PLAXIS

Material Models Manual

2011

Build 5167

TABLE OF CONTENTS

TABLE OF CONTENTS
1

Introduction
1.1
On the use of different models
1.2
Limitations

7
7
9

Preliminaries on material modelling

2.1
General denitions of stress
2.2
General denitions of strain
2.3
Elastic strains
2.4
Undrained effective stress analysis (effective stiffness parameters)
2.5
Undrained effective stress analysis with effective strength parameters
(Undrained A)
2.6
Undrained effective stress analysis with undrained strength parameters
(Undrained B)
2.7
Undrained total stress analysis with undrained parameters (Undrained C)
2.8
The initial pre-consolidation stress in advanced models
2.9
On the initial stresses

13
13
15
17
18

Linear elastic perfectly plastic model (Mohr-Coulomb model)

3.1
Linear elastic perfectly-plastic behaviour
3.2
Formulation of the Mohr-Coulomb model
3.3
Basic parameters of the Mohr-Coulomb model
3.4
Advanced parameters of the Mohr-Coulomb model

29
29
30
32
38

The Hoek-Brown model (rock behaviour)

4.1
Formulation of the Hoek-Brown model
4.2
Conversion of Hoek-Brown to Mohr-Coulomb
4.3
Parameters of the Hoek-Brown model

41
41
43
44

The Hardening Soil model (isotropic hardening)

5.1
Hyperbolic relationship for standard drained triaxial test
5.2
Approximation of hyperbola by the Hardening Soil model
5.3
Plastic volumetric strain for triaxial states of stress
5.4
Parameters of the Hardening Soil Model
5.5
On the cap yield surface in the Hardening Soil model

49
50
51
53
54
59

The Hardening Soil model with small-strain stiffness (HSsmall)

6.1
Describing small-strain stiffness with a Simple Hyperbolic Law
6.2
Applying the Hardin-Drnevich Relationship in the HS model
6.3
Virgin (initial) loading vs. unloading/reloading
6.4
Model parameters
6.5
On the parameters G0 and 0.7
6.6
Model initialization
6.7
Other differences with the Hardening Soil model

63
64
65
67
68
69
70
71

The Soft Soil model

7.1
Isotropic states of stress and strain ( '1 = '2 = '3 )
7.2
Yield function for triaxial stress state ( '2 = '3 )
7.3
Parameters of the Soft Soil model

73
73
74
76

PLAXIS 2011 | Material Models Manual

22
24
24
26
27

MATERIAL MODELS MANUAL

Soft Soil Creep model (time dependent behaviour)

8.1
Introduction
8.2
Basics of one-dimensional creep
8.3
On the variables c and c
8.4
Differential law for 1D-creep
8.5
Three-dimensional-model
8.6
Formulation of elastic 3D-strains
8.7
Review of model parameters

81
81
82
83
85
87
89
90

The Jointed Rock model (anisotropy)

9.1
Anisotropic elastic material stiffness matrix
9.2
Plastic behaviour in three directions
9.3
Parameters of the Jointed Rock model

95
96
98
101

10

Modied Cam-Clay model

107

11

The NGI-ADP model (anisotropic undrained shear strength)

11.1 Formulation of the NGI-ADP model
11.2 Parameters of the NGI-ADP model

111
111
116

12

The Sekiguchi-Ohta model

12.1 Formulation of the Sekiguchi-Ohta model
12.2 Parameters of the Sekiguchi-Ohta model

121
121
122

13

Application of advanced soil models

13.1 Hardening Soil model: Response in drained and undrained triaxial tests
13.2 Application of the Hardening Soil model on real soil tests
13.3 Application of the HS small model on real soil tests
13.4 Soft Soil Creep model : Undrained triaxial tests at different loading rates
13.5 Soft Soil Creep model: Response in one-dimensional compression test
13.6 Soft Soil model : Response in isotropic compression test
13.7 Hardening Soil model and HS small model: Excavation in Berlin sand

127
127
131
137
139
141
147
148

14

User-dened soil models

14.1 Introduction
14.2 Implementation of UD models in calculations program
14.3 Input of UD model parameters via user-interface

153
153
153
163

15

Structural behaviour
15.1 Anchors
15.2 Beams
15.3 2D geogrids
15.4 3D geogrids
15.5 2D plates
15.6 3D plates
15.7 Embedded pile

167
167
167
168
168
169
170
175

16

Hydraulic Models
16.1 Van Genuchten model
16.2 Approximate Van Genuchten model

179
179
183

Material Models Manual | PLAXIS 2011

TABLE OF CONTENTS

17

References

185

Appendix A - Symbols

191

Appendix B - Applicability of the material models

193

Appendix C - Fortran subroutines for User-dened soil models

197

Appendix D - Creating a debug-le for User-dened soil models

201

PLAXIS 2011 | Material Models Manual

MATERIAL MODELS MANUAL

Material Models Manual | PLAXIS 2011

INTRODUCTION

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. An overview of the applicability
of the material models is given in Appendix B.

1.1

ON THE USE OF DIFFERENT MODELS

Linear Elastic model (LE)

The Linear Elastic model is based on Hooke's law of isotropic elasticity. It involves two
basic elastic parameters, i.e. Young's modulus E and Poisson's ratio . Although the
Linear Elastic model is not suitable to model soil, it may be used to model stiff volumes in
the soil, like concrete walls, or intact rock formations.

Mohr-Coulomb model (MC)

The linear elastic perfectly-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 or a stiffness that
increases linearly with depth. 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, E50 , the triaxial unloading stiffness, Eur , and the oedometer loading
stiffness, Eoed . As average values for various soil types, Eur 3E50 and Eoed E50 are
suggested as default settings, but both very soft and very stiff soils tend to give other
ratios of Eoed /E50 , 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.

PLAXIS 2011 | Material Models Manual

MATERIAL MODELS MANUAL

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
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
ref
ref
parameters, i.e. G0 and 0.7 . G0 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 superseded 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 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 stratication and particular fault directions.
Plasticity can only occur in a maximum of 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.

Modied Cam-Clay model (MCC)

The Modied Cam-Clay model is a well known model from international soil modelling
literature; see for example Muir Wood (1990). It is meant primarily for the modelling of
near normally-consolidated clay-type soils. This model has been added to PLAXIS to

Material Models Manual | PLAXIS 2011

INTRODUCTION

allow for a comparison with other codes.

NGI-ADP model (NGI-ADP)

The NGI-ADP model is an anisotropic undrained shear strength model. The soil shear
strength is dened by means of su values for active, passive and direct simple shear
stress states.

Hoek-Brown model (HB)

The Hoek-Brown model is an isotropic elastic perfectly-plastic model based on the 2002
edition of the Hoek-Brown failure criterion. This non-linear stress-dependent criterion
describes shear failure and tensile failure by a continuous function, and is familiar to most
geologists and rock engineers. Besides the elastic parameters (E and ), the model
involves practical rock parameters such as the uni-axial compressive strength of the
intact rock (ci ), the Geological Strength Index (GSI ), and the disturbance factor (D ).

Analyses with different models

The Mohr-Coulomb model may be used for a relatively quick and simple rst analysis of
the problem considered.
In many cases,even if good data on dominant soil layers is limited, it is recommended to
use the Hardening Soil model or the HS small 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.

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.
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

PLAXIS 2011 | Material Models Manual

MATERIAL MODELS MANUAL

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 linear elastic perfectly-plastic 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 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 cu (su ), 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 (HSsmall) 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 HSsmall does not
incorporate the accumulation of irreversible volumetric straining nor liquefaction
behaviour with cyclic loading. The use of the HSsmall will generally result in calculation
times that are even longer than those of the Hardening Soil model.

Soft Soil model

The same limitations (including the ones for the Soft Soil Creep model) hold in the Soft
Soil model. In fact the Soft Soil model is superseded 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.

10

Material Models Manual | PLAXIS 2011

INTRODUCTION

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 seem logical 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.
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 ci and tensile strength ti . 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.

Modied Cam-Clay model

The same limitations (including these in the Soft Soil Creep model) hold in the Modied
Cam-Clay model. Moreover, the Modied Cam-Clay model may allow for unrealistically
high shear stresses. This is particularly the case for overconsolidated stress states where
the stress path crosses the critical state line. Furthermore, the Modied Cam-Clay model
may give softening behaviour for such stress paths. Without special regularization
techniques, softening behaviour may lead to mesh dependency and convergence
problems of iterative procedures. Moreover, the Modied Cam-Clay model cannot be
used in combination with Safety analysis by means of phi-c reduction. The use of the
Modied Cam-Clay model in practical applications is not recommended.

NGI-ADP model (NGI-ADP)

The NGI-ADP model is an undrained shear strength moded. It can be used in a drained
or effective stress analysis, but note that the shear strength is not automatically updated
with changes of effective stress. Also note that the NGI-ADP model does not include a
tension cut-off.

Hoek-Brown model (HB)

The Hoek-Brown model is an isotropic continuum model. Hence, the model is not
suitable for stratied or jointed rock sections with a signicant anisotropic stiffness or with
one or more dominant sliding directions. For such behaviour, the Jointed Rock model is

PLAXIS 2011 | Material Models Manual

11

MATERIAL MODELS MANUAL

available.

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 4.1.2 of the Reference Manual. In such
cases the interface stiffness is set equal to the elastic soil stiffness. Hence, E = Eur
where Eur is stress level dependent, following a power law with Eur proportional to m .
For the Soft Soil model, Soft Soil Creep model and Modied Cam-Clay model the power
m is equal to 1 and Eur 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 options to deal with undrained behaviour in an effective stress analysis, the
use of undrained shear strength (cu or su ) may be preferred 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.

12

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

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.
This Material Models Manual is a general manual for all PLAXIS programs and uses the
coordinate system as used in most programs (Figure 2.1). Please note that the new
PLAXIS 3D uses a different coordinate system where z is the vertical axis. Users should
realize this when reading this manual.

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

(2.2)

According to Terzaghi's principle, stresses in the soil are divided into effective stresses,
', and pore pressures, w :
(2.3)

= ' + w

Pore pressures are generally provided by water in the pores. Water is considered not to

yy
y

yx
xy

yz
zy
x
z

zz

zx

xx
xz

Figure 2.1 General three-dimensional coordinate system and sign convention for stresses

PLAXIS 2011 | Material Models Manual

13

MATERIAL MODELS MANUAL

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, pw :

w = pw pw pw 0 0 0

(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

(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:
(2.6)

det ' ' I = 0

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
'xx + 'yy + 'zz
3

q=

1
'xx 'yy
2

(2.8a)

+ 'yy 'zz

2
2
2
+ ( 'zz 'xx ) 2 + 6 xy + yz + zx

(2.8b)

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 .

14

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

- '1
- '1 = - '2 = - '3

- '3
- '2
Figure 2.2 Principal stress space

Principal effective stresses can be written in terms of the invariants:

'1 = p ' +

2
2
q sin
3
3

(2.9a)

'2 = p ' +

2
q sin()
3

(2.9b)

'3 = p ' +

2
2
q sin +
3
3

(2.9c)

in which is referred to as Lode 's angle (a third invariant), which is dened as:

1
27 J3
arcsin
3
2 q3

(2.10)

with
2
J3 = ( 'xx p') 'yy p' ( 'zz p') ( 'xx p')yz

2
2
'yy p' zx ( 'zz p')xy + 2xy yz zx

2.2

(2.11)

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. ij = (ui /j + uj /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 ij and ji result in shear
stress. This sum is denoted as the shear strain . Hence, instead of xy , yx , yz , zy , zx

PLAXIS 2011 | Material Models Manual

15

MATERIAL MODELS MANUAL

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

(2.13)

xx =

ux
x

(2.14a)

yy =

uy
y

(2.14b)

zz =

uz
z

(2.14c)

xy = xy + yx =

ux uy
+
y
x

(2.14d)

yz = yz + zy =

uy uz
+
z
y

(2.14e)

zx = zx + xz =

uz ux
+
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

(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
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.

16

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

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:
(2.18)

' = M

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 tabsheet for the Linear Elastic model

The simplest material model in PLAXIS is based on Hooke's law for isotropic linear
elastic 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'

=
(1 2 ')(1 + ')

1 '
'
'
0
0
0

xx

1 ' '
0
0
0 yy

' 1 '
0
0
0 zz

(2.19)

1 '
0
0
0
0 xy

2

1 '
0
0
0
0 yz

1 '
zx

0
0
0
0
2
'

'

The elastic material stiffness matrix is often denoted as D . 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

PLAXIS 2011 | Material Models Manual

17

MATERIAL MODELS MANUAL

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 ref to emphasize that it refers to a particular reference level (yref )(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 Eoed , is given by:

G=

E
2(1 + )

K =

E
3(1 2)

Eoed =

(2.20a)
(2.20b)

(1 )E
(1 2)(1 + )

(2.20c)

During the input of material parameters for the Linear Elastic model or the Mohr-Coulomb
model the values of G and Eoed 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 Eoed results in a
change of the E modulus, while remains the same.
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 a value for Einc
which is the increment of stiffness per unit of depth, as indicated in Figure 2.3.
Together with the input of Einc the input of yref becomes relevant. Above yref the stiffness
is equal to Eref . Below the stiffness is given by:

E(y) = Eref + (yref y )Einc

y < yref

(2.21)

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 (Drainage type) of a soil layer as Undrained (A) or Undrained (B). 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

18

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

supposed not to sustain any shear stress, and therefore the effective shear stresses are
equal to the total shear stresses:

xx = 'xx + pw

(2.22a)

yy = 'yy + pw

(2.22b)

zz = 'zz + pw

(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, pw is considered
negative for pressure.
A further distinction is made between steady state pore stress, psteady , and excess pore
stress, pexcess :
(2.23)

pw = psteady + pexcess

Steady state pore pressures are considered to be input data, i.e. generated on the basis
of phreatic levels. Excess pore pressures are generated during plastic calculations for the
case of undrained (A) or (B) 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:
(2.24)

pw = pexcess
Hooke's law can be inverted to obtain:

e
xx

e
yy

e
zz

e
xy

e
yz

zx
e

1 ' '

'

1 '

E' 0

1 '

' 1
0

'xx

'yy

0
0
0 'zz

2 + 2 '
0
0 'xy

0
2 + 2 '
0 'yz

0
0
2 + 2 '
'zx

(2.25)

PLAXIS 2011 | Material Models Manual

19

MATERIAL MODELS MANUAL

Substituting Eq. (2.22) gives:

e
xx

e
yy

e
zz

e
xy

e
yz

zx
e

1 ' '

'

1 '
=

E'

1 '

' 1
0

xx pw

0
0
0 yy pw

0
0
0 zz pw

2 + 2 '
0
0 xy

0
2 + 2 '
0 yz

zx

0
0
2 + 2 '

(2.26)

Considering slightly compressible water, the rate of excess pore pressure is written as:

pw =

Kw e
zz

+ e + e

n xx yy

(2.27)

in which Kw 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 Eu and u :

where:

e
xx

e
yy

e
zz

e
xy

e
yz

zx
e

1 u u
0
0
0

xx

u 1 u
0
0
0 yy

u 1

0
0
0 zz
= 1 u

Eu 0

0
0 2 + 2u
0
0 xy

0
0
0
0
2 + 2u
0 yz

0
0
0
0
0
2 + 2u
zx

Eu = 2G(1 + u )
=

1 Kw
3n K '

u =
K' =

' + (1 + ')
1 + 2(1 + ')
E'
3(1 2 ')

(2.28)

(2.29)
(2.30)

Hence, the special option for undrained behaviour in PLAXIS (Undrained (A) or
Undrained (B)) is such that the effective parameters G and ' are transferred into
undrained parameters Eu 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, Eu and u should not
be confused with Eur 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
bulk modulus for water is very large. In order to avoid numerical problems caused by an
extremely low compressibility, u is, by default, 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. Kw >> n K '. This condition is sufciently ensured by requiring

20

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

' 0.35.

Consequently, for material behaviour Undrained (A) or Undrained (B), a bulk modulus for
water is automatically added to the stiffness matrix. The value of the bulk modulus is
given by:

0.495 '
3(u ')
Kw
K ' = 300
=
K ' > 30K '
n
1 + '
(1 2u )(1 + ')

(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 Drainage type (type of material behaviour) is set to Undrained (A) or Undrained
(B), PLAXIS automatically assumes an implicit undrained bulk modulus, Ku , for the soil
as a whole (soil skeleton + water) and distinguishes between total stresses, effective
stresses and excess pore pressures (see Section 4.2 of the Reference Manual):
Total stress:

p = Ku

Effective stress:

p' = (1 B)p = K '

Excess pore pressure: pw = Bp = Kw

Note that for Undrained (A) or Undrained (B) effective stiffness parameters should be
entered in the material data set, i.e. E ' and ' and not Eu 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:

Ku =

2G(1 + u )
3(1 2u )

and

u = 0.495

or

u =

3 ' + B(1 2 ')

3 B(1 2 ')

where G =

E'
2(1 + ')

(when using the Standard setting)

when using the Manual setting with
input of Skempton's B-parameter

A particular value of the undrained Poisson's ratio, u , implies a corresponding reference

bulk stiffness of the pore uid, Kw,ref / n:

Kw,ref
= Ku K '
n

where

K' =

E'
3(1 2 ')

This value of Kw,ref / n is generally much smaller than the real bulk stiffness of pure water,
0
Kw (= 2106 kN/m2 ).

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:

PLAXIS 2011 | Material Models Manual

21

MATERIAL MODELS MANUAL

0
1
Kw Kair
Kw
=
0 n
n
SKair + (1 S)Kw

Where Kair = 100 kN/m2 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
nK '
1+
Kw

The rate of excess pore pressure is calculated from the (small) volumetric strain rate,
according to:

pw =

Kw
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 most material models in PLAXIS. This enables undrained
calculations to be executed with effective stiffness parameters, with explicit distinction
between effective stresses and (excess) pore pressures. However, shear induced
(excess) pore pressure may not be sufciently included.
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 based on
Hooke's law:

E' =

2(1 + ')
Eu
3

(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. The Soil test
facility (Section 4.3 of the Reference Manual) may be used as a convenient tool to
perform such test.

2.5

UNDRAINED EFFECTIVE STRESS ANALYSIS WITH EFFECTIVE STRENGTH

PARAMETERS (UNDRAINED A)

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 (Undrained (A)). 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 (cu or su ). However, note that most soil

22

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

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 Drainage type must be set to Undrained (A).
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.
MC failure line

q
Shear strength
increase

Shear strength:
model

(4) MC model

cu

reality
reality (2)

(3) consolidation
(1) MC model

c ', '

p'

Figure 2.4 Illustration of stress paths; reality vs. Mohr-Coulomb model

Figure 2.4 illustrates an example using the Mohr-Coulomb model. When the Drainage
type is set to Undrained (A), the model will follow an effective stress path where the mean
effective stress, p', remains constant all the way up to 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 as a result of shear
induced pore pressure (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

PLAXIS 2011 | Material Models Manual

23

MATERIAL MODELS MANUAL

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 Drainage type parameter is set to Undrained (A), 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

UNDRAINED EFFECTIVE STRESS ANALYSIS WITH UNDRAINED STRENGTH

PARAMETERS (UNDRAINED B)

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 (Drainage type = Undrained (B)), 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 = su ). 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 an 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, the
HS small model and the NGI-ADP 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 subtree in the Parameters tabsheet of the Soil 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 Drainage type parameter is set to Undrained (B), 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).

2.7

UNDRAINED TOTAL STRESS ANALYSIS WITH UNDRAINED PARAMETERS

(UNDRAINED C)

If, for any reason, it is desired not to use the Undrained (A) or Undrained (B) options 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 Eu and an undrained
Poisson ratio u , and strength is modelled using an undrained shear strength su 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

24

Material Models Manual | PLAXIS 2011

PRELIMINARIES ON MATERIAL MODELLING

Mohr-Coulomb model or the NGI-ADP model is used. In this case, one should select
Undrained (C) as the Drainage type. 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 a direct input of undrained shear strength does not automatically give the
increase of shear strength with consolidation. In fact, it does not make sense to perform a
consolidation analysis since there are no pore pressures to consolidate.
This type of approach is not possible for most advanced models.

Overview of models and allowable drainage types

Table 2.1 Material models and types
Material model

Material type
Drained

Linear Elastic model

Undrained (A)
Undrained (C)
Non-porous
Drained
Undrained (A)

Mohr-Coulomb model

Undrained (B)
Undrained (C)
Non-porous
Drained

Hardening Soil model

Undrained (A)
Undrained (B)
Drained

HS small model

Undrained (A)
Undrained (B)

Soft Soil model

Soft Soil Creep model
Jointed Rock model
Modied Cam-Clay model

Drained
Undrained (A)
Drained
Undrained (A)
Drained
Non-porous
Drained
Undrained (A)
Drained

NGI-ADP model

Undrained (B)
Undrained (C)

Hoek-Brown model
Sekiguchi-Ohta model (inviscid)

Drained
Non-porous
Drained
Undrained (A)
Drained

User-dened soil models

Undrained (A)
Non-porous

PLAXIS 2011 | Material Models Manual

25

MATERIAL MODELS MANUAL

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
eq
stress, p , but PLAXIS needs an equivalent isotropic pre-consolidation stress, pp 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 2.5),
and the in-situ effective vertical stress, '0 .
yy
OCR =

(2.34)

'0
yy

OCR=

'0
yy

p
'0
yy

POP

'0
yy

b. Using POP

a. Using OCR

Figure 2.5 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
2.5.
The vertical pre-consolidation stress p is used to compute the equivalent isotropic
eq
pre-consolidation stress, pp , which determines the initial position of a cap-type yield
eq
surface in the advanced soil models. The calculation of pp is based on the stress state:

'1 = p

and

nc
'2 = '3 = K0 p

(2.36)

nc
Where K0 is the K0 -value associated with normally consolidated states of stress, which
is an input parameter for the advanced soil models except for the Modied Cam-Clay
nc
model. For the Modied Cam-Clay model, K0 is implicitly dened as:
nc
K0 = 1 sin

Jaky (1944)

For more details about the Modied Cam-Clay model see Chapter 10.

26

Material Models Manual | PLAXIS 2011

(2.37)

PRELIMINARIES ON MATERIAL MODELLING

The stress state at pre-consolidation stress level is expressed in (p, q):

p' =

1
nc
1 + 2K0 p
3

and

nc
q = 1 K0 p

(2.38)

Depending on the model used, the equivalent isotropic pre-consolidation stress is

calculated as:
Hardening Soil model and Hardening Soil model with small-strain stiffness:
eq
pp =

(p') 2 +

q2
2

(where is an internal model parameter)

(2.39)

Soft Soil model, Soft Soil Creep model and Modied Cam-Clay model:
eq
pp = p' +

2.9

q2
M 2 p'

(where M is an internal model parameter)

(2.40)

ON THE INITIAL STRESSES

In overconsolidated soils the coefcient of lateral earth pressure for the initial stress state
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
K0 -procedure. The procedure that is followed here is described below.
- 'yy
- p
nc
K0

1- ur

1
- '0
yy

ur

- '0
xx

- 'xx

Figure 2.6 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 . During unloading the sample behaves elastically and the
yy
incremental stress ratio is, according to Hooke's law, given by (see Figure 2.6):
0
0
nc
nc
ur
'xx K0 p '0
K0 OCR 'yy 'xx
xx
=
=
=
0
0
'yy
1 ur
p 'yy
(OCR 1) 'yy

(2.41)

nc
where K0 is the stress ratio in the normally consolidated state. Hence, the default stress

PLAXIS 2011 | Material Models Manual

27

MATERIAL MODELS MANUAL

ratio of the overconsolidated soil sample is given by:

'0
ur
xx
nc
= K0 OCR
(OCR 1)
0
1 ur
'yy

(2.42)

When using POP as an alternative way to dene overconsolidation, the default initial
horizontal stress is dened by:
nc
'0 = K0 p
xx

ur
POP
1 ur

(2.43)

The use of a small Poisson's ratio will lead to a relatively large ratio of lateral stress and
vertical stress, as generally observed in overconsolidated soils. Note that Eq. (2.42) and
Eq. (2.43) are only valid in the elastic domain, because the formulas are 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.
nc
Note that the above values for K0 are only suggested (default) values, and may be
overruled by the user if more precise data is available or if other values seem more
appropriate.

28

Material Models Manual | PLAXIS 2011

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

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

LINEAR 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 :

>0

for:

f =0

and:

f T e
D 0

'
f T e
D >0

'

(Elasticity)

(3.4a)

(Plasticity)

(3.4b)

These equations may be used to obtain the following relationship between the effective
stress rates and strain rates for elastic perfectly-plastic behaviour (Smith & Grifth, 1982;
Vermeer & Borst, 1984):

' = De

e g f T e
D
D

' '
d

(3.5a)

PLAXIS 2011 | Material Models Manual

29

MATERIAL MODELS MANUAL

'

Figure 3.1 Basic idea of an elastic perfectly plastic model

where:

d=

f T e g
D
'
'

(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 full 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

g1
g2
+ 2
+ ...
'
'

(3.6)

Similarly, several quasi independent yield functions (f1 , f2 , ...) are used to determine the
magnitude of the multipliers (1 , 2 , ...).

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):

f1a =

(3.7a)

f1b =

1
1
( '3 '2 ) + ( '3 + '2 )sin c cos 0
2
2

(3.7b)

f2a =

30

1
1
( '2 '3 ) + ( '2 + '3 )sin c cos 0
2
2

1
1
( '3 '1 ) + ( '3 + '1 )sin c cos 0
2
2

(3.7c)

Material Models Manual | PLAXIS 2011

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

f2b =

1
1
( '1 '3 ) + ( '1 + '3 )sin c cos 0
2
2

(3.7d)

f3a =

1
1
( '1 '2 ) + ( '1 + '2 )sin c cos 0
2
2

(3.7e)

f3b =

1
1
( '2 '1 ) + ( '2 + '1 )sin c cos 0
2
2

(3.7f)

The two plastic model parameters appearing in the yield functions are the well-known
friction angle and the cohesion c . The condition fi = 0 for all yield functions together
(where fi is used to denote each individual yield function) represents a xed hexagonal
cone in principal stress space as shown in Figure 3.2.
-1

-3
-2
Figure 3.2 The Mohr-Coulomb yield surface in principal stress space (c = 0)

In addition to the yield functions, six plastic potential functions are dened for the
Mohr-Coulomb model:

g1a =

1
1
( '2 '3 ) + ( '2 + '3 )sin
2
2

(3.8a)

g1b =

1
1
( '3 '2 ) + ( '3 + '2 )sin
2
2

(3.8b)

g2a =

1
1
( '3 '1 ) + ( '3 + '1 )sin
2
2

(3.8c)

g2b =

1
1
( '1 '3 ) + ( '1 + '3 )sin
2
2

(3.8d)

g3a =

1
1
( '1 '2 ) + ( '1 + '2 )sin
2
2

(3.8e)

g3b =

1
1
( '2 '1 ) + ( '2 + '1 )sin
2
2

(3.8f)

The plastic potential functions contain a third plasticity parameter, the dilatancy angle .

