r5 03 21

Code_Aster
Version
default
Titre : Modlisation lasto(visco)plastique avec crouissa[...] Date : 26/06/2013 Page : 1/17
Responsable : Renaud BARGELLINI Cl : R5.03.21 Rvision : 11145
Modelization lasto (visco) plastic with isotropic
hardening in large deformation
Rsum
One describes here a thermoelastoplastic behavior model with isotropic hardening written in large deformation
and proposed by Simo and Miehe. This model is available in command STAT_NON_LINE via key word
RELATION: 'VMIS_ISOT_TRAC', ` VMIS_ISOT_PUIS' or 'VMIS_ISOT_LINE' under the key word factor
COMP_INCR and with key word DEFORMATION: 'SIMO_MIEHE'. A viscous version of this model is also
proposed: 'VISC_ISOT_TRAC' and 'VISC_ISOT_LINE'.
This model is established for the three-dimensional modelizations (3D), axisymmetric (AXIS) and in plane
strains (D_PLAN).
One presents the writing and the digital processing of this model, as well as the associated variational
formulation. It is about an eulerian variational formulation, with reactualization of the geometry and which takes
account of the rigidity of behavior and geometrical rigidity.
Warning : The translation process used on this website is a "Machine Translation". It may be imprecise and inaccurate in whole or in part and is
provided as a convenience.
Licensed under the terms of the GNU FDL (http://www.gnu.org/copyleft/fdl.html)
Code_Aster
Version
default
Contents
Table des Matires
1 Introduction3 ..........................................................................................................................................
2 Notations4 .............................................................................................................................................
3 Rappels on large the dformations5 ......................................................................................................
3.1 Cinmatique5 ..................................................................................................................................
3.2 Contraintes6 ....................................................................................................................................
3.3 Objectivit7 .....................................................................................................................................
4 Prsentation of the model of comportement7 .......................................................................................
4.1 Aspect cinmatique7 ......................................................................................................................
4.2 Relations of comportement8 ...........................................................................................................
4.3 Correction of elastic strain energy in the presence of thermique9 ..................................................
4.4 Choix of the function of crouissage10 ...........................................................................................
4.5 Stresses and variable internes11 ....................................................................................................
4.6 Domaine of utilisation11 .................................................................................................................
4.7 Intgration of the model of comportement11 ..................................................................................
5 Formulation variationnelle14 .................................................................................................................
5.1 Cas of the medium continu14 .........................................................................................................
5.2 Discretization by elements finis14 ..................................................................................................
5.3 Form of the tangent matrix of the comportement15 .......................................................................
6 Bibliographie17 ......................................................................................................................................
7 Description of the versions of the document17 .....................................................................................
Code_Aster
Version
default
1 Introduction
Nous let us present here a thermoelastoplastic constitutive law written in large deformation proposed
by SIMO J.C and MIEHE C. [bib1] which tends in small strains towards model elastoplastic behavior
with isotropic hardening and criterion of Von Mises, described in [R5.03.02]. The kinematical choices
allow, as with the simple reactualization available via key word PETIT_REAC, to treat large
displacements and large deformation but also of large rotations in an exact way.
Specificities of this model are the following ones:
just like in small strains, one supposes the existence of a slackened configuration, i.e. locally
free of stress, which makes it possible to break up the total deflection into a thermo-elastic
part and a plastic part,
the decomposition of this strain in parts thermo-elastic and plastic is not additive any more as in
small strains (or for the model large deformation written in strain rate with for example a
derivative of Jaumann) but multiplicative,
the elastic strain are measured in the present configuration (deformed) while the plastic strains
are measured in the initial setup,
as in small strains, the stresses depend only on the thermoelastic strains,
plastic strains are done with volume constant. The variation of volume is then only due to the
thermoelastic strains,
this model led during its numerical integration to a model incrmentalement objective (cf [3.3])
what makes it possible to obtain the exact solution in the presence of large rotations.
A viscous version of this model is also available (model in hyperbolic sine as in the case of the model
of Rousselier ROUSS_VISC, cf [R5.03.07]).
Thereafter, one briefly points out some concepts of mechanics in large deformation, then one presents
the behavior models of the model and his numerical integration to treat the balance equations.
One proposes an eulerian variational formulation, with reactualization of the geometry. For this
reason, one expresses the work of the internal forces and his variation (with an aim of a resolution by
the method of Newton) for the continuous problem, which respectively provide after discretization by
finite elements the vector of the internal forces and the tangent matrix.
Note Bucket :
One will find in [bib2] or [bib3] a presentation deepened on the large deformation.
This document is extracted from [bib4] where one makes a more detailed presentation of the
elastoplastic model, of his numerical integration and where some examples of validation are
given.
Code_Aster
Version
default
2 Notations
One will note by:
Id
stamp identity
tr A traces tensor A
A
T
transposed of the determining A
det A tensor of A
X positive part of X
A

