1) are dened as:
q
f
= (ccot '
3
)
2sin
1sin
and: q
a
=
q
f
R
f
(4.4)
Again it is remarked that '
3
is usually negative. The above relationship for q
f
is derived
from the Mohr-Coulomb failure criterion, which involves the strength parameters c and
. As soon as q = q
f
, the failure criterion is satised and perfectly plastic yielding occurs
as described by the Mohr-Coulomb model.
4-2 PLAXIS-GiD
The ratio between q
f
and q
a
is given by the failure ratio R
f
, which should obviously be
smaller than 1. In PLAXIS, R
f
= 0.9 is chosen as a suitable default setting.
For unloading and reloading stress paths, another stress-dependent stiffness modulus is
used:
E
ur
= E
re f
ur
_
ccos '
3
sin
ccos +
p
re f
sin
_
m
(4.5)
whereE
re f
ur
is the reference Young's modulus for unloading and reloading, corresponding
to the reference pressure p
re f
. In many practical cases it is appropriate to set E
re f
ur
equal to
3E
re f
50
; this is the default setting used in PLAXIS.
1
1
1
deviatoric stress
|
1
3
|
q
a
q
f
E
i
E
50
E
ur
asymptote
failure line
axial strain -
1
Figure 4.1 Hyperbolic stress-strain relation in primary loading for a standard drained
triaxial test
4.2 APPROXIMATION OF HYPERBOLA BY THE HARDENING SOIL
MODEL
For the sake of convenience, restriction is made here to triaxial loading conditions with
'
2
= '
3
and '
1
being the major compressive stress. Moreover, it is assumed that q
< q
f
, as also indicated in Figure 4.1. It should also be realised that compressive stress
and strain are considered negative. For a more general presentation of the Hardening Soil
model the reader is referred to Schanz, Vermeer & Bonnier (1999). In this section it will
be shown that this model gives virtually the hyperbolic stress strain curve of Eq. (4.1)
when considering stress paths of standard drained triaxial tests. Let us rst consider the
corresponding plastic strains. This stems from a shear hardening yield function of the
form:
f = f
p
(4.6)
4-3
where f is a function of stress and
p
is a function of plastic strains:
f =
2
E
i
q
1q/q
a
2q
E
ur
p
=(2
p
1

p
) 2
p
1
(4.7)
with q, q
a
, E
i
and E
ur
as dened by Eq. (4.1) to Eq. (4.5), whilst the superscript p is used
to denote plastic strains. For hard soils, plastic volume changes (
p
v
) tend to be relatively
small and this leads to the approximation
p
2
p
1
. The above denition of the strain-
hardening parameter
p
will be referred to later.
An essential feature of the above denitions for f is that it matches the well-known
hyperbolic law Eq. (4.1). For checking this statement, one has to consider primary
loading, as this implies the yield condition f = 0. For primary loading, it thus yields
p
=
f and it follows from Eq. (4.6) that:
p
1

1
2
f =
1
E
i
q
1q/q
a
q
E
ur
(4.8)
In addition to the plastic strains, the model accounts for elastic strains. Plastic strains
develop in primary loading alone, but elastic strains develop both in primary loading and
unloading / reloading. For drained triaxial test stress paths with '
2
= '
3
= constant,
the elastic Young's modulus E
ur
remains constant and the elastic strains are given by the
equations:
e
1
=
q
E
ur
e
2
=
e
3
=
ur
q
E
ur
(4.9)
where
ur
is the unloading / reloading Poisson's ratio. Here it should be realised that
restriction is made to strains that develop during deviatoric loading, whilst the strains that
develop during the very rst stage of the test (isotropic compression with consolidation)
are not considered.
For the deviatoric loading stage of the triaxial test, the axial strain is the sum of an elastic
component given by Eq. (4.9) and a plastic component according to Eq. (4.8). Hence, it
follows that:
1
=
e
1
p
1

1
E
i
q
1q/
q
a
(4.10)
This relationship holds exactly in absence of plastic volume strains, i.e. when
p
v
= 0.
In reality, plastic volumetric strains will never be precisely equal to zero, but for hard soils
plastic volume changes tend to be small when compared with the axial strain so that this
formulation yields a hyperbolic stress-strain curve under triaxial testing conditions.
For a given constant value of the hardening parameter,
p
, the yield condition f = 0, can
be visualised in p' - q-plane by means of a yield locus. When plotting such yield loci, one
has to use Eq. (4.7) as well as Eqs. (4.3) and (4.5) for E
50
and E
ur
respectively. Because
of the latter expressions, the shape of the yield loci depends on the exponent m. For m = 1,
4-4 PLAXIS-GiD
straight lines are obtained, but slightly curved yield loci correspond to lower values of the
exponent. Figure 4.2 shows the shape of successive yield loci for m = 0.5, being typical
for hard soils.
Mean effective stress
|
1
3
|
deviatoric stress
Mohr-Coulomb failure line
Figure 4.2 Successive yield loci for various constant values of the hardening parameter
p
4.3 PLASTIC VOLUMETRIC STRAIN FOR TRIAXIAL STATES OF STRESS
Having presented a relationship for the plastic shear strain,
p
, attention is now focused
on the plastic volumetric strain,
p
v
. As for all plasticity models, the Hardening Soil model
involves a relationship between rates of plastic strain, i.e. a relationship between
p
v
and

p
. This shear hardening ow rule has the linear form:

p
v
= sin
m

p
(4.11)
Clearly, further detail is needed by specifying the mobilised dilatancy angle
m
. For the
present model, the following is considered:
For sin
m
< 3/4sin :
m
= 0
For sin
m
3/4sin and > 0 sin
m
= max
_
sin
m
sin
cv
1sin
m
sin
cv
, 0
_
(4.12)
For sin
m
3/4sin and 0
m
=
If = 0
m
= 0
where
c
is the critical state friction angle, being a material constant independent of
density, and
m
is the mobilised friction angle:
sin
m
=
'
1
'
3
'
1
+'
3
2ccot
(4.13)
4-5
The above equations are a small adaptation from the well-known stress-dilatancy theory
by Rowe (1962), as explained by Schanz & Vermeer (1996). The mobilised dilatancy
angle,
m
, follows Rowe's theory for larger values of the mobilised friction angle, as
long as this results in a positive value of
m
. For small mobilised friction angles and for
negative values of
m
, as computed by Rowe's formula (as long as the dilatancy angle
is positive),
m
is taken zero. Furthermore, in all cases when = 0,
m
is set equal to
zero.
The essential property of the stress-dilatancy theory is that the material contracts for small
stress ratios
m
<
c
, whilst dilatancy occurs for high stress ratios
m
>
c
. At failure,
when the mobilised friction angle equals the failure angle, , it is found from Eq. (4.12)
that:
sin =
sin sin
cv
1sin sin
cv
(4.14a)
or equivalently:
sin
cv
=
sin sin
1sin sin
(4.14b)
Hence, the critical state angle can be computed from the failure angles and . PLAXIS
performs this computation automatically and therefore users do not need to specify a value
for
c
.
Instead, one has to provide input data on the ultimate friction angle, , and the ultimate
dilatancy angle, .
4.4 PARAMETERS OF THE HARDENING SOIL MODEL
Some parameters of the present hardening model coincide with those of the non-hardening
Mohr-Coulomb model. These are the failure parameters c, and .
Failure parameters as in Mohr-Coulomb model (see Section 3.3):
c : (Effective) cohesion [kN/m
2
]
: (Effective) angle of internal friction [
]
: Angle of dilatancy [
]
Basic parameters for soil stiffness:
E
re f
50
: Secant stiffness in standard drained triaxial test [kN/m
2
]
E
re f
oed
: Tangent stiffness for primary oedometer loading [kN/m
2
]
E
re f
ur
: Unloading / reloading stiffness (default E
re f
ur
= 3E
re f
50
) [kN/m
2
]
4-6 PLAXIS-GiD
Figure 4.3 Basic parameters for the Hardening Soil model
m : Power for stress-level dependency of stiffness [-]
Advanced parameters (it is advised to use the default setting):
ur
: Poisson's ratio for unloading-reloading (default
ur
= 0.2) [-]
p
re f
: Reference stress for stiffnesses (default p
re f
= 100 stress
units)
[kN/m
2
]
K
nc
0
: K
0
-value for normal consolidation (default K
nc
0
= 1sin) [-]
R
f
: Failure ratio q
f
/ q
a
(default R
f
= 0.9) (see Figure 4.1) [-]
tension
: Tensile strength (default
tension
= 0 stress units) [kN/m
2
]
c
increment
: As in Mohr-Coulomb model (default c
increment
= 0) [kN/m
3
]
Instead of entering the basic parameters for soil stiffness, alternative parameters can be
entered. These parameters are listed below:
C
c
: Compression index [-]
C
s
: Swelling index or reloading index [-]
e
init
: Initial void ratio [-]
4-7
Stiffness moduli E
re f
50
, E
re f
oed
& E
re f
ur
and power m
The advantage of the Hardening Soil model over the Mohr-Coulomb model is not only the
use of a hyperbolic stress-strain curve instead of a bi-linear curve, but also the control of
stress level dependency. When using the Mohr-Coulomb model, the user has to select
a xed value of Young's modulus whereas for real soils this stiffness depends on the
stress level. It is therefore necessary to estimate the stress levels within the soil and use
these to obtain suitable values of stiffness. With the Hardening Soil model, however, this
cumbersome selection of input parameters is not required.
Instead, a stiffness modulus E
re f
50
is dened for a reference minor principal stress of
'
3
= p
re f
. As a default value, the program uses p
re f
= 100 stress units.
As some PLAXIS users are familiar with the input of shear moduli rather than the above
stiffness moduli, shear moduli will now be discussed. Within Hooke's law of isotropic
elasticity conversion between E and G goes by the equation E = 2 (1+ ) G. As E
ur
is a
real elastic stiffness, one may thus write E
ur
= 2 (1+ ) G
ur
, where G
ur
is an elastic shear
modulus. Please note that PLAXIS allows for the input of E
ur
and
ur
but not for a direct
input of G
ur
. In contrast to E
ur
, the secant modulus E
50
is not used within a concept of
elasticity. As a consequence, there is no simple conversion from E
50
to G
50
.
In contrast to elasticity based models, the elastoplastic Hardening Soil model does
not involve a xed relationship between the (drained) triaxial stiffness E
50
and the
oedometer stiffness E
oed
for one-dimensional compression. Instead, these stiffnesses can
be inputted independently. Having dened E
50
by Eq. (4.3), it is now important to dene
the oedometer stiffness. Here we use the equation:
E
oed
= E
re f
oed
_
_
_
_
ccos
'
3
K
nc
0
sin
ccos +
p
re f
sin
_
_
_
_
m
(4.15)
-
-
1
1
p
re f
E
re f
oed
1
Figure 4.4 Denition of E
re f
oed
in oedometer test results
4-8 PLAXIS-GiD
where E
oed
is a tangent stiffness modulus as indicated in Figure 4.4.
Hence, E
re f
oed
is a tangent stiffness at a vertical stress of '
1
=
'
3
K
nc
0
= p
re f
. Note that we
basically use '
1
rather than '
3
and that we consider primary loading.
Alternative stiffness parameters
The alternative stiffness parameters can be calculated from the stiffness parameters and
the initial void ratio
. The relationship between these parameters and the compression

index is given by:
C
c
=
2.3(1+e
init
) p
re f
E
re f
oed
(4.16)
Changing the value of C
c
will change both the parameters E
50
and E
oed
.
The relationship between these parameters and the swelling index is given by:
C
s
=
2.3(1+e
init
)(1+)(12) p
re f
(1)E
re f
ur
(4.17)
Changing the value of C
s
will change the stiffness parameter E
ur
. The initial void ratio
e
init
can also be dened in the Advanced General Properties window.
Advanced parameters
Realistic values of
ur
are about 0.2 and this value is thus used as a default setting, as
indicated in Figure 4.5.
In contrast to the Mohr-Coulomb model, K
nc
0
is not simply a function of Poisson's ratio,
but an independent input parameter. As a default setting PLAXIS uses the correlation
K
nc
0
= 1sin. It is suggested to maintain this value as the correlation is quite realistic.
However, users do have the possibility to select different values. All possible different
input values for K
nc
0
cannot be accommodated for. Depending on other parameters, such
as E
re f
50
, E
re f
oed
, E
re f
ur
and
ur
, there happens to be a certain range of valid K
nc
0
-values. K
nc
0
values outside this range are rejected by PLAXIS. On inputting values, the program shows
the nearest possible value that will be used in the computations.
Dilatancy cut-off
After extensive shearing, dilating materials arrive in a state of critical density where
dilatancy has come to an end, as indicated in Figure 4.6. This phenomenon of soil
behaviour can be included in the Hardening Soil model by means of a dilatancy cut-off.
In order to specify this behaviour, the initial void ratio, e
init
, and the maximum void ratio,
e
max
, of the material must be entered as general parameters. As soon as the volume change
In the PLAXIS material database, these alternative parameters depend on the initial void ratio. In reality, these
parameters depend on the void ratio, which is not a constant.
4-9
Figure 4.5 Advanced Parameters window
results in a state of maximum void, the mobilised dilatancy angle,
m
, is automatically set
back to zero, as indicated in Figure 4.6.
for e < e
max
: sin
m
=
sin
m
sin
cv
1sin
m
sin
cv
where: sin
c
=
sin sin
1sin sin
(4.18a)
for e e
max
:
m
= 0 (4.18b)
The void ratio is related to the volumetric strain,
by the relationship:
init
v
_
= ln
_
1+e
1e
init
_
(4.19)
where an increment of
is positive for dilatancy.

The initial void ratio, e
init
, is the in-situ void ratio of the soil body. The maximum void
ratio is the void ratio of the material in a state of critical void (critical state). As soon as
the maximum void ratio is reached, the dilatancy angle is set to zero. The minimum void
ratio, e
min
, of a soil can also be inputted, but this general soil parameter is not used within
the context of the Hardening Soil model.
Please note that the selection of the dilatancy cut-off and the input of void ratios are done
in the General tab sheet of the Material data set window and not in the Parameters tab
4-10 PLAXIS-GiD
1 sin
2 sin
dilatancy cut - off OFF
dilatancy cut - off ON
1
maximum porosity reached
Figure 4.6 Resulting strain curve for a standard drained triaxial test when including
dilatancy cut-off
Figure 4.7 Advanced General Properties window
sheet. The selection of the Dilatancy cut-off is only available when the Hardening Soil
modelor the HSsmall model has been selected. By default, the Dilatancy cut-off is not
active.
4.5 ON THE CAP YIELD SURFACE IN THE HARDENING SOIL MODEL
Shear hardening yield surfaces as indicated in Figure 4.2 do not explain the plastic volume
strain that is measured in isotropic compression. A second type of yield surface must
therefore be introduced to close the elastic region for compressive (compaction hardening)
stress paths. Without such a cap type yield surface it would not be possible to formulate a
model with independent input of both E
re f
50
and E
re f
oed
. The triaxial modulus largely controls
the shear yield surface and the oedometer modulus controls the cap yield surface. In
fact, E
re f
50
largely controls the magnitude of the plastic strains that are associated with
the shear yield surface. Similarly, E
re f
oed
is used to control the magnitude of plastic strains
that originate from the yield cap. In this section the yield cap will be described in full
4-11
detail. To this end we consider the denition of the cap yield surface:
f
c
=
q
2
2
+ p'
2
p
2
p
(4.20)
where is an auxiliary model parameter that relates to K
nc
0
as will be discussed later.
Further more we have p' = ('
1
+'
2
+'
3
)/3 and q = '
1
+ ( 1)'
2
'
3
with
= (3 +sin)/(3 sin). q is a special stress measure for deviatoric stresses. In the
special case of triaxial compression (-'
1
> -'
2
= -'
3
) it yields q = ('
1
'
3
) and for
triaxial extension ('
1
='
2
>'
3
) q reduces to q =('
1
'
3
). The magnitude
of the yield cap is determined by the isotropic pre-consolidation stress p
p
. The hardening
law relating p
p
to volumetric cap strain
pc
v
is:
pc
v
=

1m
_
p
p
p
re f
_
1m
(4.21)
The volumetric cap strain is the plastic volumetric strain in isotropic compression. In
addition to the well known constants m and p
re f
there is another model constant . Both
and are cap parameters, but these are not used as direct input parameters. Instead,
there are relationships of the form:
K
nc
0
(default :K
nc
0
= 1sin)
E
re f
oed
(default : E
re f
oed
= E
re f
50
)
such that K
nc
0
and E
re f
oed
can be used as input parameters that determine the magnitude of
and respectively. For understanding the shape of the yield cap, it should rst of all be
realised that it is an ellipse in p - q-plane, as indicated in Figure 4.8.
The ellipse has length p
p
on the p-axis and p
p
on the q-axis. Hence, p
p
determines its
magnitude and its aspect ratio. High values of lead to steep caps underneath the Mohr-
Coulomb line, whereas small -values dene caps that are much more pointed around the
p-axis. The ellipse is used both as a yield surface and as a plastic potential. Hence:

pc
=
f
c
with: =

2p'
_
p
p
p
re f
_
m
p
p
p
re f
(4.22)
This expression for is derived from the yield condition f
c
= 0 and Eq. (4.21) for p
p
.
Input data on initial p
p
-values is provided by means of the PLAXIS procedure for initial
stresses. Here, p
p
is either computed from the inputted over-consolidation ratio (OCR) or
the pre-overburden pressure (POP) (see Section 2.8).
For understanding the yield surfaces in full detail, one should consider both Figure 4.8
and Figure 4.9. The rst gure shows simple yield lines, whereas the second one depicts
yield surfaces in principal stress space. Both the shear locus and the yield cap have the
hexagonal shape of the classical Mohr-Coulomb failure criterion. In fact, the shear yield
4-12 PLAXIS-GiD
q
p
p
ccot
elastic region
p
p
p
Figure 4.8 Yield surfaces of Hardening Soil model in p - q-plane. The elastic region can
be further reduced by means of a tension cut-off
3
Figure 4.9 Representation of total yield contour of the Hardening Soil model in principal
stress space for cohesionless soil
locus can expand up to the ultimate Mohr-Coulomb failure surface. The cap yield surface
expands as a function of the pre-consolidation stress p
p
.
4-13
4-14 PLAXIS-GiD
THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS (HSSMALL)
5 THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS
(HSSMALL)
The original Hardening Soil model assumes elastic material behaviour during unloading
and reloading. However, the strain range in which soils can be considered truly elastic,
i.e. where they recover from applied straining almost completely, is very small. With
increasing strain amplitude, soil stiffness decays nonlinearly. Plotting soil stiffness against
log(strain) yields characteristic S-shaped stiffness reduction curves. Figure 5.1 gives an
example of such a stiffness reduction curve. It outlines also the characteristic shear
strains that can be measured near geotechnical structures and the applicable strain
ranges of laboratory tests. It turns out that at the minimum strain which can be reliably
measured in classical laboratory tests, i.e. triaxial tests and oedometer tests without special
instrumentation, soil stiffness is often decreased to less than half its initial value.
Shear strain [-] g
S
Dynamic methods
Local gauges
Conventional soil testing
1e
-6
1e
-5
1e
-4
1e
-3
1e
-2
1e
-1
0
Retaining walls
Tunnels
Foundations
Small strains
Larger strains
Very
small
strains
S
h
e
a
r