PLAXIS 2011 | Material Models Manual

31

MATERIAL MODELS MANUAL

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 from one 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:

f4 = '1 t 0

(3.9a)

f5 = '2 t 0

(3.9b)

f6 = '3 t 0

(3.9c)

When this tension cut-off procedure is used, the allowable tensile stress, t , is, by
default, 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 . The model described here is ofcially called the linear elastic perfectly plastic
model with Mohr-Coulomb failure criterion. For simplicity, this model is called the
Mohr-Coulomb model in PLAXIS.

3.3

BASIC PARAMETERS OF THE MOHR-COULOMB MODEL

The linear elastic perfectly-plastic Mohr-Coulomb model requires a total of ve

parameters, which are generally familiar to most geotechnical engineers and which can
be obtained from basic tests on soil samples. These parameters with their standard units
are listed below:

[kN/m2 ]

: Poisson's ratio

[-]

: Cohesion

[kN/m2 ]

: Friction angle

[ ]

32

: Young's modulus

: Dilatancy angle

[ ]

Material Models Manual | PLAXIS 2011

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

Figure 3.3 Parameter tabsheet for Mohr-Coulomb model

Instead of using the Young's modulus as a stiffness parameter, alternative stiffness

parameters can be entered (Section 2.3). These parameters with their standard units are
listed below:

: Shear modulus

[kN/m2 ]

Eoed

: Oedometer modulus

[kN/m2 ]

In the case of dynamic applications, alternative and/or additional parameters may be

used to dene stiffness based on wave velocities. These parameters are listed below:

Vp

: Compression wave velocity

[m/s]

Vs

: Shear wave velocity

[m/s]

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 triaxial testing of soil
samples the initial slope of the stress-strain curve (tangent modulus) is usually indicated
as E0 and the secant modulus at 50% strength is denoted as E50 (see Figure 3.4). For
materials with a large linear elastic range it is realistic to use E0 , but for loading of soils
one generally uses E50 . Considering unloading problems, as in the case of tunnelling
and excavations, one needs an unload-reload modulus (Eur ) instead of E50 .
For soils, both the unloading modulus, Eur , and the rst loading modulus, E50 , 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

PLAXIS 2011 | Material Models Manual

33

MATERIAL MODELS MANUAL

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 5 and 6. 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 (Undrained (A) or
Undrained (B)), the Young's modulus has the meaning of an effective Young's modulus,
whereas PLAXIS automatically takes care of the incompressibility (Section 2.4).
|1 3 |

E0

E50

strain - 1
Figure 3.4 Denition of E0 and E50 for standard drained triaxial test results

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 K0 = 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 K0 . Hence,
is evaluated by matching K0 . This subject is treated more extensively in Section 4.1.5
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. Please note that in
this way it is not possible to create K0 values larger than 1, as may be observed in highly
overconsolidated stress states. For unloading conditions, however, it is more appropriate
to use values in the range between 0.15 and 0.25.
Further note that for material data sets where the type of material behaviour is set to
Undrained (A) or Undrained (B), the Poisson's ratio has the meaning of an effective
Poisson's ratio, whereas PLAXIS automatically takes care of the incompressibility
(Section 2.4). 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 (A) or

34

Material Models Manual | PLAXIS 2011

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

Undrained (B).

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 (A), as in both cases PLAXIS will perform an effective stress analysis.
Alternatively, the cohesion parameter may be used to model the undrained shear strength
cu (or su ) of the soil, in combination with = u = 0 when the Drainage type is set to
Undrained (B) or Undrained (C).
The disadvantage of using effective strength parameters c ' and ' in combination with the
drainage type being set to Undrained (A) 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 | 2su ).
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 (Undrained (B) or Undrained (C)) 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
in soil layers near the ground surface (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 Parameters
tabsheet in the Reference Manual for 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 (A), 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 (Undrained (B) or Undrained (C)) (Figure 3.5b).
High friction angles, as sometimes obtained for dense sands, will substantially increase
plastic computational effort. The computing time increases more or less exponentially

PLAXIS 2011 | Material Models Manual

35

MATERIAL MODELS MANUAL

1
3

shear
stress

shear
stress

'

c = su
c'

=0

normal

1 stress

a. Using effective strength parameters

normal

1 stress

b. Using undrained strength parameters

Figure 3.5 Stress circles at yield; one touches Coulomb's envelope.

3
2

Figure 3.6 Failure surface in principal stress space for cohesionless soil

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 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. However, this option is not available for the Mohr-Coulomb model. For
further information about the link between the friction angle and dilatancy, see Bolton
(1986).

36

Material Models Manual | PLAXIS 2011

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

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 stresses. In an undrained
effective stress analysis therefore the strength of the soil may be overestimated.
When the soil strength is modelled as c = cu (su ) and = 0, (Undrained (B) or Undrained
(C)) 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 (A). In that
case the model will show unlimited soil strength due to suction.

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 Eoed results in a change of the
E modulus whilst remains the same.

Oedometer modulus (Eoed )

The oedometer modulus, Eoed , 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).

Eoed =

(1 )E
(1 2)(1 + )

(3.11)

Entering a particular value for one of the alternatives G or Eoed results in a change of the
E modulus whilst remains the same.

Compression wave velocity Vp

The compression wave velocity, Vp , in a conned one-dimensional soil is a function of
stiffness, Eoed , and the mass, , as:

Vp =

Eoed

where

Eoed =

(1 )E
(1 + )(1 2)

and

unsat
g

(3.12)

in which unsat is the total unsaturated unit weight and g is the gravity acceleration (9.8
m/s2 ).

Shear wave velocity Vs

The shear wave velocity, Vs , in a conned one-dimensional soil is a function of shear
stiffness, G, and the mass, , as:

PLAXIS 2011 | Material Models Manual

37

MATERIAL MODELS MANUAL

Vs =

where

G =

E
2 (1 + )

and

unsat
g

(3.13)

in which is the total unsaturated unit weight and g is the gravity acceleration (9.8 m/s2 ).

3.4

ADVANCED PARAMETERS OF THE MOHR-COULOMB MODEL

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. These parameters are dened in the Advanced
subtree in the Parameters tabsheet of the Soil window.

Increase of stiffness (Einc )

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 Einc -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 yref
parameter, and above, the stiffness is equal to the reference Youngs modulus, Eref , as
entered in the Parameters tabsheet. Below, the stiffness is given by:

E(y) = Eref + (yref y )Einc

(y

< yref )

(3.14)

The actual value of Youngs modulus in the stress points is obtained from the reference
value and Einc . Note that during calculations a stiffness increasing with depth does not
change as a function of the stress state.

Increase of cohesion (cinc or su,inc )

PLAXIS offers an advanced option for the input of clay layers in which the cohesion, c, (or
undrained shear strength, su ) increases with depth. In order to account for the increase
of the cohesion with depth the cinc -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
yref parameter, and above, the cohesion is equal to the (reference) cohesion, cref , as
entered in the Parameters tabsheet. Below, the cohesive strength is given by :

c(y) = cref + (yref y)cinc

(y

< yref )

su (y) = su,ref + (yref y)su,inc

(y

< yref )

(3.15a)
(3.15b)

Tension cut-off
In some practical problems an area with tensile stresses may develop. According to the
Coulomb envelope shown in Figure 3.5 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

38

Material Models Manual | PLAXIS 2011

LINEAR ELASTIC PERFECTLY PLASTIC MODEL (MOHR-COULOMB MODEL)

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 the tension cut-off is, by default, selected with a tensile strength of
zero.

On the use of the Mohr-Coulomb model in dynamic calculations

When using the Mohr-Coulomb model in dynamic calculations, the stiffness parameters
need to be selected such that the model correctly predicts wave velocities in the soil
(Equation (3.13)). This generally requires a much larger small strain stiffness rather than
a stiffness at engineering strain levels. When subjected to dynamic or cyclic loading, the
Mohr-Coulomb model generally exhibits only elastic behaviour and no (hysteretic)
damping, nor accumulation of strains or pore pressure or liquefaction. In order to
simulate the soil's damping characteristics, Rayleigh damping may be dened.

PLAXIS 2011 | Material Models Manual

39

MATERIAL MODELS MANUAL

40

Material Models Manual | PLAXIS 2011

THE HOEK-BROWN MODEL (ROCK BEHAVIOUR)

THE HOEK-BROWN MODEL (ROCK BEHAVIOUR)

The material behaviour of rock differs from the behaviour of soils in the sense that it is
generally stiffer and stronger. The dependency of the stiffness on the stress level is
almost negligible, so stiffness of rocks can be considered constant. On the other hand,
the dependency of the (shear) strength on the stress level is signicant. In this respect,
heavily jointed rock can be regarded a frictional material. A rst approach is to model the
shear strength of rock by means of the Mohr-Coulomb failure criterion. However,
considering the large range of stress levels where rock may be subjected to, a linear
stress-dependency, as obtained from the Mohr-Coulomb model, is generally not
sufcient. Furthermore, rock may also show a signicant tensile strength. The
Hoek-Brown failure criterion is a better non-linear approximation of the strength of rocks.
It involves shear strength as well as tensile strength in a continuous formulation. Together
with Hooke's law of isotropic linear elastic behaviour it forms the Hoek-Brown model for
rock behaviour. The 2002 edition of this model (Hoek, Carranza-Torres & Corkum, 2002)
has been implemented in PLAXIS to simulate the isotropic behaviour of rock-type
materials. The implementation of the model, including the material strength factorization,
is based on Benz, Schwab, Vermeer & Kauther (2007). More background information on
the Hoek-Brown model and the selection of model parameters can be found in Hoek
(2006). For anisotropic rock behaviour reference is made to Chapter 9.

4.1

FORMULATION OF THE HOEK-BROWN MODEL

The generalized Hoek-Brown failure criterion can be formulated as a non-linear

relationship between the major and minor effective principal stresses (considering tension
positive and pressure negative):

1 ' = 3 ' ci mb

3 '
+s
ci

(4.1)

where mb is a reduced value of the intact rock parameter mi , which also depends on the
Geological Strength Index (GSI) and the Disturbance Factor (D):

mb = mi exp

GSI 100
28 14D

(4.2)

s and a are auxiliary material constants for the rock mass, that can be expressed as:
s = exp
a=

GSI 100
9 3D

1 1
GSI
+ exp
2 6
15

(4.3)

exp

20
3

(4.4)

ci is the uni-axial compressive strength of the intact rock material (dened as a positive
value). From this value, the uni-axial compressive strength of the specic rock under
consideration, c , can be obtained by:
c = ci sa

(4.5)

PLAXIS 2011 | Material Models Manual

41

MATERIAL MODELS MANUAL

The tensile strength of the specic rock under consideration, t , can be obtained by:

t =

sci
mb

(4.6)

The Hoek-Brown failure criterion is illustrated in Figure 4.1.

'

Figure 4.1 Hoek-Brown failure criterion in principal stresses

In the framework of plasticity theory, the Hoek-Brown failure criterion is reformulated into
the following yield function:

fHB = '1 '3 + f ( '3 )

where

f ( '3 ) = ci mb '3 + s
ci

(4.7)

For general three-dimensional stress states, more than one yield function is required to
deal with the corners of the yield contour, similar as for the full Mohr-Coulomb criterion.
Dening pressure as negative and considering ordering of principal stresses such that
'1 '2 '3 , the full criterion can be captured by two yield functions:

fHB,13 = '1 '3 + f ( '3 )

where

f ( '3 ) = ci mb '3 + s
ci

(4.8a)

fHB,12 = '1 '2 + f ( '2 )

where

f ( '2 ) = ci mb '2 + s
ci

(4.8b)

The full Hoek-Brown failure contour (fi = 0) in principal stress space is illustrated in Figure
4.2.
In addition to the two yield functions, two corresponding plastic potential functions are
dened for the Hoek-Brown model:

gHB,13 = S1
gHB,12 = S1

1 + sin mob
S3
1 sin mob
1 + sin mob
S2
1 sin mob

(4.9a)
(4.9b)

where Si are the transformed stresses, dened as:

Si =

42

1
s
+ 2
mb ci mb

for

i = 1, 2, 3

Material Models Manual | PLAXIS 2011

(4.10)

THE HOEK-BROWN MODEL (ROCK BEHAVIOUR)

Figure 4.2 The Hoek-Brown failure contour in principal stress space

mob is the mobilized dilatancy angle, varying with '3 from its input value at ( '3 = 0)
down to zero at '3 = and beyond:
mob =

+ '3
0

0 '3

(4.11)

Moreover, in order to allow for plastic expansion in the tensile zone, an increased articial
value of the mobilised dilatancy is used:

mob = +

'3
(90 )
t

(t '3 0)

(4.12)

The evolution of the mobilized dilatancy angle as a function of '3 is visualized in Figure
4.3.
mob
90

'3

Figure 4.3 Evolution of mobilized dilatancy angle

Regarding the elastic behaviour of the Hoek-Brown model, Hooke's law of isotropic linear
elastic behaviour, as described in Section 2.3, is adopted. This part of the model involves
Young's modulus, E , representing the in-situ stiffness of the jointed rock mass before
failure, and Poisson's ratio, , describing transverse straining.

4.2

CONVERSION OF HOEK-BROWN TO MOHR-COULOMB

In order to compare the Hoek-Brown failure criterion with the well-known Mohr-Coulomb
failure criterion for practical applications involving a particular stress range, a balanced t
can be made for conning stresses in the range (considering tension positive and

PLAXIS 2011 | Material Models Manual

43

MATERIAL MODELS MANUAL

compression negative):

t < '3 < 3,max

This gives the following expressions for the Mohr-Coulomb effective strength parameters
' and c ' (Carranza-Torres, 2004):

sin ' =
c' =

6a mb (s + mb '3n ) a1
2(1 + a) (2 + a) + 6amb (s + mb '3n ) a1

(4.13)

ci [(1 + 2a)s + (1 a)mb '3n ] (s + mb '3n ) a1

(1 + a) (2 + a)

(4.14)

6amb (s + mb '3n ) a1
1+
(1 + a) (2 + a)

where '3n = '3max /ci . The upper limit of the conning stress, '3,max , depends on the
application.

4.3

PARAMETERS OF THE HOEK-BROWN MODEL

The Hoek-Brown model involves a total of 8 parameters, which are generally familiar to
geologists and mining engineers. These parameters with their standard units are listed
below:
E

: Young's modulus

[kN/m2 ]

: Poisson's ratio

[-]

ci

: Uni-axial compressive strength of the intact rock (>0)

[kN/m2 ]

mi

: Intact rock parameter

[-]

GSI

: Geological Strength Index

[-]

: Disturbance factor

[-]

: Dilatancy angle (at '3 = 0)

[ ]

: Absolute value of conning pressure '3 at which

[kN/m2 ]

=0

Figure 4.4 Parameter tabsheet for Hoek-Brown model

44

Material Models Manual | PLAXIS 2011

THE HOEK-BROWN MODEL (ROCK BEHAVIOUR)

Young's modulus (E):

Young's modulus E is assumed to be a constant value for the considered rock layer. This
modulus can be estimated on the basis of the rock quality parameters in the Hoek-Brown
model (Hoek, Carranza-Torres & Corkum, 2002):

D
E (GPa) = 1
2

ci
10
pref

GSI 10
40

(4.15)

with pref = 105 kPa and considering a maximum result of the square root of 1.
Note that the input of Young's modulus in PLAXIS is generally in kN/m2
(= kPa = 106 GPa), which means that the value obtained from the above formula must
be multiplied by 106 .
More precise measurement of Young's modulus can be obtained from axial compression
tests or direct shear tests on rock samples. Note that, in contrast to most soils, the initial
stiffness, Ei , is a good representation of the rock behaviour in the elastic range.

Poisson's ratio ( ):
Poisson's ratio, , is generally in the range [0.1, 0.4]. Typical values for particular rock
types are listed in Figure 4.5.
Andesite
Basalt
Claystone
Conglomerate
Diabase
Diorite
Dolerite
Dolomite
Gneiss
Granite
Granodiorite
Greywacke
Limestone
Marble
Marl
Norite
Quartzite
Rock salt
Sandstone
Shale
Siltstone

Tuff
0

0.1

0.2
0.3
Poisson's ratio

0.4

0.5

Figure 4.5 Typical Poisson's ratio values

PLAXIS 2011 | Material Models Manual

45

MATERIAL MODELS MANUAL

Uni-axial compressive strength of intact rock (ci ):

The uni-axial strength of intact rock ci can be determined in laboratory testing, e.g. axial
compression. Laboratory testing is often conducted on intact rock so that (GSI = 100
and D = 0 hold). Typical values are shown in Table 4.1 (Hoek, 1999).
Table 4.1 Intact uniaxial compressive strength
Grade

Term

Uniaxial
Comp.
Strength
(MPa)

Field estimate of strength

Examples

R6

Extremely
strong

> 250

Specimen can be chipped with a

geological hammer

Fresh basalt, chert,

diabase,
gneiss,
granite, quartzite

R5

Very strong

100 250

Specimen requires many blows of a

geological hammer to fracture it.

Amphibolite, sandstone,
basalt, gabbro, gneiss,
granodiorite, limestone,
marble, rhyolite, tuff

R4

Strong

50 100

Specimen requires more than one blow

of a geological hammer to fracture it.

Limestone,
marble,
phyllite,
sandstone,
schist, shale

R3

Medium
strong

25 50

Cannot be scraped or peeled with

a pocket knife, specimen can be
fractured with a single blow from a
geological hammer.

Claystone,
coal,
concrete, schist, shale,
siltstone

R2

Weak

5 25

Can be peeled with a pocket knife with

difculty, shallow indentation made by
rm blow with point of a geological
hammer.

Chalk, rocksalt, potash.

R1

Very weak

15

Crumbles under rm blows with point

of a geological hammer, can be peeled
by a pocket knife.

Highly weathered
altered rock.

R0

Extremely
weak

0.25 1

Indented by thumbnail

Stiff fault gouge

or

Intact rock parameter (mi ):

The intact rock parameter is an empirical model parameter that depends on the rock
type. Typical values are given in Figure 4.6.

Geological Strength Index (GSI):

The GSI parameter can be selected on the basis of the chart depicted in Figure 4.7.

Disturbance factor (D):

The Disturbance factor, D, is a parameter that depends on the amount of disturbance of
the rock as a result of mechanical processes in open excavations, tunnels or mines, such
as blasting, tunnel boring, machine driven or manual excavation. No disturbance is
equivalent to D = 0, whereas severe disturbance is equivalent to D = 1. For more
information see Hoek (2006).

Dilatancy ( and ):
Rocks may show dilatant material behaviour when subjected to shear under relatively low
conning stress. At larger conning stress, dilatancy is suppressed. This behaviour is
modelled by means of a specied value of for 3 = 0, with a linear decrease down to
zero for '3 = , where is an additional input parameter (Figure 4.3.

46

Material Models Manual | PLAXIS 2011

THE HOEK-BROWN MODEL (ROCK BEHAVIOUR)

Rock
type

Group

Class

SEDIMENTARY

Clastic

Texture
Coarse
Conglomerate
(20)

Fine

Sandstone
Siltstone
19
9
Greywacke
(18)
Chalk
7

Very fine
Claystone
4

Organic
Coal
(8-21)

Non Clastic
Carbonate

Gypstone
16

Breccia
(22)

Chemical

METAMORPHIC

Medium

Micritic
Limestone
8
Anhydrite
13

Sparitic
Limestone
(10)

Non Foliated

Marble
9

Hornfels
(19)

Quartzite
24

Slightly foliated

Migmatite
(30)

Amphibolite
(25 - 31)

Foliated*

Gneiss
33

Schists
4-8

Mylonites
(6)
Phyllites
(10)

IGNEOUS
Extrusive pyroclastic type

Rhyolite
(16)
Dacite
(17)

Diorite
(28)

Dark

Granite
33
Granodiorite
(30)

Light

Slate
9

Andesite
19

Gabbro
27
Norite
22

Dolerite
(19)

Basalt
(17)

Agglomerate
(20)

Breccia
(18)

Tuff
(15)

Figure 4.6 Values for the constant mi for intact rock after Hoek (1999)

Figure 4.7 Selection of Geological Strength Index (Hoek (1999))

PLAXIS 2011 | Material Models Manual

47

MATERIAL MODELS MANUAL

On the use of the Hoek-Brown model in dynamic calculations

When using the Hoek-Brown model in dynamic calculations, the stiffness need to be
selected such that the model correctly predicts wave velocities in the soil (Equation
(3.13)). When subjected to dynamic or cyclic loading, the Hoek-Brown model generally
exhibits only elastic behaviour and no (hysteretic) damping, nor accumulation of strains or
pore pressure or liquefaction. In order to simulate the rock's damping characteristics,
Rayleigh damping may be dened.

48

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

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

ref
Input parameter E50

Plastic straining due to primary compression

ref
Input parameter Eoed

Elastic unloading / reloading

ref
Input parametersEur , ur

Failure according to the Mohr-Coulomb failure criterion 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
ref
the relationship Eoed = Eoed / pref 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 models for soft soil and the oedometer loading
modulus (see also Section 8.7).
ref
Eoed =

pref

(1 + e0 )

where pref is a reference pressure. Here we consider a tangent oedometer modulus at a

particular reference pressure pref . 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:
ref
Eur

2pref

(1 + e0 )

This relationship applies in combination with the input value m = 1.

PLAXIS 2011 | Material Models Manual

49

MATERIAL MODELS MANUAL

5.1

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 =

q
1
Ei 1 q/ qa

for:

q < qf

(5.1)

Where qa is the asymptotic value of the shear strength and Ei the initial stiffness. Ei is
related to E50 by:

Ei =

2E50
2 Rf

(5.2)

This relationship is plotted in Figure 5.1. The parameter E50 is the conning stress
dependent stiffness modulus for primary loading and is given by the equation:

E50 = E ref
50

c cos '3 sin

c cos + pref sin

(5.3)

ref
where E50 is a reference stiffness modulus corresponding to the reference conning
pressure pref . In PLAXIS, a default setting pref = 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, qf , and the quantity qa in Eq. (5.1) are dened as:

qf = (c cot '3 )

2 sin
1 sin

and:

qa =

qf
Rf

(5.4)

Again it is remarked that '3 is usually negative. The above relationship for qf is derived
from the Mohr-Coulomb failure criterion, which involves the strength parameters c and .
As soon as q = qf , the failure criterion is satised and perfectly plastic yielding occurs as
described by the Mohr-Coulomb model.
The ratio between qf and qa is given by the failure ratio Rf , which should obviously be
smaller than 1. In PLAXIS, Rf = 0.9 is chosen as a suitable default setting.
For unloading and reloading stress paths, another stress-dependent stiffness modulus is
used:

Eur = E ref
ur

c cos '3 sin

c cos + pref sin

(5.5)

ref
where Eur is the reference Young's modulus for unloading and reloading, corresponding
ref
to the reference pressure pref . In many practical cases it is appropriate to set Eur equal
ref
to 3E50 ; this is the default setting used in PLAXIS.

50

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

deviatoric stress
|1 3 |

qa
qf

asymptote
failure line

Ei E50
1

Eur
1

axial strain - 1
Figure 5.1 Hyperbolic stress-strain relation in primary loading for a standard drained triaxial test

5.2

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 <
qf , as also indicated in Figure 5.1. 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. (5.1)
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

(5.6)

where f is a function of stress and p is a function of plastic strains:

f =

q
2q
2

1 q/qa
Ei
Eur

p = (2p p ) 2p

1
1

(5.7)

with q , qa , Ei and Eur as dened by Eq. (5.1) to Eq. (5.5), whilst the superscript p is used
p
to denote plastic strains. For hard soils, plastic volume changes (v ) tend to be relatively
p
p
small and this leads to the approximation 21 . 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. (5.1). 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. (5.6) that:

p
1

1
q
q
1
f =

2
Ei 1 q/qa
Eur

(5.8)

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 Eur remains constant and the elastic strains are given by the
equations:

PLAXIS 2011 | Material Models Manual

51

MATERIAL MODELS MANUAL

e =
1

q
Eur

e = e = ur
2
3

q
Eur

(5.9)

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. (5.9) and a plastic component according to Eq. (5.8). Hence, it
follows that:

1 = e p
1
1

q
1
Ei 1 q/ qa

(5.10)
p

This relationship holds exactly in absence of plastic volume strains, i.e. when 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. Hence, p is associated with
mobilised friction. When plotting such yield loci, one has to use Eq. (5.7) as well as Eqs.
(5.3) and (5.5) for E50 and Eur respectively. Because of the latter expressions, the shape
of the yield loci depends on the exponent m. For m = 1, straight lines are obtained, but
slightly curved yield loci correspond to lower values of the exponent. Figure 5.2 shows
the shape of yield loci for increasing values of p considering m = 0.5, being typical for
hard soils. Hence, p can be regarded as the plastic shear strain related to the mobilised
shear resistance.
deviatoric stress

|1 3 |
Mohr-Coulomb failure line

Mean effective stress

Figure 5.2 Successive yield loci for various constant values of the hardening parameter p

52

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

5.3

PLASTIC VOLUMETRIC STRAIN FOR TRIAXIAL STATES OF STRESS

Having presented a relationship for the plastic shear strain, p , attention is now focused
p
on the plastic volumetric strain, v . As for all plasticity models, the Hardening Soil model
involves a relationship between rates of plastic strain, i.e. a relationship between p and
v
p . This shear hardening ow rule has the linear form:

p = sin m p
v

(5.11)

Clearly, further detail is needed by specifying the mobilised dilatancy angle m . For the
present model, the following is considered:
For sin m < 3/4 sin :

m = 0

For sin m 3/4 sin and > 0

sin m = max

For sin m 3/4 sin and 0

m =

If = 0

sin m sin cv
,0
1 sin m sin cv

(5.12)

m = 0

where c is the critical state friction angle, being a material constant independent of
density, and m is the mobilised friction angle:

sinm =

'1 '3
'1 + '3 2c cot

(5.13)

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.
(5.12) that:

sin =

sin sin cv
1 sin sin cv

(5.14a)

or equivalently:

sin cv =

sin sin
1 sin sin

(5.14b)

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, .