left deviatoric the tensor A defined by
A=A(
1
3
tr A) Id
:
doubly contracted product:
A: B=
i , j
A
ij
B
ij
=tr ( AB
T
)

tensor product: ( AB)
ijkl
=A
ij
B
kl
A
eq

equivalent value of von Mises defined by
A
eq
=
.
3
2

A:

A
X
A
gradient:
X
A=
A
X
div
x
A
divergence: ( div
x
A)
i
=
j
A
ij
x
j
\, j, E ,+ , K
moduli of the isotropic elasticity
c
y

elastic limit
o
coefficent of thermal dilation
T
temperature
T
ref
Par ailleurs
reference temperature, in the frame of a discretization in time, all the quantities evaluated at previous
time are subscripted by
, the quantities evaluated at time

t +At
are not subscripted and the
increments are indicated par. A One has as follows:
AQ=QQ

Code_Aster
Version
default
3 Recalls on the large deformation
3.1 Considrons
Kinematics a solid subjected to large deformation. That is to say D
0
the field occupied by solid
before strain and
D( t )
the field occupied at time t by deformed solid.
Confi gurati on actuel l e dforme
Confi gurati on i ni ti al e
F
0
( ) t

Appear 3.1-a: Representation of the initial setup and deformed shape
Dans the initial setup D
0
, the position of any particle of solid is indicated by X (Lagrangian
description). After strain, the position at the time t of the particle which occupied the position X
before strain is given by the variable x (eulerian description).
The total motion of solid is defined, with u displacement, by:
x= x( X,t )=X+u
To define the change of metric in the vicinity of a point, the tensor gradient of the transformation is
introduced F :
F=
x
X
=Id+
X
u
The transformations of the volume element and the density are worth:
d D=Jd D
o
with J=det F=
j
o
j

where j
o
and
j
are respectively the density in the configurations initial and current.
Various strain tensors can be obtained by eliminating rotation in the local transformation. For example,
by directly computing the variations length and angle (variation of the scalar product), one obtains:
E=
1
2
( CId) with C=F
T
F
A=
1
2
( Idb
1
) b=FF
T
E and A are respectively the strain tensors of Green-Lagrange and Euler-Almansi and C b , the
tensors of right and left Cauchy-Green respectively.

In Lagrangian description, one will describe the strain by the tensors C or E because they are
quantities defined on D
0
, and of eulerian description by the tensors b or A (definite on D ).

Code_Aster
Version
default
Note :
That is to say a solid undergoing two successive transformations, for example the first
transformation makes pass solid of the initial setup D
0
to a configuration D
1
(tensor
gradient F
1/ 0
and vector displacement u
1/ 0
), then the second transformation of the
configuration D
1
with D
2
(tensor gradient F
2/ 1
and vector displacement u
2/ 1
).
0

1

2
F
1/0
F
2/1
F
2/0

The transition of the configuration D
0
with D
2
is given by the tensor gradient F
2/ 0