m
o
d
u
l
u
s

G
/
G
[
-
]
0
1
Figure 5.1 Characteristic stiffness-strain behaviour of soil with typical strain ranges for
laboratory tests and structures (after Atkinson & Sallfors (1991))
The soil stiffness that should be used in the analysis of geotechnical structures is not the
one that relates to the strain range at the end of construction according to Figure 5.1.
Instead, very small-strain soil stiffness and its non-linear dependency on strain amplitude
should be properly taken into account. In addition to all features of the Hardening Soil
model, the HSsmall model offers the possibility to do so.
The HSsmall model implemented in PLAXIS is based on the Hardening Soil model and
uses almost entirely the same parameters (see Section 5.4). In fact, only two additional
parameters are needed to describe the stiffness behaviour at small strains:
the initial or very small-strain shear modulus G
0
the shear strain level
0.7
at which the secant shear modulus G
s
is reduced to
about 70% of G
0
5-1
5.1 DESCRIBING SMALL-STRAIN STIFFNESS WITH A SIMPLE
HYPERBOLIC LAW
In soil dynamics, small-strain stiffness has been a well known phenomenon for a long
time. In static analysis, the ndings from soil dynamics have long been considered not to
be applicable.
Seeming differences between static and dynamic soil stiffness have been attributed to the
nature of loading (e.g. inertia forces and strain rate effects) rather than to the magnitude of
applied strain which is generally small in dynamic conditions (earthquakes excluded). As
inertia forces and strain rate have only little inuence on the initial soil stiffness, dynamic
soil stiffness and small-strain stiffness can in fact be considered as synonyms.
The probably most frequently used model in soil dynamics is the Hardin-Drnevich
relationship. From test data, sufcient agreement is found that the stress-strain curve
for small strains can be adequately described by a simple hyperbolic law. The following
analogy to the hyperbolic law for larger strains by Kondner (1963) (see previous Section)
was proposed by Hardin & Drnevich (1972):
G
s
G
0
=
1
1+
(5.1)
where the threshold shear strain
r
is quantied as:
r
=

max
G
0
(5.2)
with
max
being the shear stress at failure. Essentially, Eqs. (5.1) and (5.2) relate large
(failure) strains to small-strain properties which often work well.
More straightforward and less prone to error is the use of a smaller threshold shear strain.
Santos & Correia (2001), for example suggest to use the shear strain
r
=
0.7
at which the
secant shear modulus G
s
is reduced to about 70 % of its initial value. Eq. (5.1) can then
be rewritten as:
G
s
G
0
=
1
1+a
0.7
where a = 0.385 (5.3)

In fact, using a = 0.385 and =
0.7
gives G
s
/G
0
= 0.722. Hence, the formulation "about
70%" should be interpreted more accurately as 72.2%.
Figure 5.2 shows the t of the modied Hardin-Drnevich relationship (Eq. 5.3) to
normalized test data.
5-2 PLAXIS-GiD
Figure 5.2 Results from the Hardin-Drnevich relationship compared to test data by Santos
& Correia (2001)
5.2 APPLYING THE HARDIN-DRNEVICH RELATIONSHIP IN THE HS
MODEL
The decay of soil stiffness at small strains can be associated with loss of intermolecular
and surface forces within the soil skeleton. Once the direction of loading is reversed,
the stiffness regains a maximum recoverable value which is in the order of the initial
soil stiffness. Then, while loading in the reversed direction is continued, the stiffness
decreases again. A strain history dependent, multi-axial extension of the Hardin-Drnevich
relationship is therefore needed in order to apply it in the HS model. Such an extension
has been proposed by Benz (2006) in the form of the small-strain overlay model. Benz
derives a scalar valued shear strain
hist
by the following projection:
hist
=
3
_
_
He
_
_
e
(5.4)
where e is the actual deviatoric strain increment and H is a symmetric tensor that
represents the deviatoric strain history of the material. Whenever a strain reversal is
detected the tensor H is partially or fully reset before the actual strain increment e
is added. As the criterion for strain reversals serves a criterion similar as in Simpson's
brick model (1992): All three principal deviatoric strain directions are checked for strain
reversals separately which resembles three independent brick models. When there is
no principal strain rotation, the criterion reduces to two independent brick-models. For
further details on the strain tensor H and its transformation at changes in the load path it
is referred to Benz (2006).
The scalar valued shear strain =
hist
calculated in Eq. (5.4) is applied subsequently
used in Eq. (5.3) . Note that in both, Eqs. (5.3) and (5.4) , the scalar valued shear strain is
5-3
dened as:
=
3
2
q
(5.5)
where
q
is the second deviatoric strain invariant. In triaxial conditions can therefore be
expressed as:
=
axial
lateral
(5.6)
Within the HSsmall model, the stress-strain relationship can be simply formulated from
the secant shear modulus (Eq. 5.3) as:
= G
s
=
G
0
1+0.385

0.7
(5.7)
Taking the derivative with respect to the shear strain gives the tangent shear modulus:
G
t
=
G
0
_
1+0.385

0.7
_
2
(5.8)
This stiffness reduction curve reaches far into the plastic material domain. In the
Hardening Soil and HSsmall model, stiffness degradation due to plastic straining is
simulated with strain hardening. In the HSsmall model, the small-strain stiffness reduction
curve is therefore bounded by a certain lower limit, determined by conventional laboratory
tests:
The lower cut-off of the tangent shear modulus G
t
is introduced at the unloading
reloading stiffness G
ur
which is dened by the material parameters E
ur
and
ur
:
G
t
G
ur
where G
ur
=
E
ur
2(1+
ur
)
(5.9)
The cut-off shear strain
cutof f
can be calculated as:
cuto f f
=
1
0.385
_
G
0
G
ur
1
_
0.7
(5.10)
Within the HSsmall model, the actual quasi-elastic tangent shear modulus is calculated
by integrating the secant stiffness modulus reduction curve over the actual shear strain
increment. An example of a stiffness reduction curve used in the HSsmall model is shown
in Figure 5.3.
5-4 PLAXIS-GiD
1E-6 1E-5 0.0001 0.001 0.01
g
0.7
[-]
0
10000
20000
30000
40000
S
h
e
a
r
m
o
d
u
l
u
s
G
[
k
N
/
m
]
HSsmall
Hardin-Drnevich
G
ur
G
0
cut-off
Figure 5.3 Cut-off of the small-strain degradation curve as used in the HS-Small model
Figure 5.4 Hysteretic material behaviour
5.3 VIRGIN (INITIAL) LOADING VS. UNLOADING/RELOADING
Masing (1926) described the hysteretic behaviour of materials in unloading / reloading
cycles in the form of the following rules:
The shear modulus in unloading is equal to the initial tangent modulus for the
initial loading curve.
The shape of the unloading and reloading curves is equal to the initial loading
curve, but twice its size. In terms of the above introduced threshold shear strain
0.7
, Masing's rule can be fullled by the following setting in the Hardin-
Drnevich relation:
5-5
Figure 5.5 HSsmall stiffness reduction in initial- or primary loading and in unloading /
reloading
0.7
reloading
= 2
0.7
virginloading
(5.11)
The HSsmall model consequently adopts Masing's rule by doubling the threshold shear
strain provided by the user for virgin loading. If hardening plasticity readily accounts
for more rapidly decaying small-strain stiffness during virgin loading, the user dened
threshold shear strain is always doubled. Next, the hardening laws of the HSsmall
model are tted such, that the small-strain stiffness reduction curve is reasonably well
approximated. Figure 5.4 and Figure 5.5 illustrate Masing's rule and the secant stiffness
reduction in virgin loading and unloading / reloading.
5.4 MODEL PARAMETERS
Compared to the standard HS model, the HSsmall model requires two additional stiffness
parameters as input: G
re f
0
and
0.7
. All other parameters, including the alternative stiffness
parameters, remain the same as in the standard HS model. G
re f
0
denes the shear modulus
at very small strains e.g. < 10
6
at a reference minor principal stress of '
3
= p
re f
.
Poisson's ratio
ur
is assumed a constant, as everywhere in PLAXIS, so that the shear
modulus G
re f
0
can also be calculated from the very small strain Young's modulus as
G
re f
0
= E
re f
0
/(2(1 +
ur
)). The threshold shear strain
0.7
is the shear strain at which the
secant shear modulus G
re f
s
is decayed to 0.722G
re f
0
. The threshold shear strain
0.7
is to
be supplied for virgin loading. In summary, the input stiffness parameters of the HSsmall
model are listed below:
m : Power for stress-level dependency of stiffness [-]
E
re f
50
: Secant stiffness in standard drained triaxial test [kN/m
2
]
5-6 PLAXIS-GiD
E
re f
oed
: Tangent stiffness for primary oedometer loading [kN/m
2
]
E
re f
ur
: unloading / reload stiffness at engineering strains (
10
3
10
2
)
[kN/m
2
]
ur
: Poisson's ratio for unloading-reloading [-]
G
re f
0
: reference shear modulus at very small strains ( < 10
6
) [kN/m
2
]
0.7
: shear strain at which G
s
= 0.722G
0
[-]
Figure 5.6 illustrates the model's stiffness parameters in a triaxial test E
50
, E
ur
, and E
0
=
2G
0
(1+
ur
). For the order of strains at which E
ur
and G
0
are dened and determined,
one may refer to e.g. Figure 5.1 and Figure 5.3. If the default setting for G
re f
0
= G
re f
ur
is
used, no small strain hardening behaviour occurs and the HSsmall model defaults back to
the standard HS model.
q =
1
3
axial strain
1
E
0
E
0
E
0
E
0
E
50
E
ur
Figure 5.6 Stiffness parameters E
50
, E
ur
, and E
0
=2G
0
(1+
ur
) of the HSsmall model in
a triaxial test
5.5 ON THE PARAMETERS G
0
AND
0.7
A number of factors inuence the small-strain parameters G
0
and
0.7
. Most importantly
they are inuenced by the material's actual state of stress and void ratio e. In the HSsmall
model, the stress dependency of the shear modulus G
0
is taken into account with the power
law:
G
0
= G
re f
0
_
ccos '
3
sin
ccos +
p
re f
sin
_
m
(5.12)
which resembles the ones used for the other stiffness parameters. The threshold shear
strain
0.7
is taken independently of the mean stress.
Assuming that within a HSsmall (or HS) computation void ratio changes are rather small,
5-7
the material parameters are not updated for changes in the void ratio. Knowledge of a
material's initial void ratio can nevertheless be very helpful in deriving its small-strain
shear stiffness G
0
. Many correlations are offered in the literature (Benz, 2006). A good
estimation for many soils is for example the relation given by Hardin & Black (1969):
G
re f
0
=
(2.97e)
2
1+e
33 [MPa] (5.13)
Alpan (1970) empirically related dynamic soil stiffness to static soil stiffness (Figure 5.7).
The dynamic soil stiffness in Alpan's chart is equivalent to the small-strain stiffness G
0
or E
0
. Considering that the static stiffness E
static
dened by Alpan equals approximately
the unloading / reloading stiffness E
ur
in the HSsmall model, Alpan's chart can be used
to guess a soil's small-strain stiffness entirely based on its unloading / reloading stiffness
E
ur
. Although Alpan suggests that the ratio E
0
/E
ur
can exceed 10 for very soft clays, the
maximum ratio E
0
/E
ur
or G
0
/G
ur
permitted in the HSsmall model is limited to 10.
Figure 5.7 Relation between dynamic (E
d
= E
0
) and static soil stiffness (E
s
E
ur
) after
Alpan (1970)
In the absence of test data, correlations are also available for the threshold shear strain
0.7
. Figure 5.8 for example gives a correlation between the threshold shear strain and
the Plasticity Index. Using the original Hardin-Drnevich relationship, the threshold shear
strain
0.7
might be also related to the model's failure parameters. Applying the Mohr-
Coulomb failure criterion in Eqs. (5.2) and (5.3) yields:
0.7
1
9G
0
[2c'(1+cos(2')) '
1
(1+K
0
)sin(2')] (5.14)
where K
0
is the earth pressure coefcient at rest and '
1
is the effective vertical stress
(pressure negative).
5-8 PLAXIS-GiD
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
10
-6
10
-5
10
-4
10
-3
Shear strain amplitude [-] gs
OCR=1-15
G

/

G
[
-
]
0
P
I
=
1
5
P
I
=
3
0
P
I
=
0
Figure 5.8 Inuence of plasticity index (PI) on stiffness reduction after Vucetic & Dobry
(1991)
5.6 MODEL INITIALIZATION
Stress relaxation erases a soil's memory of previous applied stress. Soil ageing in the
form of particle (or assembly) reorganization during stress relaxation and formation of
bonds between them can erase a soil's strain history. Considering that the second process
in a naturally deposited soil develops relatively fast, the strain history should start from
zero (H = 0) in most boundary value problems. This is the default setting in the HSsmall
model.
However, sometimes an initial strain history may be desired. In this case the strain history
can be adjusted by applying an extra load step before starting the actual analysis. Such
an additional load step might also be used to model overconsolidated soils. Usually the
over-consolidation's cause has vanished long before the start of calculation, so that the
strain history should be reset afterwards. Unfortunately, strain history is already triggered
by adding and removing a surcharge. In this case the strain history can be reset manually,
by replacing the material or applying a small reverse load step. More convenient is the use
of the initial stress procedure.
When using the HSsmall model, caution should be given to nil-steps. The strain
increments in nil-steps are purely derived from the small numerical unbalance in the
system which is due to the accepted tolerated error in the computation. The strain
increment direction in nil-steps is therefore arbitrary. Hence, a nil-step may function as
randomly reverse load step which is in most cases not desired.
5-9
5.7 OTHER DIFFERENCES BETWEEN THE HS AND THE HSSMALL
MODEL
The mobilised dilatancy angle
The shear hardening ow rule of both, the HS and the HSsmall model have the linear
form:

p
v
= sin
m

p
(5.15)
The mobilised dilatancy angle
m
in compression however, is dened differently. The HS
model assumes the following:
For sin
m
< 3/4sin
m
= 0
For sin
m
3/4sin and > 0 sin
m
= max
_
sin
m
sin
cv
1sin
m
sin

cv
, 0
_
(5.16)
For sin
m
3/4sin and 0
m
=
If = 0
m
= 0
where
c
is the critical state friction angle, being a material constant independent of
density, and
m
is the mobilised friction angle:
sin
m
=
'
1
'
3
'
1
+'
3
2ccot
(5.17)
For small mobilised friction angles and for negative values of
m
, as computed by
Rowe's formula,
m
in the HS model is taken zero. Bounding the lower value of
m
may sometimes yield too little plastic volumetric strains though. Therefore, the HSsmall
model adapts an approach by Li & Dafalias (2000) whenever
m
, as computed by
Rowe's formula,is negative. In that case, the mobilised dilatancy in the HSsmall model
is calculated by the following Equation:
sin
m
=
1
10
_
_
Mexp
_
1
15
ln
_
M
q
q
a
__
+
_
_
(5.18)
where M is the stress ratio at failure, and = q/p is the actual stress ratio. Eq. (5.18) is a
simplied version of the void ratio dependent formulation by Li & Dafalias (2000).
5-10 PLAXIS-GiD
THE SOFT SOIL MODEL
6 THE SOFT SOIL MODEL
To highlight the signicance of the Soft Soil model, it is worth mentioning that starting
from Version 7 some changes to the soil modelling strategy of PLAXIS have been
introduced. Up to Version 6 PLAXIS material models had consisted of Mohr-Coulomb
model, Soft Soil model and Hard Soil model. In Version 7, however, the idea of using
separate models for soft soil and hard soil has been excluded. Instead, the Hard Soil model
was further developed to become an advanced model for soils ranging from soft to hard.
This has resulted into the current Hardening Soil model. At the same time the Soft Soil
Creep model was implemented to capture some of the very special features of soft soil.
As a result, the Soft Soil model can be substituted by the new Hardening Soil model or
the Soft Soil Creep model. However, in order not to deny users' preferences to use models
that they have got to know well, it was decided to keep the Soft Soil model in PLAXIS 2D.
Some features of the Soft Soil model are:
Stress dependent stiffness (logarithmic compression behaviour).
Distinction between primary loading and unloading-reloading.
Memory for pre-consolidation stress.
Failure behaviour according to the Mohr-Coulomb criterion.
6.1 ISOTROPIC STATES OF STRESS AND STRAIN ('
1
= '
2
= '
3
)
In the Soft Soil model, it is assumed that there is a logarithmic relation between the
volumetric strain,
v
, and the mean effective stress, p', which can be formulated as:
0
v
=
ln
_
p'
p
0
_
(virgin compression) (6.1)
In order to maintain the validity of Eq. (6.1) a minimum value of p' is set equal to a
unit stress. The parameter
is the modied compression index, which determines the

compressibility of the material in primary loading. Note that
differs from the index

as used by Burland (1965).
The difference is that Eq. (6.1) is a function of volumetric strain instead of void ratio.
Plotting Eq. (6.1) gives a straight line as shown in Figure 6.1.
During isotropic unloading and reloading a different path (line) is followed, which can be
formulated as:
e
v
e0
v
=
ln
_
p'
p
0
_
(unloading and reloading) (6.2)
Again, a minimumvalue of p' is set equal to a unit stress. The parameter
is the modied
swelling index, which determines the compressibility of the material in unloading and
6-1
subsequent reloading. Note that
differs from the index as used by Burland. The ratio
is, however, equal to Burland's ratio /. The soil response during unloading and
reloading is assumed to be elastic as denoted by the superscript e in Eq. (6.2) . The elastic
behaviour is described by Hooke's law (see Section 2.2) and Eq. (6.2) implies linear stress
dependency on the tangent bulk modulus such that:
Kur
Eur
3(12
ur
)
=
p'
(6.3)
in which the subscript ur denotes unloading / reloading. Note that effective parameters are
considered rather than undrained soil properties, as might be suggested by the subscripts
ur. Neither the elastic bulk modulus, K
ur
, nor the elastic Young's modulus, E
ur
, is used
as an input parameter. Instead,
ur
and
are used as input constants for the part of the

model that computes the elastic strains.
p
p
v
1
1
*
*
ln p'
Figure 6.1 Logarithmic relation between volumetric strain and mean stress
An innite number of unloading / reloading lines may exist in Figure 6.1, each
corresponding to a particular value of the isotropic pre-consolidation stress p
p
. The pre-
consolidation stress represents the largest stress level experienced by the soil. During
unloading and reloading, this pre-consolidation stress remains constant. In primary
loading, however, the pre-consolidation stress increases with the stress level, causing
irreversible (plastic) volumetric strains.
6.2 YIELD FUNCTION FOR TRIAXIAL STRESS STATE ('
2
= '
3
)
The Soft Soil model is capable to simulate soil behaviour under general states of stress.
However, for clarity, in this section, restriction is made to triaxial loading conditions under
which '
2
= '
3
. For such a state of stress the yield function of the Soft Soil model is
6-2 PLAXIS-GiD
THE SOFT SOIL MODEL
dened as:
f = f
p
p
(6.4)
where f is a function of the stress state (p', q) and p
p
, the pre-consolidation stress, is a
function of plastic strain such that:
f =
q
2
M
2
(p' +ccot )
+ p' (6.5)
p
p
=
p
0
p
exp
_