PLAXIS 2011 | Material Models Manual

53

MATERIAL MODELS MANUAL

5.4

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 .

Figure 5.3 Parameters for the Hardening Soil model

Failure parameters as in Mohr-Coulomb model (see Section 3.3):

: (Effective) cohesion

[kN/m2 ]

: (Effective) angle of internal friction

[ ]

: Angle of dilatancy

[ ]

Basic parameters for soil stiffness:

ref
E50

[kN/m2 ]

ref
Eoed

: Tangent stiffness for primary oedometer loading

[kN/m2 ]

ref
Eur

ref
ref
: Unloading / reloading stiffness (default Eur = 3E50 )

[kN/m2 ]

54

: Secant stiffness in standard drained triaxial test

: Power for stress-level dependency of stiffness

[-]

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

Advanced parameters (it is advised to use the default setting):

ur

: Poisson's ratio for unloading-reloading (default ur =

0.2)

[-]

pref

: Reference stress for stiffnesses (default pref = 100

kN/m2 )

[kN/m2 ]

nc
K0

nc
: K0 -value for normal consolidation (default K0 =
1 sin )

[-]

Rf

: Failure ratio qf / qa (default Rf = 0.9) (see Figure 5.1)

[-]

tension

: Tensile strength (default tension = 0 stress units)

[kN/m2 ]

cinc

: As in Mohr-Coulomb model (default cinc = 0)

[kN/m3 ]

Instead of entering the basic parameters for soil stiffness, alternative parameters can be
entered. These parameters are listed below:

Cc

: Compression index

[-]

Cs

: Swelling index or reloading index

[-]

einit

: Initial void ratio

[-]

ref
ref
ref
Stiffness moduli E50 , Eoed & Eur 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.
ref
Instead, a stiffness modulus E50 is dened for a reference minor principal stress of
ref
'3 = p . As a default value, the program uses pref = 100 kN/m2 .

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 Eur is a
- 1

ref
Eoed

pref

- 1

Figure 5.4 Denition of

ref
Eoed

in oedometer test results

PLAXIS 2011 | Material Models Manual

55

MATERIAL MODELS MANUAL

real elastic stiffness, one may thus write Eur = 2 (1+ ) Gur , where Gur is an elastic shear
modulus. Please note that PLAXIS allows for the input of Eur and ur but not for a direct
input of Gur . In contrast to Eur , the secant modulus E50 is not used within a concept of
elasticity. As a consequence, there is no simple conversion from E50 to G50 .
In contrast to elasticity based models, the elastoplastic Hardening Soil model does not
involve a xed relationship between the (drained) triaxial stiffness E50 and the oedometer
stiffness Eoed for one-dimensional compression. Instead, these stiffnesses can be
inputted independently. Having dened E50 by Eq. (5.3), it is now important to dene the
oedometer stiffness. Here we use the equation:

Eoed = E ref
oed

c cos

'3

sin
nc

K0

ref
c cos + p sin

(5.15)

where Eoed is a tangent stiffness modulus as indicated in Figure 5.4.

ref
Hence, Eoed is a tangent stiffness at a vertical stress of '1 = '3 = pref . Note that we
nc

K0

basically use '1 rather than '3 and that we consider primary loading.

When undrained behaviour is considered in the Hardening Soil model the Drainage type
should preferably be set to Undrained (A). Alternatively, Undrained (B) can be used in
case the effective strength properties are not known or the undrained shear strength is
not properly captured using Undrained (A). However, it should be noted that the material
looses its stress-dependency of stiffness in that case. Undrained (C) is not possible since
the model is essentially formulated as an effective stress model.

Alternative stiffness parameters

When soft soils are considered, the stiffness parameters can be calculated from the
compression index, swelling index and the initial void ratio . The relationship between
these parameters and the compression index is given by:

Cc =

2.3(1 + einit )pref

ref
Eoed

(5.16)

ref
The relationship between the Eur and the swelling index is given by:

Cs

2.3(1 + einit ) (1 + ) (1 2)pref

ref
(1 )Eur

(5.17)

Hint: In this approximate relationship it is assumed that during unloading/reloading

on average the horizontal stress is equal to the vertical stress (K0 = 1).
Regardless the previous value of E50 , a new value will be automatically assigned

In the PLAXIS material database, these alternative parameters depend on the initial void ratio. In reality, these
parameters depend on the actual void ratio, which is not a constant.

56

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

according to:
ref
ref
E50 = 1.25Eoed

(5.18)

ref
ref
Although for Soft soils, E50 could be as high as 2Eoed , this high value could lead to
limitation in the modeling; therefore a lower value is used. Changing the value of Cs will
change the stiffness parameter Eur .

Note that the value of the power for stress-level dependency of stiffness (m) is
automatically set to 1.

Advanced parameters
Realistic values of ur are about 0.2 and this value is thus used as a default setting, as
indicated in Figure 5.3. Note that in the Hardening Soil model, ur is a pure elastic
parameter.
nc
In contrast to the Mohr-Coulomb model, K0 is not simply a function of Poisson's ratio,
but an independent input parameter. As a default setting PLAXIS uses the correlation
nc
K0 = 1 sin . 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
nc
input values for K0 cannot be accommodated for. Depending on other parameters, such
nc
nc
ref
ref
ref
as E50 , Eoed , Eur and ur , there happens to be a certain range of valid K0 -values. K0
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 5.5. 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, einit , and the maximum void ratio,
emax , of the material must be entered as general parameters. As soon as the volume
change results in a state of maximum void, the mobilised dilatancy angle, m , is
automatically set back to zero, as indicated in Figure 5.5.
for e < emax :

sin m =

where:

sin c =

for e emax :

sin m sin cv
1 sin m sin cv

m = 0

The void ratio is related to the volumetric strain,

v init = ln
v

1+e
1 einit

sin sin
1 sin sin

(5.19a)
(5.19b)

by the relationship:
(5.20)

where an increment of is positive for dilatancy.

The initial void ratio, einit , 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

PLAXIS 2011 | Material Models Manual

57

MATERIAL MODELS MANUAL

dilatancy cut - off OFF

dilatancy cut - off ON

1

sin

maximum porosity reached

2 sin

1
Figure 5.5 Resulting strain curve for a standard drained triaxial test when including dilatancy cut-off

Figure 5.6 General tabsheet

maximum void ratio is reached, the dilatancy angle is set to zero. The minimum void
ratio, emin , of a soil can also be input, 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 tabsheet (Figure 5.6) of the Soil window and not in the Parameters
tabsheet. The selection of the Dilatancy cut-off is only available when the Hardening Soil
model or the Hardening Soil model with small-strain stiffness has been selected. By
default, the Dilatancy cut-off is not active.

58

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

5.5

ON THE CAP YIELD SURFACE IN THE HARDENING SOIL MODEL

Shear hardening yield surfaces as indicated in Figure 5.2 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
ref
ref
formulate a model with independent input of both E50 and Eoed . The triaxial modulus
largely controls the shear yield surface and the oedometer modulus controls the cap yield
ref
surface. In fact, E50 largely controls the magnitude of the plastic strains that are
ref
associated with the shear yield surface. Similarly, Eoed 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 detail. To this end we consider the denition of the cap yield surface:
2

fc =

q
2
+ p'2 pp
2

(5.21)

nc
where is an auxiliary model parameter that relates to K0 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 pp . The hardening
pc
law relating pp to volumetric cap strain v is:

pc =
v

1m

pp
p

1m

(5.22)

ref

The volumetric cap strain is the plastic volumetric strain in isotropic compression. In
addition to the well known constants m and pref 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:
nc
K0
ref
Eoed

nc
(default : K0 = 1 sin )
ref
ref
(default : Eoed = E50 )

nc
ref
such that K0 and Eoed 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 5.7.

The ellipse has length pp on the p-axis and pp on the q -axis. Hence, pp 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'

pp
p

ref

pp
pref

(5.23)

This expression for is derived from the yield condition f c = 0 and Eq. (5.22) for pp .
Input data on initial pp -values is provided by means of the PLAXIS procedure for initial

PLAXIS 2011 | Material Models Manual

59

MATERIAL MODELS MANUAL

stresses. Here, pp is either computed from the inputted over-consolidation ratio (OCR) or
the pre-overburden pressure (POP) (see Section 2.9).

pp
elastic region

c cot

pp

Figure 5.7 Yield surfaces of Hardening Soil model in p - q -plane. The elastic region can be further
reduced by means of a tension cut-off

For understanding the yield surfaces in full detail, one should consider both Figure 5.7
and Figure 5.8. 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
locus can expand up to the ultimate Mohr-Coulomb failure surface. The cap yield surface
expands as a function of the pre-consolidation stress pp .

On the use of the Hardening Soil model in dynamic calculations

When using the Hardening Soil model in dynamic calculations, the elastic stiffness
ref
parameter Eur needs to be selected such that the model correctly predicts wave
velocities in the soil. This generally requires an even larger small strain stiffness rather
ref
than just an unloading-reloading stiffness to be entered for Eur . When subjected to
dynamic or cyclic loading, the Hardening Soil model generally exhibits only elastic
behaviour (except for the rst loading part) and no (hysteretic) damping, nor accumulation
of strains or pore pressure nor liquefaction. In order to simulate the soil's damping
characteristics, Rayleigh damping may be dened. Note that some of the limitations of
the Hardening Soil model in dynamic applications can be overcome by using the HS
small model (Chapter 6).

60

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL (ISOTROPIC HARDENING)

Figure 5.8 Representation of total yield contour of the Hardening Soil model in principal stress space
for cohesionless soil

PLAXIS 2011 | Material Models Manual

61

MATERIAL MODELS MANUAL

62

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS (HSSMALL)

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 6.1
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 modulus G/G0 [-]

Retaining walls
Foundations
Tunnels

Very
small
strains

Conventional soil testing

Small strains

Larger strains
0
1e

-6

1e

-5

1e

-4

1e

-3

1e

-2

1e

-1

Shear strain gS [-]

Dynamic methods
Local gauges

Figure 6.1 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 6.1.
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 Hardening Soil model with small-strain stiffness 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 variation of
stiffness with strain:

the initial or very small-strain shear modulus G0

the shear strain level 0.7 at which the secant shear modulus Gs is reduced to about
70% of G0

PLAXIS 2011 | Material Models Manual

63

MATERIAL MODELS MANUAL

6.1

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.
Probably the 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):

Gs
G0

(6.1)

1+
r

where the threshold shear strain r is quantied as:

r =

max

(6.2)

G0

with max being the shear stress at failure. Essentially, Eqs. (6.1) and (6.2) 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 Gs is reduced to about 70 % of its initial value. Eq. (6.1)
can then be rewritten as:

Gs
G0

1+a
0.7

where a = 0.385

In fact, using a = 0.385 and = 0.7 gives Gs /G0 = 0.722. Hence, the formulation
"about 70%" should be interpreted more accurately as 72.2%.
Figure 6.2 shows the t of the modied Hardin-Drnevich relationship (Eq. 6.3) to
normalized test data.

64

Material Models Manual | PLAXIS 2011

(6.3)

THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS (HSSMALL)

Figure 6.2 Results from the Hardin-Drnevich relationship compared to test data by Santos & Correia
(2001)

6.2

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 =

He
e

(6.4)

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. (6.4) is applied subsequently
used in Eq. (6.3) . Note that in both, Eqs. (6.3) and (6.4) , the scalar valued shear strain
is dened as:

3
q
2

(6.5)

where q is the second deviatoric strain invariant. In triaxial conditions can therefore be
expressed as:
(6.6)

= axial lateral

PLAXIS 2011 | Material Models Manual

65

MATERIAL MODELS MANUAL

Within the HS small model, the stress-strain relationship can be simply formulated from
the secant shear modulus (Eq. 6.3) as:

G0

= Gs =

1 + 0.385

(6.7)

0.7

Taking the derivative with respect to the shear strain gives the tangent shear modulus:

G0

Gt =

1 + 0.385

0.7

(6.8)
2

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 Gt is introduced at the unloading
reloading stiffness Gur which is dened by the material parameters Eur and ur :
where

Gt Gur

Gur =

The cut-off shear strain cutoff

cutoff =

1
0.385

G0
Gur

Et
Eur
and
Gt =
2(1 + ur )
2(1 + ur )
can be calculated as:

(6.9)

(6.10)

1 0.7

Within the HS small model, the 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 6.3.
G0

Shear modulus G [kN/m]

40000

30000

20000

HSsmall
10000

0
1E-6

Hardin-Drnevich

1E-5

0.0001

cut-off

0.001

Gur

0.01

g0.7 [-]

Figure 6.3 Cut-off of the small-strain degradation curve as used in the HS-Small model

Moreover, the actual tangent shear modulus Gt and corresponding Young's modulus Et

66

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS (HSSMALL)

(considering a constant Poisson's ratio ur ), is stress-dependent, and follows the same

power law as formulated in Equation (5.5). For primary loading situations, the model uses
the same hardening plasticity formulations as the Hardening Soil model, where Eur is
replaced by Et as described above.

6.3

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:
(6.11)

0.7 reloading = 20.7 virginloading

The HS small 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 HS small model
are tted such, that the small-strain stiffness reduction curve is reasonably well
approximated. Figure 6.4 and Figure 6.5 illustrate Masing's rule and the secant stiffness
reduction in virgin loading and unloading / reloading.
q [kN/m]
100

50

-510-4

-310-4

310-4

510-4

[-]

-50

-100

Figure 6.4 Hysteretic material behaviour

PLAXIS 2011 | Material Models Manual

67

MATERIAL MODELS MANUAL

4105

Esec [kN/m2 ]

reloading

primary loading

2105

plastic stiffness reduction

during primary loading

0
110-5

110-4

[-]

110-3

Figure 6.5 Stiffness reduction in initial- or primary loading and in unloading / reloading

6.4

MODEL PARAMETERS

Compared to the standard Hardening Soil model, the Hardening Soil model with
ref
small-strain stiffness requires two additional stiffness parameters as input: G0 and 0.7 .
All other parameters, including the alternative stiffness parameters, remain the same as
ref
in the standard Hardening Soil model. G0 denes the shear modulus at very small
6
strains e.g. < 10 at a reference minor principal stress of '3 = pref .

Poisson's ratio ur is assumed a constant, as everywhere in PLAXIS, so that the shear

ref
modulus G0 can also be calculated from the very small strain Young's modulus as
ref
ref
G0 = E0 /(2(1 + ur )). The threshold shear strain 0.7 is the shear strain at which the
ref
ref
secant shear modulus Gs is decayed to 0.722G0 . The threshold shear strain 0.7 is to
be supplied for virgin loading. In summary, the input stiffness parameters of the
Hardening Soil model with small-strain stiffness are listed below:

: Power for stress-level dependency of stiffness

[-]

ref
E50

: Secant stiffness in standard drained triaxial test

[kN/m2 ]

ref
Eoed

: Tangent stiffness for primary oedometer loading

[kN/m2 ]

ref
Eur

: unloading / reloading stiffness at engineering strains

( 103 to 102 )

[kN/m2 ]

: Poisson's ratio for unloading-reloading

[-]

ref
G0

: reference shear modulus at very small strains

( < 106 )

[kN/m2 ]

0.7

: shear strain at which Gs = 0.722G0

[-]

ur

Figure 6.6 illustrates the model's stiffness parameters in a triaxial test: E50 , Eur , and

68

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS (HSSMALL)

E0 = 2G0 (1 + ur ). For the order of strains at which Eur and G0 are dened and
determined, one may refer to e.g. Figure 6.1 and Figure 6.3.

E0

q = 1 3

E0
E0

E0
E50
Eur

0
axial strain 1
Figure 6.6 Stiffness parameters E50 , Eur , and E0 = 2G0 (1 + ur ) of the Hardening Soil model with
small-strain stiffness in a triaxial test

6.5

ON THE PARAMETERS G0 AND 0.7

A number of factors inuence the small-strain parameters G0 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 G0 is taken into account
with the power law:

G0 = Gref
0

c cos '3 sin

c cos + pref sin

(6.12)

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,
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 G0 . 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):
ref
G0 =

(2.97 e) 2
33 [MPa]
1+e

for

pref = 100 [kPa]

(6.13)

Alpan (1970) empirically related dynamic soil stiffness to static soil stiffness (Figure 6.7).
The dynamic soil stiffness in Alpan's chart is equivalent to the small-strain stiffness G0 or
E0 . Considering that the static stiffness Estatic dened by Alpan equals approximately the
unloading / reloading stiffness Eur in the HS small model, Alpan's chart can be used to
guess a soil's small-strain stiffness entirely based on its unloading / reloading stiffness
Eur . Although Alpan suggests that the ratio E0 /Eur can exceed 10 for very soft clays, the
maximum ratio E0 /Eur or G0 /Gur permitted in the HSsmall model is limited to 10.
In the absence of test data, correlations are also available for the threshold shear strain
0.7 . Figure 6.8 for example gives a correlation between the threshold shear strain and

PLAXIS 2011 | Material Models Manual

69

MATERIAL MODELS MANUAL

100

Ed / Es

Rocks

Soils

10

Cohesive
Granular
1

102

10

103

104
2

Static moduls of elasticity (Es) [kg/cm ]

Figure 6.7 Relation between dynamic (Ed = E0 ) and static soil stiffness (Es Eur ) after Alpan
(1970)
1.1
1.0

G / G0 [ - ]

0.9
PI = 15

0.8
0.7

PI = 30

0.6
PI = 0

0.5
OCR = 1 - 15
0.4
10-6

10-5

10-4

10-3

Shear strain amplitude s [ - ]

Figure 6.8 Inuence of plasticity index (PI) on stiffness reduction after Vucetic & Dobry (1991)

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. (6.2) and (6.3) yields:

0.7

1
9G0

[2c '(1 + cos(2')) '1 (1 + K0 ) sin(2')]

(6.14)

where K0 is the earth pressure coefcient at rest and '1 is the effective vertical stress
(pressure negative).

6.6

MODEL INITIALIZATION

Stress relaxation erases a soil's memory of previous applied stress. Soil aging 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

70

Material Models Manual | PLAXIS 2011

THE HARDENING SOIL MODEL WITH SMALL-STRAIN STIFFNESS (HSSMALL)

(H = 0) in most boundary value problems. This is the default setting in the HS small
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 HS small 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.

On the use of the HS small model in dynamic calculations

In contrast to the Hardening Soil model, the HS small model shows hysteresis in cyclic
loading (Figure 6.6). The amount of hysteresis depends on the magnitude of the
corresponding strain amplitude. However, note that the model does not generate
accumulated strains with multiple loading cycles, nor does it generate pore pressures
with undrained behaviour. When the HS small model is used wave velocities are not
shown because they vary due to the stress-dependent stiffness.
When applied in dynamic calculations, the hysteretic behaviour of the HS small model
leads to damping. The amount of damping depends on the applied load amplitude and
corresponding strain amplitudes. For more information about the hysteretic damping in
the HS small model reference is made to Brinkgreve, Kappert & Bonnier (2007).

6.7

OTHER DIFFERENCES WITH THE HARDENING SOIL MODEL

The mobilised dilatancy angle

The shear hardening ow rule of both the Hardening Soil model and the Hardening Soil
model with small-strain stiffness have the linear form:

p = sin m p
v

(6.15)

The mobilised dilatancy angle m in compression however, is dened differently. The HS

model assumes the following:
For sin m < 3/4 sin

m = 0

For sin m 3/4 sin and > 0

sin m = max

For sin m 3/4 sin and 0

m =

If = 0

sin m sin cv
,0
1 sin m sin cv
(6.16)

m = 0

PLAXIS 2011 | Material Models Manual

71

MATERIAL MODELS MANUAL

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 2c cot

(6.17)

For small mobilised friction angles and for negative values of m , as computed by Rowe's
formula, m in the Hardening Soil model is taken zero. Bounding the lower value of m
may sometimes yield too little plastic volumetric strains though. Therefore, the Hardening
Soil model with small-strain stiffness 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 Hardening Soil model with small-strain stiffness is calculated by the
following equation:

1 ln q
1
M exp 15 M qa
sin m =
10

(6.18)

where M is the stress ratio at failure, and = q/p is the actual stress ratio. Eq. (6.18) is
a simplied version of the void ratio dependent formulation by Li & Dafalias (2000).

72

Material Models Manual | PLAXIS 2011

THE SOFT SOIL MODEL

THE SOFT SOIL MODEL

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 Eoed = 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. 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.

7.1

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:

v 0 = ln
v

p'
p0

(virgin compression)

(7.1)

In order to maintain the validity of Eq. (7.1) 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. (7.1) is a function of volumetric strain
instead of void ratio. Plotting Eq. (7.1) gives a straight line as shown in Figure 12.2.
During isotropic unloading and reloading a different path (line) is followed, which can be
formulated as:

e e0 = ln
v
v

p'
p0

(unloading and reloading)

(7.2)

Again, a minimum value 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 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. (7.2) . The
elastic behaviour is described by Hooke's law. Eq. (7.2)implies linear stress dependency
on the tangent bulk modulus such that:

Kur

p'
Eur
=
3(1 2 ur )

(7.3)

PLAXIS 2011 | Material Models Manual

73

MATERIAL MODELS MANUAL

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, Kur , nor the elastic Young's modulus,
Eur , 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.

*
*
1

pp

ln p'

Figure 7.1 Logarithmic relation between volumetric strain and mean stress

An innite number of unloading / reloading lines may exist in Figure 12.2, each
corresponding to a particular value of the isotropic pre-consolidation stress pp . 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.

7.2

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
dened as:
(7.4)

f = f pp

where f is a function of the stress state (p', q ) and pp , the pre-consolidation stress, is a
function of plastic strain such that:

f =

q2
2

M (p' + c cot )

0
pp = pp exp

+ p'

p
v

(7.5)
(7.6)

The yield function f describes an ellipse in the p' - q -plane, as illustrated in Figure 7.2.
The parameter M in Eq. (7.5) determines the height of the ellipse. The height of the

74

Material Models Manual | PLAXIS 2011

THE SOFT SOIL MODEL

ellipse is responsible for the ratio of horizontal to vertical stresses in primary

one-dimensional compression.

q
1

om
oul
r-C
h
Mo

re
ailu

e
lin

bf

cap

threshold ellipse

c cot

pp

p'

Figure 7.2 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
nc
nc
K0 . In view of this, the value of M can be chosen such that a known value of K0 is
matched in primary one-dimensional compression. Such an interpretation and use of M
nc
differs from the original critical state line idea, but it ensures a proper matching of K0 .
The tops of all ellipses are located on a line with slope M in the p' - q -plane. In (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 pp
determines the extent of the ellipse along p' axis. During loading, innitely many ellipses
may exist (see Figure 7.2) each corresponds to a particular value of pp . In tension
(p' < 0), the ellipse extends to c cot (Eq. (7.5) and Figure 7.2). 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 c cot is adopted for pp . For c = 0, a minimum value of pp
equal to a stress unit is adopted. Hence, there is a 'threshold' ellipse as illustrated in
Figure 7.2.
The value of pp is determined by volumetric plastic strain following the hardening relation,
Eq. (7.6). This equation reects the principle that the pre-consolidation stress increases
0
exponentially with decreasing volumetric plastic strain (compaction). pp can be regarded
0
as the initial value of the pre-consolidation stress. The determination of pp is treated in
Section 2.8. According to Eq. (7.6) the initial volumetric plastic strain is assumed to be
zero.
In the Soft Soil model, the yield function, Eq. (7.4), 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 7.2. The slope of the
failure line is smaller than the slope of the M -line.
The total yield contour, as shown by the bold lines in Figure 7.2, is the boundary of the
elastic stress area. The failure line is xed, but the cap may increase in primary

PLAXIS 2011 | Material Models Manual

75

MATERIAL MODELS MANUAL

- '1
cap

failure surface

- '3
- '2
Figure 7.3 Representation of total yield contour of the Soft Soil model in principal stress space

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 7.3.

7.3

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

[-]

: Cohesion

[kN/m2 ]

: Friction angle

[ ]

: Dilatancy angle

[ ]

Advanced parameters (use default settings):

ur

76

: Poissons ratio for unloading / reloading

[-]

nc
K0

: Coefcient of lateral stress in normal consolidation

[-]

Material Models Manual | PLAXIS 2011

THE SOFT SOIL MODEL

nc
: K0 -parameter

[-]

Figure 7.4 shows the PLAXIS window for inputting the values of the model parameters.
nc
M is calculated automatically from the coefcient of the lateral earth pressure, K0 , by
means of Eq. (7.8). 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.

Figure 7.4 Parameters tabsheet for the Soft Soil model

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 12.2). 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 .
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, Cc and Cr (assumed as equal to Cs ). These relationships are
summarized in Table 7.1.
Remarks on Table 7.1:

PLAXIS 2011 | Material Models Manual

77

MATERIAL MODELS MANUAL

Table 7.1a Relationship to Cam-Clay parameters

1.

1+e

2.

1+e

Table 7.1b Relationship to internationally normalized parameters

3.

Cc
2.3(1 + e)

4.

2 Cs
2.3 (1 + e)

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 relation 4 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.

The factor 2.3 in relation 3 is obtained from the ratio between the logarithm of base
10 and the natural logarithm.

The ratio / (=/) ranges, in general, between 2.5 and 7.

Cohesion
The cohesion has the dimension of stresses. Any effective cohesion may be used,
including a cohesion of zero. Entering a cohesion will result in an elastic region that is
partly located in the 'tension' zone, as illustrated in Figure 7.2. The left hand side of the
ellipse crosses the p'-axis at a value of c cot . 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 pp has a minimum value of c cot . 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. The PLAXIS option to model undrained behaviour
using effective parameters may be used (Undrained (A). Please note that the resulting
effective stress path may not be accurate, which may lead to an unrealistic undrained
shear strength. Hence, when using Undrained (A) as drainage type, the resulting stress
state must be checked against a known undrained shear strength prole.

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.