(displacement u
2/ 0
=u
2/ 1
+u
1/ 0
) such as:
F
2/ 0
=F
2/ 1
F
1/ 0

One obtains then, for example, for the strain tensor of Green-Lagrange E
E
2/ 0
=F
1/ 0
T
E
2/ 1
F
1/ 0
+E
1/ 0

where E
2/ 0
, E
1/ 0
and E
2/ 1
are the strains of Green-lagrange of the configurations D
2

compared to D
0
associated with F
2/ 0
, D
1
compared to D
0
associated with F
1/ 0
and D
2

compared to D
1
associated with F
2/ 1
, respectively.
This constitutes one of the difficulties encountered at the time of the writing of a constitutive
law in large deformation because one cannot write any more one formula similar to that
written in small strains, namely e
2/ 0
=e
2/ 1
+e
1/ 0
where
e
is the linearized strain tensor
total.
To find e
2/ 0
=e
2/ 1
+e
1/ 0
in small strains starting from the statement of E
2/ 0
, it is
necessary to neglect all the terms of order 2 of u
2/ 0
, u
1/ 0
and u
2/ 1
. In this
case, one has E
2/ 0
e
2/ 0
, E
1/ 0
e
1/ 0
and F
1/ 0
T
E
2/ 1
F
1/ 0
e
2/ 1
.
3.2 Contraintes
Pour models it described here, the tensor of the stresses used is the eulerian tensor of Kirchhoff
t

defined by:
J c=t
where
c
is the eulerian tensor of Cauchy. The tensor
t
thus results from a scaling by the
variation of volume of the tensor of Cauchy
c
; this is not the case of other stress tensors used (first
and second tensor of Piola-Kirchhoff).
In eulerian description, the balance equations are given by:
div
x
c+jf =0 sur D
c. n=t
d
sur D
f

where f is the volume force applied to the field
D
, n the norm external with the border
D
f

and
D
f
the part of the border of the field
D
where are applied the surface forces
t
d
.

Code_Aster
Version
default
3.3 Objectivit
Lorsqu' one writes a constitutive law in large deformation, one must check that this model is objective,
i.e. invariant by any change of spatial reference frame of the form:
x
*
=c( t )+Q( t ) x
where Q is an orthogonal tensor which represent the rotation of the reference frame and c a vector
which represents the translation.
More concretely, if one carries out a traction test in the direction e
1
, for example, followed by a
rotation of 90 around e
3
, which amounts carrying out a traction test according to e
2
, then the
danger with a nonobjective constitutive law is not to find a uniaxial tensor of the stresses in the
direction e
2
(what is in particular the case with kinematics PETIT_REAC).
4 Presentation of the model of kinematical
4.1 Aspect behavior
This model supposes, just like in small strains, the existence of a slackened configuration
D
r
, i.e.
locally free of stress, which makes it possible then to break up the total deflection into parts elastic and
plastic, this decomposition being multiplicative.
Thereafter, one will note by F the tensor gradient which makes pass from the initial setup D
0
to the
present configuration
D( t )
, by F
p
the tensor gradient which makes pass from the configuration
D
0
to the slackened configuration
D
r
, and F
e
of the configuration D
r
with
D( t )
. The index p
refers to the plastic part, the index e with the elastic part.
Confi gurati on i ni ti al e
Confi gurati on actuel l e
Confi gurati on rel che
F
e F
p
F
0

( ) t
T T
ref =
=
0

Appear 4.1-a: Decomposition of the tensor gradient F in an elastic part F
e
and plastic F
p
Par composition of motions, one obtains following multiplicative decomposition:
F=F
e
F
p

The elastic strain are measured in the present configuration with the eulerian tensor of left Cauchy-
Green b
e
and the plastic strains in the initial setup by the tensor G
p
(Lagrangian description).
These two tensors are defined by:
Code_Aster
Version
default
b
e
=F
e
F
eT
, G
p
=( F
pT
F
p
)
1
from where b
e
=FG
p
F
T

models It presented is written in order to distinguish the isochoric terms from the terms of change of
volume. One introduces for that the two following tensors:
F=J
1/3
F
and
b
e
=J
2/3
b
e
with J=det F
Par definition, one a: det
F=1 and
det
b
e
=1
.
4.2 Behavior models
the model presented is a model thermolasto (visco) plastic with isotropic hardening which tends
under the assumption of the small strains towards the model [R5.03.02] with criterion of Von Mises (it
is the plastic model). The plastic strains are done with constant volume so that:
J
p
=det F
p
=1
from where
J=J
e
=det F
e
Les behavior models are given by:
the definition of the free energy hyper elastic:
=
ther
( T , J ) +
elas
( J ,
b
e
) +K ( o)
,
with in particular
elas
( J ,
b
e
) =
1
2
E
3( 12+)
|
1
2
( J
2
1) ln J
+
j
2
( tr
b
e
3)
and
ther
( T , J )=3oAT
(
J
1
J
)
thermo-elastic relation stress-strain:
t=j
b
e

tr t=
3K
2
( J
2
1)
9K
2
o( TT
ref
)( J+
1
J
)
threshold of plasticity (it is admitted that it is expressed with the stresses of Kirchhoff):
f =t
eq
R( p)c
y

where R is the isotropic variable of hardening, function of the cumulated plastic strain p.
models of flow:
F