p
v
_
(6.6)
The yield function f describes an ellipse in the p' - q-plane, as illustrated in Figure 6.2. The
parameter M in Eq. (6.5) determines the height of the ellipse. The height of the ellipse
is responsible for the ratio of horizontal to vertical stresses in primary one-dimensional
compression.
M
o
h
r
-
C
o
u
l
o
m
b

f
a
i
l
u
r
e

l
i
n
e
q
c cot
p
p
M
1
p'
threshold ellipse
cap
Figure 6.2 Yield surface of the Soft Soil model in p' - q-plane
As a result, the parameter M determines largely the coefcient of lateral earth pressure
K
nc
0
. In view of this, the value of M can be chosen such that a known value of K
nc
0
is
matched in primary one-dimensional compression. Such an interpretation and use of M
differs from the original critical state line idea, but it ensures a proper matching of K
nc
0
.
The tops of all ellipses are located on a line with slope M in the p' - q-plane. In the
(Burland, 1965; Burland, 1967) the M-line is referred to as the critical state line and
represents stress states at post peak failure. The parameter M is then based on the critical
state friction angle. In the Soft Soil model, however, failure is not necessarily related to
critical state. The Mohr-Coulomb failure criterion is a function of the strength parameters
and c, which might not correspond to the M-line. The isotropic pre-consolidation stress
p
p
determines the extent of the ellipse along p' axis. During loading, innitely many
ellipses may exist (see Figure 6.2) each corresponds to a particular value of p
p
. In tension
6-3
(p' < 0), the ellipse extends to ccot (Eq. (6.5) and Figure 6.2). In order to make sure
that the right hand side of the ellipse (i.e. the 'cap') will remain in the 'compression' zone
(p' > 0) a minimum value of ccot is adopted for p
p
. For c = 0, a minimum value of p
p
equal to a stress unit is adopted. Hence, there is a 'threshold' ellipse as illustrated in Figure
6.2.
The value of p
p
is determined by volumetric plastic strain following the hardening
relation, Eq. (6.6) . This equation reects the principle that the pre-consolidation stress
increases exponentially with decreasing volumetric plastic strain (compaction). p
0
p
can
be regarded as the initial value of the pre-consolidation stress. The determination of p
0
p
is treated in Section 2.8. According to Eq. (6.6) the initial volumetric plastic strain is
assumed to be zero.
In the Soft Soil model, the yield function, Eq. (6.4) , describes the irreversible volumetric
strain in primary compression, and forms the cap of the yield contour. To model the failure
state, a perfectly-plastic Mohr-Coulomb type yield function is used. This yield function
represents a straight line in p' - q-plane as shown in Figure 6.2 . The slope of the failure
line is smaller than the slope of the M-line.
- '
1
- '
2
- '
3
cap
failure surface
Figure 6.3 Representation of total yield contour of the Soft Soil modelin principal stress
space
The total yield contour, as shown by the bold lines in Figure 6.2, is the boundary of
the elastic stress area. The failure line is xed, but the cap may increase in primary
6-4 PLAXIS-GiD
THE SOFT SOIL MODEL
compression. Stress paths within this boundary give only elastic strain increments,
whereas stress paths that tend to cross the boundary generally give both elastic and plastic
strain increments.
For general states of stress, the plastic behaviour of the Soft Soil model is dened by a
total of six yield functions; three compression yield functions and three Mohr-Coulomb
yield functions. The total yield contour in principal stress space, resulting from these six
yield functions, is indicated in Figure 6.3.
6.3 PARAMETERS OF THE SOFT SOIL MODEL
The parameters of the Soft Soil model coincide with those of the Soft Soil Creep model.
However, since the Soft Soil model does not include time, the modied creep index
is
not considered. Thus, the Soft Soil model requires the following material constants:
Basic parameters:
: Modied compression index [-]
: Modied swelling index [-]

c : Cohesion [kN/m
2
]
: Friction angle [
]
: Dilatancy angle [
]
Advanced parameters (use default settings):
ur
: Poissons ratio for unloading / reloading [-]
K
nc
0
: Coefcient of lateral stress in normal consolidation [-]
M : K
nc
0
-parameter [-]
Figure 6.4 shows PLAXIS window for inputting the values of the model parameters. M is
calculated automatically fromthe coefcient of the lateral earth pressure, K
nc
0
, by means of
Eq. (6.8). Note that, physically, in the current model M differs from that in the Modied
Cam Clay model where it is related to the material friction.
Modied swelling index and modied compression index
These parameters can be obtained from an isotropic compression test including isotropic
unloading. When plotting the logarithm of the mean stress as a function of the volumetric
strain for clay-type materials, the plot can be approximated by two straight lines (see
Figure 6.1). The slope of the primary loading line gives the modied compression index,
and the slope of the unloading (or swelling) line gives the modied swelling index. Note
that there is a difference between the modied indices
and
and the original Cam-

Clay parameters and . The latter parameters are dened in terms of the void ratio e
instead of the volumetric strain
v
.
6-5
Figure 6.4 Parameters tab sheet for the Soft Soil model
Apart from the isotropic compression test, the parameters
and
can be obtained from

the one-dimensional compression test. Here a relationship exists with the internationally
recognized parameters for one-dimensional compression and recompression, C
c
and C
r
.
Another relationship exists with the Dutch parameters for one-dimensional compression,
C
p
' and A
p
. These relationships are summarized in Table 6.1.
Remarks on Table 6.1:
In relations 1 and 2, the void ratio, e, is assumed to be constant. In fact, e will
change during a compression test, but this will give a relatively small difference
in void ratio. For e one can use the average void ratio that occurs during the test
or just the initial value.
In relations 4 and 6 there is no exact relation between
and the one-dimensional

swelling indices, because the ratio of horizontal and vertical stresses changes
during one-dimensional unloading. For approximation it is assumed that the
average stress state during unloading is an isotropic stress state, i.e. the horizontal
and vertical stresses are equal.
6-6 PLAXIS-GiD
THE SOFT SOIL MODEL
Table 6.1a Relationship to Cam-Clay parameters
1.

=

1+e
2.
=

1+e
Table 6.1b Relationship to Dutch engineering practice
3.

=
1
Cp
4.
2
Ap
Table 6.1c Relationship to internationally normalized parameters
5.

=
Cc
2.3(1+e)
6.
2Cr
2.3 (1+e)
The factor 2.3 in relation 5 is obtained from the ratio between the logarithm of
base 10 and the natural logarithm.
The ratio
(=/) ranges, in general, between 3 and 7.

Cohesion
The cohesion has the dimension of stresses. Any effective cohesion may be used,
including a cohesion of zero. When using the standard setting the cohesion is set equal
to 1 kPa. Entering a cohesion will result in an elastic region that is partly located in the
'tension' zone, as illustrated in Figure 6.2. The left hand side of the ellipse crosses the
p'-axis at a value of ccot . In order to maintain the right hand side of the ellipse (i.e.
the cap) in the 'pressure' zone of the stress space, the isotropic pre-consolidation stress
p
p
has a minimum value of ccot . This means that entering a cohesion larger than zero
may result in a state of 'over-consolidation', depending on the magnitude of the cohesion
and the initial stress state. As a result, a stiffer behaviour is obtained during the onset of
loading. It is not possible to specify undrained shear strength by means of high cohesion
and a friction angle of zero. Input of model parameters should always be based on effective
values.
Friction angle
The effective angle of internal friction represents the increase of shear strength with
effective stress level. It is specied in degrees. Zero friction angle is not allowed. On
the other hand, care should be taken with the use of high friction angles. It is often
recommended to use
c
, i.e. the critical state friction angle, rather than a higher value
based on small strains.
Moreover, using a high friction angle will substantially increase the computational
requirements.
6-7
Dilatancy angle
For the type of materials, which can be described by the Soft Soil model, the dilatancy
can generally be neglected. A dilatancy angle of zero degrees is considered in the standard
settings of the Soft Soil model.
Poisson's ratio
In the Soft Soil model, the Poisson's ratio is the well known pure elastic constant rather
than the pseudo-elasticity constant as used in the Mohr-Coulomb model. Its value will
usually be in the range between 0.1 and 0.2. If the standard setting for the Soft Soil model
parameters is selected, then
ur
= 0.15 is automatically used. For loading of normally
consolidated materials, Poisson's ratio plays a minor role, but it becomes important in
unloading problems. For example, for unloading in a one-dimensional compression test
(oedometer), the relatively small Poisson's ratio will result in a small decrease of the lateral
stress compared with the decrease in vertical stress. As a result, the ratio of horizontal
and vertical stress increases, which is a well-known phenomenon in overconsolidated
materials. Hence, Poisson's ratio should not be based on the normally consolidated K
nc
0
-
value, but on the ratio of the horizontal stress increment to the vertical stress increment in
oedometer unloading and reloading test such that:
ur
1
ur
=

xx
yy
(unloading and relaoding) (6.7)
K
nc
0
-parameter
The parameter M is automatically determined based on the coefcient of lateral earth
pressure in normally consolidated condition, K
nc
0
, as entered by the user. The exact
relation between M and K
nc
0
gives (Brinkgreve, 1994):
M = 3
(1K
nc
0
)
2
(1+2K
nc
0
)
2
+
(1K
nc
0
)(12
ur
)(
1)
(1+2K
nc
0
)(12
ur
)
(1K
nc
0
)(1+
ur
)
(6.8)
The value of M is indicated in the input window. As can be seen from Eq. (6.8), M is also
inuenced by the Poisson's ratio
ur
and by the ratio
. However, the inuence of

K
nc
0
is dominant. Eq. (6.8) can be approximated by:
M 3.02.8K
nc
0
(6.9)
6-8 PLAXIS-GiD
SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)
7 SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)
7.1 INTRODUCTION
As soft soils we consider near-normally consolidated clays, clayey silts and peat. A
special feature of such materials is their high degree of compressibility. This is best
demonstrated by oedometer test data as reported for instance by Janbu in his Rankine
lecture (1985). Considering tangent stiffness moduli at a reference oedometer pressure
of 100 kPa, he reports for normally consolidated clays E
oed
= 1 to 4 MPa, depending
on the particular type of clay considered. The differences between these values and
stiffnesses for NC-sands are considerable as here we have values in the range of 10
to 50 MPa, at least for non-cemented laboratory samples. Hence, in oedometer testing
normally consolidated clays behave ten times softer than normally consolidated sands.
This illustrates the extreme compressibility of soft soils.
Another feature of the soft soils is the linear stress-dependency of soil stiffness. According
to the Hardening Soil model we have:
E
oed
= E
re f
oed
_
'
1
/p
re f
_
m
at least for c = 0, and a linear relationship is obtained for m = 1. Indeed, on using an
exponent equal to one, the above stiffness law reduces to:
E
oed
='
1
/
where
= p
re f
/E
re f
oed
For this special case of m = 1, the Hardening Soil model yields =
'
1
/'
1
, which can
be integrated to obtain the well-known logarithmic compression law =
ln('
1
)
for primary oedometer loading.
For many practical soft-soil studies, the modied compression index
will be known
and the PLAXIS user can compute the oedometer modulus from the relationship:
E
re f
oed
= p
re f
/
From the above considerations it would seem that the HS-model is quite suitable for soft
soils. Indeed, most soft soil problems can be analysed using this model, but the HS-model
is not suitable when considering creep, i.e. secondary compression. All soils exhibit some
creep, and primary compression is thus always followed by a certain amount of secondary
compression. Assuming the secondary compression (for instance during a period of 10
or 30 years) to be a certain percentage of the primary compression, it is clear that creep
is important for problems involving large primary compression. This is for instance the
case when constructing embankments on soft soils. Indeed, large primary settlements of
footings and embankments are usually followed by substantial creep settlements in later
years. In such cases it is desirable to estimate the creep from FEM-computations.
Foundations may also be founded on initially overconsolidated soil layers that yield
7-1
relatively small primary settlements. Then, as a consequence of the loading, a state
of normal consolidation may be reached and signicant creep may follow. This is a
treacherous situation as considerable secondary compression is not preceded by the
warning sign of large primary compression. Again, computations with a creep model are
desirable.
Buisman (1936) was probably the rst to propose a creep law for clay after observing
that soft-soil settlements could not be fully explained by classical consolidation theory.
This work on 1D-secondary compression was continued by other researchers including,
for example, Bjerrum (1967), Garlanger (1972), Mesri & Godlewski (1977) and
Leroueil (1977). More mathematical lines of research on creep were followed by, for
example, Sekiguchi (1977), Adachi & Oka (1982) and Borja & Kavaznjian (1985). This
mathematical 3D-creep modelling was inuenced by the more experimental line of 1D-
creep modelling, but conicts exist.
3D-creep should be a straight forward extension of 1D-creep, but this is hampered by the
fact that present 1D-models have not been formulated as differential equations. For the
presentation of the Soft Soil Creep model we will rst complete the line of 1D-modelling
by conversion to a differential form. From this 1D differential equation an extension was
made to a 3D-model. This Chapter gives a full description of the formulation of the Soft
Soil Creep model. In addition, attention is focused on the model parameters. Finally, a
validation of the 3D model is presented by considering both model predictions and data
from triaxial tests. Here, attention is focused on constant strain rate triaxial tests and
undrained triaxial creep tests. For more applications of the model the reader is referred
to Vermeer et al. (1998), Neher & Vermeer (1998) and Brinkgreve (2004).
Some basic characteristics of the Soft Soil Creep model are:
Stress-dependent stiffness (logarithmic compression behaviour)
Distinction between primary loading and unloading-reloading
Secondary (time-dependent) compression
Memory of pre-consolidation stress
Failure behaviour according to the Mohr-Coulomb criterion
7.2 BASICS OF ONE-DIMENSIONAL CREEP
When reviewing previous literature on secondary compression in oedometer tests, one is
struck by the fact that it concentrates on behaviour related to step loading, even though
natural loading processes tend to be continuous or transient in nature. Buisman (1936)
was probably the rst to consider such a classical creep test. He proposed the following
equation to describe creep behaviour under constant effective stress:
=
c
C
B
log
_
t
t
c
_
for: t >t
c
(7.1)
7-2 PLAXIS-GiD
where
c
is the strain up to the end of consolidation, t the time measured from the
beginning of loading, t
c
the time to the end of primary consolidation and C
B
is a material
constant.
Please note that we do not follow the soil mechanics convention that compression is
considered positive. Instead, compressive stresses and strains are taken to be negative.
For further consideration, it is convenient to rewrite this equation as:
=
c
CB
log
_
t
c
+t'
t
c
_
for: t' > 0 (7.2)
with t' =t t
c
being the effective creep time.
Based on the work by Bjerrum on creep, as published for instance in 1967, Garlanger
(1972) proposed a creep equation of the form:
e = e
c
C
log
_
c
+t'
c
_
with: C
=C
B
(1+e
0
) for: t' > 0 (7.3)
Differences between Garlanger's and Buisman's forms are modest. The engineering strain
is replaced by void ratio e and the consolidation time t
c
is replaced by a parameter
c
. Eqs. (7.2) and (7.3) are entirely identical when choosing
c
= t
c
. For the case that
c
=t
c
differences between both formulations will vanish when the effective creep time t'
increases.
For practical consulting, oedometer tests are usually interpreted by assuming t
c
= 24h.
Indeed, the standard oedometer test is a Multiple Stage Loading Test with loading periods
of precisely one day. Due to the special assumption that this loading period coincides to
the consolidation time t
c
, it follows that such tests have no effective creep time.
Hence one obtains t' = 0 and the log-term drops out of Eq. (7.3). It would thus seem that
there is no creep in this standard oedometer test, but this suggestion is entirely false. Even
highly impermeable oedometer samples need less than one hour for primary consolidation.
Then all excess pore pressures are zero and one observes pure creep for the other 23 hours
of the day. Therefore we will not make any assumptions about the precise values of
c
and
t
c
.
Another slightly different possibility to describe secondary compression is the form
adopted by Buttereld (1979):
H
=
H
c
C ln
_
c
+t'
c
_
(7.4)
where
H
is the logarithmic strain dened as:
H
= ln
_
V
V
0
_
= ln
_
1 +e
1 + e
0
_
(7.5)
with the subscript '0' denoting the initial values. The superscript 'H' is used to denote
7-3
logarithmic strain, as the logarithmic strain measure was originally used by Hencky. For
small strains it is possible to show that:
C =
C
(1+e
0
) ln 10
=
CB
ln 10
(7.6)
because then logarithmic strain is approximately equal to the engineering strain. Both
Buttereld (1979) and den Haan (1994) showed that for cases involving large strain, the
logarithmic small strain supersedes the traditional engineering strain.
7.3 ON THE VARIABLES
C
AND
C
In this section attention will rst be focussed on the variable
c
. Here a procedure is to be
described for an experimental determination of this variable. In order to do so we depart
from Eq. (7.4) . By differentiating this equation with respect to time and dropping the
superscript 'H' to simplify notation, one nds:
=
C
c
+t'
or inversely:
1