78

Material Models Manual | PLAXIS 2011

THE SOFT SOIL MODEL

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 linear elastic perfectly-plastic 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
nc
consolidated K0 -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 reloading)

(7.7)

nc
K0 -parameter

The parameter M is automatically determined based on the coefcient of lateral earth

nc
pressure in normally consolidated condition, K0 , as entered by the user. The exact
nc
relation between M and K0 gives (Brinkgreve, 1994):

M=3

nc
1 K0

nc
1 + 2K0

2
2

nc
1 K0 (1 2ur )( / 1)
nc
nc
1 + 2K0 (1 2ur ) / 1 K0 (1 + ur )

(7.8)

The value of M is indicated in the input window. As can be seen from Eq. (7.8), M is also
inuenced by the Poisson's ratio ur and by the ratio / . However, the inuence of
nc
K0 is dominant. Eq. (7.8) can be approximated by:
nc
M 3.0 2.8K0

(7.9)

On the use of the Soft Soil model in dynamic calculations

When using the Soft Soil model in dynamic calculations, the modied swelling index
needs to be selected such that the model correctly predicts wave velocities in the soil.
This generally requires a smaller value than just an unloading-reloading index.
When subjected to dynamic or cyclic loading, the Soft Soil model generally exhibits only
elastic behaviour (except for the rst loading part) and no (hysteretic) damping, nor
accumulation of strains or pore pressure, nor liquefaction. In order to account for the soil
damping, Rayleigh damping may be dened.

PLAXIS 2011 | Material Models Manual

79

MATERIAL MODELS MANUAL

80

Material Models Manual | PLAXIS 2011

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

8.1

INTRODUCTION

A feature of soft soils is the linear stress-dependency of soil stiffness. According to the
Hardening Soil model we have:

Eoed = E ref '1 /pref

oed

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:
ref

Eoed = '1 / where = pref /E 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:
ref
ref

Eoed = p /

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
Hardening Soil 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
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.

PLAXIS 2011 | Material Models Manual

81

MATERIAL MODELS MANUAL

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, Stolle & Bonnier (1998), Vermeer & Neher (1999) 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

8.2

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 CB log

t
tc

for:

t > tc

(8.1)

where c is the strain up to the end of consolidation, t the time measured from the
beginning of loading, tc the time to the end of primary consolidation and CB 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

tc + t '
tc

for:

t' > 0

(8.2)

with t ' = t tc 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 = ec C log

c + t '
c

with: C = CB (1 + e0 ) for: t ' > 0

Differences between Garlanger's and Buisman's forms are modest. The engineering
strain is replaced by void ratio e and the consolidation time tc is replaced by the

82

Material Models Manual | PLAXIS 2011

(8.3)

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

parameter c . Eqs. (8.2) and (8.3) are entirely identical when choosing c = tc . For the
case that c = tc , differences between both formulations will vanish when the effective
creep time t ' increases.
For practical consulting, oedometer tests are usually interpreted by assuming tc = 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 tc , it follows that such tests have no effective creep time.
Hence one obtains t ' = 0 and the log-term drops out of Eq. (8.3). 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 tc .
Another slightly different possibility to describe secondary compression is the form
adopted by Buttereld (1979):

H = H C ln
c
where

c + t '
c

(8.4)

is the logarithmic strain dened as:

H = ln

V
V0

= ln

1 +e
1 + e0

(8.5)

with the subscript '0' denoting the initial values. The superscript 'H ' is used to denote
logarithmic strain, as the logarithmic strain measure was originally used by Hencky. For
small strains it is possible to show that:

C=

CB
C
=
(1 + e0 ) ln 10 ln 10

(8.6)

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.

8.3

ON THE VARIABLES C AND

In this section attention will rst be focused 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. (8.4). 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 '
=

(8.7)

which allows one to make use of the construction developed by Janbu (1969) for
evaluating the parameters C and c from experimental data. Both the traditional way,
being indicated in Figure 8.1a, as well as the Janbu method of Figure 8.1b can be used to
determine the parameter C from an oedometer test with constant load.

PLAXIS 2011 | Material Models Manual

83

MATERIAL MODELS MANUAL

tc

-1/

ln t

tc

c
1

t ' = t tc

t'

t
a. Consolidation

b. Creep behaviour
Figure 8.1 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 8.1b, c is the
intercept with the (non-logarithmic) time axis of the straight creep line. The deviation from
a linear relation for t < tc is due to consolidation.
Considering the classical literature it is possible to describe the end-of-consolidation
strain c , by an equation of the form:

c = e + c = A ln
c
c

'
'0

B ln

pc
p0

(8.8)

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 (or
recompression) index Cs instead of A, and the compression index Cc instead of B . The
above constants A and B relate to Cr and Cc as:

A=

Cr
(1 + e0 ) ln 10

B=

Cc Cr
(1 + eo ) ln 10

(8.9)

Combining Eqs. (8.4) and (8.8) it follows that:

= e + c = A ln

'
'0

B ln

pc
p0

C ln

c + t '
c

(8.10)

where is the total logarithmic strain due to an increase in effective stress from '0 to '
and a time period of tc + t '. In Figure 8.2 the terms of Eq. (8.10) are depicted in an - ln
diagram.
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
the consolidation time is not clearly dened for transient loading conditions.

84

Material Models Manual | PLAXIS 2011

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

'0
A

p0

pc

'

ln( ')
e
c

A+B

c
c

NC - line

C ln(1 + t '/c )

Figure 8.2 Idealised stress-strain curve from oedometer test with division of strain increments into
an elastic and a creep component. For t ' +tc = 1 day, one arrives precisely on the NC-line

8.4

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. 8.10) 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 (8.10) we therefore
introduce the expression:

= e + c = A ln

'
'0

B ln

p
p0

(8.11)

where

p = p0 exp

c
B

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. (8.10) and (8.11) to obtain:

c c = B ln
c

p
pc

= C ln

c +t '
c

(8.12)

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 tc + t ' = ,

PLAXIS 2011 | Material Models Manual

85

MATERIAL MODELS MANUAL

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 ' = tc into Eq. (8.12) it is found that:

'
pc

B ln

= C ln

c + tc
c

for: OCR = 1

(8.13)

It is now assumed that (c tc ) << . This quantity can thus be disregarded with respect
to and it follows that:

B
' C
pc

=
c

or:

B
pc C
c =
'

(8.14)

Hence c depends both on the effective stress ' and the end-of-consolidation
pre-consolidation stress pc . In order to verify the assumption (c tc ) << , 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. (8.14). Hence
not only tc but also c tends to be small with respect to . It thus follows that the
assumption (c tc ) << is certainly correct.
Having derived the simple expression Eq. (8.14) for c , it is now possible to formulate the
differential creep equation. To this end Eq. (8.10) is differentiated to obtain:

= e + c = A

'

' c + t '

(8.15)

where c + t ' can be eliminated by means of Eq. (8.12) to obtain:

'

C
= + = A

' c
e

B
pc C
p

with:

p = p0 exp

c
B

(8.16)

Again it is recalled that c is a compressive strain, being considered negative in this

manual. Eq. (8.14) can now be introduced to eliminate c and pc and to obtain:

' C

= + = A

'
e

86

B
' C
p

Material Models Manual | PLAXIS 2011

(8.17)

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

8.5

THREE-DIMENSIONAL-MODEL

On extending the 1D-model to general states of stress and strain, the well-known stress
invariants for isotropic stress p and deviatoric stress q are adopted. These invariants are
used to dene a new stress measure named peq :

peq = p'

q2

(8.18)

M (p' c cot())

In Figure 8.3 it is shown that the stress measure peq 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).

deviatoric stress

M
1

peq

eq
pp

-p

isotropic stress
Figure 8.3 Diagram of peq -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 8.3. We use the general 3D-denition (Eq. 2.8b) for the deviatoric
stress q and:

M=

6 sin cv
3 sin cv

(8.19)

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 peq 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.
nc
In such situations it yields '2 = '3 = K0 '1 , and it follows from Eq. (8.18) that:
2

peq = '1

nc
nc
3 1 K0
1 + 2 K0
+ 2
nc
3
M 1 + 2 K0

eq
pp = p

nc
nc
3 1 K0
1 + 2 K0
+ 2
nc
3
M 1 + 2 K0

(8.20)

eq

where pp is a generalised pre-consolidation pressure, being simply proportional to the

nc
one-dimensional one. For known values of K0 , peq can thus be computed from ', and
eq
pp can thus be computed from p . Omitting the elastic strain in the 1D-equation (Eq.
eq
8.17), introducing the above expressions for peq and pp and writing instead of it is

PLAXIS 2011 | Material Models Manual

87

MATERIAL MODELS MANUAL

found that:

c
v

B
C

peq
eq
pp

C
=

eq
eq
pp = pp0 exp

where

c
v
B

(8.21)

For one-dimensional oedometer conditions, this equation reduces to Eq. (8.17), 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

(8.22)

On using these new parameters, Eq. (8.21) changes to become:

c =
v

peq

with

eq
pp

eq
eq
pp = Pp0 exp

c
v

(8.23)

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:
T

and

= 1 2 3

= 1 2 3

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 ' +

gc
'

(8.24)

where the elasticity matrix and the plastic potential function are dened as:

D 1 =

ur ur

ur 1 ur

Eur
ur ur 1

and

g c = peq

Hence we use the equivalent pressure peq as a plastic potential function for deriving the
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

88

Material Models Manual | PLAXIS 2011

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

behaviour. Now it follows from the above equations that:

c = c + c + c =
v 1 2 3

peq peq peq

+
+
'1
'2
'3

peq
=
p'

(8.25)

Hence we dene = peq /p'. Together with Eqs. (8.23) and (8.24) this leads to:

= D 1 ' +

c
v

peq

peq

= D 1 '

'

eq
pp

peq
'

(8.26)

where:
eq
eq
pp = pp0 exp

c
v

or inversely:

c = ( )ln
v

8.6

eq
pp
eq
pp0

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 Eur has to
been dened as a stress-dependent tangent stiffness according to:

Eur = 3(1 2ur )Kur = 3(1 2ur )

p'

(8.27)

Hence, Eur 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 Eur , the bulk modulus Kur is stress dependent according to the rule
Kur = (p + c cot )'/ . Now it can be derived for the volumetric elastic strain that:

e =
v

p'
p'
=
p' + c cot
Kur

or by integration: e = ln

p' + c cot
p'0 + c cot
(8.28)

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
nc
nc
have both 3p' = (1 + 2 K0 ) ' and 3p'0 = (1 + 2 K0 ) '0 and it follows that p'/p'0 = '/ '0 .
c
As a consequence we derive the simple rule = ln '/ '0 , whereas the 1D-model
involves c = A ln '/ '0 . It would thus seem that coincides with A. Unfortunately

this line of thinking cannot be extended toward overconsolidated states of stress and
strain. For such situations, it can be derived that:

p' 1 + ur
1
'

=
p' 1 ur 1 + 2K0 '

(8.29)

PLAXIS 2011 | Material Models Manual

89

MATERIAL MODELS MANUAL

and it follows that:

e =
v

'

p' 1 + ur

=
p' 1 ur 1 + 2K0 '

(8.30)

where K0 depends to a great extent on the degree of over-consolidation. For many

situations, it is reasonable to assume K0 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.
8.7

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 requires 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.

Figure 8.4 Parameters tabsheet for the Soft Soil Creep model

In conclusion, the Soft Soil Creep model requires the following material constants:
Failure parameters as in the Mohr-Coulomb model:

: Cohesion

[kN/m2 ]

: Friction angle

[ ]

: Dilatancy angle

[ ]

Basic stiffness parameters:

90

Material Models Manual | PLAXIS 2011

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

: 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)

[-]

nc
K0

: 'xx / 'yy stress ratio in a state of normal consolidation

[-]

nc
K0 -related

parameter (see below)

[-]

By default, M is automatically determined based on the coefcient of lateral earth

nc
pressure in normally consolidated condition, K0 , as entered by the user. The exact
nc
relationship between M and K0 can be formulated as (Brinkgreve, 1994):

M=3

nc
1 K0

nc
1 + 2K0

2
2

nc
1 K0 (1 2ur )( / 1)
nc
nc
1 + 2K0 (1 2ur ) / 1 K0 (1 + ur )

(8.31)

Hence the user cannot enter directly a particular value of M . Instead he can choose
nc
values for K0 . 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
parameters can be used. These material constants are given by:

Cc

: Compression index

[-]

Cr

: Recompression index

[-]

: Creep index for secondary compression

[-]

einit

: 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 8.2). The slope of the normal
consolidation line gives the modied compression index , and the slope of the
unloading (or recompression) line can be used to compute the modied swelling index
, as explained in Section 8.6. 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
(Table 8.1b) can be obtained by measuring the volumetric strain on the long term and
plotting it against the logarithm of time (see Figure 8.1).
In Table 8.1c, 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 einit . Entering a particular value for one of the alternatives Cc , Cr or C

PLAXIS 2011 | Material Models Manual

91

MATERIAL MODELS MANUAL

results in a change of *, * or * respectively.

As already indicated in Section 8.6, there is no exact relation between the isotropic
compression index and the one-dimensional swelling indices A and Cr , 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
Ip (%)/500, the fact that / is in the range between 15 to 25 and the general
observation / is between 2.5 and 7.
Table 8.1a Relationship to Cam-Clay parameters

1+e

1+e

---

Table 8.1b Relationship to A,B , C parameters

= B +

2A

= C

Table 8.1c Relationship to internationally normalized parameters

Cc
2.3(1 + e)

2 Cr
2.3 1 + e

C
2.3(1 + e)

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
eq
compute the initial value of the generalised pre-consolidation pressure pp (see Section
2.8).

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
nc
not be based on the normally consolidated K0 -value, but on the ratio of difference in
horizontal stress to difference in vertical stress in oedometer unloading and reloading:

ur
xx
=
yy
1 ur

(unloading and reloading)

(8.32)

On the use of the Soft Soil Creep model in dynamic calculations

When using the Soft Soil Creep model in dynamic calculations, the modied swelling
index needs to be selected such that the model correctly predicts wave velocities in
the soil. This generally requires a smaller value than just an unloading-reloading index.

92

Material Models Manual | PLAXIS 2011

SOFT SOIL CREEP MODEL (TIME DEPENDENT BEHAVIOUR)

When subjected to dynamic or cyclic loading, the Soft Soil Creep model generally
exhibits only elastic behaviour (except for the rst loading part) and no (hysteretic)
damping, nor accumulation of strains or pore pressure, nor liquefaction. In order to
account for the soil damping, Rayleigh damping may be dened.

PLAXIS 2011 | Material Models Manual

93

MATERIAL MODELS MANUAL

94

Material Models Manual | PLAXIS 2011

THE JOINTED ROCK MODEL (ANISOTROPY)

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.
rock formation

stratication
major joint
direction
Figure 9.1 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 optional stratication direction and major joint
directions. The intact rock is considered to behave as a transversely 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 E1 ,E2 , 1 , 2 ,
G2

Shear failure according to Coulomb in three directions, Parameters ci , i and i

Limited tensile strength in three directions, i

Parameters t,i

PLAXIS 2011 | Material Models Manual

95

MATERIAL MODELS MANUAL

9.1

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 y- direction, E2 , is
different from the stiffness in the rock as a continuum, E1 . 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 =

2 yy

1 zz

xx

E1
E2
E1

(9.1a)

yy =

2 xx yy

2 zz

E2
E2
E2

(9.1b)

zz =

2 yy zz

1 xx

+
E1
E2
E1

(9.1c)

xy =

yz =

zx =

xy

(9.1d)

G2
yz

(9.1e)

G2
2(1 + 1 ) zx

E1

(9.1f)

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 9.3). 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
nst = R xyz

96

Material Models Manual | PLAXIS 2011

xyz = R 1 nst

(9.2a)

R 1 nst

(9.2b)

xyz =

THE JOINTED ROCK MODEL (ANISOTROPY)

where

and

2
nx

2
ny

2
nz

2 nx ny

2 ny nz

2 nx nz

2
2
2
sx
2 sx sy
2 sy sz
2 sx sz
sy
sz

2
2
2

tx
2 tx ty
2 ty tz
2 tx tz
ty
tz

R =

nx sx ny sy nz sz nx sy + ny sx ny sz + nz sy nz sx + nx sz

sx tx sy ty sz tz sx ty + sy tx sy tz + sz ty sx tz + sz tx

+ ny tx ny tz + nz ty nz tx + nx tz
nx tx ny ty nz tz nx ty

2
nx

2
ny

2
nz

nx ny

ny nz

nx nz

(9.3)

2
2
2

sx
sx sy
sy sz
sx sz
sz
sy

2
2
2

tx
ty
tz
tx ty
ty tz
tx tz

R =

2 nx sx 2 ny sy 2 nz sz nx sy + ny sx ny sz + nz sy nz sx + nx sz

2 sx tx 2 sy ty 2 sz tz sx ty + sy tx sy tz + sz ty sx tz + sz tx

2 nx tx 2 ny ty 2 nz tz nx ty + ny tx ny tz + nz ty nz tx + nx tz

(9.4)

nx , ny , nz , sx , sy , sz , tx , ty and tz are the components of the normalized n, s and

t -vectors in global (x ,y ,z )-coordinates (i.e. 'sines' and 'cosines'; see Section 9.3). For
plane condition nz = sz = tx = ty = 0 and tz = 1.
It further holds that :

R T = R 1

R T = R 1

(9.5)

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

R xyz = D R xyz
nst = R xyz
nst

=R
nst

Hence,

(9.6)

xyz

xyz = R 1 D R xyz

nst

(9.7)

Using to above condition (Eq. 9.5):

xyz = R T D R xyz = D xyz

nst
xyz

or

D = RT D Re
xyz

nst

PLAXIS 2011 | Material Models Manual

(9.8)

97

MATERIAL MODELS MANUAL

Actually, not the D *-matrix is given in local coordinates but the inverse matrix (D )1 .

nst = D
nst

nst

nst = R xyz
nst = R xyz
Hence,

D
xyz

= RT D

nst

xyz = R 1 D

nst

or

R xyz = R T D
nst

D = RT D
xyz

nst

R xyz

(9.9)

(9.10)

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 .
9.2

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 ci , 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

(9.11)

where

i = n s t

= xx yy zz xy yz zx

(9.12a)
T

(9.12b)

T T = transformation matrix (3x6), for plane i

i
As usual in PLAXIS, tensile (normal) stresses are dened as positive whereas
compression is dened as negative.
Consider a plane strain situation as visualized in Figure 9.2. Here a sliding plane is
considered under an angle 1 (= dip angle) with respect to the x-axis. In this case the

98

Material Models Manual | PLAXIS 2011

THE JOINTED ROCK MODEL (ANISOTROPY)

y
s

sliding plane

1
1

Figure 9.2 Plane strain situation with a single sliding plane and vectors n, s

transformation matrix T

where

becomes:

s2 c 2 0

2sc

T
T = sc sc 0 s2 + c 2 0 0

0 0 0
0
c s

(9.13)

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 9.3):

2
nx

2
ny

2
nz

2 nx ny

2 ny nz

2 nz nx

T T = nx sx ny sy nz sz nx sy + ny sx nz sy + ny sz nz sx + nx sz

nx tx ny ty nz tz ny tx + nx ty ny tz + nz ty nz tx + nx tz

(9.14)

Note that the general transformation matrix, T , for the calculation of local stresses
corresponds to rows 1, 4 and 6 of R (see Eq. 9.3).
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:

fic =|s |+ n tan i ci

t
fi

= n t,i (t,i ci coti )

(Coulomb)

(9.15a)

(Tension cut-off)

(9.15b)

Figure 9.3 visualizes the full yield criterion on a single plane.

PLAXIS 2011 | Material Models Manual

99

MATERIAL MODELS MANUAL

| |

ci

t,i

Figure 9.3 Yield criterion for individual plane

The local plastic strains are dened by:

p = j
j

gjc

(9.16)

where gj is the local plastic potential function for plane j :

gjc = j + n tanj cj
gjt

(Coulomb)
(Tension cut-off)

= n t,j

(9.17a)
(9.17b)

The transformation matrix, T , is also used to transform the local plastic strain increments
p
of plane j , j , into global plastic strain increments, p :

p = T j p
j

(9.18)

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.
np
c
fi

= fic(e)

< c >
j
j=1

T
gjc
fic
T
D Tj

Ti

np

< tj >
j=1

T
gjt
fic
T
D Tj
Ti

(9.19a)
T

np
t(e)
t
fi = fi

< c >
j
j=1

gjc
fit
T

Ti D Tj

np

< tj >
j=1

gjt
fit
T
Ti D Tj

(9.19b)

This means nding up to 6 values of i 0 such that all fi 0 and i fi = 0

When the maximum of 3 planes are used, there are 26 = 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.

100 Material Models Manual | PLAXIS 2011

THE JOINTED ROCK MODEL (ANISOTROPY)

9.3

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):

E1

: Young's modulus for rock as a continuum

[kN/m2 ]

: Poisson's ratio for rock as a continuum

[-]

Anisotropic elastic parameters 'Plane 1' direction (e.g. stratication direction):

E2

: Young's modulus perpendicular on 'Plane 1' direction

[kN/m2 ]

G2

: Shear modulus perpendicular on 'Plane 1' direction

[kN/m2 ]

: Poisson's ratio perpendicular on 'Plane 1' direction

[-]

Strength parameters in joint directions (Plane i=1, 2, 3):

ci

: Cohesion

[kN/m2 ]

: Friction angle

[ ]

: Dilatancy angle

[ ]

t,i

: Tensile strength

[kN/m2 ]

Denition of joint directions (Plane i=1, 2, 3):

n
1,i
2,i

: Number of joint directions (1 n 3)

[-]

: Dip direction (2,i = 90 in PLAXIS 2D)

[ ]

: Dip angle (180 1,i 180)

[ ]

Elastic parameters
The elastic parameters E1 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 E2 , 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 E2 and 2 .
Elastic shearing in the stratication direction is also considered to be 'weaker' than elastic
shearing in other directions. In general, the shear stiffness in the anisotropic direction can
explicitly be dened by means of the elastic shear modulus G2 . In contrast to Hooke's
law of isotropic elasticity, G2 is a separate parameter and is not simply related to Young's
modulus by means of Poisson's ratio (see Eq. 9.1d and e).
If the elastic behaviour of the rock is fully isotropic, then the parameters E2 and 2 can be

PLAXIS 2011 | Material Models Manual 101

MATERIAL MODELS MANUAL

Figure 9.4 Parameters for the Jointed Rock model (PLAXIS 3D)

simply set equal to E1 and 1 respectively, whereas G2 should be set to E1 /(1 + 1 ).

Strength parameters
Each sliding direction (plane) has its own strength properties ci , i and t,i and dilatancy
angle i . The strength properties ci 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,

102 Material Models Manual | PLAXIS 2011

THE JOINTED ROCK MODEL (ANISOTROPY)

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 9.5.
Consider a sliding plane, as indicated in Figure 9.5. 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 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 ought to be entered in the
range [0 , 90 ], but negative values as well as values larger than 90 can also be entered.

z
y

x
2
1

s
s
Figure 9.5 Denition of dip angle and dip direction

The orientation of the sliding plane is further dened by the dip direction, 2 , which is
dened in PLAXIS as the orientation of the vector t with respect to the x -direction. The
dip direction is dened as the positive angle from the x -direction, measured clockwise to
the t -axis. The dip direction is entered in the range [-180 , 180 ].
From the denitions as given above, it follows for PLAXIS 3D that:

nx

sin 1 sin 2

n = ny = sin 1 cos 2

nz
cos 1

sx
cos 1 sin 2

s = sy = cos 1 cos 2

sz
sin 1

(9.20a)

(9.20b)

PLAXIS 2011 | Material Models Manual 103

MATERIAL MODELS MANUAL

tx

t = ty

tz

cos 2

= sin 2

(9.20c)

whereas for PLAXIS 2D 2 is taken by denition as 2 = 90 , such that:

nx

sin 1

n = ny =

cos 1
nz

sx
cos 1

s = sy = 0

sz
sin 1

t
0
x

t = ty = 1

0
tz

(9.21a)

(9.21b)

(9.21c)

Figure 9.6 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 PLAXIS 2D) only 1 is required. By default, 2 is xed at 90 and the
declination is set to 0 .

On the use of the Jointed Rock model in dynamic calculations

When using the Jointed Rock model in dynamic calculations, the stiffness need to be
selected such that the model correctly predicts wave velocities in the soil (Equation
(3.13)). When subjected to dynamic or cyclic loading, the Jointed Rock model generally
exhibits only elastic behaviour and no (hysteretic) damping, nor accumulation of strains or
pore pressure or liquefaction. In order to simulate the rock's damping characteristics,
Rayleigh damping may be dened.

104 Material Models Manual | PLAXIS 2011

THE JOINTED ROCK MODEL (ANISOTROPY)

PLAXIS 3D

z
1
1 = 30
2 = 0
x

2 = 90
1

1 = 30
2 = 90
x

PLAXIS 2D

y
1

1 = 30
2 = 90
x
Figure 9.6 Examples of failure directions dened by 1 and 2

PLAXIS 2011 | Material Models Manual 105

MATERIAL MODELS MANUAL

106 Material Models Manual | PLAXIS 2011

MODIFIED CAM-CLAY MODEL

10

MODIFIED CAM-CLAY MODEL

The Modied Cam-Clay model is described in several textbooks on critical state soil
mechanics (for example Muir Wood (1990)). In this chapter a short overview is given of
the basic equations.
In the Modied Cam-Clay model, a logarithmic relation is assumed between void ratio e
and the mean effective stress p' in virgin isotropic compression, which can be formulated
as:

e e0 = ln

p'
p0

(virgin isotropic compression)

(10.1)

The parameter is the Cam-Clay compression index, which determines the

compressibility of the material in primary loading. When plotting relation (Eq. 10.1) in a
e - ln p' diagram one obtains a straight line. During unloading and reloading, a different
line is followed, which can be formulated as:

e e0 = ln

p'
p0

(isotropic unloading and reloading)

(10.2)

The parameter is the Cam-Clay swelling index, which determines the compressibility of
material in unloading and reloading. In fact, an innite number of unloading and reloading
lines exists in p' - e-plane each corresponding to a particular value of the preconsolidation
stress pc .
The yield function of the Modied Cam-Clay model is dened as:

f =

q2
+ p' (p' pc )
M2

(10.3)

The yield surface (f = 0) represents an ellipse in p' - q -plane as indicated in Figure 10.1.
The yield surface is the boundary of the elastic stress states. Stress paths within this
boundary only give elastic strain increments, whereas stress paths that tend to cross the
boundary generally give both elastic ans plastic strain increments.
In p' - q -plane, the top of the ellipse intersects a line that we can be written as:

q = Mp'

(10.4)

This line is called the critical state line (CSL) and gives the relation between p' and q in a
state of failure (i.e. the critical state). The constant M is the tangent of the critical state
line and determines the extent to which the ultimate deviatoric stress, q , depends on the
mean effective stress, p'. Hence, M can be regarded as a friction constant. Moreover, M
determines the shape of the yield surface (height of the ellipse) and inuences the
nc
coefcient of lateral earth pressure K0 in a normally consolidated stress state.
The preconsolidation stress, pc , determines the size of the ellipse. In fact, an innite
number of ellipses exist, each corresponding to a particular value of pc .
The left hand side of the yield ellipse (often described as the 'dry side' of the critical state
line) may be thought of as a failure surface. In this region plastic yielding is associated
with softening, and therefore failure. The values of q can become unrealistically large in

PLAXIS 2011 | Material Models Manual 107

MATERIAL MODELS MANUAL

Critical State Line

q
'dry side'

'wet side'

nc
K0 -line

pc

p'

Figure 10.1 Yield surface of the Modied Cam-Clay model in p' - q -plane

this region.
For more detailed information on Cam-Clay type models, the reader is referred to Muir
Wood (1990).
In conclusion, the Modied Cam-Clay model is based on ve parameters:

ur :

Poisson's ratio

Cam-Clay swelling index

Cam-Clay compression index

M:

Tangent of the critical state line

einit :

Initial void ratio

Figure 10.2 Parameters for the Modied Cam-Clay model

Poisson's ratio
Poisson's ratio ur is a real elastic parameter and not 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.