G
p
F
T
=
\
3
t
eq
t
b
e
=3

\
(
1
3
tr
b
e
+
t
j
)
t
t
eq

p=
\

For the model of plasticity, the plastic multiplier is obtained by writing the condition of coherence
f =0
and one a:
p0, f 0 et p f =0
Dans the viscous case, one takes p equalizes with:
Code_Aster
Version
default
p=
e
0
|
sh
(
f
0
)
m

where e
0
, c
0
and m are the viscosity coefficients. Let us announce that this model is reduced to a
model of the Norton type when the 2 parameters materials e
0
and c
0
are very large.
It is reminded the meeting that:
b
e
=J
2/ 3
b
e

F=J
1/3
F

and that the partition of the strains is written:
b
e
=
FG
p
F
T

For metallic materials where the report t
eq
/ j is small in front of 1, the form of the flow model can be
approximate by:
F

G
p
F
T
=
\tr
b
e
t
t
eq
+O
(
t
eq
j
)
q.4.2-1
where
O
(
t
eq
j
)
is negligible in front of the first term.
It is this last statement which is established in Code_Aster.
Note :
If the strains are small, one a:

J 1+tr e
b
e
Id+2e
e
G
p
Id2e
p

where
e
is the total deflection,
e
e
elastic strain and
e
p
plastic strain in small strains.
As a substitute these three statements in the equations of the constitutive law presented here, one
finds well the thermoelastoplastic conventional model with isotropic hardening and criterion of Von
Mises.
4.3 Correction of elastic strain energy in the presence of thermal
Code_Aster
Version
default
the statement of energy hyper elastic
elas
raises some difficulties. Indeed, it depends on J ; in the
presence of thermal strain, J included a thermal component which disturbs the statement and which
it is advisable to correct.
The step of the correction of the energy of Simo-Miehe in the presence of thermal is the following one:
1. cancellation of energy in pure thermal when the stress is null;
2. resolution of the equation in J obtained, which one calls J
0
the solution.
3. final computation of the elastic strain energy of Simo-Miehe in the presence of thermal:
elas
corrigee
=
elas
( J ,
b
e
) +
ther
( T , J ) +
elas
( J ,
b
e
)
elas
(
J
0,
b
e
=I
d
)
ther
(
T , J
0
)

The cancellation of the stress in pure thermal leads to the equation:
tr t=0( J
2
1) 3o( TT
ref
)( J +
1
J
)=0
J
3
J 3oAT ( J
2
1)
J 0

This equation, under the assumption
3oAT 1
, has solutions close to -1.0 and 1. Only largest is
physically acceptable. One thus poses:
J
0
=MAX J tq J
3
J3oAT ( J
2
1) =0

The energy corrected in the presence of thermal is written finally:
elas
corrigee
=
elas
( J ,
b
e
) +
ther
( T , J ) +
elas
( J ,
b
e
)
elas
(
J
0,
b
e
=I
d
)
ther
(
T , J
0
)

4.4 Choice of the function of hardening
Cette behavior model is available in operator STAT_NON_LINE, under the key word factor
COMP_INCR and the argument ` SIMO_MIEHE' key word factor DEFORMATION. One can choose for
the function of hardening, a linear hardening or provide a traction diagram. Five relations can be used.
RELATION = / ` VMIS_ISOT_TRAC'
/ ` VMIS_ISOT_PUIS'
/ ` VMIS_ISOT_LINE'
/ ` VISC_ISOT_TRAC'
/ ` VISC_ISOT_LINE'
For a purely thermo-elastic behavior, the user chooses the argument ` VMIS_ISOT_LINE' for
example, with SY very large (the behavior is then hyper elastic); for an isotropic hardening given by a
traction diagram, the user chooses the argument 'VMIS_ISOT_TRAC' in the plastic case or `
VISC_ISOT_TRAC' in the viscous case and for a linear isotropic hardening, the argument
'VMIS_ISOT_LINE' in the plastic case or ` VISC_ISOT_LINE' in the viscous case. For an
elastoplastic behavior whose curve of hardening (diagram traction rational) is given by a model in
power, form