=

c
+t'
C
(7.7)
which allows one to make use of the construction developed by Janbu (1969) for
evaluating the parameters C and
c
fromexperimental data. Both the traditional way, being
indicated in Figure 7.1a, as well as the Janbu method of b can be used to determine the
parameter C from an oedometer test with constant load.
1
t
c
C
lnt
t
a. Consolidation
1
t
c
C
-1/
c
t'
t' =t t
c
t
b. Creep behaviour
Figure 7.1 Standard oedometer test
The use of the Janbu method is attractive, because both
c
and C follow directly when
tting a straight line through the data. In Janbu's representation of Figure 7.1b,
c
is the
intercept with the (non-logarithmic) time axis of the straight creep line. The deviation
from a linear relation for t <t
c
is due to consolidation.
7-4 PLAXIS-GiD
Considering the classical literature it is possible to describe the end-of-consolidation strain
c
, by an equation of the form:
c
=
e
c
+
c
c
=A ln
_
'
'
0
_
Bln
_
pc
p0
_
(7.8)
Note that is a logarithmic strain, rather than a classical small strain although we
conveniently omit the subscript 'H'. In the above equation '
0
represents the initial
effective pressure before loading and ' is the nal effective loading pressure. The values
p0
and
pc
represent the pre-consolidation pressure corresponding to before-loading
and end-of-consolidation states respectively. In most literature on oedometer testing, one
adopts the void ratio e instead of , and log instead of ln, and the swelling index C
s
instead
of A, and the compression index C
c
instead of B. The above constants A and B relate to C
r
and C
c
as:
A =
Cs
(1+e
0
) ln10
B =
C
c
C
s
(1+e
o
) ln10
(7.9)
Combining Eqs. (7.4) and (7.8) it follows that:
=
e
+
c
=A ln
_
'
'
0
_
B ln
_
pc
p0
_
Cln
_
c
+t'
c
_
(7.10)
where is the total logarithmic strain due to an increase in effective stress from '
0
to '
and a time period of t
c
+t'. In Figure 7.2 the terms of Eq. (7.10) are depicted in an - ln
diagram.
1
1
'
0

p0

pc
'
A
A+B
e
c
c
c
Cln(1+t'/
c
)
ln(')
-
NC - line
Figure 7.2 Idealised stress-strain curve from oedometer test with division of strain
increments into an elastic and a creep component. For t' +t
c
= 1 day, one
arrives precisely on the NC-line
Up to this point, the more general problem of creep under transient loading conditions
has not yet been addressed, as it should be recalled that restrictions have been made to
creep under constant load. For generalising the model, a differential form of the creep
model is needed. No doubt, such a general equation may not contain t' and neither
c
as
7-5
the consolidation time is not clearly dened for transient loading conditions.
7.4 DIFFERENTIAL LAW FOR 1D-CREEP
The previous equations emphasize the relation between accumulated creep and time, for
a given constant effective stress. For solving transient or continuous loading problems, it
is necessary to formulate a constitutive law in differential form, as will be described in
this section. In a rst step we will derive an equation for
c
. Indeed, despite the use of
logarithmic strain and ln instead of log, equation (Eq. 7.10) is classical without adding
new knowledge. Moreover, the question on the physical meaning of
c
is still open. In
fact, we have not been able to nd precise information on
c
in the literature, apart from
Janbu's method of experimental determination.
In order to nd an analytical expression for the quantity
c
, we adopt the basic idea that
all inelastic strains are time dependent. Hence total strain is the sum of an elastic part
e
and a time-dependent creep part
c
. For non-failure situations as met in oedometer
loading conditions, we do not assume an instantaneous plastic strain component, as used
in traditional elastoplastic modelling. In addition to this basic concept, we adopt Bjerrum's
idea that the pre-consolidation stress depends entirely on the amount of creep strain being
accumulated in the course of time. In addition to Eq. (7.10) we therefore introduce the
expression:
=
e
+
c
=Aln
_
'
'
0
_
Bln
_

p
p0
_
p
=
p0
exp
_
c
B
_
(7.11)
Please note that
c
is negative, so that
p
exceeds
p0
. The longer a soil sample is left to
creep the larger
p
grows. The time-dependency of the pre-consolidation pressure
p
is
now found by combining Eqs. (7.10) and (7.11) to obtain:
c
c
=Bln
_

p
pc
_
=C ln
_
c
+t'
c
_
(7.12)
This equation can now be used for a better understanding of
c
, at least when adding
knowledge from standard oedometer loading. In conventional oedometer testing the load
is stepwise increased and each load step is maintained for a constant period of t
c
+t' = ,
where is precisely one day.
In this way of stepwise loading the so-called normal consolidation line (NC-line) with
p
= ' is obtained. On entering
p
= ' and t' = t
c
into Eq. (7.12) it is found that:
B ln
_
'
pc
_
=Cln
_
c
+ t
c
c
_
for: OCR = 1 (7.13)
It is now assumed that (
c
t
c
) << . This quantity can thus be disregarded with respect
7-6 PLAXIS-GiD
to and it follows that:
c
=
_
'
pc
_B
C
or:
c
=
_
pc
'
_B
C
(7.14)
Hence
c
depends both on the effective stress ' and the end-of-consolidation pre-
consolidation stress
pc
. In order to verify the assumption (
c
t
c
) << , it should be
realised that usual oedometer samples consolidate for relatively short periods of less than
one hour. Considering load steps on the normal consolidation line, we have OCR=1 both
in the beginning and at the end of the load step. During such a load step
p
increases from
p0
up to
pc
during the short period of (primary) consolidation. Hereafter
p
increases
further from
pc
up to ' during a relatively long creep period. Hence, at the end of the
day the sample is again in a state of normal consolidation, but directly after the short
consolidation period the sample is under-consolidated with
p
< '. For the usually very
high ratios of B/C 15, we thus nd very small
c
-values from Eq. (7.14) . Hence not
only t
c
but also
c
tends to be small with respect to . It thus follows that the assumption
(
c
t
c
) << is certainly correct.
Having derived the simple expression Eq. (7.14) for
c
, it is now possible to formulate the
differential creep equation. To this end Eq. (7.10) is differentiated to obtain:
=
e
+
c
=A
'
'
c
+t'
(7.15)
where
c
+t' can be eliminated by means of Eq. (7.12) to obtain:
=
e
+
c
=A
'
'
c
_
pc
p
_B
C
with:
p
=
p0
exp
_
c
B
_
(7.16)
Again it is recalled that
c
is a compressive strain, being considered negative in this
manual. Eq. (7.14) can now be introduced to eliminate
c
and
pc
and to obtain:
=
e
+
c
=A
'
'
_
'
p
_B
C
(7.17)
7.5 THREE-DIMENSIONAL-MODEL
On extending the 1D-model to general states of stress and strain, the well-known stress
invariants for pressure p and deviatoric stress q are adopted. These invariants are used to
7-7
dene a new stress measure named p
eq
:
p
eq
= p'
q
2
M
2
(p' ccot())
(7.18)
In Figure 7.3 it is shown that the stress measure p
eq
is constant on ellipses in p - q-plane.
In fact we have the ellipses from the Modied Cam-Clay-Model as introduced by Roscoe
& Burland (1968).
1
deviatoric stress
isotropic stress
M
q
p
eq
-p
p
eq
p
Figure 7.3 Diagram of p
eq
-ellipse in a p-q-plane
The soil parameter M represents the slope of the so-called 'critical state line' as also
indicated in Figure 7.3. We use the general 3D-denition (Eq. 2.8b) for the deviatoric
stress q and:
M =
6sin
cv
3sin
cv
(7.19)
where
cv
is the critical-void friction angle, also referred to as critical-state friction angle.
On using Eq. (2.8b) for q, the equivalent pressure p
eq
is constant along ellipsoids in
principal stress space. To extend the 1D-theory to a general 3D-theory, attention is now
focused on normally consolidated states of stress and strain as met in oedometer testing.
In such situations it yields '
2
= '
3
= K
nc
0
'
1
, and it follows from Eq. (7.18) that:
p
eq
= '
1
_
1+2K
nc
0
3
+
3(1K
nc
0
)
2
M
2
(1+2K
nc
0
)
_
,
p
eq
p
=
p
_
1+2K
nc
0
3
+
3(1K
nc
0
)
2
M
2
(1+2K
nc
0
)
_
(7.20)
where p
eq
p
is a generalised pre-consolidation pressure, being simply proportional to the
one-dimensional one. For known values of K
nc
0
, p
eq
can thus be computed from ', and
p
eq
p
can thus be computed from
p
. Omitting the elastic strain in the 1D-equation (Eq.
7.17) , introducing the above expressions for p
eq
and p
eq
p
and writing
instead of it is
7-8 PLAXIS-GiD
found that:

c
v
=
C
_
p
eq
p
eq
p
_B
C
where:
p
eq
p
=
p
eq
p0
exp
_
c
v
B
_
(7.21)
For one-dimensional oedometer conditions, this equation reduces to Eq. (7.17) , so that
one has a true extension of the 1D-creep model. It should be noted that the subscript '0' is
once again used in the equations to denote initial conditions and that
c
=0 for time t =0.

Instead of the parameters A, B and C of the 1D-model, we will now change to the material
parameters
and
, who t into the framework of critical-state soil mechanics.

Conversion between constants follows the rules:
2A B =
=C (7.22)
On using these new parameters, Eq. (7.21) changes to become:

c
v
=

_
p
eq
p
eq
p
_
with:
p
eq
p
= P
eq
p0
exp
_

c
v
_
(7.23)
As yet the 3D-creep model is incomplete, as we have only considered a volumetric creep
strain
c
, whilst soft soils also exhibit deviatoric creep strains.

For introducing general creep strains, we adopt the view that creep strain is simply a time-
dependent plastic strain. It is thus logic to assume a ow rule for the rate of creep strain,
as usually done in plasticity theory. For formulating such a ow rule, it is convenient to
adopt the vector notation and considering principal directions:
=
_
1

2

3
_
T
and: =
_
1

2

3
_
T
where T is used to denote a transpose. Similar to the 1D-model we have both elastic and
creep strains in the 3D-model. Using Hooke's law for the elastic part, and a ow rule for
the creep part, one obtains:
=
e
+
c
= D
1
' +

g
c
'
(7.24)
where the elasticity matrix and the plastic potential function are dened as:
D
1
=
1
Eur
_
_
1
ur

ur
ur
1
ur
ur

ur
1
_
_
and:
g
c
=
p
eq
Hence we use the equivalent pressure p
eq
as a plastic potential function for deriving the
7-9
individual creep strain-rate components. The subscripts 'ur' are introduced to emphasize
that both the elasticity modulus and Poisson's ratio will determine unloading-reloading
behaviour. Now it follows from the above equations that:

c
v
=
c
1
+
c
2
+
c
3
=
_
p
eq
'
1
+

p
eq
'
2
+

p
eq
'
3
_
=
p
eq
p'
= (7.25)
Hence we dene = p
eq
/ p'. Together with Eqs. (7.23) and (7.24) this leads to:
= D
1
' +

c
v
p
eq
'
= D
1
'
1
_
p
eq
p
eq
p
_

p
eq
'
(7.26)
where: (7.27)
p
eq
p
=
p
eq
p0
exp
_

c
v
_
or inversely:
c
v
= (
)ln
_
p
eq
p
p
eq
p0
_
7.6 FORMULATION OF ELASTIC 3D-STRAINS
Considering creep strains, it has been shown that the 1D-model can be extended to obtain
the 3D-model, but as yet this has not been done for the elastic strains.
To get a proper 3D-model for the elastic strains as well, the elastic modulus E
ur
has to
been dened as a stress-dependent tangent stiffness according to:
Eur
= 3(12
ur
)K
ur
=3(12
ur
)
p'
(7.28)
Hence, E
ur
is not a new input parameter, but simply a variable quantity that relates to
the input parameter
. On the other hand

ur
is an additional true material constant.
Hence similar to E
ur
, the bulk modulus K
ur
is stress dependent according to the rule K
ur
=
p'/
. Now it can be derived for the volumetric elastic strain that:

e
v
=
p'
Kur
=
p'
p'
or by integration:
e
ln
_
p'
p'
0
_
(7.29)
Hence in the 3D-model the elastic strain is controlled by the mean stress p', rather than
by principal stress ' as in the 1D-model. However mean stress can be converted into
principal stress. For one-dimensional compression on the normal consolidation line, we
have both 3p' = (1+2K
nc
0
)' and 3p'
0
= (1+2K
nc
0
)'
0
and it follows that p'/p'
0
='/'
0
.
As a consequence we derive the simple rule
c
ln'/'
0
, whereas the 1D-model
involves
c
= Aln'/'
0
. It would thus seem that
coincides with A. Unfortunately

this line of thinking cannot be extended towards overconsolidated states of stress and
7-10 PLAXIS-GiD
strain. For such situations, it can be derived that:
p'
p'
=
1+
ur
1
ur
1
1+2K
0
'
'
(7.30)
and it follows that:

e
v
=
p'
p'
=
1+
ur
1
ur
1+2K
0
'
'
(7.31)
where K
0
depends to a great extent on the degree of over-consolidation. For many
situations, it is reasonable to assume K
0
1 and together with
ur
0.2 one obtains
2
c
ln('/'
0
). Good agreement with the 1D-model is thus found by taking
2A.
7.7 REVIEW OF MODEL PARAMETERS
As soon as the Mohr-Coulomb failure yield criterion f (', c, ) = 0 is met, instantaneous
plastic strain rates develop according to the ow rule
p
= g/' with g = g(', ).
For details see Chapter 3 on the Mohr-Coulomb model. This criterion gives additional soil
parameters such as the effective cohesion, c, the Mohr-Coulomb friction angle, , and the
dilatancy angle . For ne grained, cohesive soils, the dilatancy angle tends to be small,
it may often be assumed that is equal to zero.
In conclusion, the Soft Soil Creep model requires the following material constants:
Failure parameters as in the Mohr-Coulomb model:
c : Cohesion [kN/m
2
]
: Friction angle [
]
: Dilatancy angle [
]
Basic stiffness parameters:
: Modied swelling index [-]
: Modied compression index [-]
: Modied creep index [-]

Advanced parameters (it is advised to use the default setting):
ur
: Poisson's ratio for unloading-reloading (default 0.15) [-]
K
nc
0
: '
xx
/'
yy
stress ratio in a state of normal consolidation [-]
M : K
nc
0
-related parameter (see below) [-]
7-11
Figure 7.4 Parameters tab for the Soft Soil Creep model
By default, M is automatically determined based on the coefcient of lateral earth pressure
in normally consolidated condition, K
nc
0
, as entered by the user. The exact relationship
between M and K
nc
0
can be formulated as (Brinkgreve, 1994):
M = 3
(1K
nc
0
)
2
(1+2K
nc
0
)
2
+
(1K
nc
0
)(12
ur
)(
1)
(1+2K
nc
0
)(12
ur
)
(1K
nc
0
)(1+
ur
)
(7.32)
Figure 7.5 Advanced parameters for Soft Soil Creep model
Hence the user can not enter directly a particular value of M. Instead he can choose values
for K
nc
0
. Note that the particular selection of M has an inuence on lateral deformation of
pseudo-vertical loading problems. For details, see Brinkgreve (2004).
Instead of dening the stiffness by the basic stiffness parameters, alternative stiffness
7-12 PLAXIS-GiD
parameters can be used. These material constants are given by:
C
c
: Compression index [-]
C
s
: Swelling index [-]
C
: Creep index for secondary compression [-]

e
init
: Initial void ratio [-]
Modied swelling index, modied compression index and modied creep index
These parameters can be obtained both from an isotropic compression test and an
oedometer test. When plotting the logarithm of stress as a function of strain, the plot
can be approximated by two straight lines (see Figure 7.2). The slope of the normal
consolidation line gives the modied compression index
, and the slope of the unloading

(or swelling) line can be used to compute the modied swelling index
, as explained in
Section 7.6. Note that there is a difference between the modied indices
and
and
the original Cam-Clay parameters and . The latter parameters are dened in terms of
the void ratio e instead of the volumetric strain
. The parameter
can be obtained by
measuring the volumetric strain on the long term and plotting it against the logarithm of
time (see Figure 7.1).
In Table 7.1c, the value 2.3 is in fact ln10 and stems from the conversion from
10
log to
natural logarithm. The alternative stiffness parameters can also be calculated from this
table. Since the void ratio e is not a constant, in the conversion from the alternative
parameters to the original model parameters in PLAXIS the void ratio e is dened as the
initial void ratio e
init
. Entering a particular value for one of the alternatives C
c
, C
s
or C
results in a change of *, * or * respectively. The parameter e

init
can also be dened
in the Advanced General Properties window.
As already indicated in Section 7.6, there is no exact relation between the isotropic
compression index
and the one-dimensional swelling indices A and C

s
, because the
ratio of horizontal and vertical stress changes during one-dimensional unloading. For the
approximation it is assumed that the average stress state during unloading is an isotropic
stress state, i.e. the horizontal and vertical stresses are equal.
For a rough estimate of the model parameters, one might use the correlation

I
p
(%)/500, the fact that
is in the range between 15 to 25 and the general observation
is in the range around 5.

For characterising a particular layer of soft soil, it is also necessary to know the initial pre-
consolidation pressure
p0
. This pressure may, for example, be computed from a given
value of the over-consolidation ratio (OCR). Subsequently
p0
can be used to compute
the initial value of the generalised pre-consolidation pressure
p
eq
p
(see Section 2.8).
7-13
Table 7.1a Relationship to Cam-Clay parameters
=

1+e

=

1+e
- - -
Table 7.1b Relationship to A,B, C parameters
= B+
2A
=C
Table 7.1c Relationship to internationally normalized parameters
=
Cc
2.3(1+e)

2
2.3
Cs
1+e

=
C
2.3(1+e)
Poisson's ratio
In the case of the Soft Soil Creep model, Poisson's ratio is purely an elasticity constant
rather than a pseudo-elasticity constant as used in the Mohr-Coulomb model. Its value
will usually be in the range between 0.1 and 0.2. If the standard setting for the Soft
Soil Creep model parameters is selected, then the value
ur
= 0.15 is automatically
adopted. For loading of normally consolidated materials, Poisson's ratio plays a minor
role, but it becomes important in unloading problems. For example, for unloading in a
one-dimensional compression test (oedometer), the relatively small Poisson's ratio will
result in a small decrease of the lateral stress compared with the decrease in vertical
stress. As a result, the ratio of horizontal and vertical stress increases, which is a well-
known phenomenon for overconsolidated materials. Hence, Poisson's ratio should not be
based on the normally consolidated K
nc
0
-value, but on the ratio of difference in horizontal
stress to difference in vertical stress in oedometer unloading and reloading:
ur
1
ur
=

xx
yy
(unloading and reloading) (7.33)
7.8 VALIDATION OF THE 3D-MODEL
This section briey compares the simulated response of undrained triaxial creep behaviour
of Haney clay with test data provided by Vaid & Campanella (1977), using the material
parameters summarized below. An extensive validation of the Soft Soil Creep model is
also provided in Stolle, Bonnier & Vermeer (1997). All triaxial tests considered were
completed by initially consolidating the samples under an effective isotropic conning
pressure of 525 kPa for 36 hours and then allowing them to stand for 12 hours under
undrained conditions before starting a shearing part of the test.
The material properties for Haney Clay are:
7-14 PLAXIS-GiD
= 0.016
= 0.105
= 0.004
mc
= 32
= 0
c = 0 kPa
K
nc
0
= 0.616 = 0.25
The end-of-consolidation pre-consolidation pressure, p
eq
p
, was found to be 373 kPa.
This value was determined by simulating the consolidation part of the test. The pre-
consolidation pressure p
eq
p
of 373 kPa is less than 525 kPa, which would have been
required for an OCR
of 1. It is clear that the pre-consolidation pressure not only depends

on the applied maximum consolidation stress, but also on creep time as discussed in
previous sections.
In Figure 7.6 we can see the results of Vaid & Campanella's tests (1977) for different strain
rates and the computed curves, which were calculated with the present creep model. It is
observed that the model describes the tests very well.
Constant strain rate shear tests
Undrained triaxial compression tests, as considered in Figure 7.6, are performed under
constant rates of vertical strain
1
and constant horizontal pressure
3
where the shear
stress q can move. This behaviour is shown in Figure 7.7.
0
100
200
300
400
0.00 -0.04 -0.08 -0.12
axial strain
1
deviatoric stress q
1.10 % per minute
0.15 % per minute
0.00094 % per minute
Figure 7.6 Results of the undrained triaxial tests (CU-tests) with different rates of strain.
The faster the test the higher the undrained shear strength
7-15
deviatoric stress
q
isotropic stress
"Fast" tests
"Slow" tests
p
Figure 7.7 Strain rate dependency of the effective stress path in undrained triaxial tests
During these undrained tests it yields
v
=
e
v
+
c
v
= 0 or equivalently
e
v
=
c
v
.
Hence, creep compaction is balanced by elastic swelling of the sample. The slower a
test is performed, the larger the creep compaction is and nally the elastic swelling. The
expression p' = K
ur