108 Material Models Manual | PLAXIS 2011

MODIFIED CAM-CLAY MODEL

Compression index and swelling index

These parameters can be obtained from an isotropic compression test including isotropic
unloading. When plotting the natural logarithm of the mean stress as a function of the
void ratio for clay-type materials, the plot can be approximated by two straight lines. The
slope of the primary loading line gives the compression index and the slope of the
unloading line gives the swelling index. These parameters can be obtained from a
one-dimensional compression test, as discussed in Section 7.3.

Tangent of the critical state line

In order to obtain the correct shear strength, the parameter M should be based on the
friction angle . The critical state line is comparable with the Drucker-Prager failure line,
and represents a (circular) cone in principle stress space. Hence, the value of M can be
obtained from in a similar way as the Drucker-Prager friction constant is obtained
from . In addition to determining the shear strength, the parameter M has an important
nc
inuence on the value of the coefcient of lateral earth pressure, K0 , in a state of normal
consolidation. In general, when M is chosen such that the model predicts the correct
nc
shearing strength, the value of K0 is too high.

Warning
The Modied Cam-Clay model may allow for extremely large shear stresses. This is
particularly the case for stress paths that cross the critical state line. Furthermore, the
Modied Cam-Clay model may give softening behaviour for particular stress paths.
Without special regularization techniques, softening behaviour may lead to mesh
dependency and convergence problems of iterative procedures. Moreover, the Modied
Cam-Clay model| cannot be used in combination with Safety analysis by means of phi-c
reduction. The use of the Modied Cam-Clay model in practical applications is not
recommended.

On the use of the Modied Cam-Clay model in dynamic calculations

When using the Modied Cam-Clay model in dynamic calculations, the swelling index
needs to be selected such that the model correctly predicts wave velocities in the soil.
This generally requires a smaller value than just an unloading-reloading index.
When subjected to dynamic or cyclic loading, the Modied Cam-Clay model generally
exhibits only elastic behaviour (except for the rst loading part) and no (hysteretic)
damping, nor accumulation of strains or pore pressure, nor liquefaction. In order to
account for the soil damping, Rayleigh damping may be dened.

PLAXIS 2011 | Material Models Manual 109

MATERIAL MODELS MANUAL

110 Material Models Manual | PLAXIS 2011

THE NGI-ADP MODEL (ANISOTROPIC UNDRAINED SHEAR STRENGTH)

11

THE NGI-ADP MODEL (ANISOTROPIC UNDRAINED SHEAR STRENGTH)

The NGI-ADP model may be used for capacity, deformation and soil-structure interaction
analyses involving undrained loading of clay. The basis of the material model is:

Input parameters for (undrained) shear strength for three different stress paths/
states (Active, Direct Simple Shear, Passive).

A yield criterion based on a translated approximated Tresca criterion.

Elliptical interpolation functions for plastic failure strains and for shear strengths in
arbitrary stress paths.

Isotropic elasticity, given by the unloading/reloading shear modulus, Gur .

11.1

FORMULATION OF THE NGI-ADP MODEL

The NGI-ADP model is formulated for a general stress state, matching both undrained
failure shear strengths and strains to that of selected design proles (Andresen & Jostad
(1999), Andresen (2002), Grimstad, Andresen & Jostad (2010)). The model formulation
is presented in steps, starting with 1D anisotropy in triaxial test condition. In Section
11.1.2 a simplied expression for plane strain is presented. Thereafter the formulation is
extended to full 3D stress state. In this formulation compressive stresses are positive.
In the NGI-ADP model the Tresca approximation after Billington (1988) together with a
modied von Mises plastic potential function (von Mises (1913)) is used to circumvent the
possible corner problems. The yield and plastic potential function are independent of the
mean stress hence zero plastic volume strain develops.
11.1.1

1D MODEL PRESENTATION

C
Under triaxial tests condition two undrained shear strengths can be determined, i.e. su
E
andsu . The test measures the response in vertical stress 'v and horizontal stress 'h for
applied shear strain . The Tresca yield criteria can be modied, Eq. (11.1), to account
for the difference in undrained shear strength in compression and extension:

f = (1 )0

E
C
E
su su
sC + su
=0
u
2
2

(11.1)

where = 0.5( 'v 'h ) and the initial in situ maximum shear stress 0 is then dened as
0 = 0.5( 'v 0 'h0 ) = 0.5 'v 0 (1 K0 ).
To account for difference in failure shear strain a stress path dependent hardening
parameter is introduced. The stress path dependent hardening is made possible by
p
different plastic failure shear strain f in compression and extension. The hardening
function is given by:

=2

p /fp
1 + p /fp

when p < f

else = 1

(11.2)

where p and f are the plastic shear strain and the failure (peak) plastic shear strain

PLAXIS 2011 | Material Models Manual 111

MATERIAL MODELS MANUAL

respectively.
stress path

stress-strain

max = 1(1 3 )
2

cs line
TXC

C
su

0
K0 line

fC
fE

p'

E
su

TXE

cs line

Figure 11.1 Typical stress paths and stress strain curves for triaxial compression and triaxial
extension

11.1.2

THE NGI-ADP MODEL IN PLANE STRAIN

The yield criterion for the NGI-ADP model in plane strain is dened by:

f =

P
yy xx
s A su
(1 )0 u
2
2

+ xy

A
P
su + su
DSS
2su

A
P
su + su
=0
2

(11.3)
Restriction to clays with horizontal surfaces are made to simplify the presentation.
Further y is taken as the vertical (depositional) direction. For isotropy in hardening (i.e.
independent of stress orientation) Eq. (11.3) plots as an elliptical shaped curve in a plane
strain deviatoric stress plot. When equals 1.0, the criterion in Eq. (11.3) reduces to the
formulation given by Davis & Christian (1971). While hardening the yield curves are
characterized by slightly distorted elliptical shapes. The shape is dependent on the
interpolation function used and values of failure strain. The NGI-ADP model uses
elliptical interpolation between failure strain in passive stress state, direct simple shear
and active stress state. In the implementation of the NGI-ADP model the yield surface is
ensured to remain convex by restricting the input.
11.1.3

THE NGI-ADP MODEL IN 3D STRESS SPACE

This section describes the actual implementation of the NGI-ADP model inPLAXIS,
whereas the previous sections should be regarded as an introduction using simplied
formulation. For the general stress condition a modied deviatoric stress vector is dened

112 Material Models Manual | PLAXIS 2011

THE NGI-ADP MODEL (ANISOTROPIC UNDRAINED SHEAR STRENGTH)

Figure 11.2 Typical deviatoric plane strain plot of equal shear strain contours for the NGI-ADP model

as:

s
xx

syy

szz

sxy

sxz

syz

P
A

'xx 'xx0 (1 ) + 1 (su su ) p

3
' ' (1 ) 2 (sA sP ) p
yy

yy0
u

3 u

A
P
'zz 'zz0 (1 ) + 1 (su su ) p


3

=
A
P

su + su

xy

DSS

2su

xz

A
P

su + su
yz
DSS
2su

(11.4)

where 'xx0 , 'yy 0 and 'xx0 are the initial stresses and p is the modied mean stress.
The modied mean stress is dened as:
( 'xx 'xx0 (1 )) + ( 'yy 'yy0 (1 )) + ( 'zz 'zzx0 (1 ))
3
= p' (1 )p'0

p=

(11.5)

where p' is the mean stress. Modied second and third deviatoric invariants are dened
accordingly in Eqs. (11.6) and (11.7).

2
2
2
J2 = xx syy sxx szz syy szz + sxz + sxz + syz
s

(11.6)

2
2
2
J3 = sxx syy szz + 2 xy syz sxz sxx syz syy sxz szz sxy
s

(11.7)

The yield criterion is expressed as:

f =

H()J2

A
P
su + su
=0
2

(11.8)

PLAXIS 2011 | Material Models Manual 113

MATERIAL MODELS MANUAL

where, to approximate the Tresca criterion, the term H() is dened as:

1
arccos(1 2a1 )
6

H() = cos2

with =

2
27 J3
3
4 J2

(11.9)

By letting the value of a1 go to 1.0 an exact Tresca criterion is obtained. The parameter
C
A
a1 can be directly linked to the rounding ratio su /su . This ratio takes typically a value just
below 1.0 and a value of 0.99 is chosen as an appropriate default. Figure 11.3 shows the
failure criterion of the NGI-ADP model model in the -plane (for Cartesian stresses) with
default rounding ratio. This criterion is continuous and differentiable and it is described by
a single function.

Figure 11.3 Failure criterion of the NGI-ADP model in the -plane

A
C
The combinations of strength ratios are limited by lower limit for combinations of su /su
P
A
and su /su .
p

The value of f is given by elliptical interpolation:

fp () =

RB RD

2
2
2
2

(RD RC )cos2 (2) + RC RD RA cos(2)

2
2
2

RB (RB RD )cos2 (2)

(11.10)

where

RA =

RB =

RC =

RD =
p

fp,E fp,C
fp,E

2
+ fp,C
2

fp,E fp,C

fp,DSS RB

RC
p

and f ,C , f ,DSS and f ,E are the failure plastic maximum shear strain in triaxial

114 Material Models Manual | PLAXIS 2011

(11.11)
(11.12)
(11.13)
(11.14)

THE NGI-ADP MODEL (ANISOTROPIC UNDRAINED SHEAR STRENGTH)

compression, direct simple shear and triaxial extension respectively. Note that is not the
Lode angle, but is dened as:

cos(2) =

3 syy
2

J2

(11.15)

A non-associated ow rule is used such that the derivative of the plastic potential g is:

1 p
g
=
I+
' 2
'

J2

J2

(11.16)

where I is a modied unit vector:

=
I

1
1
1
A
P
su + su
DSS
2su

A
P
su + su
DSS
2su

(11.17)

The increment in plastic shear strain is dened as:

p =

H()

2
dp dp
xx
yy
3

+ dp dp
xx
zz

+ dp dp
yy
zz

1/2
p
dxy

p
+ dxz

p
+ dyz

(11.18)

where the plastic strains are dened as:

dp = d

g
'

(11.19)

with d being the plastic multiplier. Hence, Eq. (11.18) gives the relation between the
hardening parameter p and the plastic multiplier d.
11.1.4

THE NGI-ADP MODEL TRACTION CRITERION FOR INTERFACES

For plane strain conditions a traction criterion, corresponding to the plane strain failure
criterion is formulated. This criterion is intended to be used on interface elements in nite
element calculations. The interface strength is controlled by a lower and upper limit,
which are dependent on the direction of the interface, .
Let a plane being oriented by the direction to the horizontal. The plane has a tangential
direction t and a normal direction n and the adjacent continuum denes the stresses nn ,

PLAXIS 2011 | Material Models Manual 115

MATERIAL MODELS MANUAL

tt and n by:

tt
xx

nn = A yy

xy
tn

(11.20)

where A is the transformation matrix.

In a local coordinate system three strains nn , tt and tn are dened in the plane strain
condition. Due to the requirement of no volume change for a perfect plastic mechanism
with shearing in tangential direction, will give that nn = tt = 0, resulting in:

f
=0
(nn tt )

(11.21)

The plain strain formulation for the NGI-ADP model is dened as follows:

f =

nn tt
cos(2) tn sin(2) RA
2
RB = 0

RB
RD

nn tt
sin(2) + tn cos(2) 2
2
(11.22)

where
A
P
su su
2
A
P
su + su
RB =
2
DSS
RD =su

(11.23)

RA =

11.2

(11.24)
(11.25)

PARAMETERS OF THE NGI-ADP MODEL

Stiffness parameters:
A
Gur /su

: Ratio unloading/reloading shear modulus over (plane

strain) active shear strength

[-]

fC

: Shear strain at failure in triaxial compression

[%]

fE

: Shear strain at failure in triaxial extension

[%]

fDSS

: Shear strain at failure in direct simple shear

[%]

Strength parameters:
A
su,ref

: Reference (plane strain) active shear strength

C,TX
A
su /su : Ratio triaxial compressive shear strength over (plane
strain) active shear strength (default = 0.99)

yref

: Reference depth

116 Material Models Manual | PLAXIS 2011

[kN/m2 /m]
[-]
[m]

THE NGI-ADP MODEL (ANISOTROPIC UNDRAINED SHEAR STRENGTH)

A
su,inc

: Increase of shear strength with depth

[kN/m2 /m]

P
A
su /su

: Ratio of (plane strain) passive shear strength over

(plane strain) active shear strength

[-]

A
0 /su

: Initial mobilization (default = 0.7)

[-]

DSS
A
su /su

: Ratio of direct simple shear strength over (plain strain)

active shear strength

[-]

Advanced parameter:

'

: Poisson's ratio

[-]

Ratio unloading / reloading shear modulus over plane strain active shear
A
strength (Gur /su )
Ratio unloading / reloading shear stiffness as a ratio of the plane strain active shear
A
strength. If the shear strength is increasing with depth the constant ratio for Gur /su gives
a shear stiffness increasing linearly with depth.

Shear strain at failure in triaxial compression (fC )

This parameter fC (%) denes the shear strain at which failure is obtained in undrained
triaxial compression mode of loading, i.e. fC = 3/2 C from triaxial testing.
1

Shear strain at failure in triaxial extension (fE )

This parameter fE (%) denes the shear strain at which failure is obtained in undrained
triaxial extension mode of loading, i.e. fE = 3/2 C from triaxial testing.
1

Shear strain at failure in direct simple shear (fDSS )

This parameter fDSS (%) denes the shear strain at which failure is obtained in undrained
direct simple shear mode of loading (DSS device).
Hint: For near normally consolidated clays, the failure strain in compression
loading fC is generally the lowest value and the failure strain in extension
loading fE is the highest value. The failure strain from direct simple shear
loading takes an intermediate value, i.e. fC < fDSS < fE . From laboratory
test results reported in literature one nd typically fE in the range 3-8 %,
fDSS in the range 2-8 % and fC in the range 0.5 - 4 %.

A
Reference active shear strength(su,ref )

The reference active shear strength is the shear strength obtained in (plane strain)
undrained active stress paths for the reference depth yref , expressed in the unit of stress.

PLAXIS 2011 | Material Models Manual 117

MATERIAL MODELS MANUAL

Hint: If stress-strain curves from undrained triaxial and/or DSS laboratory tests are
available it is recommended to choose the elastic shear modulus and failure
strains such that a good t to the curves are obtained. This is in particular
important for deformation and SLS assessments. However, for pure capacity
and stability (e.g. factor of safety) analyses the values for shear strains at
failure is not important and one may set all three values equal to e.g. 5 % for
simplicity.

Hint: Note that it is the failure strains from triaxial loading that is input because
they are the most readily available. When the NGI-ADP model is used for
plane strain conditions the failure strains will automatically be slightly
adjusted for that loading condition. See Grimstad, Andresen & Jostad (2010)
for more details

Ratio triaxial compressive shear strength over active shear

C,TX
A
strength(su /su )
C,TX
A
This ratio su /su denes the shear strength in undrained triaxial compression mode of
loading in relation to the shear strength in plane strain undrained active mode of loading.
The default value is 0.99 giving practically the same strengths in triaxial and plane strain
conditions.

Reference depth (yref )

A
This is the reference depth yref at which the reference active shear strength su,ref is
dened. Below this depth the shear strength and stiffness may increase linearly with
A
increasing depth. Above the reference depth the shear strength is equal to su,ref .

A
Increase of shear strength with depth (su,inc )
A
This parameter su,inc denes the increase (positive) or decrease (negative) of the
undrained active shear strength with depth, expressed in the unit of stress per unit of
A
depth. Above the reference depth the shear strength is equal to su,ref , below the
reference depth the shear strength is dened as:
A
A
A
su (y ) = su,ref + (yref y )su,inc

(11.26)

P
A
Ratio of passive shear strength over active shear strength (su /su )
P
A
This ratio su /su denes the undrained shear strength for (plane strain) passive mode of
loading.

DSS
A
Ratio of direct simple shear strength over active shear strength (su /su )
DSS
A
This ratio su /su denes the undrained shear strength for direct simple shear mode of
loading.

118 Material Models Manual | PLAXIS 2011

THE NGI-ADP MODEL (ANISOTROPIC UNDRAINED SHEAR STRENGTH)

Hint: Please note that active / passive strength input is dened for plane strain
conditions. However, it is generally acceptable and only slightly conservative
to use the strength obtained from a triaxial compression test as input for the
C,TX
A
active plane strain condition (i.e. su = su ) and the strength obtained from
a triaxial extension test as input for the passive plane strain condition (i.e.
E,TX
P
su = su ). More control over the strength difference between triaxial and
plane strain loading conditions can be obtained by using the advanced
C,TX
A
parameter su /su .

P
Hint: For near normally consolidated clays, the passive strength su is generally the
lowest strength value, while the direct simple shear strength takes an
P
DSS
A
intermediate value, i.e. su < su
< su . From laboratory results reported in
A
DSS
A
P
literature one nd typically su /su in the range 0.2 - 0.5 and su /su in the
DSS
can be
range 0.3 - 0.8. If direct simple shear strengths are not available su
A
P
A
DSS
estimated from: su /su = (1 + su /su )/2.

A
Initial mobilization (0 /su )
A
The initial mobilization 0 /su is clearly dened for nearly horizontally deposited normally
consolidated or lightly overconsolidated clay layers where the vertical stress is the major
principle stress '1 . As dened in Figure 11.4, the initial mobilization can be calculated
from the earth pressure coefcient at rest K0 by the following equation:
A
A
0 /su = 0.5(1 K0 ) 'yy0 /su , where 'yy0 is the initial (in situ) vertical effective stress
A
(compression negative). A default value 0.7 of 0 /su is given which represent a typical
value for a near normally consolidated clay deposit (e.g. K0 = 0.55 and
A
A
'yy0 /su = 3.11, or K0 = 0.6 and 'yy0 /su = 3.5).

Figure 11.4 Denition of initial mobilized maximum shear stress 0 = 1/2 | 'yy0 'xx0 | for a soil
element in a horizontal deposited layer.

PLAXIS 2011 | Material Models Manual 119

MATERIAL MODELS MANUAL

Hint: A more detailed evaluation of the initial (in situ) mobilization can be done by
assessing the in situ K0 value and use the relationship:
A
A
0 /su = 0.5(1 K0 ) 'v 0 /su . Changing the default value for the initial
mobilization should be considered in particular for over-consolidated
materials where K0 generally is higher than 0.6, however the NGI-ADP model
is not intended used for heavily overconsolidated clays and should be used
A
with care for K0 > 1.0 (i.e. negative 0 /su ).

Hint: For non-horizontal layering (e.g. sloping ground) a K0 procedure is normally

not recommended. In such cases it is recommended to establish the initial
stress condition by gravity loading using a material model suited for such a
purpose (e.g. drained behavior with the Mohr-Coulomb model or the
Hardening Soil model). After the equilibrium initial stresses are established
for gravity loading in the rst phase, one should switch to the NGI-ADP
model in the relevant clusters for the next phase and run a NIL step (i.e.
without changing the external loads). The hardening parameter of the
NGI-ADP model will then be adjusted such that equilibrium is obtained (f=0).
Then in the third phase the external loading can be applied.

Poisson's ratio ( ')

Similar as in the Mohr-Coulomb model, Poisson's ratio is generally between 0.3 and 0.4
for loading conditions considering effective stress analysis. For unloading conditions a
lower value is more appropriate. See the linear elastic perfectly plastic model
(Mohr-Coulomb model) for more details.
Hint: When the Drainage type = Undrained (B) option is used an effective
Poisson's ratio should be entered that is less than 0.35. Excess pore
pressures are generated using the bulk modulus of water as described in
Section 2.4. In this case, Undrained (B), the model will calculate excess pore
pressures due to mean stress changes according to (2.32).
When the Undrained (C) drainage option is used a pure total stress analysis
is carried out where no distinction between effective stresses and pore
pressures is made and all stress changes should be considered as changes
in total stress. A Poisson's ratio close to 0.5 should be entered. = 0.495 is
given as default.

120 Material Models Manual | PLAXIS 2011

THE SEKIGUCHI-OHTA MODEL

12

THE SEKIGUCHI-OHTA MODEL

The Sekiguchi-Ohta model has been developed to formulate a constitutive law for
normally consolidated clay. Particular emphasis is placed on taking the effect of time and
stress-induced anisotropy into consideration. A complete description of the model has
been presented in Sekiguchi & Ohta (1977) and Iizuka & Ohta (1987).

12.1

FORMULATION OF THE SEKIGUCHI-OHTA MODEL

The Sekiguchi-Ohta model combines the concepts lying behind the well known Cam Clay
model (Roscoe, Schoeld & Thurairajah (1963)) and the rheological model developed by
Murayama & Shibata (1966). The Cam Clay model was further developed by Ohta &
Hata (1973) counting for the stress induced anisotropy for anisotropically consolidated
clays. However due to the fact that this model deals with the stress-strain behaviour of
the soil in equilibrium, the time effect is not considered. The rheological model is further
developed by Sekiguchi (1977) to describe the time-dependent and elastoplastic
behaviour for normally consolidated clays.
12.1.1

INVISCID FORMULATION

The yield function of the inviscid model in triaxial space is expressed by the following
equation:

f = MD ln(

q
p
)+D
pc
p

(12.1)

where M is the critical state frictional parameter, p is the mean stress, pc is the isotropic

hardening stress parameter, q is the relative deviatoric stress and D is the coefcient of
dilatancy.
The isotropic hardening stress parameter of the model is dened as:

(
pc = pc0 e

p
v

p0 )
v
(MD)

(12.2)

where MD = , v is the current plastic volumetric strain and v 0 is the initial plastic
volumetric strain. Hence, the parameter D is an auxiliary parameter implicitly dened as
D = ( )/M .
12.1.2

VISCID FORMULATION

The ow function of the visco-plastic (viscid) model is written as:

F = ln 1 +

f ()
v0 t

exp( )

vp
v

=0

(12.3)

where

f () = MD ln

p
q
+D
p0
p

(12.4)

PLAXIS 2011 | Material Models Manual 121

MATERIAL MODELS MANUAL

The ow function F can be transformed to a function of stress and hardening parameter

g as follows:

g( ', h) = f ( ') h(

vp
v , t)

(12.5)

=0

where the hardening parameter h is dened as:

h(

vp
v , t)

= ln

vp

exp( v ) 1

v0 t

(12.6)

The initial time to calculate the hardening parameter h should not be zero, because it is
not determined due to 1/t in Eq. (12.6). To be able to calculate the initial volumetric
visco-plastic strain, it is assumed that the hardening parameter h is equal to zero. Hence,
the initial visco-plastic volumetric strain can be calculated as follows:

h(
vp
v0

vp
v , t)

= ln

= ln

vp

exp( v ) 1

v0 t

v0 t

+1

(12.7)

=0

(12.8)

=0

vp
v0

is used as the initial visco-plastic volumetric strain to calculate the current

visco-plastic volumetric strain.
Hint: The Viscid Sekiguchi-Ohta model is available on request as a User-dened
soil model.

12.2

PARAMETERS OF THE SEKIGUCHI-OHTA MODEL

12.2.1

INVISCID MODEL

The inviscid Sekiguchi-Ohta model requires a total of seven parameters (Figure 12.1).
Basic parameters for soil stiffness:

: Modied compression index

[-]

: Modied swelling index

[-]

Instead of entering the basic parameters for soil stiffness, alternative parameters can be
entered. These parameters are listed below:

Cc

: Compression index

[-]

Cr

: Recompression index

[-]

einit

: Initial void ratio

[-]

The relationship of the alternative and the basic parameters for soil stiffness is given in
Table 7.1.

122 Material Models Manual | PLAXIS 2011

THE SEKIGUCHI-OHTA MODEL

Parameters for soil strength:

: Slope of the critical state line

[-]

Advanced parameters (use default settings):

ur

: Poissons ratio for unloading / reloading

[-]

nc
K0

: Coefcient of lateral stress in normal consolidation

[-]

Figure 12.1 shows the PLAXIS window for inputting the values of the model parameters.

Figure 12.1 Parameters for the Sekiguchi-Ohta model

Modied compression index and modied swelling index ( and )

These parameters can be obtained from an isotropic compression test including isotropic
unloading. When plotting the logarithm of the mean effective stress as a function of the
volumetric strain for clay type materials, the plot can be approximated by two straight
lines, see Figure 12.2. 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 .

*
*

pp

ln p'

Figure 12.2 Logarithmic relation between volumetric strain and mean stress

PLAXIS 2011 | Material Models Manual 123

MATERIAL MODELS MANUAL

Poisson's ratio (ur )

The poisson's ratio ur is a real elastic parameter and not 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.
NC
Earth pressure coefcient at rest (K0 )
nc
The K0 parameter is dened as the stress ratio in one-dimensional compression in a
state of normal consolidation:
NC
K0 =

'xx
'yy

Slope of the critical state line (M )

In order to obtain the correct shear strength, the parameter M should be based on the
friction angle . The critical state line is comparable with the Drucker-Prager failure
criteria, and represents a (circular) cone in the principal stress space. Hence, the value of
M can be obtained from in a similar way as the Drucker-Prager friction constant is
obtained from . For details see also the Modied Cam-Clay model in the Material
Models Manual.