Code_Aster
Version
default
R( p)=c
y
+c
y
(
E
ac
y
p
)
1
n
, the user chooses the argument 'VMIS_ISOT_PUIS'.
The various characteristics of the material are indicated in operator DEFI_MATERIAU ([U4.23.01])
under the key words:
ELAS some is the model (one gives the Young modulus, the Poisson's ratio and possibly the
thermal coefficient of thermal expansion),
TENSION for ` VMIS_ISOT_TRAC' and ` VISC_ISOT_TRAC' (the rational traction diagram is
given),
ECRO_PUIS for ` VMIS_ISOT_PUIS' (one gives the parameters of the model power),
ECRO_LINE for ` VMIS_ISOT_LINE' and ` VISC_ISOT_LINE' (the elastic limit and the
slope of hardening are given),
VISC_SINH for ` VISC_ISOT_TRAC' and ` VISC_ISOT_LINE' (one gives the three
viscosity coefficients).

Note :
The user must make sure well that the experimental traction diagram used, either directly,
or to deduct the slope of it of hardening is well given in the plane forced rational
c=F/ S
-
logarithmic strain ln( 1+Al / l
0
) where l
0
is the initial length of the useful part of the test-
tube,
Al
the variation length after strain, F the applied force and S current surface. It will
be noticed that c=F/ S=
F
S
0
l
0
l
1
J
from where t=J c=
F
S
0
l
0
l
. In general, it is well the
quantity
F
S
0
l
0
l
which is measured by the research workers and this gives the stress of
Kirchhoff directly used in the model of Simo and Miehe.
4.5 Forced stresses and
Les intern variables are the stresses of Cauchy
c
, thus calculated on the present configuration (six
components in 3D, four in 2D).
The intern variables produced in Code_Aster are:
V1 , the cumulated plastic strain p ,
V2 the indicator of plasticity (0 if the last computed increment is elastic, 1 if not).
V3 with V8 : opposite of the elastic strain
b
e

Remarque :
If the user wants to possibly recover strains in postprocessing of his computation, it is
necessary to trace the strains of Green-Lagrange E , which represents a measurement of
the strains in large deformation (options EPSG_ELGA or EPSG_ELNO ). The conventional
linearized
e
strains measure strains under the assumption of the small strains and do not
have a meaning in large deformation.
4.6 Field of application
the choice of kinematics DEFORMATION: ` PETIT_REAC' also allows to treat a thermoelastoplastic
constitutive law with isotropic hardening and criterion of Von Mises in large deformation. The model is
written in small strains and the taking into account of the large deformation is done by reactualizing
the geometry.
Between the model presented here (SIMO_MIEHE ) and PETIT_REAC,
there is no difference if the strains are small
it does not have there a difference if the strains are large but small rotations
there are differences if rotations are important.
Code_Aster
Version
default
In particular, the solution obtained with kinematics PETIT_REAC can deviate notably from the exact
solution in the presence of large rotations and this whatever the size of the time steps chosen by the
user, contrary to kinematics SIMO_MIEHE.
4.7 Integration of the constitutive law
Dans the case of an incremental behavior, key word factor COMP_INCR, knowing the stress
c
, the
cumulated plastic strain
p
, the trace divided by three of the strain tensor elastics

1
3
tr
b
e
,
displacements
u
and
Au
the temperatures
T
and T , one seeks to determine

( c , p ,
1
3
tr
b
e
) .
Displacements being known, the gradients of the transformation of D
0
with
D
, noted
F
, and of
D
with
D
, noted
AF
, are known.
The implicit discretization of the model gives:
F=AFF