e
v
, where K
ur
is the elastic bulk modulus, shows that elastic swelling
implies a decrease of the mean stress.
For extremely fast tests there is no time for creep; it yields
c
v
= 0 and consequently also

e
v
= 0. Hence in this extreme case there is no elastic volume change and consequently
neither a change of the mean stress. This implies a straight vertical path for the effective
stress in p - q-plane. On inspecting all numerical results, it appears that the undrained
shear strength, c
u
, may be approximated by the equation:
c
u
c
u
1.02+0.09log (7.34)
where c
u
is the undrained shear strength in an undrained triaxial test with a strain rate
of 1% per hour. This agrees well with the experimental data summarized by Kulhawy &
Mayne (1990).
Undrained triaxial creep tests
These tests begin with isotropic consolidation up to a mean stress of 525 kPa. Then a
deviatoric stress q is applied in undrained loading. Finally all external stresses are kept
constant and the sample is subjected to undrained creep.
Samples were rst consolidated under the same isotropic stress. Then undrained samples
were loaded up to different deviatoric stresses. Creep under constant deviatoric stress is
observed, being well predicted by the Soft Soil Creep model.
The amount of creep depends on the applied deviatoric stress q or rather on the applied
stress ratio q/p. For relatively small stress ratios, creep rates are small and they decrease
7-16 PLAXIS-GiD
0.00
0.02
0.04
0.06
0.08
1 10 100 10000
axial strain
1
time [minutes]
experimental
1000
q = 323.4 q = 300.3 q = 278.3
Figure 7.8 Results of triaxial creep tests
in course of time. For large stress ratios, however, creep rates increase with time and
samples will nally fail, i.e. strain rates become innitely large.
1
10
100
1000
10000
260 280 300 340
Creep rupture time [minutes]
deviatoric stress q [kPa]
experimental
320
Figure 7.9 Results of triaxial creep tests
Figure 7.8 shows the actual creep strain evolution for samples with three different
deviatoric stresses.
All tests have different constant deviatoric stress. The creep rupture time is the creep time
7-17
up to a creep rate
1
=, as indicated by the asymptotes in Figure 7.8. Figure 7.9 shows
the experimental and the calculated relation between the creep rupture time and different
deviatoric stresses.
7-18 PLAXIS-GiD
THE JOINTED ROCK MODEL (ANISOTROPY)
8 THE JOINTED ROCK MODEL (ANISOTROPY)
Materials may have different properties in different directions. As a result, they may
respond differently when subjected to particular conditions in one direction or another.
This aspect of material behaviour is called anisotropy. When modelling anisotropy,
distinction can be made between elastic anisotropy and plastic anisotropy. Elastic
anisotropy refers to the use of different elastic stiffness properties in different directions.
Plastic anisotropy may involve the use of different strength properties in different
directions, as considered in the Jointed Rock model. Another form of plastic anisotropy
is kinematic hardening. The latter is not considered in PLAXIS program.
rock formation
major joint
direction
stratication
Figure 8.1 Visualization of concept behind the Jointed Rock model
The Jointed Rock model is an anisotropic elastic perfectly-plastic model, especially
meant to simulate the behaviour of stratied and jointed rock layers. In this model it is
assumed that there is intact rock with an eventual stratication direction and major joint
directions. The intact rock is considered to behave as a transversly anisotropic elastic
material, quantied by ve parameters and a direction. The anisotropy may result from
stratication or from other phenomena. In the major joint directions it is assumed that
shear stresses are limited according to Coulomb's criterion. Upon reaching the maximum
shear stress in such a direction, plastic sliding will occur. A maximum of three sliding
directions ('planes') can be dened, of which the rst plane is assumed to coincide with the
direction of elastic anisotropy. Each plane may have different shear strength properties.
In addition to plastic shearing, the tensile stresses perpendicular to the three planes are
limited according to a predened tensile strength (tension cut-off).
The application of the Jointed Rock model is justied when families of joints or joint
sets are present. These joint sets have to be parallel, not lled with fault gouge, and their
spacing has to be small compared to the characteristic dimension of the structure.
Some basic characteristics of the Jointed Rock model are:
Anisotropic elastic behaviour for intact rock Parameters E
1
,E
2
,
1
,
2
,
G
2
Shear failure according to Coulomb in three directions, i Parameters c
i
,
i
and
i
8-1
Limited tensile strength in three directions, i Parameters
t,i
8.1 ANISOTROPIC ELASTIC MATERIAL STIFFNESS MATRIX
The elastic material behaviour in the Jointed Rock model is described by an elastic
material stiffness matrix, D
. In contrast to Hooke's law, the D
-matrix as used in the

Jointed Rock model is transversely anisotropic. Different stiffnesses can be used normal to
and in a predened direction ('plane 1'). This direction may correspond to the stratication
direction or to any other direction with signicantly different elastic stiffness properties.
Consider, for example, a horizontal stratication, where the stiffness in horizontal
direction, E
1
, is different from the stiffness in vertical direction, E
2
. In this case the 'plane
1' direction is parallel to the x-z-plane and the following constitutive relations exist (See:
Zienkiewicz & Taylor: The Finite Element Method, 4th Ed.):

xx
=

xx
E
1
2

yy
E
2
1

zz
E
1
(8.1a)

yy
=
2

xx
E
2
+

yy
E
2
2

zz
E
2
(8.1b)

zz
=
1

xx
E
1
2

yy
E
2
+

zz
E
1
(8.1c)

xy
=

xy
G
2
(8.1d)

yz
=

yz
G
2
(8.1e)

zx
=
2(1+
1
)
zx
E
1
(8.1f)
The inverse of the anisotropic elastic material stiffness matrix, (D
)
1
, follows from the
above relations. This matrix is symmetric. The regular material stiffness matrix D
can
only be obtained by numerical inversion.
In general, the stratication plane will not be parallel to the global x-z-plane, but the above
relations will generally hold for a local (n,s,t) coordinate system where the stratication
plane is parallel to the s-t-plane. The orientation of this plane is dened by the dip
angle and dip direction (see Section 8.3). As a consequence, the local material stiffness
matrix has to be transformed from the local to the global coordinate system. Therefore we
consider rst a transformation of stresses and strains:
nst
= R

xyz

xyz
= R
1

nst
(8.2a)
nst
= R

xyz

xyz
= R
1

nst
(8.2b)
8-2 PLAXIS-GiD
where
R
=
_
_
n
2
x
n
2
y
n
2
z
2n
x
n
y
2n
y
n
z
2n
x
n
z
s
2
x
s
2
y
s
2
z
2s
x
s
y
2s
y
s
z
2s
x
s
z
t
2
x
t
2
y
t
2
z
2t
x
t
y
2t
y
t
z
2t
x
t
z
n
x
s
x
n
y
s
y
n
z
s
z
n
x
s
y
+n
y
s
x
n
y
s
z
+n
z
s
y
n
z
s
x
+n
x
s
z
s
x
t
x
s
y
t
y
s
z
t
z
s
x
t
y
+s
y
t
x
s
y
t
z
+s
z
t
y
s
x
t
z
+s
z
t
x
n
x
t
x
n
y
t
y
n
z
t
z
n
x
t
y
+n
y
t
x
n
y
t
z
+n
z
t
y
n
z
t
x
+n
x
t
z
_
_
(8.3)
and
R
=
_
_
n
2
x
n
2
y
n
2
z
n
x
n
y
n
y
n
z
n
x
n
z
s
2
x
s
2
y
s
2
z
s
x
s
y
s
y
s
z
s
x
s
z
t
2
x
t
2
y
t
2
z
t
x
t
y
t
y
t
z
t
x
t
z
2n
x
s
x
2n
y
s
y
2n
z
s
z
n
x
s
y
+n
y
s
x
n
y
s
z
+n
z
s
y
n
z
s
x
+n
x
s
z
2s
x
t
x
2s
y
t
y
2s
z
t
z
s
x
t
y
+s
y
t
x
s
y
t
z
+s
z
t
y
s
x
t
z
+s
z
t
x
2n
x
t
x
2n
y
t
y
2n
z
t
z
n
x
t
y
+n
y
t
x
n
y
t
z
+n
z
t
y
n
z
t
x
+n
x
t
z
_
_
(8.4)
n
x
, n
y
, n
z
, s
x
, s
y
, s
z
, t
x
, t
y
and t
z
are the components of the normalized n, s and t-vectors in
global (x,y,z)-coordinates (i.e. 'sines' and 'cosines'; see Section 8.3). For plane condition
n
z
= s
z
=t
x
=t
y
= 0 and t
z
= 1.
It further holds that :
R
T
= R
1
R
T
= R
1
(8.5)
A local stress-strain relationship in (n,s,t)-coordinates can be transformed to a global
relationship in (x,y,z)-coordinates in the following way:
nst
= D
nst
nst
nst
= R
xyz
nst
= R
xyz
_
_
R
xyz
= D
nst
R
xyz
(8.6)
Hence,
xyz
= R
1
nst
R

xyz
(8.7)
Using to above condition (Eq. 8.5):
8-3
xyz
= R
T
nst
R

xyz
= D
xyz

xyz
or D
xyz
= R
T
nst
R
e
(8.8)
Actually, not the D*-matrix is given in local coordinates but the inverse matrix (D
)
1
.
nst
= D
nst
1
nst
nst
= R
xyz
nst
= R
xyz
_
xyz
= R
1
nst
1
R

xyz
= R
T
nst
1
R

xyz
(8.9)
Hence,
D
xyz
1
= R
T
nst
1
R
or D
xyz
=
_
R
T
nst
1
R
_
1
(8.10)
Instead of inverting the (D
nst
)
1
-matrix in the rst place, the transformation is considered
rst, after which the total is numerically inverted to obtain the global material stiffness
matrix D
xyz
.
8.2 PLASTIC BEHAVIOUR IN THREE DIRECTIONS
A maximum of 3 sliding directions (sliding planes) can be dened in the Jointed
Rock model. The rst sliding plane corresponds to the direction of elastic anisotropy.
In addition, a maximum of two other sliding directions may be dened. However, the
formulation of plasticity on all planes is similar. On each plane a local Coulomb condition
applies to limit the shear stress, ||. Moreover, a tension cut-off criterion is used to limit
the tensile stress on a plane. Each plane, i, has its own strength parameters c
i
,
i
,
i
and
t,i
.
In order to check the plasticity conditions for a plane with local (n,s,t)-coordinates it is
necessary to calculate the local stresses from the Cartesian stresses. The local stresses
involve three components, i.e. a normal stress component,
n
, and two independent shear
stress components,
s
and
t
.
i
= T
T
i
(8.11)
where
i
=
_
n

s

t
_
T
(8.12a)
=
_
xx

yy

zz

xy

yz

zx
_
T
(8.12b)
T
T
i
= transformation matrix (3x6), for planei
8-4 PLAXIS-GiD
As usual in PLAXIS, tensile (normal) stresses are dened as positive whereas compression
is dened as negative.
1
y
s
n
sliding plane
x
Figure 8.2 Plane strain situation with a single sliding plane and vectors n, s
Consider a plane strain situation as visualized in Figure 8.2. Here a sliding plane is
considered under an angle
1
(= dip angle) with respect to the x-axis. In this case the
transformation matrix T
T
becomes:
T
T
=
_
_
s
2
c
2
0 2sc 0 0
sc sc 0 s
2
+c
2
0 0
0 0 0 0 c s
_
_
(8.13)
where
s = sin
1
c = cos
1
In the general three-dimensional case the transformation matrix is more complex, since it
involves both the dip angle and the dip direction (see Section 8.3):
T
T
=
_
_
n
2
x
n
2
y
n
2
z
2n
x
n
y
2n
y
n
z
2n
z
n
x
n
x
s
x
n
y
s
y
n
z
s
z
n
x
s
y
+n
y
s
x
n
z
s
y
+n
y
s
z
n
z
s
x
+n
x
s
z
n
x
t
x
n
y
t
y
n
z
t
z
n
y
t
x
+n
x
t
y
n
y
t
z
+n
z
t
y
n
z
t
x
+n
x
t
z
_
_
(8.14)
Note that the general transformation matrix, T
T
, for the calculation of local stresses
corresponds to rows 1, 4 and 6 of R
(see Eq. 8.3).

8-5
After having determined the local stress components, the plasticity conditions can be
checked on the basis of yield functions. The yield functions for plane i are dened as:
f
c
i
=|
s
|+
n
tan
i
c
i
(Coulomb) (8.15a)
f
t
i
=
n
t,i
(
t,i
c
i
cot
i
) (Tension cut-off) (8.15b)
Figure 8.3 visualizes the full yield criterion on a single plane.
||
t,i
c
i
n
Figure 8.3 Yield criterion for individual plane
The local plastic strains are dened by:
p
j
=
j
g
c
j

j
(8.16)
where g
j
is the local plastic potential function for plane j:
g
c
j
=
+
n
tan
j
c
j
(Coulomb) (8.17a)
g
t
j
=
n
t, j
(Tension cut-off) (8.17b)
The transformation matrix, T, is also used to transform the local plastic strain increments
of plane j,
p
j
, into global plastic strain increments,
p
:
p
= T
j
p
j
(8.18)
The consistency condition requires that at yielding the value of the yield function must
remain zero for all active yield functions. For all planes together, a maximum of 6 yield
functions exist, so up to 6 plastic multipliers must be found such that all yield functions
are at most zero and the plastic multipliers are non-negative.
f
c
i
= f
c(e)
i

np
j=1
<

c
j
>
f
c
i
T
T
T
i
DTj
g
c
j

np
j=1
<

t
j
>
f
c
i
T
T
T
i
DTj
g
t
j
8-6 PLAXIS-GiD
(8.19a)
f
t
i
= f
t(e)
i

np
j=1
<

c
j
>
f
t
i
T
T
T
i
DTj
g
c
j

np
j=1
<

t
j
>
f
t
i
T
T
T
i
DTj
g
t
j
(8.19b)
This means nding up to 6 values of
i
0 such that all f
i
0 and
i
f
i
= 0
When the maximum of 3 planes are used, there are 2
6
= 64 possibilities of (combined)
yielding. In the calculation process, all these possibilities are taken into account in order
to provide an exact calculation of stresses.
8.3 PARAMETERS OF THE JOINTED ROCK MODEL
Most parameters of the jointed rock model coincide with those of the isotropic Mohr-
Coulomb model. These are the basic elastic parameters and the basic strength parameters.
Elastic parameters as in Mohr-Coulomb model (see Section 3.3):
E
1
: Young's modulus for rock as a continuum [kN/m
2
]
1
: Poisson's ratio for rock as a continuum [-]
Anisotropic elastic parameters 'Plane 1' direction (e.g. stratication direction):
E
2
: Young's modulus in 'Plane 1' direction [kN/m
2
]
G
2
: Shear modulus in 'Plane 1' direction [kN/m
2
]
2
: Poisson's ratio in 'Plane 1' direction [-]
Strength parameters in joint directions (Plane i=1, 2, 3):
c
i
: Cohesion [kN/m
2
]
i
: Friction angle [
i
: Dilatancy angle [
t,i
: Tensile strength [kN/m
2
]
Denition of joint directions (Plane i=1, 2, 3):
n : Numer of joint directions (1 n 3) [-]
1,i
: Dip angle [
2,i
: Dip direction [
]
8-7
Figure 8.4 Parameters for the Jointed Rock model
Elastic parameters
The elastic parameters E
1
and
1
are the (constant) stiffness (Young's modulus) and
Poisson's ratio of the rock as a continuum according to Hooke's law, i.e. as if it would
not be anisotropic.
Elastic anisotropy in a rock formation may be introduced by stratication. The stiffness
perpendicular to the stratication direction is usually reduced compared with the general
stiffness. This reduced stiffness can be represented by the parameter E
2
, together with a
second Poisson's ratio,
2
. In general, the elastic stiffness normal to the direction of elastic
anisotropy is dened by the parameters E
2
and
2
.
Elastic shearing in the stratication direction is also considered to be 'weaker' than elastic
8-8 PLAXIS-GiD
shearing in other directions. In general, the shear stiffness in the anisotropic direction can
explicitly be dened by means of the elastic shear modulus G
2
. In contrast to Hooke's
law of isotropic elasticity, G
2
is a separate parameter and is not simply related to Young's
modulus by means of Poisson's ratio (see Eq. 8.1d and e).
If the elastic behaviour of the rock is fully isotropic, then the parameters E
2
and
2
can be
simply set equal to E
1
and
1
respectively, whereas G
2
should be set to E
1
/(1+
1
).
Strength parameters
Each sliding direction (plane) has its own strength properties c
i
,
i
and
t,i
and dilatancy
angle
i
. The strength properties c
i
and
i
determine the allowable shear strength
according to Coulomb's criterion and
t
determines the tensile strength according to
the tension cut-off criterion. The latter is displayed after pressing Advanced button. By
default, the tension cut-off is active and the tensile strength is set to zero. The dilatancy
angle,
i
, is used in the plastic potential function g, and determines the plastic volume
expansion due to shearing.
Denition of joint directions
It is assumed that the direction of elastic anisotropy corresponds with the rst direction
where plastic shearing may occur ('plane 1'). This direction must always be specied. In
the case the rock formation is stratied without major joints, the number of sliding planes
(= sliding directions) is still 1, and strength parameters must be specied for this direction
anyway. A maximum of three sliding directions can be dened. These directions may
correspond to the most critical directions of joints in the rock formation.
The sliding directions are dened by means of two parameters: The Dip angle (
1
) (or
shortly Dip) and the Dip direction (
2
). Instead of the latter parameter, it is also common
in geology to use the Strike. However, care should be taken with the denition of Strike,
and therefore the unambiguous Dip direction as mostly used by rock engineers is used in
PLAXIS. The denition of both parameters is visualized in Figure 8.5.
Consider a sliding plane, as indicated in Figure 8.5. The sliding plane can be dened by
the vectors (s, t), which are both normal to the vector n. The vector n is the 'normal' to
the sliding plane, whereas the vector s is the 'fall line' of the sliding plane and the vector
t is the 'horizontal line' of the sliding plane. The sliding plane makes an angle
1
with
respect to the horizontal plane, where the horizontal plane can be dened by the vectors
(s
, t), which are both normal to the vertical y-axis. The angle
1
is the dip angle, which is
dened as the positive 'downward' inclination angle between the horizontal plane and the
sliding plane. Hence,
1
is the angle between the vectors s
and s, measured clockwise

from s
to s when looking in the positive t-direction. The dip angle must be entered in the
range [0
, 90