Initial overconsolidation ratio (OCR0 )

The initial overconsolidation ratio OCR0 is dened as the highest vertical effective stress
experienced in the past, 'p , divided by the current stress 'yy . A soil which is currently
experiencing its highest stress is said to be normally consolidated and to have an OCR of
1.

Initial pre-overburden pressure (POP0 )

The initial pre-overburden pressure POP0 , expressed in the unit of stress, is dened as:

POP 0 = 'p 'yy

where 'p is the vertical pre-consolidation stress (the greatest vertical stress reached
previously) and 'yy is the in situ effective vertical stress.
12.2.2

MODEL PARAMETERS OF THE VISCID MODEL

Compared to the inviscid Sekiguchi-Ohta model, the viscid Sekiguchi-Ohta model

requires the coefcient of secondary compression and the initial volumetric strain rate
v0 as two additional parameters of input. All other parameters remain the same as in the

inviscid Sekiguchi-Ohta model model.

: Modied compression index

[-]

: Modied swelling index

[-]

ur

: Poissons ratio for unloading / reloading

[-]

nc
K0

: Coefcient of lateral stress in normal consolidation

[-]

: Slope of the critical state line

[-]

124 Material Models Manual | PLAXIS 2011

THE SEKIGUCHI-OHTA MODEL

OCR0

: Initial overconsolidation ratio

[-]

POP0

: Initial pre-overburden pressure

[kN/m2 ]

: Coefcient of secondary compression

[-]

v0

: Initial volumetric strain rate

[day-1 ]

Coefcient of secondary compression ( )

The coefcient of secondary compression is dened as:

d v
d(ln t)

at time tc (the end of primary consolidation).

Initial volumetric strain rate (v0 )

The initial volumetric strain rate v0 at reference state is expressed as:

v0 =

tc

Hint: The Viscid Sekiguchi-Ohta model is available on request as a User-dened

soil model.

PLAXIS 2011 | Material Models Manual 125

MATERIAL MODELS MANUAL

126 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

13

APPLICATION OF ADVANCED SOIL MODELS

In this chapter, advanced soil models will be utilised in various applications in order to
illustrate the particular features of these models. For applications on the standard
Mohr-Coulomb model, the reader is referred to the Tutorial Manual.

13.1

HARDENING SOIL MODEL: RESPONSE IN DRAINED AND UNDRAINED

TRIAXIAL TESTS

In this section, the Hardening Soil model is utilised for the simulations of drained and
undrained triaxial tests. Arbitrary sets of model parameters, Table 13.1, representing
sands of different properties, are considered.
A triaxial test can simply be modelled by means of an axisymmetric geometry of unit
dimensions (1m x 1m), that represent a quarter of the soil specimen, Figure 13.1. These
dimensions are not realistic, but they are selected for simplicity. The dimension of the
model does not inuence the results, provided that the soil weight is not taken into
account. In this conguration the stresses and strains are uniformly distributed over the
geometry. The deformation magnitudes in x - and y -direction of the top right hand corner
correspond to the horizontal and vertical strains respectively.
The left hand side and the bottom of the geometry are axes of symmetry. At these
boundaries the displacements normal to the boundary are xed and the tangential
displacements are kept free to allow for 'smooth' movements. The remaining boundaries
are fully free to move.
The value of the applied loads can be controlled by the load multipliers such as MloadA
and MloadB . However, in PLAXIS 2D, and as described in the Reference Manual, the
load congurations and magnitudes can be specied in the Input program. Then in the
calculation program these loads can be activated or deactivated by means of the Staged
construction option. For this case, and to simulate the conning pressure p', distributed
loads of 100 kN/m2 representing the principal stresses '1 (load A) and '3 (load B) are
Table 13.1 Arbitrary Hardening Soil parameters for sands of different densities

Parameter

Loose

Medium

Dense

Unit

ref
E50 (for pref = 100 kPa)
ref
Eur (for pref = 100 kPa)
ref
Eoed (for pref = 100 kPa)

20000

30000

40000

kN/m2

60000

90000

120000

kN/m2

20000

30000

40000

kN/m2

Cohesion c

0.0

0.0

0.0

kN/m2

Friction angle

30

35

40

Dilatancy angle

10

Poisson's ratio ur

0.2

0.2

0.2

Power m

0.5

0.5

0.5

(using Cap)

0.5

0.43

0.36

Tensile strength

0.0

0.0

0.0

kN/m2

Failure ratio

0.9

0.9

0.9

nc
K0

PLAXIS 2011 | Material Models Manual 127

MATERIAL MODELS MANUAL

applied in the Input program, as shown in Figure 13.1.

Figure 13.1 Simplied conguration of a triaxial test

A very course mesh is sufcient for this simple geometry. Initial stresses and steady pore
pressures are not taken into account.
400
dense

|1 3 | [kN/m2 ]

300

medium

loose

200

100

0
0

0.01

0.02

0.03

0.04

0.05

0.06

Figure 13.2 Results of drained triaxial tests using the Hardening Soil model, Principal stress
difference versus axial strain

In the Calculation program, the calculation of all phases can be done by means of the
Staged construction process. In the rst phase, the connement pressure p' is applied by
activating load A and B. In the second phase the displacements are reset to zero and the
sample is vertically loaded up to failure while the horizontal load is kept constant. This
implies modication of load A by double clicking the load in the geometry model. As a
result a load window appears in which the input values of the load can be changed.
(Details of the procedure can be found in the Reference and Tutorial manuals.) The latter
phase is carried out for drained as well as undrained conditions.
These calculations are performed for the three different sets of material parameters,

128 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

0.015

dense

0.010
0.005

medium

0.000
loose

-0.005
-0.010
0

0.01

0.02

0.03

0.04

0.05

0.06

Figure 13.3 Results of drained triaxial tests using the Hardening Soil model, Volumetric strain versus
axial strain
200
dense

|1 3 | [kN/m2 ]

160
medium
120
loose
80

40

0.005

0.010

0.015

0.020

0.025

0.030

Figure 13.4 Results of undrained triaxial tests using the Hardening Soil model, Principal stress
difference versus axial strain

Table 13.1. The computational results are presented in the gures on the following pages.
Figure 13.2 shows the principal stress difference versus the axial strain for the drained
condition. This shows a hyperbolic relationship between the stress and the strain, which
is typical for the Hardening Soil model. Obviously, the failure level is higher when the
sand is denser. The Hardening Soil model does not include softening behaviour, so after
reaching failure the stress level does not reduce, at least in the drained tests.
Figure 13.3 shows the axial strain versus the volumetric strain for the drained test. This
graph clearly shows the inuence of dilatancy in the denser sands. In contrast to the
Mohr-Coulomb model, the transition from elastic behaviour to failure is much more
gradual when using the Hardening Soil model. In fact, in the Hardening Soil model,
plastic strain occurs immediately after load application.
In the undrained tests, Figure 13.4, the failure level is, in principle, lower than that of the
drained tests. However, for the medium and dense sands the stress level continues to
increase after reaching the failure level due to the fact that dilatancy occurs which causes
reduction of excess pore pressures and thus increase of the effective stresses. This can

PLAXIS 2011 | Material Models Manual 129

MATERIAL MODELS MANUAL

-60
loose

-50

medium

pexcess [kPa]

-40

-30
dense
-20

-10

0.005

0.010

0.015

0.020

0.025

0.030

Figure 13.5 Results of undrained triaxial tests using the Hardening Soil model, Excess pore
pressure vs axial strain
-400

-300

yy [kPa]

undrained

drained

-200

-100

isotropic loading
0
0

-100

-200

xx [kPa]

Figure 13.6 Stress paths for drained and undrained triaxial tests using the Hardening Soil model

be seen in Figure 13.5.

Figure 13.6 shows the effective stress paths, for the medium sand, during both the
drained and undrained tests. During rst phase (isotropic loading), both tests were
drained. In the second phase there is a clear distinction between the two tests. In the
undrained test the effective horizontal stress reduces while the vertical stress increases
due to the development of excess pore pressures. The decrease in horizontal effective
stress is more than when if the Mohr-Coulomb model would have been used. This is
attributed to the plastic compaction (Cap hardening) that occurs in the Hardening Soil

130 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

model.

13.2

APPLICATION OF THE HARDENING SOIL MODEL ON REAL SOIL TESTS

In this section the ability of the Hardening Soil model to simulate laboratory tests on sand
is examined by comparing PLAXIS calculation results with those obtained from laboratory
tests. Extensive laboratory tests were conducted on loose and dense Hostun sand. On
the basis of these tests the model parameters for the Hardening Soil model were
determined, Table 13.2.
Table 13.2 Hardening Soil parameters for loose and dense Hostun sand

Parameter

Loose sand

Dense sand

Unit

Volumetric weight

17

17.5

kN/m3

ref
E50 (pref = 100 kPa)
ref
Eur (pref = 100 kPa)
ref
Eoed (pref = 100 kPa)

20000

37000

kN/m2

60000

90000

kN/m2

16000

29600

kN/m2

Cohesion c

0.0

0.0

kN/m2

Friction angle

34

41

Dilatancy angle

14

Poisson's ratio ur

0.20

0.20

Power m

0.65

0.50

nc
K0

0.44

0.34

Tensile strength

0.0

0.0

kN/m2

Failure ratio

0.9

0.9

Triaxial test
Standard drained triaxial tests were performed on loose and dense sand specimens. In
PLAXIS the procedure for the simulation of the triaxial tests has been described in
Section 13.1. In the rst phase the sample is isotropically compressed up to a conning
pressure of p' = 300 kN/m2 . In the second phase the sample is vertically loaded up to
failure while the horizontal stress (conning pressure) is kept constant. The
computational results and the measured data are presented in Figure 13.7, Figure 13.8,
Figure 13.9 and Figure 13.10.
The gures show that the computational results match reasonably with the test data. It
can be seen that the material response (measured and computed) show gradual
transition from elastic to plastic behaviour. As such the relation between the deviatoric
stress and the axial strain can be approximated by a hyperbola.
The failure level is fully controlled by the friction angle (the cohesion is zero). The test
results on dense sand show softening behaviour after the peak load has been reached.
Modelling of the softening behaviour, however, is not incorporated in the Hardening Soil
model, and thus, the deviatoric stress remains constant. It can also be seen from the test
data that the dilatancy reduces during softening. However, in the Hardening Soil model
the dilatancy continues to innity, unless the dilatancy cut-off option has been used.

PLAXIS 2011 | Material Models Manual 131

MATERIAL MODELS MANUAL

Figure 13.7 Results of drained triaxial tests on loose Hostun sand, principal stress ratio versus axial
strain

Figure 13.8 Results of drained triaxial tests on loose Hostun sand, volumetric strain versus axial
strain
|1-3| [kPa]
1400
1200
1000
800

Hardening soil model

600

test data
400
200
0
0

10

15

-1 [%]

Figure 13.9 Results of drained triaxial tests on dense Hostun sand, principal stress ratio versus axial
strain

Oedometer test
As for the triaxial test, a set of oedometer test on both loose and dense sands, Table
13.2, was conducted. In PLAXIS the oedometer test is simulated as an axisymmetric
geometry with unit dimensions, Figure 13.11. A coarse mesh is sufcient for this case.
The computational results as compared with those obtained from the laboratory tests are

132 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

v [%]
8
7
6
5
4
3
Hardening soil model
2

test data

1
0
-1
-2
0

10

15

-1 [%]

Figure 13.10 Results of drained triaxial tests on dense Hostun sand, volumetric strain versus axial
strain

Figure 13.11 Simplied conguration of an oedometer test

shown in Figure 13.12 and Figure 13.13.

From a stress free state the loose sand sample is loaded consecutively to 25 kPa, 50
kPa, 100 kPa and 200 kPa with intermediate unloading. The dense sand sample is
loaded to 50 kPa, 100 kPa, 200 kPa and 400 kPa with intermediate unloading.
As it can be seen, the computational results show a reasonable agreement with the test
data. No doubt, distinction should be made between loose and dense soil, but it seems
that for a soil with a certain density the stiffness behaviour under different stress paths
can be well captured with a single set of model parameters. (A small offset of 0.15% has
been applied to the computational results of the loose sample in order to account for the
relative soft response at the beginning of the test.)

PLAXIS 2011 | Material Models Manual 133

MATERIAL MODELS MANUAL

-yy [kPa]
400

Hardening soil model

300
test data

200

100

0
0

0.005

0.01

0.015

0.02

0.025

-yy

Figure 13.12 Results of oedometer test on loose Hostun sand, axial stress versus axial strain
-yy [kPa]
400

Hardening soil model

300

test data

200

100

0
0

0.005

0.01

0.015

0.02

0.025

-yy

Figure 13.13 Results of oedometer test on dense Hostun sand, axial stress versus axial strain

Pressiometer test
In this section the Pressiometer test is simulated and results from PLAXIS and laboratory
experimentation are compared. Laboratory testing results on dense sand with material
parameters listed in Table 13.2 are used.
In the eld, the pressiometer with 44 mm in diameter covered with a membrane with 160
mm in height is attached to the Cone penetration shaft. In the laboratory, the
pressiometer is attached to a 44 mm pipe and placed in a circular calibration chamber
with a diameter of 1.2 m and a height of 0.75 m. A large overburden pressure of 500 kPa
is applied at the surface to simulate the stress state at larger depths. In PLAXIS only half

134 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

of the geometry is simulated by an axisymmetric model, Figure 13.14. The overburden

pressure is simulated by load A, and the expansion of the pressiometer is simulated by
imposing a horizontal distributed load, load B. Therefore the initial standard boundary
conditions have to be changed near the pressiometer in order to allow for free horizontal
displacements.

Figure 13.14 Geometry model for pressiometer test

To allow for a discontinuity in horizontal displacements, a vertical interface along the

shaft of the pressiometer borehole and a horizontal interface just above the pressiometer
are introduced. Both interfaces are set rigid (Rinter = 1.0). Extra geometry lines are
created around the pressiometer to locally generate a ner mesh.
After the generation of initial stresses, the vertical overburden load (load A) is applied
using the standard boundary xities. From the calculations, the lateral stress around the
pressiometer appears to be 180 kPa. Subsequently, the horizontal xity near the
pressiometer is removed, in the Input program, and replaced by Load B with a magnitude
of 180 kPa. In the next calculation the pressure (load B) is further increased by use of
Staged construction in an Updated mesh analysis. The results of this calculation are
presented in Figure 13.15 and Figure 13.16.
Figure 13.15 shows details of the deformations and the stress distribution when the
pressure in the pressiometer was 2350 kPa. The high passive stresses appear very
locally near the pressiometer. Just above the pressiometer the vertical stress is very low
due to arching effects. Away from the pressiometer, a normal K0 -like stress state exists.
Figure 13.16 shows a comparison of the numerical results with those obtained from the
laboratory test. In the gure the pressiometer pressure is presented as a function of the
relative volume change. The latter quantity cannot directly be obtained from PLAXIS and
was calculated from the original radius R0 and the lateral expansion ux of the
pressiometer:
2
V
(R0 + ux )2 R0
=
2
V0
R0

PLAXIS 2011 | Material Models Manual 135

MATERIAL MODELS MANUAL

Figure 13.15 Stress distribution in deformed geometry around the pressiometer at a pressure of
2350 kPa

Up to a pressure of 1600 kPa the results match quite well. Above 1600 kPa there is a
sudden decrease in stiffness in the real test data, which cannot be explained.
Nevertheless, the original set of parameters for the dense sand that were derived from
triaxial testing also seem to match the pressiometer data quite well.

Figure 13.16 Comparison of numerical results and pressiometer test data

Conclusion
The above results indicate that by use of the Hardening Soil model it is possible to
simulate different laboratory tests with different stress paths. This cannot be obtained
with simple models such as Mohr-Coulomb without changing input parameters. Hence,
the parameters in the Hardening Soil model are consistent and more or less independent
from the particular stress path. This makes the Hardening Soil model a powerful and an
accurate model, which can be used in many applications.

136 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

13.3

APPLICATION OF THE HS SMALL MODEL ON REAL SOIL TESTS

In this section, the ability of the HS small model to simulate laboratory tests is examined.
Both, the laboratory test data and the basic HS parameters are identical to those
presented in the previous section. The two additional small strain parameters used in the
Hardening Soil model are quantied in Table 13.3.
Table 13.3 Additional HS small model parameters for loose and dense Hostun sand

Parameter
ref
G0

(pref = 100 kPa)

Shear strain 0.7

Loose sand

Dense sand

Unit

70000

112500

kN/m2

0.0001

0.0002

Triaxial tests on loose and dense Hostun sand are presented in Figure 13.17 and Figure
13.18 respectively. As a reference, the previously obtained results from the Hardening
Soil model are plotted as well.

Figure 13.17 Excavation Drained triaxial tests on loose Hostun sand at conning pressures of 100,
300, and 600 kPa. Left: Stress-strain data. Right: Shear modulus reduction.

PLAXIS 2011 | Material Models Manual 137

MATERIAL MODELS MANUAL

The overall stress-strain data obtained from both models seems almost identical. Only a
closer look at the small-strain domain shows a clear difference: The HS small model
follows a S-shaped stiffness reduction curve with much higher initial stiffness than the
one of the Hardening Soil model. Generally, both models match the test data at different
conning pressures reasonably well.

Figure 13.18 Drained triaxial tests on dense Hostun sand at conning pressures of 100, 300, and
600 kPa. Left: Stress-strain data. Right:Shear modulus reduction.

Figure 13.19 presents results from a cyclic triaxial test by Rivera & Bard on dense sand.
The HS small model simulation of the test shows material damping which could not be
obtained when simulating the test with the Hardening Soil model. As virgin loading is
conducted in triaxial compression, the unloading sequence in triaxial extension gives
some plasticity. Therefore the rst unloading / reloading loop is not closed.

138 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

Figure 13.19 Hysteresis loop in a drained triaxial test on dense Hostun sand. Test data published
in Biarez & Hicher (1994).

13.4

SOFT SOIL CREEP MODEL : UNDRAINED TRIAXIAL TESTS AT DIFFERENT

LOADING RATES

In this section the Soft Soil Creep model (see Chapter 8) is utilised for the simulation of
clay in an undrained triaxial test at different strain rates. The model parameters are
obtained from test results on Haney Clay and are listed in Table 13.4.
Table 13.4 Soft Soil Creep model parameters for Haney clay

Parameter

Symbol

Value

Unit

Modied compression index

0.105

Modied swelling index

0.016

Secondary compression index

0.004

Poisson's ratio

ur

0.15

Cohesion

0.0

kN/m2

Friction angle

32

Dilatancy angle

0.0

Coefcient of lateral stress

nc
K0

0.61

Permeability

kx , ky

0.0001

m/day

Modelling of the triaxial test is as described in Section 13.1. However, here, the real
dimension of the test set-up is simulated (17.5 x 17.5 mm2 ), Figure 13.20. The specimen
surfaces (top and right hand side in Figure 13.20) are assumed drained whereas the
other boundaries are assumed closed.
In addition to isotropic loading, prescribed displacements are also applied. Both types of
loading are simulated using the Staged construction option. During isotropic loading,
horizontal and vertical loads (both System A) are applied. The calculation phases for
isotropic loading consist of undrained plastic and consolidation analyses.

PLAXIS 2011 | Material Models Manual 139

MATERIAL MODELS MANUAL

Figure 13.20 Modelling of triaxial test on Haney clay. Left, Initial conguration. Right, conguration
for phase 9 - 11

After the isotropic loading phases, the displacements are reset to zero. The vertical load
is deactivated and the prescribed displacement is activated. Rate of loading is simulated
by applying prescribed displacements at different velocities. As such a total of 12% axial
strain (2.1 mm) is applied in 8.865 days (0.00094%/minute), 0.0556 days (0.15%/minute)
and 0.00758 days (1.10%/minute) respectively. Each of the prescribed displacement
loading phase starts from the end of the isotropic loading phase. The calculation scheme
is listed in Table 13.5.
Table 13.5 Loading scheme for triaxial tests at different loading rates

Phase Start from:

Calculation

Load

Displacement

Time interval

[kPa]

[mm]

[day]

Plastic

65

Inactive

0.00

Consolidation

65

Inactive

0.01

Plastic

130

Inactive

0.00

Consolidation

130

Inactive

0.01

Plastic

260

Inactive

0.00

Consolidation

260

Inactive

0.01

Plastic

520

Inactive

0.00

Consolidation

520

Inactive

0.01

Plastic

520

0.0021

8.865

10

Plastic

520

0.0021

0.0556

11

Plastic

520

0.0021

0.00758

The computational results are presented in Figure 13.21 and Figure 13.22. Figure 13.21
shows the stress-strain curves of the prescribed displacement loading phases. It can be
seen that the shear strength highly depends on the strain rate; the higher strain rate the
higher the shear strength.
Figure 13.22 shows the p - q stress paths from the prescribed displacement loading
phases. For higher strain rates there is a smaller reduction of the mean effective stress,
which allows for a larger ultimate deviatoric stress. It should be noted that the stress state
is not homogeneous at all, because of the inhomogeneous (excess) pore pressure

140 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

Figure 13.21 Average deviatoric stress versus axial strain for different rates of straining

distribution. This is due to the fact that points close to draining boundaries consolidate
faster than points at a larger distance.
400

q [kN/m2 ]

300

A
200

A = 0.00094%/m
100

B = 0.15%/m
C = 1.10%/m

-100

-200

-300

-400

p [kN/m2 ]

Figure 13.22 p - q stress paths for different rates of straining for a point at position (0.01, 0.01)

13.5

SOFT SOIL CREEP MODEL: RESPONSE IN ONE-DIMENSIONAL

COMPRESSION TEST

In this section the behaviour of the Soft Soil Creep model is illustrated on the basis of a
one-dimensional compression test on clay. Two types of analysis are performed. First,
the test is simulated assuming drained conditions in order to demonstrate the logarithmic
stress-strain relationship and the logarithmic time-settlement behaviour on the long term
(secondary compression). Second, the test is simulated more realistically by including
undrained conditions and consolidation. Since the consolidation process depends on the
drainage length, it is important to use actual dimensions of the test set-up. In this case an
axisymmetric conguration with specimen height of 0.01 m, Figure 13.23, is used. The
material parameters are shown in Table 13.6. The parameter values are selected
arbitrarily, but they are realistic for normally consolidated clay. The vertical
preconsolidation stress is xed at 50 kPa (POP = 50 kPa).

PLAXIS 2011 | Material Models Manual 141

MATERIAL MODELS MANUAL

Figure 13.23 One-dimensional compression test

Table 13.6 Soft Soil Creep model parameters for one-dimensional compression test

Parameter

Symbol

Value

Unit

Unit weight

19

kN/m3

Permeability

kx , ky

Modied compression index

0.0001

m/day

0.10

Modied swelling index

0.02

Secondary compression index

0.005

Poisson's ratio

ur

0.15

Cohesion

1.0

kN/m2

Friction angle

30

Dilatancy angle

0.0

Coefcient of lateral stress

nc
K0

0.5

Drained analysis
In the rst analysis successive plastic loading steps are applied using drained conditions.
The load is doubled in every step using Staged construction with time increments of 1
day. After the last loading step an additional creep period of 100 days is applied. The
calculation scheme is listed in Table 13.7. All calculations are performed with a tolerance
of 1%.

Undrained analysis
In the second analysis the loading steps are instantaneously applied using undrained
conditions. After each loading step a consolidation of 1 day is applied to let the excess
pore pressures fully dissipate. After the last loading step, an additional creep period of
100 days is again introduced. The calculation scheme for this analysis is listed in Table
13.8. All calculations are performed with a reduced tolerance of 1%.
Figure 13.24 shows the load-settlement curves of both analyses. It can be seen that,
after consolidation, the results of the undrained test match those of the drained test. The
inuence of the preconsolidation stress can clearly be seen, although the transition
between reloading and primary loading is not as sharp as when using the Soft Soil

142 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

Table 13.7 Calculation scheme for the drained case

Phase Calculation type

Loading input

Load

Time interval

End time

[kPa]

[day]

[day]

Plastic

Staged
construction

10

Plastic

Staged
construction

20

Plastic

Staged
construction

40

Plastic

Staged
construction

80

Plastic

Staged
construction

160

Plastic

Staged
construction

320

Plastic

Staged
construction

640

Plastic

Staged
construction

640

100

107

model. In fact, the results presented here are more realistic. The transition is indeed
around 50 kPa.
From the slope of the primary loading line one can back-calculate the modied
compression index = 1 / ln((1 + 1 )/1 0.10. Note that 1 mm settlement
corresponds to 1 = 10%. For an axial strain of 30% one would normally use an Updated
mesh analysis, which has not been done in this simple analysis. If, however, the Soft Soil
Creep model would have been used in an Updated mesh analysis with axial strains over
15% one would observe a stiffening effect as indicated by line C in Figure 13.24.
Figure 13.25 shows the time-settlement curves of the drained and the undrained
analyses. From the last part of the curve one can back-calculate the secondary
compression index = 1 / ln(t/t0 ) 0.005 (with t0 = 1 day).

Another interesting phenomenon is the development of lateral stresses. During primary

nc
loading, the lateral stress is determined by K0 , appropriate for normally consolidated
soil. During unloading, the lateral stress decreases much less than the vertical stress, so
that the ratio 'xx / 'yy increases.
To show these effects the calculation is continued after with a new drained unloading
phase that starts from phase 7 (see Table 13.7) in which the vertical stress is reduced to
80 kPa.