J=det F
F=J
1/3
F

b
e
=
FG
p
F
T

J c=t

t=j
b
e

1
3
tr t=
1
2
K( J
2
1)
3
2
K o( TT
ref
)( J+
1
J
)
f =t
eq
R( p
+A p)c
y

F( G
p
G
p
)
F
T
=tr
b
e
t
t
eq
Ap from where
b
e
=
FG
p
F
T
tr
b
e
t
t
eq
Ap
Dans the plastic case:
Ap0, f 0 et f A p=0
In the viscous case: t
eq
R( p
+A p)c
y
c
0
sh
1
|
(
A p
e
0
At
)
1
m
=0
Note :
This formulation is incrmentalement objective because the only incremental tensorial
quantity which intervenes in the discretization is
G
p
. As
G
p
and
G
p
are measured on
the same configuration, i.e. the initial setup, the discretization of
G
p
, is
AG
p
=G
p
G
p
,
is incrmentalement objective.
One introduces
t
Tr
, the tensor of Kirchhoff which results from an elastic prediction (Tr: trial, in
English test):
t
Tr
=j
b
eTr

where
b
eTr
=
FG
p
F
T
=A
b
e
A
F
T
,
A
F=( AJ )
1/3
AF
and
AJ =det ( AF)
One obtains
b
e
starting from the stresses
t
by the thermo-elastic relation stress-strain and of

the trace of the tensor of the elastic strain.
Code_Aster
Version
default
b
e
=
t
+
1
3
tr
b
e

Note :
The interest of this formulation is that it is not necessary of compute the plastic strain
G
p

which would oblige us to reverse the gradient of the transformation
F . One needs to only
know
FG
p
F
T
.
If t
eq
Tr
R( p
)+c
y
, one remains elastic. In this case, one a:
p=p
, t= t
Tr
+
1
3
tr t
Tr
Id and
1
3
tr
b
e
=
1
3
tr
b
eTr
if not, one obtains:

tr
b
e
=tr
b
eTr
, thanks to simplification on the flow model:
b
e
=
b
e
Tr
tr
b
e
t
t
eq
Ap
By taking the deviatoric parts of this equation, and by multiplying them by
j
one leads to:
t
Tr
= t( 1+
jtr
b
eTr
A p
t
eq
)
While computing the equivalent stress, one brings back oneself to a nonlinear scalar equation in
Ap

:
t
eq
Tr
t
eq
jtr
b
eTr
A p=0
In the plastic case : t
eq
=c
y
+R( p
+Ap) , which leads to

A p
solution of the equation:
t
eq
Tr
c
y
R( p
+Ap )jtr
b
eTr
Ap=0
In the viscoplastic case: t
eq
=c
y
+R( p
+Ap)+c
0
sh
1
|
(
Ap
e
0
At
)
1
m
, which leads to
A p

solution of the equation:
t
eq
Tr
c
y
R( p
+Ap )c
0
sh
1
|
(
A p
e
0
At
)
1
m
jtr

b
eTr
A p=0
If hardening linear, or is given by a point by point defined traction diagram, therefore closely
connected per pieces, the equation to be solved is linear. The solution directly is obtained.
A p
In the
other cases, the resolution is carried out in Code_Aster by a secant method with search interval (cf
[R5.03.05]). Integration can be controlled by parameters RESI_INTE_RELA and ITER_INTE_MAXI.
Once
Ap
known, one can then deduct the tensor of it of Kirchhoff, that is to say:
t=
t
eq
t
eq
Tr
t
Tr
+
|
K
2
( J
2
1)
3K
2
( TT
ref
)( J +
1
J
)
Id

Once computed the cumulated plastic strain, the tensor of the stresses and the tangent matrix, one
carries out a correction on the trail of tensor of the elastic strain
b
e
to take account of the plastic
Code_Aster
Version
default
incompressibility, which is not preserved with the simplification made on the flow model [q 4.2.1].
This correction is carried out by using a relation between the invariants of
b
e
and
b
e
by exploiting
the plastic condition of incompressibility
J
p
=1
(or in an equivalent way
det
b
e
=1
). This relation is
written:
x
3
J
2
e
x( 1
J
3
e
)=0
with
J
2
e
=
1
2
b
eq
e
2
=
t
eq
2
2j
2
,
J
3
e
=det

b
e
=det
t
j
and x=
1
3
tr
b
e
the solution of this cubic equation makes it possible to obtain
tr
b
e
and consequently the thermo-
elastic strain
b
e
with the following time step. If this equation admits several solutions, one takes the
solution nearest to the solution of the preceding time step. It is besides why one stores in an intern
variable
1
3
tr
b
e
.
5 Variational formulation
Dans la mesure o the stresses provided by the constitutive law are eulerian, one chooses a
variational formulation written on the present configuration (eulerian) and not on the initial setup, that
is to say:
D
c
x
6v d D
_
F
int
. 6v
=
D
jf 6v d D+
D
f
t
d
6v dS
_
F
ext
. 6 v