].
The orientation of the sliding plane is further dened by the dip direction,
2
, which is
the orientation of the vector s
with respect to the North direction (N). The dip direction

is dened as the positive angle from the North direction, measured clockwise to the
8-9
2
y
N
t
s
n
s
sliding plane
Figure 8.5 Denition of dip angle and dip direction
horizontal projection of the fall line (=s
-direction) when looking downwards. The dip

direction is entered in the range [0
, 360
].
In addition to the orientation of the sliding planes it is also known how the global
(x,y,z) model coordinates relate to the North direction. This information is contained in
the Declination parameter, as dened in the General settings in the Input program. The
Declination is the positive angle from the North direction to the positive z-direction of the
model.
y
s
x
z
2
N
declination
Figure 8.6 Denition of various directions and angles in the horiziontal plane
In order to transform the local (n,s,t) coordinate system into the global (x,y,z) coordinate
system, an auxiliary angle
3
is used internally, being the difference between the Dip
8-10 PLAXIS-GiD
direction and the Declination:
3
=
2
Declination (8.20)
Hence,
3
is dened as the positive angle from the positive z-direction clockwise to the
s*-direction when looking downwards.
From the denitions as given above, it follows that:
n =
_
_
n
x
n
y
n
z
_
_
=
_
_
sin
1
sin
3
cos
1
sin
1
cos
3
_
_
(8.21a)
s =
_
_
s
x
s
y
s
z
_
_
=
_
_
cos
1
sin
3
sin
1
cos
1
cos
3
_
_
(8.21b)
t =
_
_
t
x
t
y
t
z
_
_
=
_
_
cos
3
0
sin
3
_
_
(8.21c)
Figure 8.7 shows some examples of how sliding planes occur in a 3D models for different
values of
1
,
2
and Declination. As it can be seen, for plane strain conditions (the cases
considered in 2D) only
1
is required. By default,
2
is xed at 90
and the declination is

set to 0
.
8-11
1
= 45
1
= 45
1
= 45
2
= 0
2
= 0
Declination = 0
Declination = 0
Declination = 90
2
= 90
x
x
x
y
y
y
z
z
z
Figure 8.7 Examples of failure directions dened by
1
,
2
and Declination
8-12 PLAXIS-GiD
USER-DEFINED SOIL MODELS
9 USER-DEFINED SOIL MODELS
9.1 INTRODUCTION
PLAXIS has a facility for user-dened (UD) soil models. This facility allows users to
implement a wide range of constitutive soil models (stress-strain-time relationship) in
PLAXIS. Such models must be programmed in FORTRAN (or another programming
language), then compiled as a Dynamic Link Library (DLL) and then added to the PLAXIS
program directory.
In principle the user provides information about the current stresses and state variables
and PLAXIS provides information about the previous ones and also the strain and time
increments. In the material data base of the PLAXIS input program, the required model
parameters can be entered in the material data sets.
t+t
i j
,
t+t
current stresses and state variables
t
i j
,
t
previous stresses and state variables
i j
, t strain and time increments
As an example, a UD subroutine based on the Drucker-Prager material model is provided
in the user-dened soil models directory, which is included in the program CD. In this
section, a step-by-step description on how a user-dened soil model can be formed and
utilised in PLAXIS is presented.
Hint: Please note that the PLAXIS organization cannot be held responsible for any
malfunctioning or wrong results due to the implementation and/or use of user-
dened soil models.
9.2 IMPLEMENTATION OF UD MODELS IN CALCULATIONS PROGRAM
The PLAXIS calculations program has been designed to allow for User-dened soil
models. There are mainly four tasks (functionalities) to be performed in the calculations
program:
Initialisation of state variables
Calculation of constitutive stresses (stresses computed from the material model
at certain step)
Creation of effective material stiffness matrix
Creation of elastic material stiffness matrix
These main tasks (and other tasks) have to be dened by the user in a subroutine called
'User_Mod'. In this subroutine more than one UD soil model can be dened. If a UD soil
9-1
model is used in an application, the calculation program calls the corresponding task from
the subroutine User_Mod. To create a UDsoil model, the User_Mod subroutine must have
the following structure:
Subroutine User_Mod (IDTask, iMod, IsUndr, iStep, iTer, Iel,
Int, X, Y, Z, Time0, dTime, Props, Sig0,
Swp0, StVar0, dEps, D, Bulk_W, Sig,
Swp, StVar, ipl, nStat, NonSym, StrsDep,
iTimeDep, iTang, iPrjDir, iPrjLen, iAbort)
where:
IDTask = Identication of the task (1 = Initialise state variables; 2 =
Calculate constitutive stresses; 3 = Create effective material
stiffness matrix; 4 = Return the number of state variables; 5
= Return matrix attributes (NonSym, iStrsDep, iTimeDep); 6 =
Create elastic material stiffness matrix)
iMod = User-dened soil models number (This option allows for more
than one UD model.)
IsUndr = Drained condition (IsUndr = 0) or undrained condition (IsUndr
= 1)
iStep = Current calculation step number
iter = Current iteration number
Iel = Current element number
Int = Current local stress point number (1..3 for 6-noded elements,
or 1..12 for 15-noded elements)
X,Y,Z = Global coordinates of current stress point
Time0 = Time at the start of the current step
dTime = Time increment of current step
Props = Array(1..50) with User-dened model parameters for the
current stress point
Sig0 = Array(1..6) with previous (= at the start of the current step)
effective stress components of the current stress point ('
0
xx
,
'
0
yy
, '
0
zz
, '
0
xy
, '
0
yz
, '
0
zx
). In 2D calculations
yz
and
zx
should be zero.
9-2 PLAXIS-GiD
Swp0 = Previous excess pore pressure of the current stress point
StVar0 = Array(1..nStat) with previous values of state variables of the
dEps = Array(1..6) with strain increments of the current stress point in
the current step (
xx
,
yy
,
zz
,
xy
,
yz
,
zx
)
D = Effective material stiffness matrix of the current stress point
(1..6, 1..6)
Bulk_W = Bulk modulus of water for the current stress point (for
undrained calculations and consolidation)
Sig = Array (1..6) with resulting constitutive stresses of the current
stress point ('
xx
, '
yy
, '
zz
, '
xy
, '
yz
, '
zx
)
Swp = Resulting excess pore pressure of the current stress point
StVar = Array(1..nStat) with resulting values of state variables for the
ipl = Plasticity indicator: 0 = no plasticity, 1 = Mohr-Coulomb
(failure) point; 2 = Tension cut-off point, 3 = Cap hardening
point, 4 = Cap friction point, 5 = Friction hardening point.
nStat = Number of state variables (unlimited)
NonSym = Parameter indicating whether the material stiffness matrix is
non-symmetric (NonSym = 1) or not (NonSym = 0) (required
for matrix storage and solution).
iStrsDep = Parameter indicating whether the material stiffness matrix is
stress-dependent (iStrsDep = 1) or not (iStrsDep = 0).
iTimeDep = Parameter indicating whether the material stiffness matrix is
time-dependent (iTimeDep = 1) or not (iTimeDep = 0).
iTang = Parameter indicating whether the material stiffness matrix is a
tangent stiffness matrix, to be used in a full Newton-Raphson
iteration process (iTang = 1) or not (iTang = 0).
iPrjDir = Project directory (for debugging purposes)
iPrjLen = Length of project directory name (for debugging purposes)
9-3
iAbort = Parameter forcing the calculation to stop (iAbort = 1).
In the above, 'increment' means 'the total contribution within the current step' and not per
iteration. 'Previous' means 'at the start of the current step', which is equal to the value at
the end of the previous step.
In the terminology of the above parameters it is assumed that the standard type of
parameters is used, i.e. parameters beginning with the characters A-H and O-Z are double
(8-byte) oating point values and the remaining parameters are 4-byte integer values.
The parameters IDTask to dEps and iPrjDir and iPrjLen are input parameters; The
values of these parameters are provided by PLAXIS and can be used within the subroutine.
These input parameters should not be modied (except for StVar0 in case IDTask =
1). The parameters D to iTang and iAbort are output parameters. The values of these
parameters are to be determined by the user. In case IDTask = 1, StVar0 becomes output
parameter.
The user subroutine should contain program code for listing the tasks and output
parameters (IDTask = 1 to 6). After the declaration of variables, the User_Mod subroutine
must have the following structure (here specied in pseudo code):
Case IDTask of
1 Begin
{ Initialise state variables StVar0 }
End
2 Begin
{ Calculate constitutive stresses Sig (and Swp) }
End
3 Begin
{ Create effective material stiffness matrix D }
End
4 Begin
{ Return the number of state variables nStat }
End
5 Begin
{ Return matrix attributes NonSym, iStrsDep,
iTimeDep }
End
6 Begin
{ Create elastic material stiffness matrix De }
End
End Case
If more than one UD model is considered, distinction should be made between different
models, indicated by the UD model number iMod.
9-4 PLAXIS-GiD
Initialise state variables (IDTask = 1)
State variables (also called the hardening parameters) are, for example, used in hardening
models to indicate the current position of the yield loci. The update of state variables
is considered in the calculation of constitutive stresses based on the previous value of
the state variables and the new stress state. Hence, it is necessary to know about the
initial value of the state variables, i.e. the value at the beginning of the calculation step.
Within a continuous calculation phase, state variables are automatically transferred from
one calculation step to another. The resulting value of the state variable in the previous
step, StVar, is stored in the output les and automatically used as the initial value in the
current step, StVar0. When starting a new calculation phase, the initial value of the state
variables is read from the output le of the previous calculation step and put in the StVar0
array. In this case it is not necessary to modify the StVar0 array.
However, if the previous calculation step does not contain information on the state
variables (for example in the very rst calculation step), the StVar0 array would contain
zeros. For this case the initial value has to be calculated based on the actual conditions
(actual stress state) at the beginning of the step. Consider, for example, the situation
where the rst state variable is the minimum mean effective stress, p' (considering
that compression is negative). If the initial stresses have been generated using the K
0
-
procedure, then the initial effective stresses are non-zero, but the initial value of the state
variable is zero, because the initialization of this user-dened variable is not considered in
the K
0
-procedure. In this case, part 1 of the user subroutine may look like:
1 Begin
{ Initialise state variables StVar0}
p = (Sig0[1] + Sig0[2] + Sig0[3] ) / 3.0
StVar0[1] = Min(StVar0[1] ,p)
End
Calculate constitutive stresses (IDTask = 2)
This task constitutes the main part of the user subroutine in which the stress integration
and correction are performed according to the user-dened soil model formulation. Let us
consider a simple example using a linear elastic D-matrix as created under IDTask = 3.
In this case the stress components, Sig, can directly be calculated from the initial stresses,
Sig0, the material stiffness matrix, D, and the strain increments, dEps: Sig[i] = Sig0[i]
+ (D[i, j]*dEps[ j] ). In this case, part 2 of the user subroutine may look like:
2 Begin
{ Calculate constitutive stresses Sig (and Swp) }
For i=1 to 6 do
Sig[i] = Sig0[i]
For j=1 to 6 do
Sig[i] = Sig[i] + D[i,j]
*
dEps[j]
9-5
End for {j}
End for {i}
End
Create effective material stiffness matrix (IDTask = 3)
The material stiffness matrix, D, may be a matrix containing only the elastic components
of the stress-strain relationship (as it is the case for the existing soil models in PLAXIS),
or the full elastoplastic material stiffness matrix (tangent stiffness matrix). Let us consider
the very simple example of Hooke's law of isotropic linear elasticity. There are only two
model parameters involved: Young's modulus, E, and Poisson's ratio, . These parameters
are stored, respectively, in position 1 and 2 of the model parameters array, Props(1..50).
In this case, part 3 of the user subroutine may look like:
3 Begin
{ Create effective material stiffness matrix D}
{ Create effective material stiffness matrix D }
E = Props[1]
v = Props[2]
G = 0.5
*
E/(1.0+v)
Fac = 2
*
G/(1.0-2
*
v) { make sure that v < 0.5 !! }
Term1 = Fac
*
(1-v)
Term2 = Fac
*
v
D[1,1] = Term1
D[1,2] = Term2
D[1,3] = Term2
D[2,1] = Term2
D[2,2] = Term1
D[2,3] = Term2
D[3,1] = Term2
D[3,2] = Term2
D[3,3] = Term1
D[4,4] = G
D[5,5] = G
D[6,6] = G
End
(By default, D will be initialized to zero, so the remaining terms are still zero; however, it
is a good habit to explicitly dene zero terms as well.)
If undrained behaviour is considered (IsUndr = 1), then a bulk stiffness for water
(Bulk_W) must be specied at the end of part 3. After calling the user subroutine with
IDTask = 3 and IsUndr = 1, PLAXIS will automatically add the stiffness of the water to the
material stiffness matrix Dsuch that: D[i=1..3, j=1..3] = D[i, j]+ Bulk_W. If Bulk_W
is not specied, PLAXIS will give it a default value of 100*Avg(D[i=1..3, j=1..3]).
9-6 PLAXIS-GiD
Return the number of state variables (IDTask = 4)
This part of the user subroutine returns the parameter nStat, i.e. the number of state
variables. In the case of just a single state parameter, the user subroutine should look
like:
4 Begin
{ Return the number of state variables nStat }
nStat = 1
End
Return matrix attributes (IDTask = 5)
The material stiffness matrix may be stress-dependent (such as in the Hardening Soil
model) or time-dependent (such as in the Soft Soil Creep model). When using a tangent
stiffness matrix, the matrix may even be non-symmetric, for example in the case of
non-associated plasticity. The last part of the user subroutine is used to initialize the
matrix attributes in order to update and store the global stiffness matrix properly during
the calculation process. For the simple example of Hooke's law, as described earlier,
the matrix is symmetric and neither stress- nor time-dependent. In this case the user
subroutine may be written as:
5 Begin
{ Return matrix attributes NonSym, iStrsDep, }
{ iTimeDep, iTang }
NonSym = 0
iStrsDep = 0
iTimeDep = 0
iTang = 0
End
For NonSym = 0 only half of the global stiffness matrix is stored using a prole structure,
whereas for Nonsym = 1 the full matrix prole is stored.
For iStrsDep = 1 the global stiffness matrix is created and decomposed at the beginning
of each calculation step based on the actual stress state (modied Newton-Raphson
procedure).
For iTimeDep = 1 the global stiffness matrix is created and decomposed every time when
the time step changes.
For iTang = 1 the global stiffness matrix is created and decomposed at the beginning of
each iteration based on the actual stress state (full Newton-Raphson procedure; to be used
in combination with iStrsDep=1).
9-7
Create elastic material stiffness matrix (IDTask = 6)
The elastic material stiffness matrix, D
e
, is the elastic part of the effective material stiffness
matrix as described earlier.
In the case that the effective material stiffness matrix was taken to be the elastic stiffness
matrix, this matrix may just be adopted here. However in the case that an elastoplastic or
tangent matrix was used for the effective stiffness matrix, then the matrix to be created
here should only contain the elastic components.
The reason that an elastic material stiffness matrix is required is because PLAXIS
calculates the current relative global stiffness of the nite element model as a whole (CSP
= Current Stiffness Parameter). The CSP parameter is dened as:
CSP =
Total elastic work
Total work
The elastic material stiffness matrix is required to calculate the total elastic work in
the denition of the CSP. The CSP equals unity if all the material is elastic whereas it
gradually reduces to zero when failure is approached.
The CSP parameter is used in the calculation of the global error. The global error is dened
as:
Global error =
|unbalance f orce|
|currently activated load|+CSP|previously activated load|
The unbalance force is the difference between the external forces and the internal
reactions. The currently activated load is the load that is being activated in the current
calculation phase, whereas the previously activated load is the load that has been activated
in previous calculation phases and that is still active in the current phase.
Using the above denition for the global error in combination with a xed tolerated
error results in an improved equilibrium situation when plasticity increases or failure is
approached. The idea is that a small out-of-balance is not a problem when a situation is
mostly elastic, but in order to accurately calculate failure state, safety factor or bearing
capacity, a stricter equilibrium condition must be adopted.
Part 6 of the user subroutine looks very similar to part 3, except that only elastic
components are considered here. It should be noted that the same variable D is used to
store the elastic material stiffness matrix, whereas in Part 3 this variable is used to store
the effective material stiffness matrix.
6 Begin
{ Create elastic material stiffness matrix D }
D[1,1] =
D[1,2] =
D[1,3] =
.....
9-8 PLAXIS-GiD
D[6,6] =
End
Using predened subroutines from libraries
In order to simplify the creation of user subroutines, a number of FORTRAN subroutines
and functions for vector and matrix operations are available in PLAXIS in specic compiler
libraries (LFUsrLib.lib or DFUsrLib.lib) and in the source code (to be included in the
le with the user subroutine). The available subroutines may be called in by User_Mod
subroutine to shorten the code. An overview of the available subroutines is given in
Appendix B.
Denition of user-interface functions
In addition to the user-dened model itself it is possible to dene functions that will
facilitate its use within the Plaxis user-interface. If available, Plaxis Input will retrieve
information about the model and its parameters using the procedures described hereafter.
procedure GetModelCount(var C:longint) ;
C = number of models (return parameter)
This procedure retrieves the number of models that have been dened in the DLL. Plaxis
assumes that model IDs are successive starting at model ID = 1.
procedure GetModelName(var iModel : longint;
var Name : shortstring) ;
iModel = User-dened soil models number to retrieve the name for (input
parameter)
Name = model name (return parameter)
This procedure retrieves the names of the models dened in the DLL.
procedure GetParamCount(var iModel : longint; var C : longint) ;
iModel = User-dened soil models number (input parameter)
C = number of parameters for the specied model (return
parameter)
This procedure retrieves the number of parameters of a specic model.
procedure GetParamName(var iModel,iParam : longint;
var Name : shortstring);
iParam = Parameter number (input parameter)
9-9
Name = parameter name (return parameter)
This procedure retrieves the parameter name of a specic parameter.
Procedure GetParamUnit(var iModel,iParam : longint;
var Units : shortstring) ;
iParam = Parameter number (input parameter)
Units = parameter units (return parameter)
This procedure retrieves the parameter units of a specic parameter. Since the chosen
units are dependent on the units of length, force and time chosen by the user the following
characters should be used for dening parameter units:
'L' or 'l' for units of length
'F' or 'f' for units of force
'T' or 't' for units of time.
For model names, model parameter names and model parameter units special characters
can be used for indicating subscript, superscript or symbol font (for instance for Greek
characters).
^ : From here characters will be superscript
_ : From here characters will be subscript
@ : From here characters will be in symbol font
# : Ends the current superscript or subscript. Pairs of '^..#', '_. . . #'and '@. . . #'can be
nested.
For example:
A UD model parameter uses the oedometer stiffness as parameter. The parameter name
can be dened as 'E_oed#'and its units as 'F/L^2#'.
All procedures are dened in Pascal but equivalent procedures can be created, for
instance in a Fortran programming language. Please make sure that the data format of the
parameters in the subroutine headers is identical to those formulated before. For instance,
the procedures mentioned above use a "shortstring" type; a "shortstring" is an array of 256
characters where the rst character contains the actual length of the shortstring contents.
Some programming languages only have null-terminated strings; in this case it may be
necessary to use an array of 256 bytes representing the ASCII values of the characters to
return names and units. An example of Fortran subroutines is included on the program
CD.
9-10 PLAXIS-GiD
Compiling the user subroutine
The user subroutine User_Mod has to be compiled into a DLL le using an appropriate
compiler. Note that the compiler must have the option for compiling DLL les. Below are
examples for two different FORTRAN compilers. It is supposed that the user subroutine
User_Mod is contained in the le USRMOD.FOR.
After creating the user subroutine User_Mod, a command must be included to export data
to the DLL.
The following statement has to be inserted in the subroutine just after the declaration of
variables:
Using Lahey Fortran (LF90, LF95, ...): DLL_Export User_Mod
Using Digital Visual Fortran: !DEC$ ATTRIBUTES DLLExport :: User_Mod
In order to compile the USRMOD.FOR into a DLL le, the following command must be
executed:
Using Lahey Fortran 90: LF90 -win -dll USRMOD.FOR -lib LFUsrLib
Using Lahey Fortran 95: LF95 -win -dll USRMOD.FOR -lib LFUsrLib -ml
LF90
Using Digital Visual Fortran: DF /winapp USRMOD.FOR DFUsrLib.lib /dll
In all cases USRMOD.DLL le will be created. This le should be placed in
the "db" folder, "calculate" folder as well as the "output" folder in the "PLAXIS
program directory\problemtypes\Plaxis#\plaxis.gid\exec" path. Thereafter, it can be
used together with the existing PLAXIS material database (MATDB.EXE), the calculations
program (PLASW3DF.EXE) and Output program (GIDOUT.EXE). Once the UD model
is used, PLAXIS will execute the commands as listed in the USRMOD.DLL le.
Debugging possibilities
When making computer programs, usually some time is spent to 'debug' code earlier
written source code. In order to be able to effectively debug the user subroutine, there
should be a possibility for the user to write any kind of data to a le. Such a 'debug-le' is
not automatically available and has to be created in the user subroutine.
In Appendix C a suggestion on how this can be done is given. After the debug-le is
created, data can be written to this le from within the user subroutine. This can be done
by using, for example, the availably written subroutines (see Appendix B).
9.3 INPUT OF UD MODEL PARAMETERS VIA USER-INTERFACE
Input of the model parameters for user-dened soil models can be done using PLAXIS
material data base. In fact, the procedure is very similar to the input of parameters for the
existing PLAXIS models.
9-11
When creating a new material data set for soil and interfaces in the material data base,
a window appears with three tab sheets: General, Parameters, Interface, Figure 9.1. A
user-dened model can be selected from the Material model combo box in the General
tab sheet.
After inputting general properties, the appropriate UD model can be chosen from the
available models that have been found by PLAXIS Input.
The Parameters tab sheet shows two combo boxes at the top; the left combo box lists all
the DLLs that contain valid UD models and the right combo box shows the models dened
in the selected DLL. Each UD model has its own set of model parameters, dened in the
same DLL that contains the model denition.
When an available model is chosen PLAXIS will automatically read its parameter names
and units from the DLL and ll the parameter table below.
Interfaces
The Interfaces tab sheet, Figure 9.2, contains the material data for interfaces.
Normally, this tab sheet contains the R
inter
parameter. For user-dened soil models the
interface tab sheet is slightly different and contains the interface oedometer modulus,
E
re f
oed
, and the interface strength parameters c
inter
,
inter
and
inter
. Hence, the interface
shear strength is directly given in strength parameters instead of using a factor relating the
interface shear strength to the soil shear strength, as it is the case in PLAXIS models.
After having entered values for all parameters, the data sets can be assigned to the
corresponding soil clusters, in a similar way as for the existing material models in PLAXIS.
The user-dened parameters are transmitted to the calculation program and appear for the
appropriate stress points as Props(1..50) in the User_Mod subroutine.
9-12 PLAXIS-GiD
a. Selection of user-dened soil models
b. Input of parameters
Figure 9.1 Selection window
9-13
Figure 9.2 Interface tab sheet
9-14 PLAXIS-GiD
STRUCTURAL BEHAVIOUR
10 STRUCTURAL BEHAVIOUR
10.1 ANCHORS
The elastic behaviour of an anchor involves only a relationship between axial force N and
displacement (elongation) u of the form:
N =
EA
L
u (10.1)
The anchor stiffness EA is dened by the user based on the material stiffness E and cross
section A.
In case of elastoplastic behaviour of the anchor the maximum tension force is bound by
F
max,tens
and the maximum compression force is bound by F
max,comp
.
10.2 BEAMS
The local system of axes of a beam element is such that the rst axis corresponds with
the axial beam direction. The second and third axis are always perpendicular to the beam
axis.
1
2
3
a. Local Axes
N