Figure 13.26 shows the stress state for two different calculation phases, both at a vertical
stress level of 80 kPa. The plot in the left hand side shows the stress state after primary
loading. As expected the horizontal stress is found to be approximately 40 kPa
nc
(corresponding to K0 = 0.5). The plot in the right hand side shows the nal situation
after unloading down to 80 kPa. In this case the horizontal stress is decreased from
320 kPa to approximately 220 kPa, ( 'xx = 100 kPa), i.e, much less than the
PLAXIS 2011 | Material Models Manual 143

MATERIAL MODELS MANUAL

Table 13.8 Calculation scheme for second analysis

Phase Calculation type

Loading input

Load

Time interval

End time

[kPa]

[day]

[day]

Plastic

Staged
construction

10

Consolidation

Staged
construction

10

Plastic

Staged
construction

20

Consolidation

Staged
construction

20

Plastic

Staged
construction

40

Consolidation

Staged
construction

40

Plastic

Staged
construction

80

Consolidation

Staged
construction

80

Plastic

Staged
construction

160

10

Consolidation

Staged
construction

160

11

Plastic

Staged
construction

320

12

Consolidation

Staged
construction

320

13

Plastic

Staged
construction

640

14

Consolidation

Staged
construction

640

15

Consolidation

Staged
construction

640

100

107

144 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

settlement [m]

0.0

-0.001

-0.002

-0.003

10

1 [kPa]

100

1000

Figure 13.24 Load-settlement curve of oedometer test with Soft Soil Creep model. A) Transient
loading with doubling of loading within one day. B) Instantaneous loading with
doubling of load at the beginning of a new day. C) As 'A' using Updated Mesh
calculation

Figure 13.25 Time-settlement curve of oedometer test with Soft Soil Creep model. A) Transient
loading with doubling of loading within one day. B) Instantaneous loading with
doubling of load at the beginning of a new day

decrease of the vertical stress ( 'yy = 560 kPa). Thus, a situation where 'xx is larger
than 'yy is obtained.
During sudden unloading in a one-dimensional compression test, the behaviour is purely
elastic. Hence, the ratio of the horizontal and vertical stress increments can be
determined as:

'xx
ur
=
'yy 1 ur

(13.1)

It is easy to verify that the results correspond to Poisson's ratio ur = 0.15 as listed in

PLAXIS 2011 | Material Models Manual 145

MATERIAL MODELS MANUAL

Figure 13.26 Stress states at a vertical stress level of 80 kPa. Left, after primary loading 'xx 40
kPa. Right, after unloading from 640 kPa 'xx 220 kPa

Table 13.3.

146 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

13.6

SOFT SOIL MODEL : RESPONSE IN ISOTROPIC COMPRESSION TEST

In this section it will be demonstrated that the Soft Soil model obeys a logarithmic
relationship between the volumetric strain and the mean stress in isotropic compression.
For this purpose the test set up is simulated as that presented in Figure 13.1. The vertical
load (A) and the horizontal load (B) are simultaneously applied to the same level so that a
fully isotropic stress state occurs. The parameters of the Soft Soil model are chosen
arbitrarily, but the values are realistic for normally consolidated clay. The parameters are
presented in Table 13.9.
From a stress-free state, the model is isotropically loaded to a mean stress of p' = 100
kPa, after which the displacements are reset to zero. As a result, the material becomes
'normally consolidated', i.e., the preconsolidation stress is equivalent to the current
state-of-stress. After that, the isotropic pressure is increased to p' = 1000 kPa. This
loading path is denoted as 'primary loading'. Then, the sample is isotropically 'unloaded'
to p' = 100 kPa. Finally, the sample is loaded up to p' = 10000 kPa. In the last loading
path, the maximum preload of 1000 kPa is exceeded. and hence, it consists of two parts:
the part of the loading path for which p' < 1000 kPa is referred to as 'reloading', and the
part of the loading path for p' > 1000 kPa consists of further primary loading. The
calculation phases are indicated in Table 13.10.
The computational results are presented in Figure 13.27, which shows the relation
between the vertical strain yy and the vertical stress 'yy .
The latter quantity is plotted on a logarithmic scale. The plot shows two straight lines,
which indicates that there is indeed a logarithmic relation for loading and unloading. The
vertical strain is 1/3 of the volumetric strain, v , and the vertical stress is equal to the
mean stress, p'. The volumetric strains obtained from the calculation are given in Table
13.11.
From these strains and corresponding stresses, the parameters and can be
back-calculated using Eqs. (8.1) and (8.2).
Phase 1

1 0
0.235
v
v
=
= 0.102
1 p0
ln(1000/100)
ln p /

Phase 2

2 1
0.188 0.235
v
v
=
= 0.020
ln(100/1000)
ln p2 / p1

Phase 3

3 2
0.235 0.188
v
v
=
= 0.020
ln(1000/100)
ln p3 / p2

Table 13.9 Soft Soil model parameters for isotropic compression test

0.10

Modied swelling index

0.02

Poisson's ratio

ur

0.15

Friction angle

30

Cohesion

1.0 kPa

Normally consolidated K0

nc
K0

0.5

Modied compression index

PLAXIS 2011 | Material Models Manual 147

MATERIAL MODELS MANUAL

Table 13.10 Calculation phases for isotropic compression test on clay

Stage

Initial stress

Final stress

Initial situation

p0 = 100 kPa

Primary loading

p0 = 100 kPa

p1 = 1000 kPa

Unloading

p1 = 1000 kPa

p2 = 100 kPa

Reloading

p2 = 100 kPa

p3 = 1000 kPa

Primary loading

p3 = 1000 kPa

p4 = 10000 kPa

Table 13.11 Volumetric strains from various calculation phases

Phase

Initial strain

Final strain

0 = 0.000
v

0 = 0.000
v

1 = 0.235
v

1 = 0.235
v
2 = 0.188
v

2 = 0.188
v

3 = 0.235
v

3 = 0.235
v

4 = 0.471
v

Figure 13.27 Results of isotropic compression test

Phase 4

4 3
v
v

ln p4 /p3

0.471 0.235
= 0.102
ln(10000/1000)

The back-calculated values correspond to the input values as given in Table 13.9.
Note that the Soft Soil model does not include time effects such as in the secondary
compression. Such behaviour is included in the Soft Soil Creep model.

13.7

HARDENING SOIL MODEL AND HS SMALL MODEL: EXCAVATION IN BERLIN

SAND

In the previous example, the advantage of the Hardening Soil model's distinct loading
and unloading stiffness was highlighted. With those, the calculated excavation heave
could be reduced to a more realistic, but in most cases still too high value. In the Berlin

148 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

excavation example, now the further advantage of considering small-strain stiffness in the
analysis is demonstrated.
The working group 1.6 Numerical methods in Geotechnics of the German Geotechnical
Society (DGGT) has organized several comparative nite element studies (benchmarks).
One of these benchmark examples is the installation of a triple anchored deep excavation
wall in Berlin sand. The reference solution by Schweiger (2002) is used here as the
starting point for the next validation example: Both, the mesh shown in Figure 13.28, and
the soil parameters given in Table 13.12 are taken from this reference solution. However,
the bottom soil layer 3 dened by Schweiger (2002) is assigned the parameters of layer 2
in the HSsmall analysis. In the reference solution this layer's only purpose is the
simulation of small-strain stiffness due to a lack of small-strain stiffness constitutive
models back then.
0.00

30.00

Excavation step 1

Sand (layer 1)

-4.80

Excavation step 2

40

-14.35

Excavation step 4

Sand (layer 2)

-9.30

Excavation step 3

20

-16.80

-3.00(GWT)
27
27

8.0
0

27

8.0
0

A1

8.0
0

A2

-17.90(GWT)
hydraulic barrier

60

Anchor Prestress

80

A3
-30.00

-32.00

80 diaphragm wall
Distance
Length l

A1

2.30 m

15 cm2 19.80 m

945 kN

1.35 m

15 cm2 23.30 m

A3

100

768 kN

A2

Sand (layer 3)

980 kN

1.35 m

15 cm2 23.80 m

Excavation and anchor detail (1:1000)

-20

20

40

60

80

100

120

Figure 13.28 Excavation in Berlin sand: plane strain mesh (left) and geometry detail (right).

Figure 13.29 shows results from the nite element calculation using the original
Hardening Soil model and the HS small model. The small-strain stiffness formulation in
the HS small model accumulates more settlements right next to the wall, whereas the
settlement trough is smaller. The triple anchored retaining wall is deected less when
using the HSsmall model, almost tting the measured deection. Calculated excavation
heave at the end of excavation is shown in Figure 13.30. Compared to the HS results, the
heave which is due to elastic unloading, is roughly halved when using the HS small
model.

PLAXIS 2011 | Material Models Manual 149

MATERIAL MODELS MANUAL

Table 13.12 Hardening Soil modeland HS small modelparameters for the three sand layers in the
excavation project

Parameter

Layer 1

Layer 2

Layer 3

Unit

Unit weight above/below

phreatic level

19 / 20

19 / 20

19 / 20

kN/m3

ref
E50 (pref = 100 kPa)

45000

75000

105000

kN/m2

ref
Eur (pref = 100 kPa)
ref
Eoed (pref = 100 kPa)

45000

75000

105000

kN/m2

180000

300000

315000

kN/m2

ref
G0 (pref = 100 kPa)
Shear strain 0.7

168750

281250

NA

kN/m2

0.0002

0.0002

NA

Cohesion c

1.0

1.0

1.0

kN/m2

Friction angle

35

38

38

Dilatancy angle

5.0

6.0

6.0

Poisson's ratio ur

0.2

0.2

0.2

Power m

0.55

0.55

0.55

nc
K0

0.43

0.38

0.38

Tensile strength

0.0

0.0

0.0

kN/m2

Failure ratio

0.9

0.9

0.9

Figure 13.29 Hardening Soil model and HS small model predictions versus measured
displacements after the nal excavation step. Left: Surface settlement trough. Right:
Lateral wall deection.

150 Material Models Manual | PLAXIS 2011

APPLICATION OF ADVANCED SOIL MODELS

0.01

0.02

0.03

0.04
Vertical heave [m]

-20
Section A-A

20.00 m

-40

-60

-80
HS (reference)
HSsmall
-100
Depth below surface [m]

Figure 13.30 Vertical displacements in the excavation pit at a distance of 10 m from the retaining
wall (Section A-A).

PLAXIS 2011 | Material Models Manual 151

MATERIAL MODELS MANUAL

152 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

14

USER-DEFINED SOIL MODELS

14.1

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
ij , t+t

current stresses and state variables

t
ij ,

previous stresses and state variables

ij , 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 installation
package. 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.

14.2

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 user-dened soil model can be dened. If a
UD soil model is used in an application, the calculation program calls the corresponding
task from the subroutine User_Mod. To create a UD soil model, the User_Mod subroutine
must have the following structure:

PLAXIS 2011 | Material Models Manual 153

MATERIAL MODELS MANUAL

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, iStrsDep, 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, iTang);
6 = Create elastic material stiffness matrix)

iMod

User-dened soil model number (This option allows for more

than one UD model, up to 10.)

IsUndr

Drained condition (IsUndr = 0) or undrained condition (IsUndr =

1). In the latter case, PLAXIS will add a large bulk stiffness for
water.

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..20) with previous (= at the start of the current step)

effective stress components of the current stress point ( '0 , '0 ,
xx
yy
'0 , '0 , '0 , '0 , psteady , Mstage0 , Mstage, Sat , Sat 0 ,
zz
xy
yz
zx
Suc , Suc 0 , Msf 0 , Msf , 0, 0, 0, 0, 0). In 2D calculations yz
and zx should be zero.

Swp0

Previous excess pore pressure of the current stress point

StVar0

Array(1..nStat) with previous values of state variables of the

current stress point

154 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

dEps

Array(1..12) with strain increments of the current stress point in

the current step ( xx , yy , zz , xy , yz , zx , 0 , 0 ,
xx
yy
0
0
0
0
0
0
, xy , yz , zx ). In 2D calculations yz , zx , yz and zx
zz
should be zero.

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

current stress point

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 nonsymmetric (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)

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

PLAXIS 2011 | Material Models Manual 155

MATERIAL MODELS MANUAL

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.

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.

156 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

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
K0 -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 K0 -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]
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}
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 !! }

PLAXIS 2011 | Material Models Manual 157

MATERIAL MODELS MANUAL

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 D such 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]).

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

158 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

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).

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 force|
|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

PLAXIS 2011 | Material Models Manual 159

MATERIAL MODELS MANUAL

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] =
.....
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 by User_Mod
subroutine to shorten the code. An overview of the available subroutines is given in
Appendix C.

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 model 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 model number (input parameter)

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);

160 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

iModel

User-dened soil modelnumber (input parameter)

iParam

Parameter number (input parameter)

Name

parameter name (return parameter)

This procedure retrieves the parameter name of a specic parameter.

Procedure GetParamUnit(var iModel,iParam : longint;
var Units : shortstring) ;
iModel

User-dened soil model number (input parameter)

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#'.
When dening a unit containing one of the letters 'l', 'f' or 't', like 'cal/mol', these letters will
be replaced by the unit of length, the unit of force or the unit of time respectively. To avoid
this, these letters should be preceded by a backslash. For example 'cal/mol' should be
dened as 'ca\l/mo\l' to avoid getting 'cam/mom'.
The state variables to be displayed in the Output program can be dened.
procedure GetStateVarCount(var iModel : longint; var C :longint) ;
iModel

User-dened soil model number (input parameter)

number of state variables for the specied model (return

parameter)

This procedure retrieves the number of state variables of a specic model.

procedure GetStateVarName(var iModel,iParam : longint;
var Name : shortstring);

PLAXIS 2011 | Material Models Manual 161

MATERIAL MODELS MANUAL

iModel

Used-dened soil model number (input parameter)

iParam

Parameter number (input parameter)

Name

parameter name (return parameter)

This procedure retrieves the state parameter name of a specic parameter.

Procedure GetStateVArUnit(var iModel,iParam : longint;
var Units : shortstring) ;
iModel

User-dened soil model number (input parameter)

iParam

Parameter number (input parameter)

Units

Parameter units (return parameter)

This procedure retrieves the state parameter units of a specic parameter.

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 in the software package.

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 (LF95, ...): DLL_Export User_Mod

Using Intel Visual Fortran: !DEC$ ATTRIBUTES DLLExport,StdCall,Reference ::

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 -ml bd

Using Lahey Fortran 95: LF95 -win -dll USRMOD.FOR -lib LFUsrLib -ml bd

Using Intel Visual Fortran: ifort /winapp USRMOD.FOR DFUsrLib.lib /dll

Using GCC complier: g55 USRMOD.FOR -o usermod.dll -shared -fcase -upper

-fno-underscoring -mrtd

162 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

In all cases USRMOD.DLL le will be created. It can be renamed to 'any' .dll. This le
should be placed in the PLAXIS program directory, thereafter it can be used together with
the existing PLAXIS calculations program (PLASW.EXE in PLAXIS 2D or
PLASW3DF.EXE in PLAXIS 3D). Once the UD model is used, PLAXIS will execute the
commands as listed in the USRMOD.DLL le.
In order to compile as 64-bit, you need both a 32-bit compiled 'USRMOD.DLL' and a
64-bit compiled ' USRMOD64.DLL' in the program installation directory. The last one only
needs to contain the 'User_Mod' subroutine.

Debugging possibilities
When making computer programs, usually some time is spent to 'debug' 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 D 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 (Section C).

14.3

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.
When creating a new material data set for soil and interfaces in the material data base, a
window appears with three tabsheets: General, Parameters, Interface, Figure 14.1. A
user-dened model can be selected from the Material model combo box in the General
tabsheet.
After inputting general properties, the appropriate UD model can be chosen from the
available models that have been found by PLAXIS Input.
The Parameters tabsheet shows two combo boxes; the top combo box lists all the DLLs
that contain valid UD models and the next 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 tabsheet, Figure 14.2, contains the material data for interfaces.
Normally, this tabsheet contains the Rinter parameter. For user-dened soil models the
interface tabsheet is slightly different and contains the interface oedometer modulus,
ref
Eoed , and the interface strength parameters cinter ,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

PLAXIS 2011 | Material Models Manual 163

MATERIAL MODELS MANUAL

a. Selection of user-dened soil models

b. Input of parameters
Figure 14.1 Selection window

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.

164 Material Models Manual | PLAXIS 2011

USER-DEFINED SOIL MODELS

Figure 14.2 Interface tabsheet

PLAXIS 2011 | Material Models Manual 165

MATERIAL MODELS MANUAL

166 Material Models Manual | PLAXIS 2011

STRUCTURAL BEHAVIOUR

15

STRUCTURAL BEHAVIOUR

15.1

ANCHORS

The elastic behaviour of an anchor involves only a relationship between axial force N and
displacement (elongation) u of the form:

N=

EA
u
L

(15.1)

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
Fmax,tens and the maximum compression force is bound by Fmax,comp .
15.2

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
a. Local Axes

b. Axial force N

Q13

Q12
c. Shear force Q12

d. Shear force Q13

Figure 15.1 Axial force and shear forces

I3 M3 3

I2 M2 2

a. Bending moment M3

b. Bending moment M2

Figure 15.1 Bending moments

Elastic behaviour of beam elements is dened by the following parameters:

PLAXIS 2011 | Material Models Manual 167

MATERIAL MODELS MANUAL

A:

Beam cross section area

E:

Young's modulus in axial direction

I2 :

Moment of inertia against bending around the second axis

I3 :

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:

N = EA

M2 = EI2 2

Q12 = kGA12

M3 = EI3 3

(15.2)

Q13 = kGA13
In which k is the shear correction factor, which is taken as 5 6 . The shear modulus is
/
taken as G = 1 2 E .
/
15.3

2D GEOGRIDS

The PLAXIS 2D 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 Np .
The relationship between the force and the strain is dened as:

N = EA

15.4

(15.3)

3D GEOGRIDS

The PLAXIS 3D program allows for orthotropic as well as anisotropic material behaviour
in geogrid elements, which is dened by the following parameters:

N1 = EA1 1
N2 = EA2 2

(15.4)

Q12 = GA12
/
In the case of orthotropic behaviour EA1 = EA2 and GA = 1 2 EA1 in the general three
dimensional case.
When plasticity is considered, the maximum tensile forces can be dened:

Np,1 :

Maximum tensile force in 1-direction

Np,2 :

Maximum tensile force in 2-direction

Axial forces are calculated at the stress points of the geogrid elements. If Np is
exceeded, stresses are redistributed according to the theory of plasticity, so that the
maximum forces are complied with. This will result in irreversible deformations. Output of

168 Material Models Manual | PLAXIS 2011

STRUCTURAL BEHAVIOUR

axial forces is given in the nodes, which requires extrapolation of the values at the stress
points. Nodal forces are not checked against the maximum forces. Hence, it is possible
that the nodal values of the axial force may slightly exceed Np .
15.5

2D PLATES

The PLAXIS 2D program allows for elasto-plastic material behaviour in plate elements.
The elastic behaviour is dened by the following parameters:

EA:

Normal stiffness

EI :

Bending stiffness

Poisson's ratio

The material behaviour in plate elements is dened by the following relationship between
structural forces and strains:

N = EA
kEA

Q=
2(1 + )
M = EI

(15.5)
(15.6)
(15.7)

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. In which k is the
shear correction factor, which is taken as 5 6 . This implies that the shear stiffness is
/
determined from the assumption that the plate has a rectangular cross section. In the
case of modelling a solid wall, this will give the correct shear deformation. However, in
the case of steel prole elements, like sheet-pile walls, the computed shear deformation
may be too large. You can check this by judging the value of deq , which can be computed
as 12EI/EA. For steel prole elements, deq should be at least of the order of a factor
10 times smaller than the length of the plate to ensure negligible shear deformations.
When plasticity is considered, the maximum bending moment and maximum normal
force can be dened:

Mp :

Maximum bending moment

Np :

Maximum normal force

The maximum bending moment is given in units of force times length per unit width. The
maximum axial force, Np , is specied in units of force per unit width. When the
combination of a bending moment and an axial force occur in a plate, then the actual
bending moment or axial force at which plasticity occurs is lower than respectively Mp or
Np . The relationship between Mp and Np is visualised in (Figure 15.2). The diamond
shape represents the ultimate combination of forces for which plasticity will occur. Force
combinations inside the diamond will result in elastic deformations only.
By default the maximum moment is set to 11015 units if the material type is set to elastic
(the default setting).
Bending moments and axial forces are calculated at the stress points of the beam
elements. If Mp or Np is exceeded, stresses are redistributed according to the theory of
plasticity, so that the maxima are complied with. This will result in irreversible

PLAXIS 2011 | Material Models Manual 169

MATERIAL MODELS MANUAL

deformations. Output of bending moments and axial forces is given in the nodes, which
requires extrapolation of the values at the stress points. Nodal forces are not checked
against the maximum forces. Hence, it is possible that the nodal values of the axial force
may slightly exceed Mp .

N
Np

M
Mp

Mp

Np
Figure 15.2 Combinations of maximum bending moment and axial force

15.6

3D PLATES

The PLAXIS 3D program allows for orthotropic material behaviour in plate elements,
which is dened by the following parameters:

E1 :

Young's modulus in rst axial direction

E2 :

Young's modulus in second axial direction

G12 :

In-plane shear modulus

G13 :

Out-of-plane shear modulus related to shear deformation over rst direction

G23 :

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

1/E1

22 12 /E1

12 =
0

13
0

23
0

12 /E1

1/E2

1/G12

1/kG13

170 Material Models Manual | PLAXIS 2011

11

22

12

13

1/kG23
23
0

(15.8)

STRUCTURAL BEHAVIOUR

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

E1

22 12 E2

12 = 0

13 0

0
23

12 E2 0

E2

G12

0 kG13

11

22
0

0 12

0 13

23
kG23

(15.9)

This approximation holds as long as the Poisson's ratio is small. These stress-strain
relationships can be transformed into relationships for structural forces:

N1

12 E2 d

12 E2 d E2 d
2

G12 d
0
0
12
Q12

Q13 = 0 kG13 d
0 13

0
0
kG23 d
23
Q23

E d3 E d3

12 2
1
0
M11
11

12 3
12

E2 d 3
M22 = 12 E2 d

0 22

12
12

G12 d 3
M12
12
0
0
12
N2

E1 d

(15.10a)

(15.10b)

(15.10c)

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 15.3).
When geometric orthotropy is considered rather than material orthotropy, the following
relationships for structural forces apply:

EA1 EA1
1
EA1 EA1

2
2

= 1 1

1

EA

EA2
1
N2
2
EA1 EA2
2
1 2 1 2

EA12
0
0
2(1 + )
12
Q12

EA13

0
0
Q13 =
13

2(1 + )

EA23
Q23
0
0
23
2(1 + )

N1

(15.11a)

(15.11b)

PLAXIS 2011 | Material Models Manual 171

MATERIAL MODELS MANUAL

N1 E1 1

Q12 G12 12

M12 12

N2 E2 2

Q13 G13 13

Q23 G23 23

M11 11

M22 22

Figure 15.3 Denition of positive normal forces (N), shear forces (Q) and bending moments (M) for
a plate based on local system of axes

EI1
EI1
1 2 1 2

EI1
EI2

M22 =
1 2 1 2

M12
0
0

M11

0
0
EI12
2(1 + )

EI1 EI1
11

EI EI

1
2
22

0
0
12

11

0
22

EI12
12
2(1 + )
(15.11c)

where the approximations hold for a small Poisson's ratio. In these relations the following
alternative parameters are used:

A1 :

Effective material cross section area for axial forces in the rst direction

A2 :

Effective material cross section area for axial forces in the second direction

A12 :

Effective material cross section area for shear forces Q12

A13 :

Effective material cross section area for shear forces Q13

172 Material Models Manual | PLAXIS 2011

STRUCTURAL BEHAVIOUR

A23 :

Effective material cross section area for shear forces Q23

I1 :

Moment of inertia against bending over the rst axis

I2 :

Moment of inertia against bending over the second axis

I12 :

Moment of inertia against torsion

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 E1 A is
equal to the normal stiffness EA1 of the plate. Here E1 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 A1 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:

E1 = 12EI1 /d 3
6EI12
G12 =
(1 + )d 3
12 =

E2 = 12EI2 /d 3
EA13
G13 =
2(1 + )d

G23 =

EA23
2(1 + )d

(15.12)

In this case the resulting exural rigidities E1 d 3 /12 and E2 d 3 /12 and shear stiffnesses
G13 d and G23 d prove to be independent of the chosen value for equivalent plate
thickness. Only the normal stiffnesses E1 A and E2 A and shear stiffness G12 d 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 15.4.
From the sheet-pile manufacturer, the following properties are known: t (wall thickness),
h (total height), A (per m wall width), I1 , Esteel and steel .
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 (E1 > E2 ). Moreover, the exural rigidity
against bending over the vertical direction, I1 , is much larger than the stiffness against
bending over the horizontal direction, I2 , (I1 >> I2 say I1 20I2 and I1 >> I12 say
I1 10I12 ).
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,

A factor of 20 is used here to move the bending stiffness over the rst direction sufciently small compared to
the bending stiffness over the second direction, whilst the matrix condition is still OK. Note that in reality bending
stiffness differences in order of 1000 may exist.