6v
Kinematically admissible
Nous we interest only here in the work of the internal forces and its variation in the optics of a
resolution by the method of Newton. One will find in [bib 4] the demonstration of the statements
presented.
5.1 Case of the continuum
One rewrites here the work of the internal forces in indicielle form, that is to say:
F
int
. 6 v=
D
c
ij
6v
i
x
j
d D
We need also to express the variation of the internal forces in the present configuration
D
is:
6F
int
.6u. 6 v=
D
|
c
ij
6u
p
x
p
c
ik
6u
j
x
k

6v
i
x
j
d D rigidit gomtrique
+
D
|
c
ij
AF
pq
6u
p
x
q

6 v
i
x
j
d D rigidit de comportement
i

where
x
are the punctual coordinates on the configuration

D
.
5.2 Discretization by finite elements
Code_Aster
Version
default
One discretizes displacements u and virtual displacements v by finite elements. The notations are
the following ones, by adopting the summation convention of the repeated indices:
u
i
( x )=N
n
( x) U
i
n
u
i
x
j
=D
j
n
( x) U
i
n
u
i
x
j
=D
j
n
( x) U
i
n

where:
N
n
( x )
is the shape function associated with the node n
U
i
n
, component I of the nodal displacement of the node n
D
j
n
( x ) , the components of the gradient of the shape functions on the configuration
D
D
j
n
( x ) , the components of the gradient of the shape functions on the configuration D
One obtains for the vector of the internal forces:

( F
int
)
i
n
=
D
c
ij
D
j
n
d D

and for the tangent matrix, which is not a priori symmetric:
K
i p
nm
=
D
|
D
p
m
c
ij
D
j
n
D
k
m
c
ik
D
p
n
d D
+
D
|
D
q
n
c
ij
AF
pq
D
j
n
d D

In the case of a two-dimensional modelization (plane strain), the statements of the vector of the
internal forces and tangent matrix are identical to this ready that the indices corresponding to the
components only vary from 1 to 2.
In the case of an axisymmetric modelization, by numbering the axes in the order
(r , z , 0 ) ,
the
vector of the internal forces is written:
( F
int
axi
)
o
n
=
D
|
c
o
D
n
+c
33
N
n
r
6
o1
d D o1,2 , 1,2
and the tangent matrix:
| K
axi
=| K +| K
corr

with:
|
K(1 )
corr
1
nm
=
D
N
n
r
c
m
d D+
D
N
n
r
AF
33
D
m
d D
|
K( 2)
corr
o1
nm
=
D
D
o
n
c
33
N
m
r
d D+
D
D
- n
c
33
AF
o
N
m
r
d D
|
K( 3)
corr
11
nm
=
D
N
n
r
c
33
AF
33
N
m
r

From an algorithmic point of view, the tangent elementary matrix K is not symmetric a priori. The
total resolution will thus be made by default with an asymmetric solver. It is however possible to
symmetrize the total matrix tangent before resolution (key word SOLVER), which makes it possible to
save time computation but can degrade convergence.

Code_Aster
Version
default
5.3 Form of the tangent matrix of the behavior
One gives the form of the tangent matrix here (option FULL_MECA during iterations of Newton, option
RIGI_MECA_TANG for the first iteration). The aforementioned is obtained by linearizing the system of
equations which governs the constitutive law. We give here the final result of this linearization. One
will find in [bib4] the detail of this computation.
One poses:
J=det F ,
J
=det F
and
AJ =det AF
Pour option FULL_MECA , one a:
A=
c
AF
=
( AJ )
1/3
J
H
1
3 J AJ
( HA
F)B
J
J
2
tB
+
J
J
|
KJ
3
2
K o( TT
ref
)( 1J
2
)
IdB