b. Axial force N
Q
12
c. Shear force Q
12
Q
13
d. Shear force Q
13
Figure 10.1 Axial force and shear forces
Elastic behaviour of beam elements is dened by the following parameters:
A: Beam cross section area
E: Young's modulus in axial direction
I
2
: Moment of inertia against bending around the second axis
I
3
: Moment of inertia against bending around the third axis
The relationships between the different force components and the strain / gradient /
curvature components are dened as:
10-1
1
2
I
3
M
3

3
a. Bending moment M
3
1
3
I
2
M
2

2
3
b. Bending moment M
2
Figure 10.2 Bending moments
N = EA M
2
= EI
2
2
Q
12
= kGA
12
M
3
= EI
3
3
(10.2)
Q
13
= kGA
13
In which k is the shear correction factor, which is taken as
5
/
6
. The shear modulus is taken
as G =
1
/
2
E.
10.3 GEOGRIDS
The PLAXIS program allows for isotropic behaviour of geogrid elements, which is dened
by the parameter EA. This stiffness is dened by the user and is based on the material
tension stiffness E and the cross section A. Geogrid elements cannot sustain compression
forces. In case of elastoplastic behaviour the maximum tension force in any direction is
bound by N
p
.
The relationship between the force and the strain is dened as:
N
1
= EA
1
N
2
= EA
2
(10.3)
Q
12
= GA
12
Where GA =
1
/
2
EA.
10-2 PLAXIS-GiD
10.4 PLATES
The PLAXIS program allows for orthotropic material behaviour in plate elements, which
is dened by the following parameters:
E
1
: Young's modulus in rst axial direction
E
2
: Young's modulus in second axial direction
G
12
: In-plane shear modulus
G
13
: Out-of-plane shear modulus related to shear deformation over rst direction
G
23
: Out-of-plane shear modulus related to shear deformation over second direction
12
: Poisson's ratio
The material behaviour in plate elements is dened by the following relationship
between strains and stresses, which is based on the general three-dimensional continuum
mechanics theory and the assumption that
33
= 0.
_
11
22
12
13
23
_
_
=
_
_
1/E
1

12
/E
1
0 0 0
12
/E
1
1/E
2
0 0 0
0 0 1/G
12
0 0
0 0 0 1/kG
13
0
0 0 0 0 1/kG
23
_
_
_
11
22
12
13
23
_
_
(10.4)
In which k is the shear correction factor, which is taken as
5
/
6
.
Inverting this relationship and ignoring the higher order terms in gives the following
stress-strain relationship:
_
11
22
12
13
23
_
_
=
_
_
E
1

12
E
2
0 0 0
12
E
2
E
2
0 0 0
0 0 G
12
0 0
0 0 0 kG
13
0
0 0 0 0 kG
23
_
_
_
11
22
12
13
23
_
_
(10.5)
This approximation holds as long as the Poisson's ratio is small.
These stress-strain relationships can be transformed into relationships for structural forces:
_
_
N
1
N
2
_
_
=
_
_
E
1
d
12
E
2
d
12
E
2
d E
2
d
_
_
_
_
2
_
_
(10.6a)
10-3
_
_
Q
12
Q
13
Q
23
_
_
=
_
_
G
12
d 0 0
0 kG
13
d 0
0 0 kG
23
d
_
_
_
12
13
23
_
_
(10.6b)
_
_
M
11
M
22
M
12
_
_
=
_
_
E
1
d
3
12
12
E
2
d
3
12
0
12
E
2
d
3
12
E
2
d
3
12
0
0 0
G
12
d
3
12
_
_
_
11
22
12
_
_
(10.6c)
In which d is the thickness of the plate, which also determines the distributed weight of the
plate together with the unit weight of the plate material: d. The modied shear strain
takes into account the shear strain and some additional terms in order to give a more
accurate approximation of the problem.
The local system of axes in a plate element is such that the rst and the second local axis
lie in the plane of the plate whereas the third axis is perpendicular to the plane of the plate
(Figure 10.3).
When geometric orthotropy is considered rather than material orthotropy, the following
relationships for structural forces apply:
_
_
N
1
N
2
_
_
=
_
_
EA
1
1
2
EA
1
1
2
EA
1
1
2
EA
2
1
2
_
_
_
_
2
_
_
_
_
EA
1
EA
1
EA
1
EA
2
_
_
_
_
2
_
_
(10.7a)
_
_
Q
12
Q
13
Q
23
_
_
=
_
_
EA
12
2(1+)
0 0
0
EA
13
2(1+)
0
0 0
EA
23
2(1+)
_
_
_
12
13
23
_
_
(10.7b)
_
_
M
11
M
22
M
12
_
_
=
_
_
EI
1
1
2
EI
1
1
2
0
EI
1
1
2
EI
2
1
2
0
0 0
EI
12
2(1+)
_
_
_
11
22
12
_
_
EI
1
EI
1
0
EI
1
EI
2
0
0 0
EI
12
2(1+)
_
_
_
11
22
12
_
_
(10.7c)
where the approximations hold for a small Poisson's ratio. In these relations the following
alternative parameters are used:
A
1
: Effective material cross section area for axial forces in the rst direction
A
2
: Effective material cross section area for axial forces in the second direction
10-4 PLAXIS-GiD
1
2 3
N
1
E
1

1
N
2
E
2

2
Q
12
G
12

12
Q
13
G
13

13
Q
23
G
23

23
M
12

12
M
11

11
M
22

22
Figure 10.3 Denition of positive normal forces (N), shear forces (Q) and bending
moments (M) for a plate based on local system of axes
A
12
: Effective material cross section area for shear forces Q
12
A
13
: Effective material cross section area for shear forces Q
13
A
23
: :Effective material cross section area for shear forces Q
23
I
1
: Moment of inertia against bending over the rst axis
I
2
: Moment of inertia against bending over the second axis
I
12
: Moment of inertia against torsion
10-5
In order to use the available plate elements for geometric orthotropy, the basic material
parameters should be chosen in such a way that the resulting normal stiffness E
1
A is equal
to the normal stiffness EA
1
of the plate. Here E
1
is the input value for the Young's modulus
in the rst direction, A is the internally calculated area of the plate, E is the actual Young's
modulus of the material and A
1
is the cross sectional area of the element to be modelled.
Similar parameter choices should be made for the other normal stiffness, exural rigidities
and shear stiffnesses. However, for a given choice of the equivalent plate thickness d it will
not be possible to match all stiffness components exactly. The thickness d is the equivalent
plate thickness such that the average distributed weight of the plate corresponds to d.
When bending is considered as the most important type of deformation, the following
choices are recommended:
E
1
= 12EI
1
/d
3
E
2
= 12EI
2
/d
3
G
12
=
6EI
12
(1+)d
3
G
13
=
EA
13
2(1+)d
G
23
=
EA
23
2(1+)d
(10.8)
12
=
In this case the resulting exural rigidities E
1
I
1
and E
2
I
2
and shear stiffnesses G
13
A
13
and
G
23
A
23
prove to be independent of the chosen value for equivalent plate thickness. Only
the normal stiffnesses E
1
A and E
2
A and shear stiffness G
12
A
12
are not independent of the
chosen value of the equivalent plate thickness, and a suitable selection for d has to be
made. What is the most suitable selection for d depends on the construction that is to be
modelled. Two examples are given below.
Engineering examples:
In the following, two types of applications are given that frequently occur in the
engineering practice. The rst example is a sheet-pile wall, as depicted in Figure 10.4.
From the sheet-pile manufacturer, the following properties are known: t (wall thickness),
h (total height), A (per m wall width), I
1
, E
steel
and
steel
.
1
1
1
2
2
2
3
3
3
I
1
,
E
1
I
2
, E
2
M
11
,
11
M
22
,
22
Figure 10.4 Example of sheet-pile wall with its major quantities
10-6 PLAXIS-GiD
The structure is geometrically orthotropic with signicant different stiffnesses in
horizontal and vertical direction. It is known that the axial stiffness in vertical direction is
larger than the effective stiffness in horizontal direction (E
1
> E
2
). Moreover, the exural
rigidity against bending over the vertical direction, I
1
, is much larger than the stiffness
against bending over the horizontal direction, I
2
, (I
1
>> I
2
say I
1
20I
2
and I
1
>> I
12
say I
1
10I
12
).
Furthermore, it is assumed that the cross section area that is effective against shear
deformation over the vertical direction is about 1/3 of the total cross section area, whereas
the area that is effective against shear deformation over the horizontal direction is about
1/10 of the total cross section area. Finally, the Poisson's ratio's for sheet pile walls can be
assumed zero. With these assumptions, the situation could be modelled by selecting the
model parameters in the following way:
d = h (which is sightly more than t)
E
1
= 12E
steel
I
1
/d
3
E
2
= 12E
steel
I
2
/d
3
I
2
/
I
1
E
1
E
1
/20
G
12
=
6E
steel
I
12
(1+
steel
)d
3
6E
steel
I
1
/10d
3
G
13
=
E
steel
A
13
2(1+
steel
)d
E
steel
A/6d E
steel
A/6d
G
23
=
E
steel
A
23
2(1+
steel
)d
E
steel
A/20d E
steel
A/20d
12
=
steel
0
= A
steel
/d
The second example is a concrete T-shaped oor prole, as depicted in Figure 10.5. In
addition to the precise geometry dimensions, the following properties are known:
E
concrete
,
concrete
The structure is again geometrically orthotropic with signicant different stiffnesses in the
two major oor directions. The exural rigidity against bending over the second direction,
I
2
, is larger than the stiffness against bending over the rst direction, I
1
, (I
2
>I
1
), since I
2
is
10-7
1 1
1
2 2
22
3
3
I
1
E
1
I
2
E
2
M
11
11
M
22
22
w
1
w
2
h
1
h
2
Figure 10.5 Example of concrete oor prole with its major quantities
dominated by h
1
+ h
2
whereas I
1
only depends on h
1
. Furthermore, the cross section area
(per unit of width) that is effective against shear deformation over the second direction
(G
23
) is equal to the total cross section area A, whereas the area that is effective against
shear deformation over the rst direction (G
13
) is equal to h
1
.
With these assumptions, the situation could be modelled by selecting the model
parameters in the following way:
d =
h
1
w
1
+(h
1
+h
2
)w
2
w
1
+w
2
E
1
= 12E
concrete
I
1
/d
3
= 12E
concrete
1
12
h
3
1
/d
3
= E
concrete
h
3
1
/d
3
E
2
= 12E
concrete
I
2
/d
3
where I
2
=
1
12
h
3
1
w
1
+
1
12
(h
1
+h
2
)
3
w
2
w
1
+w
2
G
12
=
6E
concrete
I
12
(1+
concrete
)d
3
where I
12
I
1
=
1
12
h
3
1
G
13
=
E
concrete
A
13
2(1+
concrete
)d
=
E
concrete
h
1
2(1+
concrete
)d
G
23
=
E
concrete
A
23
2(1+
concrete
)d
=
E
concrete
d
2(1+
concrete
)d
=
E
concrete
2(1+
concrete
)
12
=
concrete
10-8 PLAXIS-GiD
REFERENCES
11 REFERENCES
[1] Adachi, T., Oka, F. (1982). Constitutive equation for normally consolidated clays
based on elasto-viscoplasticity. Soils and Foundations, 22, 5770.
[2] Alpan, I. (1970). The geotechnical properties of soils. Earth-Science Reviews, 6,
549.
[3] Atkinson, J.H., Bransby, P.L. (1978). The Mechanics of Soils. McGraw-Hill,
Londen.
[4] Atkinson, J.H., Sallfors, G. (1991). Experimental determination of soil properties.
In Proc. 10th ECSMFE, 3, 915956.
[5] Bairez, J., Hicher, P.Y. (1994). Elementary Mechanics of Soil Behaviour. A A
Balkema, Rotterdam, The Netherlands.
[6] Belytschko, T., Lasry, D. (1989). Localization limiters and numerical strategies for
strain-softening materials. In Proc. France-US Workshop on Strain localization and
size effect due to cracking and Damage editor= (Mazars & Bazant). 349362.
[7] Benz, T. (2006). Small-Strain Stiffness of Soils and its Numerical Consequences.
Ph.d. thesis, Universitt Stuttgart.
[8] Bjerrum, L. (1967). Engineering geology of norwegian normally-consolidated
marine clays as related to settlements of buildings. Seventh Rankine Lecture,
Geotechnique 17, 81118.
[9] Bolton, M.D. (1986). The strength and dilatancy of sands. Gotechnique, 36(1),
6578.
[10] Borja, R.I., Kavaznjian, E. (1985). A constitutive model for the --t behaviour of
wet clays. Geotechnique, 35, 283298.
[11] Borja, R.I., Lee, S.R. (1990). Cam-clay plasticity, part 1: implicit integration of
elasto-plastic constitutive relations. Computer Methods in Applied Mechanics and
Engineering, 78, 4872.
[12] Brinkgreve, R.B.J. (1994). Geomaterial Models and Numerical Analysis of
Softening. Dissertation. Delft University of Technology.
[13] Brinkgreve, R.B.J. (2004). Time-dependent behaviour of soft soils during
embankment construction - a numerical study. Proc. NUMOG IX, 631637.
[14] Brinkgreve, R.B.J., Vermeer, P.A. (1992). On the use of cam-clay models. In Proc.
IV Int. Symposium on Numerical Models in Geomechanics (eds. G.N. Pande, S.
Pietruszczak). Balkema, Rotterdam, volume 2, 557565.
[15] Buisman, K. (1936). Results of long duration settlement tests. Proceedings 1st
International Conference on Soil Mechanics and Foundation Engineering, Mass.
Vol. 1, 103107.
11-1
[16] Burland, J.B. (1965). The yielding and dilation of clay. Gotechnique, 15, 211214.
(Correspondence).
[17] Burland, J.B. (1967). Deformation of Soft Clay. Dissertation. Cambridge
University.
[18] Buttereld, R. (1979). A natural compression law for soils (an advance on e-log p').
Geotechnique, 29, 469480.
[19] Chen, W.F. (1975). Limit analysis and soil plasticity. Elsevier, Amsterdam.
[20] den Haan, E.J. (1994). Vertical Compression of Soils. Thesis, Delft University.
[21] Drucker, D.C., Prager, W. (1952). Soil mechanics and plastic analysis or limit
design. Quart. Appl. Math., 10(2), 157165.
[22] Duncan, J.M., Chang, C.Y. (1970). Nonlinear analysis of stress and strain in soil.
ASCE J. of the Soil Mech. and Found. Div., 96, 16291653.
[23] Fung, Y.C. (1965). Foundations of Solid Mechanics. Prentice-Hall, New Jersey,
USA.
[24] Garlanger, J.E. (1972). The consolidation of soils exhibiting creep under constant
effective stress. Gotechnique, 22, 7178.
[25] Hardin, B.O., Black, W.L. (1969). Closure to vibration modulus of normally
consolidated clays. Proc. ASCE: Journal of the Soil Mechanics and Foundations
Division, 95(SM6), 15311537.
[26] Hardin, B.O., Drnevich, V.P. (1972). Shear modulus and damping in soils: Design
equations and curves. Proc. ASCE: Journal of the Soil Mechanics and Foundations
Division, 98(SM7), 667692.
[27] Hill, R. (1950). The Mathematical Theory of Plasticity. Oxford University Press,
London, U.K.
[28] Janbu, N. (1963). Soil compressibility as determined by oedometer and triaxial
tests. Proc. ECSMFE Wiesbaden, 1, 1925.
[29] Janbu, N. (1969). The resistance concept applied to soils. Proceedings of the 7h
ICSMFE, Mexico City, 1, 191196.
[30] Janbu, N. (1985). Soil models in offshore engineering (25
th
rankine lecture).
Gotechnique, 35, 241280.
[31] Koiter, W.T. (1960). General theorems for elastic-plastic solids. In I.N. Sneddon,
R. Hill (eds.), Progress in Solid Mechanics. North-Holland, Amsterdam, volume 1,
165221.
[32] Kondner, R.L. (1963). A hyperbolic stress strain formulation for sands. 2. Pan. Am.
ICOSFE Brazil, 1, 289324.
[33] Kulhawy, F.H., Mayne, P.W. (1990). Manual on Estimating Soil Properties for
Foundation Design. Cornell University, Ithaca, New York.
11-2 PLAXIS-GiD
REFERENCES
[34] Leroueil, S. (1977). Quelques considrations sur le comportement des argiles
sensibles. Ph.d. thesis, Laval University, Qubec.
[35] Li, X.S., Dafalias, Y.F. (2000). Dilatancy for cohesionless soils. Geotechnique,
50(4), 449460.
[36] Masing, G. (1926). Eigenspannungen und verfestigung beim messing. In In Proc.
2nd Int. Congr. Appl. Mech. Zurich.
[37] Mesri, G., Godlewski, P.M. (1977). Time and stress-compressibility inter-
relationship. Journal of the Geotechnical Engineering Division, ASCE, 103(GT5),
417430.
[38] Muir Wood, D. (1990). Soil Behaviour and Critical State Soil Mechanics.
Cambridge University Press.
[39] Niemunis, A., Herle, I. (1997). I.: Hypoplastic model for cohesionless soils with
elastic strain range. Mechanics of Cohesive Frictional Materials, 2(3), 279299.
[40] Prevost, J.H. (1976). Undrained stress-strain-time behaviour of clays. Journal of
the Geotechnical Engineering Division, GT12, 12451259.
[41] Roscoe, K.H., Burland, J.B. (1968). On the generalized stress-strain behaviour of
wet clay. In In: Heyman & Leckie, Engineering Plasticity, Cambridge University
Press. 535609.
[42] Rowe, P.W. (1962). The stress-dilatancy relation for static equilibrium of an
assembly of particles in contact. In Proc. Roy. Soc. A., No. 269. 500527.
[43] Santos, J.A., Correia, A.G. (2001). reference threshold shear strain of soil. its
application to obtain a unique strain-dependent shear modulus curve for soil. In
Proceedings 15th International Conference on Soil Mechanics and Geotechnical
Engineering. Istanbul, Turkey, volume 1, 267270.
[44] Schanz, T. (1998). Zur Modellierung des Mechanischen Verhaltens von
Reibungsmaterialen. Habilitation, Stuttgart Universitt.
[45] Schanz, T., Vermeer, P.A. (1996). Angles of friction and dilatancy of sand.
Gotechnique, 46, 145151.
[46] Schanz, T., Vermeer, P.A. (1998). Special issue on pre-failure deformation
behaviour of geomaterials. Gotechnique, 48, 383387.
[47] Schanz, T., Vermeer, P.A., Bonnier, P.G. (1999). The hardening-soil model:
Formulation and verication. In R.B.J. Brinkgreve, Beyond 2000 in Computational
Geotechnics, Balkema, Rotterdam. 281290.
[48] Schweiger, H.F. (2002). Results from numerical benchmark exercises in
geotechnics. In P. Mestat (ed.), 5th European Conference Numerical Methods in
Geotechnical Engineering. Numge 2002, Paris, volume 1, 305314.
[49] Sekiguchi, H. (1977). Rheological characteristics of clays. In Proceedings of the
9th ICSMFE. Tokyo, volume 1, 289292.
11-3
[50] Simpson, B. (1992). "retaining structures: displacement and design", the 32
nd
rankine lecture. Geotechnique, 42(4), 541576.
[51] Smith, I.M., Grifth, D.V. (1982). Programming the Finite Element Method. John
Wiley & Sons, Chisester, U.K, second edition.
[52] Stolle, D.F.E. (1991). An interpretation of initial stress and strain methods, and
numerical stability. International Journal for Numerical and Analytical Methods in
Geomechanics, 15, 399416.
[53] Stolle, D.F.E., Bonnier, P.G., Vermeer, P.A. (1997). A soft soil model and
experiences with two integration schemes. In Numerical Models in Geomechanics.
Numog 1997, 123128.
[54] Vaid, Y., Campanella, R.G. (1977). Time-dependent behaviour of undisturbed clay.
ASCE Journal of the Geotechnical Engineering Division, 103(GT7), 693709.
[55] van Langen, H., Vermeer, P.A. (1990). Automatic step size correction for non-
associated plasticity problems. Int. J. Num. Meth. Engng., 29, 579598.
[56] Vermeer, P.A., Borst, R. (1984). Non-associated plasticity for soils, concrete and
rock. Heron, 29(3).
[57] Vermeer, P.A., Neher, H. (1999). A soft soil model that accounts for creep. In R.B.J.
Brinkgreve, Beyond 2000 in Computational Geotechnics, Balkema, Rotterdam.
249261.
[58] Vermeer, P.A., Stolle, D.F.E., Bonnier, P.G. (1998). From the classical theory of
secondary compression to modern creep analysis. Proc. 9th Int. Conf. Comp. Meth.
and Adv. Geomech. , Wuhan, China, 4, 24692478.
[59] Vermeer, P.A., van Langen, H. (1989). Soil collapse computations with nite
elements. In Ingenieur-Archive 59. 221236.
[60] von Soos, P. (1990). Properties of soil and rock (in german). In In:
Grundbautaschenbuch Part 4. Ernst & Sohn, Berlin.
[61] Vucetic, M., Dobry, R. (1991). Effect of soil plasticity on cyclic response. Journal
of Geotechnical Engineering, ASCE, 117(1), 89107.
11-4 PLAXIS-GiD
APPENDIX A - SYMBOLS
APPENDIX A - SYMBOLS
A : Cross section area
c : Cohesion
csp : Current stiffness parameter
C
u
, S
u
: Undrained shear-strength
d : Thickness
D
e
: Elastic material matrix representing Hooke's law
e : Void ratio
E : Young's modulus
E
oed
: Oedometer modulus
f : Yield function
g : Plastic potential function
G : Shear modulus
I : Moment of inertia
K : Bulk modulus
K
0
: Coefcient of lateral earth pressure (initial stress state)
K
nc
0
: Coefcient of lateral earth pressure for a normally
consolidated stress state
m : Power in stress-dependent stiffness relation
M : Slope of critical state line in p' q space
M : Bending moment
n : Porosity
N : Normal force
OCR : Over-consolidation ratio
p : Isotropic stress or mean stress,
negative for pressure; positive for tension
p
p
: Isotropic pre-consolidation stress, negative for pressure
POP : Pre overburden pressure, positive for (over)pressure
q : Equivalent shear stress or deviatoric stress
Q : Shear force
R
f
: Failure ratio
t : Time
u : Vector with displacement components
: Unit weight
: Increment
: Vector with Cartesian strain components, normal components
positive for extension; negative for compression
v
: Volumetric strain,
negative for compression; positive for extension
: Cam-Clay swelling index
: Modied swelling index