PLAXIS 2011 | Material Models Manual 173

MATERIAL MODELS MANUAL

3
2

E1

I1 , M11 ,
11
3
E2

3
2

I2 , M22 ,
22

Figure 15.4 Example of sheet-pile wall with its major quantities

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 considerably larger than t)

E1 = 12Esteel I1 /d 3
E2 = 12Esteel I2 /d 3 I2 /I1 E1 E1 /20
G12 =

6Esteel I12 6E
3
steel I1 /10d
(1 + steel )d 3

G13 =

Esteel A13 E
steel (A/3)/2d Esteel A/6d
2(1 + steel )d

G23 =

Esteel A23 E
steel (A/10)/2d Esteel A/20d
2(1 + steel )d

12 0
= Asteel /d
The second example is a concrete T-shaped oor prole, as depicted in Figure 15.5. In
addition to the precise geometry dimensions, the following properties are known:

Econcrete , 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,
I2 , is larger than the stiffness against bending over the rst direction, I1 , (I2 > I1 ), since I2

174 Material Models Manual | PLAXIS 2011

STRUCTURAL BEHAVIOUR

I2
M22
3 22

E2

h1
h2
w1w2

E1

I1
M11
11

Figure 15.5 Example of concrete oor prole with its major quantities

is dominated by h1 + h2 whereas I1 only depends on h1 . Furthermore, the cross section

area (per unit of width) that is effective against shear deformation over the second
direction (G23 ) is equal to the total cross section area A, whereas the area that is
effective against shear deformation over the rst direction (G13 ) is equal to h1 . With these
assumptions, the situation could be modelled by selecting the model parameters in the
following way:

d=

h1 w1 + (h1 + h2 )w2
w1 + w2

3
3
E1 = 12Econcrete I1 /d 3 = 12Econcrete 1 h1 /d 3 = Econcrete h1 /d 3
12

E2 = 12Econcrete I2 /d 3

1
1 3
h1 w1 +
(h1 + h2 )3 w2
12
12
where I2 =
w1 + w2

G12 =

6Econcrete I12 where I I = 1 h3

12
1
12 1
(1 + concrete )d 3

G13 =

Econcrete h1
Econcrete A13 =
2(1 + concrete )d 2(1 + concrete )d

G23 =

Econcrete d
Econcrete
Econcrete A23 =
=
2(1 + concrete )d 2(1 + concrete )d 2(1 + concrete )

12 = concrete

15.7

EMBEDDED PILE

An embedded pile consists of beam elements with embedded interface elements to

describe the interaction with the soil at the pile skin and at the pile foot (bearing capacity).
The material parameters of the embedded pile distinguish between the parameters of the

PLAXIS 2011 | Material Models Manual 175

MATERIAL MODELS MANUAL

beam and the parameters of the skin resistance and foot resistance. The beam elements
are considered to be linear elastic and are dened by the same material parameters as a
regular beam element (Section 15.2).
The interaction of the pile with the soil at the skin of the pile is described by linear elastic
behaviour with a nite strength and is dened by the following parameter:

Tmax :

Maximum traction allowed at the skin of the embedded pile (can vary along the
pile)

The constitutive equation at the skin of the pile is dened by (see Figure 15.6):

ts

Ks 0 0

p
s
us us

p

s

tn = 0 Kn 0 un un

utp uts
0 0 Kt
tt

(15.13)

where u p denotes the displacement of the pile and u s denotes the displacement of the
soil. Ks denotes the elastic shear stiffness (against parallel displacement differences) of
the embedded interface elements and Kn and Kt denote the elastic normal stiffness
(against perpendicular displacement differences) of the embedded interface elements.
By default these values are dened such that the stiffness of the embedded interface
elements does not inuence the total elastic stiffness of the pile-soil structure:

Ks >> Gsoil
2(1 )
Kn = Kt =
Ks
1 2

(15.14)

Kt

Kn
Ks
s

Kt

Kn
Ks
n
Figure 15.6 Stiffness of the embedded interface elements at the skin of the pile

The normal stresses tn and tt will always remain elastic. For the shear stress in axial
direction ts to remain elastic it is given by:

|ts |< Tmax

176 Material Models Manual | PLAXIS 2011

(15.15)

STRUCTURAL BEHAVIOUR

For plastic behaviour the shear force ts is given by:

(15.16)

|ts |= Tmax

In case of dening a layer dependent skin resistance the shear force ts will remain elastic
as long as (see also Section 3.5.4 of the Reference Manual):
avg

|ts |< n

tan i + ci D

and

|ts | < Tmax

(15.17)

where D denotes the diameter or the equivalent diameter (in the case alternative beam
avg
properties have been specied) of the embedded pile and n is the average lateral
(perpendicular) stress of the soil around the pile:
avg

1
(xx + zz )(in
2

the case of a vertical pile)

(15.18)

The parameters i and ci are the friction angle and cohesion of the embedded interface.
The strength properties of embedded interfaces with layer dependent skin resistance are
linked to the strength properties of a soil layer. Each data set has an associated strength
reduction factor for interfaces Rinter . The embedded interface properties are calculated
from the soil properties in the associated data set and the strength reduction factor by
applying the following rules:

tan i = Rinter tan soil

(15.19)

ci = Rinter csoil
For plastic behaviour ts is given by:
avg

|ts |= n

tan i + ci D

provided that

avg

(n

tan i + ci )D Tmax

(15.20)

In the case of a layer dependent skin resistance where the actual bearing capacity is not
known, Tmax can be used as an ultimate cut-off value. The interaction of the pile with the
soil at the foot of the pile is described by a linear elastic perfectly plastic interface
element. The strength of the base is described by the following parameter:

Fmax :

Maximum force allowed at the foot of the embedded pile

In addition, no tension forces are allowed. The constitutive relationship at the foot of the
pile and its failure criterion are dened by (see Figure 15.7):
p
s
Ffoot = Kfoot (ufoot ufoot ) < Fmax

(15.21)

where Kfoot denotes the stiffness of the spring which is dened in the same way as the
stiffness of the embedded interface elements:
(15.22)

Kfoot >> Gsoil

In case of plastic behaviour, the foot force Ffoot is given by:

(15.23)

Ffoot = Fmax
In order to ensure that a realistic pile bearing capacity as specied can actually be

PLAXIS 2011 | Material Models Manual 177

MATERIAL MODELS MANUAL

reached, a zone in the soil volume elements surrounding the beam is identied where
any kind of soil plasticity is excluded (elastic zone). The size of this zone is determined by
the embedded pile's equivalent radius Req (see Section 4.1.4 of the Reference Manual).
The elastic zone makes the embedded pile almost behave like a volume pile. However,
installation effects of piles are not taken into account and the pile-soil interaction is
modelled at the centre rather than at the circumference.

s
t

Kfoot
Figure 15.7 Stiffness of the embedded interface element at the foot of the pile

178 Material Models Manual | PLAXIS 2011

HYDRAULIC MODELS

16

HYDRAULIC MODELS

16.1

VAN GENUCHTEN MODEL

Soil Water Characteristic Curve (SWCC) is introduced to describe hydraulic parameters

of the groundwater ow in unsaturated zones (usually above the phreatic surface). The
SWCC describes the capacity of the soil to keep water at different stresses. There are
many models which describe the hydraulic behaviour of unsaturated soils. The most
common in the groundwater literature is the model proposed by Van Genuchten (1980),
which is used in PLAXIS. Van Genuchten function is a three-parameter equation and
relates the saturation to the pressure head p :

S(p ) = Sres + (Ssat Sres ) 1 + ga |p |

gn gc

(16.1)

Where

p =

pw
w

(16.2)

pw

Suction pore pressure.

Unit weight of the pore uid.

Sres

A residual saturation which describes a part of the uid that

remains in the pores even at high suction heads.

Ssat

In general at saturated conditions the pores will not be

completely lled with water as air can get trapped and the
saturation in this situation, Ssat , will be less than one.

ga

A tting parameter which is related to the air entry value of the

soil and has to be measured for a specic material. It is in the
unit of 1/L and is a positive value.

gn

A tting parameter which is a function of the rate of water

extraction from the soil once the air entry value has been
exceeded. This parameter has to be measured for a specic
material.

gc

A tting parameter which is used in the general Van Genuchten

equation. In PLAXIS the following assumption is made to convert
the Van Genuchten to a two-parameter equation (Eq. (16.3)).

gc =

1 gn
gn

(16.3)

The Van Genuchten relationship provides reasonable results for low and intermediate
suctions. For very high suction values, saturation remains at the residual saturation.
Figure 16.1 and Figure 16.2 show the effect of the parameters ga and gn on the shape of
the SWCC.
Relative permeability is related to the saturation via the effective saturation. The effective

PLAXIS 2011 | Material Models Manual 179

MATERIAL MODELS MANUAL

Figure 16.1 Effect of the parameter ga on the SWCC

saturation Se is expressed as:

Se =

S Sres

(16.4)

Ssat Sres

The relative permeability according to Van Genuchten now reads:

gn g 1
n

g 1 gn

krel (S) = (Se )gl 1 1 Se n

(16.5)

gl is a tting parameter and has to be measured for a specic material. Note that using
the above expressions, the relative permeability can be related to the suction pore
pressure directly.
The derivative of the degree of saturation with respect to the suction pore pressure reads:

1 gn
ga
S(pw )
gn
= (Ssat Sres )
gn
w
pw

gn

pw (gn 1) 1 + ga

pw
w

gn

1 2gn
gn
(16.6)

Figure 16.3 and Figure 16.4 present the Van Genuchten relations for a sandy material
with parameters Ssat = 1.0, Sres = 0.027, ga = 2.24 m1 , gl = 0.0 and gn = 2.286
graphically.

180 Material Models Manual | PLAXIS 2011

HYDRAULIC MODELS

Figure 16.2 Effect of the parameter gn on the SWCC

Figure 16.3 Van Genuchten pressure-saturation

PLAXIS 2011 | Material Models Manual 181

MATERIAL MODELS MANUAL

Figure 16.4 Van Genuchten pressure-relative permeability

182 Material Models Manual | PLAXIS 2011

HYDRAULIC MODELS

16.2

APPROXIMATE VAN GENUCHTEN MODEL

As an alternative, the PlaxFlow program supports a linearized Van Genuchten model for
which the approximate Van Genuchten parameters can be derived. According to this
concept saturation relates to the pore pressure head as:

S(p ) = 1 + p
|ps |

if p 0

if ps < p < 0

(16.7)

if p ps

The variable ps is a material dependent pressure head which species the extent of the
unsaturated zone under hydrostatic conditions. Below this threshold value the saturation
is assumed to be zero. For saturated conditions the degree of saturation equals one. The
relation between relative permeability and pressure head is written as:

4p
krel (p ) =
10 |pk |

10

if p 0
if pk < p < 0

(16.8)

if p pk

According to this formulation the permeability in the transition zone is described as a

log-linear relation of pressure head where pk is the pressure head at which the relative
permeability is reduced to 104 . The permeability remains constant for higher values of
the pressure head. Under saturated conditions the relative permeability equals one and
the effective permeability is equal to the saturated permeability which is assumed to be
constant.
The input parameters of the approximate Van Genuchten model are derived from the
classical Van Genuchten model. These parameters are translated into approximately
equivalent process parameters for the numerically more robust linearized model. For ps
the translation is as follows:

ps =

1
Sp =1.0 m Ssat

(16.9)

The parameter pk is set equal to the pressure head at which the relative permeability
according to Van Genuchten is 102 , with a lower limit of -0.5 m. Figure 16.5 presents
the functional relation between pressure and saturation according to the approximate Van
Genuchten model using ps = 1.48 m. The corresponding pressure-relative saturation
relation pk = 1.15 m is given in Figure 16.6.

PLAXIS 2011 | Material Models Manual 183

MATERIAL MODELS MANUAL

Figure 16.5 Approximate Van Genuchten pressure-saturation

Figure 16.6 Approximate Van Genuchten pressure-relative permeability

184 Material Models Manual | PLAXIS 2011

REFERENCES

17

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]

Andresen, L. (2002). Capacity analysis of anisotropic and strain-softening clay.

Ph.d. thesis, University of Oslo, Institute of Geology.

[4]

Andresen, L., Jostad, H.P. (1999). Application of an anisotropic hardening model for
undrained response of saturated clay. Proc. NUMOG VII, 581585.

[5]

Atkinson, J.H., Bransby, P.L. (1978). The Mechanics of Soils. McGraw-Hill, Londen.

[6]

Atkinson, J.H., Sallfors, G. (1991). Experimental determination of soil properties. In

Proc. 10th ECSMFE, 3, 915956.

[7]

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.

[8]

Benz, T. (2006). Small-Strain Stiffness of Soils and its Numerical Consequences.

Ph.d. thesis, Universitt Stuttgart.

[9]

Benz, T., Schwab, R., Vermeer, P.A., Kauther, R.A. (2007). A hoek-brown criterion
with intrinsic material strength factorization. Int.J. of Rock Mechanics and Mining
Sci., 45(2), 210222.

[10] Biarez, J., Hicher, P.Y. (1994). Elementary Mechanics of Soil Behaviour. A A
Balkema, Rotterdam, The Netherlands.
[11] Billington, E.W. (1988). Generalized isotropic yield criterion for incompressible
materials. Acta Mechanica, 72, 120.
[12] Bjerrum, L. (1967). Engineering geology of norwegian normally-consolidated
marine clays as related to settlements of buildings. Seventh Rankine Lecture,
Geotechnique 17, 81118.
[13] Bolton, M.D. (1986). The strength and dilatancy of sands. Gotechnique, 36(1),
6578.
[14] Borja, R.I., Kavaznjian, E. (1985). A constitutive model for the --t behaviour of
wet clays. Geotechnique, 35, 283298.
[15] 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.
[16] Brinkgreve, R.B.J. (1994). Geomaterial Models and Numerical Analysis of
Softening. Dissertation. Delft University of Technology.
[17] Brinkgreve, R.B.J. (2004). Time-dependent behaviour of soft soils during
embankment construction - a numerical study. Proc. NUMOG IX, 631637.
[18] Brinkgreve, R.B.J., Kappert, M.H., Bonnier, P.G. (2007). Hysteretic damping in a
small-strain stiffness model. Proc. NUMOG X, 737742.

PLAXIS 2011 | Material Models Manual 185

MATERIAL MODELS MANUAL

[19] 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.
[20] Buisman, K. (1936). Results of long duration settlement tests. Proceedings 1st
International Conference on Soil Mechanics and Foundation Engineering, Mass.
Vol. 1, 103107.
[21] Burland, J.B. (1965). The yielding and dilation of clay. Gotechnique, 15, 211214.
(Correspondence).
[22] Burland, J.B. (1967). Deformation of Soft Clay. Dissertation. Cambridge University.
[23] Buttereld, R. (1979). A natural compression law for soils (an advance on e-log p').
Geotechnique, 29, 469480.
[24] Carranza-Torres, C. (2004). Elastoplastic solution of tunnel problems using the
generalized for of the hoek-brown failure criterion. Int.J. of Rock Mechanics and
Mining Sci., 41(3), 480481.
[25] Chen, W.F. (1975). Limit analysis and soil plasticity. Elsevier, Amsterdam.
[26] Davis, E.H., Christian, J.T. (1971). Bearing capacity of anisotropic cohesive soil.
Journal of Soil Mechanics and Foundation Division, 97, 753769.
[27] den Haan, E.J. (1994). Vertical Compression of Soils. Thesis, Delft University.
[28] Drucker, D.C., Prager, W. (1952). Soil mechanics and plastic analysis or limit
design. Quart. Appl. Math., 10(2), 157165.
[29] 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.
[30] Fung, Y.C. (1965). Foundations of Solid Mechanics. Prentice-Hall, New Jersey,
USA.
[31] Garlanger, J.E. (1972). The consolidation of soils exhibiting creep under constant
effective stress. Gotechnique, 22, 7178.
[32] Grimstad, G., Andresen, L., Jostad, H.P. (2010). Anisotropic shear strength model
for clay. International Journal for Numerical and Analytical Methods in
Geomechanics, (Accepted for publication).
[33] 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.
[34] 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.
[35] Hill, R. (1950). The Mathematical Theory of Plasticity. Oxford University Press,
London, U.K.
[36] Hoek, E. (1999). Putting numbers to geology-an engineerss viewpoint. Quarterly
Journal of Engineering Geology, 32, 19.
[37] Hoek, E. (2006). Practical Rock Engineering. E-book.
[38] Hoek, E., Carranza-Torres, C., Corkum, B. (2002). Hoek-brown failure criterion -

186 Material Models Manual | PLAXIS 2011

REFERENCES

2002 edition. volume 1, 267273.

[39] Iizuka, A., Ohta, H. (1987). A determination procedure of input parameters in
elasto-viscoplastic nite element analysis. Soils and Foundations, 27, 7187.
[40] Jaky, J. (1944). The coefcient of earth pressure at rest. J. Soc. Hung. Eng. Arch.,
1.
[41] Janbu, N. (1963). Soil compressibility as determined by oedometer and triaxial
tests. Proc. ECSMFE Wiesbaden, 1, 1925.
[42] Janbu, N. (1969). The resistance concept applied to soils. Proceedings of the 7h
ICSMFE, Mexico City, 1, 191196.
[43] Janbu, N. (1985). Soil models in offshore engineering (25th rankine lecture).
Gotechnique, 35, 241280.
[44] Koiter, W.T. (1953). Stress-strain relations, uniqueness and variational theorems for
elasto-plastic materials with a singular yield surface. Quart. Appl. Math., 11,
350354.
[45] 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.
[46] Kondner, R.L. (1963). A hyperbolic stress strain formulation for sands. 2. Pan. Am.
ICOSFE Brazil, 1, 289324.
[47] Kulhawy, F.H., Mayne, P.W. (1990). Manual on Estimating Soil Properties for
Foundation Design. Cornell University, Ithaca, New York.
[48] Leroueil, S. (1977). Quelques considrations sur le comportement des argiles
sensibles. Ph.d. thesis, Laval University, Qubec.
[49] Li, X.S., Dafalias, Y.F. (2000). Dilatancy for cohesionless soils. Geotechnique,
50(4), 449460.
[50] Masing, G. (1926). Eigenspannungen und verfestigung beim messing. In In Proc.
2nd Int. Congr. Appl. Mech. Zurich.
[51] Mesri, G., Godlewski, P.M. (1977). Time and stress-compressibility
inter-relationship. Journal of the Geotechnical Engineering Division, ASCE,
103(GT5), 417430.
[52] Muir Wood, D. (1990). Soil Behaviour and Critical State Soil Mechanics. Cambridge
University Press.
[53] Murayama, S., Shibata, T. (1966). Flow and stress relaxation of clays. In I.U.T.A.M.
Symp. on Rheology and Soil Mechanics. Grenoble, 99129.
[54] Niemunis, A., Herle, I. (1997). I.: Hypoplastic model for cohesionless soils with
elastic strain range. Mechanics of Cohesive Frictional Materials, 2(3), 279299.
[55] Ohta, H., Hata, S. (1973). Immediate and consolidation deformations of clay. In 8th.
Int. Conf. S.M.F.E. volume 1, 193196.
[56] Pipatpongsa, T., Iizuka, A., Kobayashi, I., Ohta, H. (2002). Fem formulation for
analysis of soil constitutive model with a corner on the yield surface. Journal of

Structural Engineering, 48, 185U194.

PLAXIS 2011 | Material Models Manual 187

MATERIAL MODELS MANUAL

[57] Prevost, J.H. (1976). Undrained stress-strain-time behaviour of clays. Journal of the
Geotechnical Engineering Division, GT12, 12451259.
[58] 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.
[59] Roscoe, K.H., Schoeld, A.N., Thurairajah, A. (1963). Yielding of clays in states
wetter than critical. Geotechnique, 13(3), 211240.
[60] 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.
[61] 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.
[62] Schanz, T. (1998). Zur Modellierung des Mechanischen Verhaltens von
Reibungsmaterialen. Habilitation, Stuttgart Universitt.
[63] Schanz, T., Vermeer, P.A. (1996). Angles of friction and dilatancy of sand.
Gotechnique, 46, 145151.
[64] Schanz, T., Vermeer, P.A. (1998). Special issue on pre-failure deformation
behaviour of geomaterials. Gotechnique, 48, 383387.
[65] 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.
[66] 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.
[67] Sekiguchi, H. (1977). Rheological characteristics of clays. In Proceedings of the 9th
ICSMFE. Tokyo, volume 1, 289292.
[68] Sekiguchi, H., Ohta, H. (1977). Induced anisotropy and time dependency in clays.
In Proceedings of the 9th ICSMFE. Tokyo, volume 3, 542544.
[69] Simpson, B. (1992). "retaining structures: displacement and design", the 32nd
rankine lecture. Geotechnique, 42(4), 541576.
[70] Smith, I.M., Grifth, D.V. (1982). Programming the Finite Element Method. John
Wiley & Sons, Chisester, U.K, second edition.
[71] 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.
[72] 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.
[73] Taiebat, H.A., Carter, J.P. (2008). Flow rule effects in the tresca model. Computers
and Geotechnics, 35, 500503.
[74] Takeyama, T., Ohno, S., Pipatpongsa, T., Iizuka, A., Ohta, H. (2010). The stress

188 Material Models Manual | PLAXIS 2011

REFERENCES

update using implicit integration for the viscid version of sekiguchi-ohta model.
Technical report, Technical report.
[75] Ukritchon, B., Whittle, A.J., Sloan, S.W. (2003). Undrained stability of braced
excavations in clay. Journal of Geotechnical and Geoenvironmental Engineering,
129(8), 738755.
[76] Vaid, Y., Campanella, R.G. (1977). Time-dependent behaviour of undisturbed clay.
ASCE Journal of the Geotechnical Engineering Division, 103(GT7), 693709.
[77] van Langen, H., Vermeer, P.A. (1990). Automatic step size correction for
non-associated plasticity problems. Int. J. Num. Meth. Engng., 29, 579598.
[78] Vermeer, P.A., Borst, R. (1984). Non-associated plasticity for soils, concrete and
rock. Heron, 29(3).
[79] 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.
[80] 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.
[81] Vermeer, P.A., van Langen, H. (1989). Soil collapse computations with nite
elements. In Ingenieur-Archive 59. 221236.
[82] von Mises, R. (1913). Mechanik der festen krper in plastisch deformablem
zustand. Gttinger Nachrichten Math.-Phys. Klasse, 1, 582592.
[83] von Soos, P. (1990). Properties of soil and rock (in german). In In:
Grundbautaschenbuch Part 4. Ernst & Sohn, Berlin.
[84] Vucetic, M., Dobry, R. (1991). Effect of soil plasticity on cyclic response. Journal of
Geotechnical Engineering, ASCE, 117(1), 89107.

PLAXIS 2011 | Material Models Manual 189

MATERIAL MODELS MANUAL

190 Material Models Manual | PLAXIS 2011

APPENDIX A - SYMBOLS

APPENDIX A - SYMBOLS

K0
nc
K0

:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:

m
M
M
n
N
OCR
p

:
:
:
:
:
:
:

pp
POP
q
Q

Rf
t
u

:
:
:
:
:
:
:
:

:
:

:
:
:
:
:
:
:

A
c
csp
Cu , Su
d
De
e
E
Eoed
f
g
G
I
K

Cross section area

Cohesion
Current stiffness parameter
Undrained shear-strength
Thickness
Elastic material matrix representing Hooke's law
Void ratio
Young's modulus
Oedometer modulus
Yield function
Plastic potential function
Shear modulus
Moment of inertia
Bulk modulus
Coefcient of lateral earth pressure (initial stress state)
Coefcient of lateral earth pressure for a normally
consolidated stress state
Power in stress-dependent stiffness relation
Slope of critical state line in p' q space
Bending moment
Porosity
Normal force
Over-consolidation ratio
Isotropic stress or mean stress,
negative for pressure; positive for tension
Isotropic pre-consolidation stress, negative for pressure
Pre overburden pressure, positive for (over)pressure
Equivalent shear stress or deviatoric stress
Shear force
Failure ratio
Time
Vector with displacement components
Unit weight
Increment
Vector with Cartesian strain components,
components
positive for extension; negative for compression
Volumetric strain,
negative for compression; positive for extension
Cam-Clay swelling index
Modied swelling index
Plastic multiplier
Cam-Clay compression index
Modied compression index
Modied creep index
Poisson's ratio

normal

PLAXIS 2011 | Material Models Manual 191

MATERIAL MODELS MANUAL

c
e
p
ref
u
ur

:
:
:
:
:
:
:
:
:

Vector with Cartesian stress components, normal

components
positive for tension; negative for pressure
Vertical pre-consolidation stress, negative for pressure
Friction angle
Dilatancy angle
denotes creep component
denotes elastic component
denotes plastic component
denotes reference value
denotes undrained
denotes unloading and reloading

192 Material Models Manual | PLAXIS 2011

NC clay

A*

A*

A*

Peat (org)

** : Jointed Rock model in case of anisotropy and stratication; Hoek-Brown model for rock in general

* : Soft Soil Creep model in case time-dependent behaviour is important; NGI-ADP model for short-term analysis, in case only undrained strength is known

C : First order (crude) approximation

B : Reasonable modelling

A : The best standard model in PLAXIS for this application

Hoek-Brown model

A*

NGI-ADP model
A**

Modied Cam-Clay model

Jointed Rock model

A*

OC clay

A*

Silt

Soft Soil model

Sand

Soft Soil Creep model

A**

Mohr-Coulomb model

Gravel

HS small model

Linear Elastic model

Rock

Hardening Soil model

Concrete

Model

Considering different types of soils

APPENDIX B - APPLICABILITY OF THE MATERIAL MODELS

APPENDIX B - APPLICABILITY OF THE MATERIAL MODELS

In this appendix, an overview of the applicability of the material models is given.

PLAXIS 2011 | Material Models Manual 193

194 Material Models Manual | PLAXIS 2011

B

Hoek-Brown model

C : First order (crude) approximation

B : Reasonable modelling

A : The best standard model in PLAXIS for this application

Jointed Rock model

Soft Soil model

NGI-ADP model

Soft Soil Creep model

Modied Cam-Clay model

HS small model

Hardening Soil model

Tunnel

Mohr-Coulomb model

Excavation
C

Foundation

Linear Elastic model

Model

Considering different types of applications (consider also type of soil!)

Embankment

Slope

Dam

Offshore

Other

MATERIAL MODELS MANUAL

A
B

Soft Soil model

Jointed Rock model

Unloading /
Reloading

C : First order (crude) approximation

B : Reasonable modelling

A : The best standard model in PLAXIS for this application

Soft Soil Creep model

Hoek-Brown model

HS small model

Hardening Soil model

Mohr-Coulomb model

NGI-ADP model

Linear Elastic model

Modied Cam-Clay model

Primary
compression

Model

Shear / Deviatoric
loading

Considering different types of loading and soils (consider also type of soil!)

Undrained
loading

Cyclic

Compression
+ Shear

Extension
+ Shear

APPENDIX B - APPLICABILITY OF THE MATERIAL MODELS

PLAXIS 2011 | Material Models Manual 195

MATERIAL MODELS MANUAL

196 Material Models Manual | PLAXIS 2011

APPENDIX C - FORTRAN SUBROUTINES FOR USER-DEFINED SOIL MODELS

APPENDIX C - 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 code 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 Vec1i + R2 Vec2i

MatMat( xMat1, id1, xMat2, id2, nR1, nC2,

nC1, xMatR, idR):

Matrix multiplication

xMatR ij = xMat 1ik xMat 2kj

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 ):
Matrix multiplication xMatR ij = xMat 1ik xMat 2kj

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 ):

PLAXIS 2011 | Material Models Manual 197

MATERIAL MODELS MANUAL

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 ):
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 ):

198 Material Models Manual | PLAXIS 2011

APPENDIX C - FORTRAN SUBROUTINES FOR USER-DEFINED SOIL MODELS

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

PLAXIS 2011 | Material Models Manual 199

MATERIAL MODELS MANUAL

200 Material Models Manual | PLAXIS 2011

APPENDIX D - CREATING A DEBUG-FILE FOR USER-DEFINED SOIL MODELS

APPENDIX D - 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
If (.Not. IsOpen) Then
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(*)
Logical IsOpen
Data IsOpen / .FALSE. /
Save IsOpen
If (.Not. IsOpen) Then
fName = ' '
Do i=1, iPrjLen
fName(i:i) = Char( iPrjDir(i) )
End Do

PLAXIS 2011 | Material Models Manual 201

MATERIAL MODELS MANUAL

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
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.

202 Material Models Manual | PLAXIS 2011