where B is worth:
B
11
=AF
22
AF
33
AF
23
AF
32
B
22
=AF
11
AF
33
AF
13
AF
31
B
33
=AF
11
AF
22
AF
12
AF
21
B
12
=AF
31
AF
23
AF
33
AF
21
B
21
=AF
13
AF
32
AF
33
AF
12
B
13
=AF
21
AF
32
AF
22
AF
31
B
31
=AF
12
AF
23
AF
22
AF
13
B
23
=AF
31
AF
12
AF
11
AF
32
B
32
=AF
13
AF
21
AF
11
AF
23

H and
HA
F
are given by:
In the elastic case ( f 0 ):
H
ijkl
=j
(
6
ik
b
lp
e
A
F
jp
+A
F
ip
b
pl
e
6
jk
)
2j
3
6
ij
A
F
kp
b
lp
e

and
HA
F=2j
b
eTr

Dans the plastic case ( f =0 ) or viscoplastic:
H
ijkl
=
j
a
( 6
ik
b
lp
e
A

F
jp
+A
F
ip
b
pl
e
6
jk
)
2j
|
6
ij
3a
+
RAp t
ij
t
eq
(
R+jtr
b
eTr
)
F
kp
b
lp
e
+
3j
2
tr
b
eTr
(
RA pt
eq
)
at
eq
3
(
R+mu tr
b
eTr
)
t
ij
t
kq
A
F
qp
b
lp
e

and
Code_Aster
Version
default
HA
F=
2j
a

b
eTr
2jtr
b
eTr
|
Id
3a
+

RA p t
t
eq
(
R+jtr
b
eTr
)
+
3j
2
tr
b
eTr
(
RA pt
eq
)
at
eq
3
(
R+jtr

b
eTr
)
( t:
b
eTr
) t
where a=
t
eq
Tr
t
eq

and
R=R' ( p)+c
0
(
1+
(
Ap
e
0
At
)
2
)
1
2
1
m
( e
0
At
)
1
m
( Ap)
1
m
1
_
uniquement cas visqueux
,
R
'
( p)
being the derivative of isotropic hardening compared to the cumulated plastic strain p .
For option RIGI_MECA_TANG, they is the same statements as those given for FULL_MECA but with
Ap=0
and all the variables and coefficients of the material taken to time
t
(in theory, it
would be necessary in the viscous case, to take the statements of FULL_MECA in the elastic case,
all the variables being taken at time
t
). In particular, one will have

A
F=Id
.

6 Bibliography
1) SIMO J.C., MIEHE C.: Associative coupled thermoplasticity At finite strains:
Formulation, numerical analysis and implementation, Comp. Meth. Appl.
Mech. Eng., 98, pp. 41-104, North Holland, 1992.
2) SIDOROFF F.: Elastoplastic Formulations in large deformation, Rapport Greco
n29, 1981.
3) SIDOROFF F.: Course over the large deformation, Rapport Greco n51, 1982.
4) CANO V., LORENTZ E.: Introduction into Code_Aster of a model of behavior in
elastoplastic large deformation with isotropic hardening, internal Note EDF
DER, HI-74/98/006/0, 1998
7 Description of the versions of the document
Version
Aster
Auteur (S) or contributor (S),
Description
organization of the amendments
4.4 V.Cano, E.Lorentz EDF/R & D/AMA initial Texte
6.3 V.Cano, E.Lorentz EDF/R & D/AMA Card-index 6396
7.4 S.Michel-Ponnelle EDF/R & D/AMA Card-indexes 8000: addition of viscosity
9.4 J.M.Proix EDF/R & D/AMA Ajout of VMIS_ISOT_PUIS.
11.4 J.M.Proix EDF/R & D/AMA Card-index 19650 order of the intern
variables

r5 03 21

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

r5 03 21

Uploaded by

Copyright:

Available Formats

Code_Aster

, the quantities evaluated at time

, the trace divided by three of the strain tensor elastics

and T , one seeks to determine

by the thermo-elastic relation stress-strain and of

+Ap) , which leads to

are the punctual coordinates on the configuration

One obtains for the vector of the internal forces:

). In particular, one will have

You might also like