: Plastic multiplier
A-1
: Cam-Clay compression index
: Modied compression index
: Modied creep index

: Poisson's ratio
: Vector with Cartesian stress components, normal components
positive for tension; negative for pressure
p
: Vertical pre-consolidation stress, negative for pressure
: Friction angle
: Dilatancy angle
c
: denotes creep component
e
: denotes elastic component
p
: denotes plastic component
re f
: denotes reference value
u
: denotes undrained
ur
: denotes unloading and reloading
A-2 PLAXIS-GiD
APPENDIX B - FORTRAN SUBROUTINES FOR USER-DEFINED SOIL MODELS
APPENDIX B - FORTRAN SUBROUTINES FOR USER-DEFINED SOIL
MODELS
In this appendix, a listing is given of the subroutines and functions which are provided by
PLAXIS in libraries and source cod in the User-dened soil models directory. These can
be called by the User_Mod subroutine:
Subroutines
MZeroR( R, K ):
To initialize K terms of double array R to zero
MZeroI( I, K ):
To initialize K terms of integer array I to zero
SetRVal( R, K, V ):
To initialize K terms of double array R to V
SetIVal( I, K, IV ):
To initialize K terms of integer array I to IV
CopyIVec( I1, I2, K ):
To copy K values from integer array I1 to I2
CopyRVec( R1, R2, K ):
To copy K values from double array R1 to R2
MulVec( V, F, n ):
To multiply a vector V by a factor F, n values
MatVec( xMat, im, Vec, n, VecR ):
Matrix (xMat)-vector(Vec) operation.
First dimension of matrix is im; resulting vector is VecR
AddVec( Vec1, Vec2, R1, R2, n, VecR ):
To add n terms of two vectors; result in VecR
VecR
i
= R1 Vec1
i
+R2 Vec2
i
MatMat( xMat1, id1, xMat2, id2, nR1, nC2,
nC1, xMatR, idR):
Matrix multiplication xMatR
i j
= xMat
1
ik
xMat
2
k j
id1, id2, idR : rst dimension of matrices
nR1 number of rows in xMat1 and resulting xMatR
nC2 number of column in xMat2 and resulting xMatR
nC1 number of columns in xMat2 =rows in xMat2
MatMatSq( n, xMat1, xMat2, xMatR ):
B-1
Matrix multiplication xMatR
i j
= xMat
1
ik
xMat
2
k j
Fully lled square matrices with dimensions n
MatInvPiv( AOrig, B, n ):
Matrix inversion of square matrices AOrig and B with dimensions n.
AOrig is NOT destroyed, B contains inverse matrix of AOrig.
Row-pivoting is used.
WriVal( io, C, V ):
To write a double value V to le unit io (when io > 0)
The value is preceded by the character string C.
WriIVl( io, C, I ):
As WriVal but for integer value I
WriVec( io, C, V, n ):
As WriVal but for n values of double array V
WriIVc( io, C, iV, n ):
As WriVal but for n values of integer array iV
WriMat( io, C, V, nd, nr, nc ):
As WriVal but for double matrix V. nd is rst dimension of V, nr and nc are the
number of rows and columns to print respectively.
PrnSig( iOpt, S, xN1, xN2, xN3, S1, S2, S3, P, Q ):
To determine principal stresses and (for iOpt=1) principal directions.
iOpt = 0 to obtain principal stresses without directions
iOpt = 1 to obtain principal stresses and directions
S array containing 6 stress components (XX, YY, ZZ, XY, YZ, ZX)
xN1, xN2, xN3 array containing 3 values of principal normalized directions
only when iOpt=1.
S1, S2, S3 sorted principal stresses (S S2 S3 )
P isotropic stress (negative for compression)
Q deviatoric stress
CarSig( S1, S2, S3, xN1, xN2, xN3, SNew ):
To calculate Cartesian stresses from principal stresses and principal directions.
S1, S2, S3 principal stresses
xN1, xN2, xN3 arrays containing principal directions (from PrnSig)
SNew contains 6 stress components (XX, YY, ZZ, XY, YZ, ZX)
CrossProd( xN1, xN2, xN3 ):
B-2 PLAXIS-GiD
APPENDIX B - FORTRAN SUBROUTINES FOR USER-DEFINED SOIL MODELS
Cross product of vectors xN1 and xN2
SetVecLen( xN, n, xL ):
To multiply the n components of vector xN such that the length of xN becomes
xL (for example to normalize vector xN to unit length).
Functions
Logical Function LEqual( A, B, Eps ):
Returns TRUE when two values A and B are almost equal, FALSE otherwise.
LEqual = |A-B| < Eps * ( |A| + |B| + Eps ) / 2
Logical Function Is0Arr( A, n ):
Returns TRUE when all n values of real (double) array A are zero, FALSE
otherwise
Logical Function Is0IArr( IArr, n ):
Returns TRUE when all n values of integer array IArr are zero, FALSE otherwise
Double Precision Function DInProd( A, B, n ):
Returns the dot product of two vectors with length n
B-3
B-4 PLAXIS-GiD
APPENDIX C - CREATING A DEBUG-FILE FOR USER-DEFINED SOIL MODELS
APPENDIX C - CREATING A DEBUG-FILE FOR USER-DEFINED SOIL
MODELS
A debug le is not automatically created and opened in PLAXIS. The user should do this
by including the corresponding source code in the user subroutine. The debug le needs
only to be created and opened once. Since the user subroutine is used many times, it must
be checked whether the le, i.e. the corresponding IO unit number, is already open.
When writing a FORTRAN user subroutine and compiling it as a DLL, les are not shared
with the main program. This means that any IO unit number can be used without conicts
between the debug le and existing les used by PLAXIS.
Here suggestions are given on how the debug le can be created and opened:
1: Inquire if a unit number is opened. If not, open it.
Logical IsOpen
Inquire( unit = 1, Opened = IsOpen)
If (.Not. IsOpen) Then
Open( Unit = 1, File = ... )
End If
2: Use a DATA statement
Logical IsOpen
Data IsOpen / .FALSE. /
Save IsOpen
Open( Unit = 1, File = ... )
IsOpen = .TRUE.
End If
The above suggestions assume that the debug le is located in the currently active
directory, which is not necessarily the proper location. It is suggested that the debug
le is stored in the DTA-directory of the corresponding PLAXIS project. Therefore it is
necessary to include also the pathname in the File = ... . The project directory is
passed to the user subroutine by means of the parameters iPrjDir and iPrjLen. The iPrjDir
array contains the ASCII numbers of the characters of the project directory string and
iPrjLen is the length of the string (max. 255). This is to avoid character passing conicts
(Fortran - C conicts). The project directory string will always end with character 92 (\).
The user has to rebuild the character string and can directly add the actual name of the
debug le.
The example underneath shows how a debug le called 'usrdbg' can be created and opened
in the current project directory for debugging purposes:
Character fName
*
255
Dimension iPrjDir(
*
)
C-1
Logical IsOpen
Data IsOpen / .FALSE. /
Save IsOpen
fName =
Do i=1, iPrjLen
fName(i:i) = Char( iPrjDir(i) )
End Do
fName = fName(:iPrjLen) // usrdbg
Open( Unit = 1, File = fName )
IsOpen = .TRUE.
End If
In the user subroutine, values can be written to IO unit 1, using for example the available
writing subroutines in Appendix B.
Debugging hints
When developing and debugging a constitutive soil model in the user subroutine, it is very
useful to start by testing it with a simple nite element model in which a homogeneous
stress state should occur (for example an axisymmetric, 1 x 1 unit model of a one-
dimensional compression test or a triaxial test with zero soil weight). The nite element
model will still contain many stress points, but the stress state in each point should be the
same.
In any case, it is useful to write output for a limited number of stress points only (or for
certain step numbers or iteration numbers), in order to avoid large debug les. Examples
of writing useful but limited debug information are given below:
io = 0
If ( iEl .Eq. 1 .And. Int.Eq.1 .And. iStep.Gt.10 )
io = 1
...
Call WriIVl( io, Step, iStep )
Call WriIVl( io, Iter, iTer )
Call WriVec( io, Sig0, Sig0, 6 )
Call WriVec( io, dEps, dEps, 6 )
Call WriMat( io, D, D, 6, 6, 6 )
...
Call WriVec( io, Sig, Sig, 6 )
The available writing routines do not write when io is zero or negative.
Alternatively:
If ( io .Eq. 1 ) Then
C-2 PLAXIS-GiD
APPENDIX C - CREATING A DEBUG-FILE FOR USER-DEFINED SOIL MODELS
Write( 2,
*
) StVar:,(StVar(j),j=1,2)
End If
Note that here le 2 is used. This le must be opened before. When the le has not been
opened before, Lahey Fortran will give a Run-Time Error (File not open). Digital Fortran
on the other hand will open a le with the name FORT.2 in the current directory or checks
the environment variable FORT2.
C-3
C-4 PLAXIS-GiD

Material Models in Plaxis

Uploaded by

Copyright:

Available Formats

Material Models in Plaxis

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Material Models in Plaxis

Uploaded by

Copyright:

Available Formats

PLAXIS-GiD

and the cohesion c set to

, which is dened as the sum of all

Effective stress: p' = (1B)p = K'

Excess pore pressure: p

, the stiffness moduli in the model are no

Figure 3.1 Basic idea of an elastic perfectly plastic model

= / (1) for one-dimensional

. For -values of less than 30

, however, the angle of dilatancy is mostly zero. A

or to the standard Cam-Clay compression index .

. The relationship between these parameters and the compression

is positive for dilatancy.

where a = 0.385 (5.3)

is the modied compression index, which determines the

differs from the index

differs from the index as used by Burland. The ratio

are used as input constants for the part of the

: Modied compression index [-]

: Modied swelling index [-]

and the original Cam-

can be obtained from

and the one-dimensional

(=/) ranges, in general, between 3 and 7.

. However, the inuence of

=0 for time t =0.

, who t into the framework of critical-state soil mechanics.

, whilst soft soils also exhibit deviatoric creep strains.

. On the other hand

. Now it can be derived for the volumetric elastic strain that:

coincides with A. Unfortunately

: Modied swelling index [-]

: Modied compression index [-]

: Modied creep index [-]

: Creep index for secondary compression [-]

, and the slope of the unloading

results in a change of *, * or * respectively. The parameter e

and the one-dimensional swelling indices A and C

is in the range between 15 to 25 and the general observation

is in the range around 5.

of 1. It is clear that the pre-consolidation pressure not only depends

. In contrast to Hooke's law, the D

-matrix as used in the

(see Eq. 8.3).

and s, measured clockwise

with respect to the North direction (N). The dip direction

-direction) when looking downwards. The dip

and the declination is

: Modied swelling index

: Modied compression index

: Modied creep index

You might also like

results in a change of , or * respectively. The parameter e